# Paul Kedrosky > Thinking in systems. Informed by the latest research, topics include: risk, complexity, finance, technology, AI, and economic history. Newsletters and more. Public Ghost content for AI and LLM tooling. This file includes a bounded export of public pages first, then recent public posts. Append `.md` to any post or page URL to get the content in Markdown (for example, `/example-post.md`). ## Pages ### About URL: https://paulkedrosky.com/about/ Last updated: 2026-04-15T18:04:19.000Z ### Dr. Paul Kedrosky Paul Kedrosky is an investor, writer, and researcher. He has been a regular on-air contributor at CNBC and Bloomberg, and has written for the New York Times, the Wall Street Journal, and publications around the world. He speaks at private and public events on risk, finance, complexity, and artificial intelligence, and consults with some of the world's largest asset managers. Paul started as an engineer, became a quota-carrying salesperson, then a technology equity analyst on Wall Street, and eventually a venture capitalist. He has started companies, held academic positions in the U.S. and Canada, has a popular newsletter, and is a research fellow at MIT's Institute for the Digital Economy. His Ph.D. research focused on risk, adoption, and path dependence in finance and technology. He co-hosts The Nick, Dick, Paul Show, which has been described as "Top Gear meets The Economist". It is available on [YouTube](#) and [Spotify](https://open.spotify.com/show/6mxUS2hFE2hdaNx1sjhdYu?si=IJZ5DBLUTruGbtLmFj9Dtw), with clips on [TikTok](https://www.tiktok.com/@nickdickpaul?%5Fr=1&%5Ft=ZT-95YzomDnLQc) and [Instagram](https://www.instagram.com/nickdickpaul?igsh=NTc4MTIwNjQ2YQ%3D%3D&utm%5Fsource=qr). Paul lives in California. ### Publications URL: https://paulkedrosky.com/publications/ Last updated: 2023-06-07T14:21:42.000Z A sample of some academic and non-academic publications. - Kedrosky, P. (2004, June). Feeding time. Harvard Business Review, 18-19. - Kedrosky, P. (2008). Entrepreneurs and Recessions: Do Downturns Matter? Kauffman Foundation Education Research. Available at SSRN: https://ssrn.com/abstract=1352591 or http://dx.doi.org/10.2139/ssrn.1352591 - Kedrosky, P. (2009). Right-sizing the US venture capital industry. Venture Capital, 11(4), 287–293 http://doi.org/10.1080/1369106090319025 - Kedrosky, P. (2009). Challenge anyone? Quantifying the challenge system in professional tennis. NY Times, p. A23, , September 20, 2009\. (Available at http://www.nytimes.com/2009/09/21/opinion/21kedrosky.html) - Kedrosky, P. & Feld, B. (2009). “Start-Up Visas Can Jump-Start the Economy,” The Wall Street Journal, December 2, 2009, http://online.wsj.com/article/SB10001424052748704402404574525772299940870.html - Kedrosky, P., & Stangler, D. (2010). Exploring Firm Formation: Why is the Number of New Firms Constant? SSRN Electronic Journal. http://doi.org/10.2139/ssrn.1585380 - Kedrosky, P. & Stangler, D. (2010). The structural dynamics of startups, young firms, and job creation. SSRN Electronic Journal. http://doi.org/10.2139/ssrn.1657167 - Kedrosky, P. & Stangler, D. (2011). Financialization and Its Consequences. SSRN Electronic Journal. http://doi.org/10.2139/ssrn.1798605 - Kedrosky, P. (2013). Is Growth Entrepreneurship Contagious? Available at SSRN: https://ssrn.com/abstract=2341122 or http://dx.doi.org/10.2139/ssrn.2341122 ### Contact Us URL: https://paulkedrosky.com/contact/ Last updated: 2026-10-20T18:28:15.000Z Use the following form and select the type of request from the dropdown. Note: 1. This is not an investment advisory, and **requests for financial advice** will not be responded to. 2. **Media and interview requests** require information about the forum, audience, etc. 3. General questions or comments are read but **don't always generate a response**. ### Privacy Policy URL: https://paulkedrosky.com/privacy/ Last updated: 2023-06-01T23:39:59.000Z **PRIVACY POLICY** **Last updated May 26, 2023** This privacy notice for paulkedrosky.com ("**Company**," "**we**," "**us**," or "**our**"), describes how and why we might collect, store, use, and/or share ("**process**") your information when you use our services ("**Services**"), such as when you: - Visit our website at [https://paulkedrosky.com](https://paulkedrosky.com/), or any website of ours that links to this privacy notice - Engage with us in other related ways, including any sales, marketing, or events **Questions or concerns?** Reading this privacy notice will help you understand your privacy rights and choices. If you do not agree with our policies and practices, please do not use our Services. If you still have any questions or concerns, please contact us at privacy@paulkedrosky.com. **SUMMARY OF KEY POINTS** ***This summary provides key points from our privacy notice, but you can find out more details about any of these topics by clicking the link following each key point or by using our*** [***table of contents***](#toc) ***below to find the section you are looking for.*** **What personal information do we process?** When you visit, use, or navigate our Services, we may process personal information depending on how you interact with paulkedrosky.com and the Services, the choices you make, and the products and features you use. Learn more about [personal information you disclose to us](#personalinfo). **Do we process any sensitive personal information?** We do not process sensitive personal information. **Do we receive any information from third parties?** We do not receive any information from third parties. **How do we process your information?** We process your information to provide, improve, and administer our Services, communicate with you, for security and fraud prevention, and to comply with law. We may also process your information for other purposes with your consent. We process your information only when we have a valid legal reason to do so. Learn more about [how we process your information](#infouse). **In what situations and with which parties do we share personal information?** We may share information in specific situations and with specific third parties. Learn more about [when and with whom we share your personal information](#whoshare). **How do we keep your information safe?** We have organizational and technical processes and procedures in place to protect your personal information. However, no electronic transmission over the internet or information storage technology can be guaranteed to be 100% secure, so we cannot promise or guarantee that hackers, cybercriminals, or other unauthorized third parties will not be able to defeat our security and improperly collect, access, steal, or modify your information. Learn more about [how we keep your information safe](#infosafe). **What are your rights?** Depending on where you are located geographically, the applicable privacy law may mean you have certain rights regarding your personal information. Learn more about [your privacy rights](#privacyrights). **How do you exercise your rights?** The easiest way to exercise your rights is by submitting a [data subject access request](https://app.termly.io/notify/90a94bd2-66a9-4cf2-9b9d-773f28fdbfd9), or by contacting us. We will consider and act upon any request in accordance with applicable data protection laws. Want to learn more about what paulkedrosky.com does with any information we collect? [Review the privacy notice in full](#toc). **TABLE OF CONTENTS** [1\. WHAT INFORMATION DO WE COLLECT?](#infocollect) [2\. HOW DO WE PROCESS YOUR INFORMATION?](#infouse) [3\. WHAT LEGAL BASES DO WE RELY ON TO PROCESS YOUR PERSONAL INFORMATION?](#legalbases) [4\. WHEN AND WITH WHOM DO WE SHARE YOUR PERSONAL INFORMATION?](#whoshare) [5\. DO WE USE COOKIES AND OTHER TRACKING TECHNOLOGIES?](#cookies) [6\. HOW LONG DO WE KEEP YOUR INFORMATION?](#inforetain) [7\. HOW DO WE KEEP YOUR INFORMATION SAFE?](#infosafe) [8\. DO WE COLLECT INFORMATION FROM MINORS?](#infominors) [9\. WHAT ARE YOUR PRIVACY RIGHTS?](#privacyrights) [10\. CONTROLS FOR DO-NOT-TRACK FEATURES](#DNT) [11\. DO CALIFORNIA RESIDENTS HAVE SPECIFIC PRIVACY RIGHTS?](#caresidents) [12\. DO WE MAKE UPDATES TO THIS NOTICE?](#policyupdates) [13\. HOW CAN YOU CONTACT US ABOUT THIS NOTICE?](#contact) [14\. HOW CAN YOU REVIEW, UPDATE, OR DELETE THE DATA WE COLLECT FROM YOU?](#request) **1\. WHAT INFORMATION DO WE COLLECT?** **Personal information you disclose to us** ***In Short:*** *We collect personal information that you provide to us.* We collect personal information that you voluntarily provide to us when you register on the Services, express an interest in obtaining information about us or our products and Services, when you participate in activities on the Services, or otherwise when you contact us. **Personal Information Provided by You.** The personal information that we collect depends on the context of your interactions with us and the Services, the choices you make, and the products and features you use. The personal information we collect may include the following: - email addresses - names **Sensitive Information.** We do not process sensitive information. **Payment Data.** We may collect data necessary to process your payment if you make purchases, such as your payment instrument number, and the security code associated with your payment instrument. All payment data is stored by Stripe. You may find their privacy notice link(s) here: . All personal information that you provide to us must be true, complete, and accurate, and you must notify us of any changes to such personal information. **Information automatically collected** ***In Short:*** *Some information — such as your Internet Protocol (IP) address and/or browser and device characteristics — is collected automatically when you visit our Services.* We automatically collect certain information when you visit, use, or navigate the Services. This information does not reveal your specific identity (like your name or contact information) but may include device and usage information, such as your IP address, browser and device characteristics, operating system, language preferences, referring URLs, device name, country, location, information about how and when you use our Services, and other technical information. This information is primarily needed to maintain the security and operation of our Services, and for our internal analytics and reporting purposes. Like many businesses, we also collect information through cookies and similar technologies. You can find out more about this in our Cookie Notice: paulkedrosky.com/cookies. The information we collect includes: - *Log and Usage Data.* Log and usage data is service-related, diagnostic, usage, and performance information our servers automatically collect when you access or use our Services and which we record in log files. Depending on how you interact with us, this log data may include your IP address, device information, browser type, and settings and information about your activity in the Services (such as the date/time stamps associated with your usage, pages and files viewed, searches, and other actions you take such as which features you use), device event information (such as system activity, error reports (sometimes called "crash dumps"), and hardware settings). - *Location Data.* We collect location data such as information about your device's location, which can be either precise or imprecise. How much information we collect depends on the type and settings of the device you use to access the Services. For example, we may use GPS and other technologies to collect geolocation data that tells us your current location (based on your IP address). You can opt out of allowing us to collect this information either by refusing access to the information or by disabling your Location setting on your device. However, if you choose to opt out, you may not be able to use certain aspects of the Services. **2\. HOW DO WE PROCESS YOUR INFORMATION?** ***In Short:*** *We process your information to provide, improve, and administer our Services, communicate with you, for security and fraud prevention, and to comply with law. We may also process your information for other purposes with your consent.* **We process your personal information for a variety of reasons, depending on how you interact with our Services, including:** - **To facilitate account creation and authentication and otherwise manage user accounts.** We may process your information so you can create and log in to your account, as well as keep your account in working order. - **To deliver and facilitate delivery of services to the user.** We may process your information to provide you with the requested service. - **To send you marketing and promotional communications.** We may process the personal information you send to us for our marketing purposes, if this is in accordance with your marketing preferences. You can opt out of our marketing emails at any time. For more information, see "[WHAT ARE YOUR PRIVACY RIGHTS?](#privacyrights)" below. - **To save or protect an individual's vital interest.** We may process your information when necessary to save or protect an individual’s vital interest, such as to prevent harm. **3\. WHAT LEGAL BASES DO WE RELY ON TO PROCESS YOUR INFORMATION?** **In Short:* We only process your personal information when we believe it is necessary and we have a valid legal reason (i.e., legal basis) to do so under applicable law, like with your consent, to comply with laws, to provide you with services to enter into or fulfill our contractual obligations, to protect your rights, or to fulfill our legitimate business interests.* **If you are located in the EU or UK, this section applies to you.** The General Data Protection Regulation (GDPR) and UK GDPR require us to explain the valid legal bases we rely on in order to process your personal information. As such, we may rely on the following legal bases to process your personal information: - **Consent.** We may process your information if you have given us permission (i.e., consent) to use your personal information for a specific purpose. You can withdraw your consent at any time. Learn more about [withdrawing your consent](#withdrawconsent). - **Performance of a Contract.** We may process your personal information when we believe it is necessary to fulfill our contractual obligations to you, including providing our Services or at your request prior to entering into a contract with you. - **Legitimate Interests.** We may process your information when we believe it is reasonably necessary to achieve our legitimate business interests and those interests do not outweigh your interests and fundamental rights and freedoms. For example, we may process your personal information for some of the purposes described in order to: - Send users information about special offers and discounts on our products and services - **Legal Obligations.** We may process your information where we believe it is necessary for compliance with our legal obligations, such as to cooperate with a law enforcement body or regulatory agency, exercise or defend our legal rights, or disclose your information as evidence in litigation in which we are involved. - **Vital Interests.** We may process your information where we believe it is necessary to protect your vital interests or the vital interests of a third party, such as situations involving potential threats to the safety of any person. ***If you are located in Canada, this section applies to you.*** We may process your information if you have given us specific permission (i.e., express consent) to use your personal information for a specific purpose, or in situations where your permission can be inferred (i.e., implied consent). You can [withdraw your consent](#withdrawconsent) at any time. In some exceptional cases, we may be legally permitted under applicable law to process your information without your consent, including, for example: - If collection is clearly in the interests of an individual and consent cannot be obtained in a timely way - For investigations and fraud detection and prevention - For business transactions provided certain conditions are met - If it is contained in a witness statement and the collection is necessary to assess, process, or settle an insurance claim - For identifying injured, ill, or deceased persons and communicating with next of kin - If we have reasonable grounds to believe an individual has been, is, or may be victim of financial abuse - If it is reasonable to expect collection and use with consent would compromise the availability or the accuracy of the information and the collection is reasonable for purposes related to investigating a breach of an agreement or a contravention of the laws of Canada or a province - If disclosure is required to comply with a subpoena, warrant, court order, or rules of the court relating to the production of records - If it was produced by an individual in the course of their employment, business, or profession and the collection is consistent with the purposes for which the information was produced - If the collection is solely for journalistic, artistic, or literary purposes - If the information is publicly available and is specified by the regulations **4\. WHEN AND WITH WHOM DO WE SHARE YOUR PERSONAL INFORMATION?** ***In Short:*** *We may share information in specific situations described in this section and/or with the following third parties.* We may need to share your personal information in the following situations: - **Business Transfers.** We may share or transfer your information in connection with, or during negotiations of, any merger, sale of company assets, financing, or acquisition of all or a portion of our business to another company. **5\. DO WE USE COOKIES AND OTHER TRACKING TECHNOLOGIES?** ***In Short:*** *We may use cookies and other tracking technologies to collect and store your information.* We may use cookies and similar tracking technologies (like web beacons and pixels) to access or store information. Specific information about how we use such technologies and how you can refuse certain cookies is set out in our Cookie Notice: paulkedrosky.com/cookies. **6\. HOW LONG DO WE KEEP YOUR INFORMATION?** ***In Short:*** *We keep your information for as long as necessary to fulfill the purposes outlined in this privacy notice unless otherwise required by law.* We will only keep your personal information for as long as it is necessary for the purposes set out in this privacy notice, unless a longer retention period is required or permitted by law (such as tax, accounting, or other legal requirements). No purpose in this notice will require us keeping your personal information for longer than the period of time in which users have an account with us. When we have no ongoing legitimate business need to process your personal information, we will either delete or anonymize such information, or, if this is not possible (for example, because your personal information has been stored in backup archives), then we will securely store your personal information and isolate it from any further processing until deletion is possible. **7\. HOW DO WE KEEP YOUR INFORMATION SAFE?** ***In Short:*** *We aim to protect your personal information through a system of organizational and technical security measures.* We have implemented appropriate and reasonable technical and organizational security measures designed to protect the security of any personal information we process. However, despite our safeguards and efforts to secure your information, no electronic transmission over the Internet or information storage technology can be guaranteed to be 100% secure, so we cannot promise or guarantee that hackers, cybercriminals, or other unauthorized third parties will not be able to defeat our security and improperly collect, access, steal, or modify your information. Although we will do our best to protect your personal information, transmission of personal information to and from our Services is at your own risk. You should only access the Services within a secure environment. **8\. DO WE COLLECT INFORMATION FROM MINORS?** ***In Short:*** *We do not knowingly collect data from or market to children under 18 years of age.* We do not knowingly solicit data from or market to children under 18 years of age. By using the Services, you represent that you are at least 18 or that you are the parent or guardian of such a minor and consent to such minor dependent’s use of the Services. If we learn that personal information from users less than 18 years of age has been collected, we will deactivate the account and take reasonable measures to promptly delete such data from our records. If you become aware of any data we may have collected from children under age 18, please contact us at privacy@paulkedrosky.com. **9\. WHAT ARE YOUR PRIVACY RIGHTS?** ***In Short:*** *In some regions, such as the European Economic Area (EEA), United Kingdom (UK), and Canada, you have rights that allow you greater access to and control over your personal information. You may review, change, or terminate your account at any time.* In some regions (like the EEA, UK, and Canada), you have certain rights under applicable data protection laws. These may include the right (i) to request access and obtain a copy of your personal information, (ii) to request rectification or erasure; (iii) to restrict the processing of your personal information; and (iv) if applicable, to data portability. In certain circumstances, you may also have the right to object to the processing of your personal information. You can make such a request by contacting us by using the contact details provided in the section "[HOW CAN YOU CONTACT US ABOUT THIS NOTICE?](#contact)" below. We will consider and act upon any request in accordance with applicable data protection laws. If you are located in the EEA or UK and you believe we are unlawfully processing your personal information, you also have the right to complain to your [Member State data protection authority](https://ec.europa.eu/justice/data-protection/bodies/authorities/index%5Fen.htm) or [UK data protection authority](https://ico.org.uk/make-a-complaint/data-protection-complaints/data-protection-complaints/). If you are located in Switzerland, you may contact the [Federal Data Protection and Information Commissioner](https://www.edoeb.admin.ch/edoeb/en/home.html). **Withdrawing your consent:** If we are relying on your consent to process your personal information, which may be express and/or implied consent depending on the applicable law, you have the right to withdraw your consent at any time. You can withdraw your consent at any time by contacting us by using the contact details provided in the section "[HOW CAN YOU CONTACT US ABOUT THIS NOTICE?](#contact)" below or updating your preferences. However, please note that this will not affect the lawfulness of the processing before its withdrawal nor, when applicable law allows, will it affect the processing of your personal information conducted in reliance on lawful processing grounds other than consent. **Opting out of marketing and promotional communications:**You can unsubscribe from our marketing and promotional communications at any time by clicking on the unsubscribe link in the emails that we send, or by contacting us using the details provided in the section "[HOW CAN YOU CONTACT US ABOUT THIS NOTICE?](#contact)" below. You will then be removed from the marketing lists. However, we may still communicate with you — for example, to send you service-related messages that are necessary for the administration and use of your account, to respond to service requests, or for other non-marketing purposes. **Account Information** If you would at any time like to review or change the information in your account or terminate your account, you can: - Log in to your account settings and update your user account. Upon your request to terminate your account, we will deactivate or delete your account and information from our active databases. However, we may retain some information in our files to prevent fraud, troubleshoot problems, assist with any investigations, enforce our legal terms and/or comply with applicable legal requirements. **Cookies and similar technologies:** Most Web browsers are set to accept cookies by default. If you prefer, you can usually choose to set your browser to remove cookies and to reject cookies. If you choose to remove cookies or reject cookies, this could affect certain features or services of our Services. You may also [opt out of interest-based advertising by advertisers](http://www.aboutads.info/choices/) on our Services. For further information, please see our Cookie Notice: paulkedrosky.com/cookies. If you have questions or comments about your privacy rights, you may email us at privacy@paulkedrosky.com. **10\. CONTROLS FOR DO-NOT-TRACK FEATURES** Most web browsers and some mobile operating systems and mobile applications include a Do-Not-Track ("DNT") feature or setting you can activate to signal your privacy preference not to have data about your online browsing activities monitored and collected. At this stage no uniform technology standard for recognizing and implementing DNT signals has been finalized. As such, we do not currently respond to DNT browser signals or any other mechanism that automatically communicates your choice not to be tracked online. If a standard for online tracking is adopted that we must follow in the future, we will inform you about that practice in a revised version of this privacy notice. **11\. DO CALIFORNIA RESIDENTS HAVE SPECIFIC PRIVACY RIGHTS?** ***In Short:*** *Yes, if you are a resident of California, you are granted specific rights regarding access to your personal information.* California Civil Code Section 1798.83, also known as the "Shine The Light" law, permits our users who are California residents to request and obtain from us, once a year and free of charge, information about categories of personal information (if any) we disclosed to third parties for direct marketing purposes and the names and addresses of all third parties with which we shared personal information in the immediately preceding calendar year. If you are a California resident and would like to make such a request, please submit your request in writing to us using the contact information provided below. If you are under 18 years of age, reside in California, and have a registered account with Services, you have the right to request removal of unwanted data that you publicly post on the Services. To request removal of such data, please contact us using the contact information provided below and include the email address associated with your account and a statement that you reside in California. We will make sure the data is not publicly displayed on the Services, but please be aware that the data may not be completely or comprehensively removed from all our systems (e.g., backups, etc.). **CCPA Privacy Notice** The California Code of Regulations defines a "resident" as: (1) every individual who is in the State of California for other than a temporary or transitory purpose and (2) every individual who is domiciled in the State of California who is outside the State of California for a temporary or transitory purpose All other individuals are defined as "non-residents." If this definition of "resident" applies to you, we must adhere to certain rights and obligations regarding your personal information. **What categories of personal information do we collect?** We have collected the following categories of personal information in the past twelve (12) months: **Category** **Examples** **Collected** A. Identifiers Contact details, such as real name, alias, postal address, telephone or mobile contact number, unique personal identifier, online identifier, Internet Protocol address, email address, and account name YES B. Personal information categories listed in the California Customer Records statute Name, contact information, education, employment, employment history, and financial information NO C. Protected classification characteristics under California or federal law Gender and date of birth NO D. Commercial information Transaction information, purchase history, financial details, and payment information NO E. Biometric information Fingerprints and voiceprints NO F. Internet or other similar network activity Browsing history, search history, online behavior, interest data, and interactions with our and other websites, applications, systems, and advertisements NO G. Geolocation data Device location NO H. Audio, electronic, visual, thermal, olfactory, or similar information Images and audio, video or call recordings created in connection with our business activities NO I. Professional or employment-related information Business contact details in order to provide you our Services at a business level or job title, work history, and professional qualifications if you apply for a job with us NO J. Education Information Student records and directory information NO K. Inferences drawn from other personal information Inferences drawn from any of the collected personal information listed above to create a profile or summary about, for example, an individual’s preferences and characteristics NO L. Sensitive Personal Information NO We will use and retain the collected personal information as needed to provide the Services or for: - Category A - As long as the user has an account with us We may also collect other personal information outside of these categories through instances where you interact with us in person, online, or by phone or mail in the context of: - Receiving help through our customer support channels; - Participation in customer surveys or contests; and - Facilitation in the delivery of our Services and to respond to your inquiries. **How do we use and share your personal information?** More information about our data collection and sharing practices can be found in this privacy notice and our Cookie Notice: paulkedrosky.com/cookies. You may contact us by visiting paulkedrosky.com/contact, or by referring to the contact details at the bottom of this document. If you are using an authorized agent to exercise your right to opt out we may deny a request if the authorized agent does not submit proof that they have been validly authorized to act on your behalf. **Will your information be shared with anyone else?** We may disclose your personal information with our service providers pursuant to a written contract between us and each service provider. Each service provider is a for-profit entity that processes the information on our behalf, following the same strict privacy protection obligations mandated by the CCPA. We may use your personal information for our own business purposes, such as for undertaking internal research for technological development and demonstration. This is not considered to be "selling" of your personal information. paulkedrosky.com has not disclosed, sold, or shared any personal information to third parties for a business or commercial purpose in the preceding twelve (12) months. paulkedrosky.com will not sell or share personal information in the future belonging to website visitors, users, and other consumers. **Your rights with respect to your personal data** Right to request deletion of the data — Request to delete You can ask for the deletion of your personal information. If you ask us to delete your personal information, we will respect your request and delete your personal information, subject to certain exceptions provided by law, such as (but not limited to) the exercise by another consumer of his or her right to free speech, our compliance requirements resulting from a legal obligation, or any processing that may be required to protect against illegal activities. Right to be informed — Request to know Depending on the circumstances, you have a right to know: - whether we collect and use your personal information; - the categories of personal information that we collect; - the purposes for which the collected personal information is used; - whether we sell or share personal information to third parties; - the categories of personal information that we sold, shared, or disclosed for a business purpose; - the categories of third parties to whom the personal information was sold, shared, or disclosed for a business purpose; - the business or commercial purpose for collecting, selling, or sharing personal information; and - the specific pieces of personal information we collected about you. In accordance with applicable law, we are not obligated to provide or delete consumer information that is de-identified in response to a consumer request or to re-identify individual data to verify a consumer request. Right to Non-Discrimination for the Exercise of a Consumer’s Privacy Rights We will not discriminate against you if you exercise your privacy rights. Right to Limit Use and Disclosure of Sensitive Personal Information We do not process consumer's sensitive personal information. Verification process Upon receiving your request, we will need to verify your identity to determine you are the same person about whom we have the information in our system. These verification efforts require us to ask you to provide information so that we can match it with information you have previously provided us. For instance, depending on the type of request you submit, we may ask you to provide certain information so that we can match the information you provide with the information we already have on file, or we may contact you through a communication method (e.g., phone or email) that you have previously provided to us. We may also use other verification methods as the circumstances dictate. We will only use personal information provided in your request to verify your identity or authority to make the request. To the extent possible, we will avoid requesting additional information from you for the purposes of verification. However, if we cannot verify your identity from the information already maintained by us, we may request that you provide additional information for the purposes of verifying your identity and for security or fraud-prevention purposes. We will delete such additionally provided information as soon as we finish verifying you. Other privacy rights - You may object to the processing of your personal information. - You may request correction of your personal data if it is incorrect or no longer relevant, or ask to restrict the processing of the information. - You can designate an authorized agent to make a request under the CCPA on your behalf. We may deny a request from an authorized agent that does not submit proof that they have been validly authorized to act on your behalf in accordance with the CCPA. - You may request to opt out from future selling or sharing of your personal information to third parties. Upon receiving an opt-out request, we will act upon the request as soon as feasibly possible, but no later than fifteen (15) days from the date of the request submission. To exercise these rights, you can contact us by visiting paulkedrosky.com/contact, or by referring to the contact details at the bottom of this document. If you have a complaint about how we handle your data, we would like to hear from you. **12\. DO WE MAKE UPDATES TO THIS NOTICE?** **In Short:* Yes, we will update this notice as necessary to stay compliant with relevant laws.* We may update this privacy notice from time to time. The updated version will be indicated by an updated "Revised" date and the updated version will be effective as soon as it is accessible. If we make material changes to this privacy notice, we may notify you either by prominently posting a notice of such changes or by directly sending you a notification. We encourage you to review this privacy notice frequently to be informed of how we are protecting your information. **13\. HOW CAN YOU CONTACT US ABOUT THIS NOTICE?** If you have questions or comments about this notice, you may email us at privacy@paulkedrosky.com or contact us by post at: paulkedrosky.com \_\_\_\_\_\_\_\_\_\_ \_\_\_\_\_\_\_\_\_\_, CA United States **14\. HOW CAN YOU REVIEW, UPDATE, OR DELETE THE DATA WE COLLECT FROM YOU?** Based on the applicable laws of your country, you may have the right to request access to the personal information we collect from you, change that information, or delete it. To request to review, update, or delete your personal information, please fill out and submit a [data subject access request](https://app.termly.io/notify/90a94bd2-66a9-4cf2-9b9d-773f28fdbfd9). This privacy policy was created using Termly's [Privacy Policy Generator](https://termly.io/products/privacy-policy-generator/). ### Terms of Service URL: https://paulkedrosky.com/terms-of-service/ Last updated: 2023-06-01T21:35:50.000Z # **1\. Terms** By accessing this Website, accessible from paulkedrosky.com, you are agreeing to be bound by these Website Terms and Conditions of Use and agree that you are responsible for the agreement with any applicable local laws. If you disagree with any of these terms, you are prohibited from accessing this site. The materials contained in this Website are protected by copyright and trade mark law. ## **2\. Use License** Permission is granted to temporarily download one copy of the materials on paulkedrosky.com for personal, non-commercial transitory viewing only. This is the grant of a license, not a transfer of title, and under this license you may not: - modify or copy the materials; - use the materials for any commercial purpose or for any public display; - attempt to reverse engineer any software contained on paulkedrosky.com; - remove any copyright or other proprietary notations from the materials; or - transferring the materials to another person or "mirror" the materials on any other server. This will let paulkedrosky.com terminate upon violations of any of these restrictions. Upon termination, your viewing right will also be terminated and you should destroy any downloaded materials in your possession whether it is printed or electronic format. ## **3\. Disclaimer** All the materials on paulkedrosky.com are provided “as is”. paulkedrosky.com makes no warranties, may it be expressed or implied, therefore negates all other warranties. Furthermore, paulkedrosky.com does not make any representations concerning the accuracy or reliability of the use of the materials on its Website or otherwise relating to such materials or any sites linked to this Website. ## **4\. Limitations** paulkedrosky.com or its suppliers will not be hold accountable for any damages that will arise with the use or inability to use the materials on paulkedrosky.com, even if paulkedrosky.com or an authorize representative of this Website has been notified, orally or written, of the possibility of such damage. Some jurisdiction does not allow limitations on implied warranties or limitations of liability for incidental damages, these limitations may not apply to you. ## **5\. Revisions and Errata** The materials appearing on paulkedrosky.com may include technical, typographical, or photographic errors. paulkedrosky.com will not promise that any of the materials in this Website are accurate, complete, or current. paulkedrosky.com may change the materials contained on its Website at any time without notice. paulkedrosky.com does not make any commitment to update the materials. ## **6\. Links** paulkedrosky.com has not reviewed all of the sites linked to its Website and is not responsible for the contents of any such linked site. The presence of any link does not imply endorsement by paulkedrosky.com. The use of any linked website is at the user's own risk. ## **7\. Site Terms of Use Modifications** paulkedrosky.com may revise these Terms of Use for its Website at any time without prior notice. By using this Website, you are agreeing to be bound by the current version of these Terms and Conditions of Use. ## **8\. Governing Law** Any claim related to paulkedrosky.com shall be governed by the laws of the United States and the State of California without regards to its conflict of law provisions. ### Premium Services URL: https://paulkedrosky.com/premium/ Last updated: 2025-05-15T21:19:51.000Z Premium subscribers get the same thoughtful, context-creating commentary on risk, complexity, AI, and finance. But premium subscribers also get specific, biweekly ideas that can potentially be incorporated into investment theses and ideas. As former equity analysts, this is serious, empirical, and extrapolative work, not merely "horse race analysis" about which chipset vendor is currently getting more orders. Specifically, we think through the broader societal consequences of what is going on in economies, how it is likely to play out in the near and longer term, and who will benefit and who will be hurt. We consider risk models, long/short trades, macro consequences, and currency implications. We also make ourselves available, via appointment, for one premium subscriber consultation a month. Here is a summary of some of the main service items: | Item | Description | | -------------- | ---------------------------------------------- | | Biweekly ideas | Biweekly ideas with actionable implications | | Calls | By appointment, one monthly call with analysts | | Archives | Fully searchable access to research archives | The premium service is currently $500 a year and is capped with respect to the number of subscribers. We will never be a wide-open service and will remain a boutique, idea-driven research shop. We have had many requests to launch this, so if it sounds like something that you would find valuable, sign up before we reach our subscriber cap. You can subscribe by clicking the link below: [Subscribe](https://paulkedrosky.com/#/portal/signup) ### Media/Talks URL: https://paulkedrosky.com/appearances/ Last updated: 2024-02-04T20:57:16.000Z We are regularly quoted in media, as well as giving private and public talks. Here are some recent examples: - [Milken 2023: AI ethics and the future of work ](https://milkeninstitute.org/panel/14501/governing-ai-ethics-regulation-and-practical-applications) - [TechCrunch: Is ChatGPT a "virus that has been released into the wild"? ](https://news.google.com/articles/CBMiWmh0dHBzOi8vdGVjaGNydW5jaC5jb20vMjAyMi8xMi8wOS9pcy1jaGF0Z3B0LWEtdmlydXMtdGhhdC1oYXMtYmVlbi1yZWxlYXNlZC1pbnRvLXRoZS13aWxkL9IBXmh0dHBzOi8vdGVjaGNydW5jaC5jb20vMjAyMi8xMi8wOS9pcy1jaGF0Z3B0LWEtdmlydXMtdGhhdC1oYXMtYmVlbi1yZWxlYXNlZC1pbnRvLXRoZS13aWxkL2FtcC8?hl=en-US&gl=US&ceid=US%3Aen) - [ Axios: ChatGPT is Changing the Software-Making Game](https://news.google.com/articles/CBMiSGh0dHBzOi8vd3d3LmF4aW9zLmNvbS8yMDIzLzAzLzI3L2FpLWNoYXRncHQtc29mdHdhcmUtcHJvZ3JhbW1pbmctcGx1Z2luc9IBAA?hl=en-US&gl=US&ceid=US%3Aen) - [TTI/Vanguard: Demographics, Automation, AI, and the Future of Work ](https://youtu.be/mZ49-QiX3wE) ### Members URL: https://paulkedrosky.com/account/ Last updated: 2025-09-25T19:49:57.000Z _This page is for paying subscribers only._ ### Thanks For Signing Up! URL: https://paulkedrosky.com/plus-thanks/ Last updated: 2025-11-11T05:19:53.000Z Thanks for signing up at the **Plus** level. You will start receiving weekly newsletters, Kedrosky Daily (two graphs + an idea), and will have full access to **Plus** and **Free** archives. ### Thanks for Signing Up! URL: https://paulkedrosky.com/basic-thanks/ Last updated: 2025-11-11T05:19:29.000Z Thanks for signing up at the **Basic** level. You will start receiving free letters, excerpts from premium letters, Kedrosky Daily (Sunday Edition), and access to free letters in the archives. ### Subhead on main URL: https://paulkedrosky.com/home-subscribe/ Last updated: 2025-12-23T04:41:36.000Z Complexity & linkagesacross **AI technologies**, **finance**, **human physiology**, **energy**, **climate**, and more. Thoughtful, specific, data-driven, and generally **uncorrelated** with current events. **Consultant** to some of the largest **asset managers** and **investment banks** in the U.S. and Europe. As **quoted** in the Wall Street Journal, BBC, Financial Times, NY Times, CNBC, Bloomberg, and many others. ### topic-list URL: https://paulkedrosky.com/topic-list/ Last updated: 2025-05-16T15:36:20.000Z _No content available._ ### As seen on URL: https://paulkedrosky.com/home-logos/ Last updated: 2026-01-24T04:16:09.000Z ![](https://paulkedrosky.com/content/images/2025/05/brief-nyt.svg) ![](https://paulkedrosky.com/content/images/2025/05/New_Bloomberg_Logo.svg-1.png) ![](https://paulkedrosky.com/content/images/2025/12/2025-12-23_19-27-51.jpg) ![](https://paulkedrosky.com/content/images/2025/12/2025-12-23_19-28-23.jpg) ![](https://paulkedrosky.com/content/images/2026/01/2026-01-23_19-47-51.jpg) ![](https://paulkedrosky.com/content/images/2025/12/2025-12-23_19-29-57.jpg) ![](https://paulkedrosky.com/content/images/2025/12/2025-12-23_19-30-49.jpg) ![](https://paulkedrosky.com/content/images/2025/05/2025-05-16_09-26-50.png) ### Tags URL: https://paulkedrosky.com/tags/ Last updated: 2025-05-16T22:44:07.000Z # Topics Explore the most important topics we cover ### Consulting/Speaking URL: https://paulkedrosky.com/consulting-speaking/ Last updated: 2025-11-11T05:23:25.000Z I selectively consult for some of the largest asset managers and sell-side firms in the U.S. and Europe. If you would like to discuss that further, please feel free to contact me [here](https://paulkedrosky.com/contact/). ### Institutional URL: https://paulkedrosky.com/institutional/ Last updated: 2025-11-10T23:38:42.000Z Here is all recent writing, including clients-only. ### Members URL: https://paulkedrosky.com/plus-members/ Last updated: 2025-11-10T23:38:33.000Z Here is all recent writing, both members-only and free, for non-institutional subscribers: ### Rough Notes URL: https://paulkedrosky.com/thenotes/ Last updated: 2026-02-04T05:16:48.000Z We publish Rough Notes on topics of interest on a continuous basis. You can see the latest notes on the front page, but the full scrollable archive is for subscribers only. _This page is for paying subscribers only._ ### Note URL: https://paulkedrosky.com/n/ Last updated: 2026-01-02T23:48:27.000Z _No content available._ ### Token Deflation Simulator URL: https://paulkedrosky.com/token-sim/ Last updated: 2026-02-03T22:50:31.000Z _This page is for paying subscribers only._ ## Posts ### Ezra Klein Appearance: Is There an AI Bubble? URL: https://paulkedrosky.com/ezra-klein-appearance-is-there-an-ai-bubble/ Last updated: 2026-10-16T13:34:03.000Z I was a guest earlier this week on the Ezra Klein show, discussing the current AI moment. As ever, I was focused on the underlying financial and funding architecture, the implicit assumptions, and the stressors, some of which exploded into broader awareness this week as credit cracks yawned wide. Getting mired in whether AI works or is useful is, of course, a mug's gam (it does, and it can be very), and a sideshow when being thoughtful about a historically epic economic capex cycle and its far-reaching consequences. The episode is out [now](https://www.nytimes.com/2026/10/16/opinion/ezra-klein-podcast-azeem-azhar-paul-kedrosky.html). Thanks to Ezra for having me. ### Five Things from Goldman's New China AI Report URL: https://paulkedrosky.com/five-things-from-goldmans-new-china-ai-report/ Last updated: 2026-10-11T22:35:10.000Z _This post is for paying subscribers only._ ### Roundup: Capital, Confidence, and Contagion URL: https://paulkedrosky.com/roundup-capital-confidence-and-contagion/ Last updated: 2026-10-09T21:24:28.000Z The following Friday roundup describes an increasingly narrow system: **capital and confidence are concentrating** around AI and its infrastructure, while **strain builds ... everywhere**, from public markets, to sovereign debt, to the consumer economy. Here is what's ahead: 1. Tech Dominates the S&P 500 — Concentration surpasses the dot-com peak 2. AI's Growing Debt Appetite — Hyperscalers reshape bond markets 3. AI Scaling Approaches Human Efficiency — The cost of longer autonomous tasks collapses 4. Anthropic IPO Uncertainty — Prediction markets price further delays 5. IPO Cancellations Surge — September withdrawals hit a five-year high 6. Getty's AI Reckoning — Generative AI undermines image licensing economics 7. Consumer Confidence Collapses — Household sentiment approaches historic lows 8. France's Sovereign Debt Stress — Rising spreads threaten euro-area contagion --- ![](https://paulkedrosky.com/content/images/2026/10/2026-10-09_14-11-19.jpeg) Source: Reuters **Tech was already the market; now it is even more so.** The sector has reached roughly 40% of S&P 500 capitalization, surpassing the dot-com peak. This is not merely a valuation story. Index performance, retirement assets, corporate investment, and the wealth effect are becoming increasingly dependent on one interconnected technology trade. --- ![](https://paulkedrosky.com/content/images/2026/10/2026-10-09_13-19-22.jpeg) **That concentration requires enormous amounts of capital.** Hyperscalers dominate the year’s largest bond offerings, sometimes returning to the market within months. The AI buildout is no longer being financed from cash flow alone: its capital requirements are beginning to reshape the investment-grade debt market. --- ![](https://paulkedrosky.com/content/images/2026/10/2026-10-09_13-55-45.jpeg) Sourc: Toby Ord **AI capability is also becoming radically cheaper to extend.** Early models required roughly 10,000 times more resources to achieve a tenfold increase in task horizon; recent models need closer to 20 times, versus the human benchmark of 10\. The remaining efficiency gap is now small enough that longer autonomous work becomes an engineering and deployment question, not a distant theoretical one. --- ![](https://paulkedrosky.com/content/images/2026/10/2026-10-09_11-29-13.jpeg) **Markets remain skeptical about Anthropic’s IPO timetable.** Despite an announcement being expected within weeks, prediction markets still price a substantial chance that it will not happen by then. The odds rise only as the deadline extends into 2027—a market pricing slippage, not inevitability. --- ![](https://paulkedrosky.com/content/images/2026/10/2026-10-09_13-54-44.jpeg) **IPO withdrawals are at record levels.** September produced nine cancellations, the most for that month in at least five years and 4.5 times the 2025 total. Delays and withdrawals are accumulating even as a small group of AI companies commands extraordinary private valuations and investor attention. _This post is for paying subscribers only._ ### The Glacial Lake Outburst Problem in AI IPOs URL: https://paulkedrosky.com/the-glacial-lake-outburst-problem-in-ai-ipos/ Last updated: 2026-10-09T18:54:30.000Z ### What Happened The IPO market has abruptly slowed, with a growing list of companies postponing or abandoning planned public offerings. Here is just a partial list: **Holtec, Oura, SB Energy, Bamboo Insurance, and EG Group,** all of which have delayed their US listings. In Australia, Nvidia-backed **AI data center operator Firmus** just scrapped a planned $5 billion IPO that would have valued the company at roughly $30 billion. This has come as a **huge surprise to many observers**, who noisily predicted a torrent of IPOs would come to market after SpaceX, and before the main events: **OpenAI and Anthropic**. ### What It Means News [reports](https://www.ft.com/content/b8924d77-364b-46c1-b783-5db73a91f351?syn-25a6b1a6=1) wrongly, we think, attribute the slowdown to **weakening investor demand, excessive valuations, and growing skepticism about AI**. The right answer is in something we have been [writing here for months](https://paulkedrosky.com/the-coming-mega-ipo-flow-funding-problem-of-2026/), and even created a tool to model: **the mega-IPO flow and its consequences**. Or as one banker told the FT, “Right now no one cares about anything that isn’t Anthropic.” Anthropic and OpenAI are, in size, **historically unprecedented IPOs**. With Anthropic reportedly targeting a valuation above $2 trillion and OpenAI potentially following with another enormous offering, institutional investors are **reserving capital, managing exposure, and waiting** for the two transactions that will define the AI market. The capital market effect is to **suppress everything else**. Smaller, more speculative companies are unable to attract attention or get the valuations they want. Issuance slows. Bankers (and journalists) interpret the slowdown as deteriorating sentiment. **Both are wrong**. _This post is for paying subscribers only._ ### Roundup: Capex, Trade, Risk, Climate, Call Centers, Concentration, etc. URL: https://paulkedrosky.com/roundup-capex-trade-risk-climate-call-centers-concentration-etc/ Last updated: 2026-10-06T23:01:20.000Z **What's ahead:** 1. **AI Is Carrying the S&P 500** 2. **Tech Imports Surge With the AI Buildout** 3. **Call-Center Employment Is Shrinking Fast** 4. **Pacific Temperatures Move Far Outside the Historical Range** 5. **The S&P 500 Without AI Looks Much Weaker** 6. **AI Fear Reaches the Supermarket Tabloids** 7. **US External Liabilities Keep Growing** 8. **Semiconductor Revenue Is Outrunning Unit Growth** 9. **US Investment Returns to Dot-Com-Era Levels** ![](https://paulkedrosky.com/content/images/2026/10/2026-10-06_15-24-05.jpeg) SourceL Bespoke Premium **AI is carrying the S&P 500 more than ever.** The index is at new highs even as the share of stocks making 52-week highs has fallen sharply: the breadth disconnect is at a record. **The gap between the index and the broader market is an AI story.** --- ![](https://paulkedrosky.com/content/images/2026/10/2026-10-06_15-18-06.jpeg) Source: Bespoke Premium **Tech imports have surged from about 1% of GDP to nearly 3%.** That puts them close to the historical peak for all consumer-goods imports combined. **AI infrastructure spending is now large enough to alter the composition of US trade and national investment data.** --- ![](https://paulkedrosky.com/content/images/2026/10/call-centers.jpg) **Global call-center employment is now shrinking quickly after more than a decade of steady growth.** Year-over-year employment growth has fallen from roughly 4% to about -4% since **the ChatGPT Moment**. **This is one of the clearest large-scale signs so far of AI replacing routine white-collar service work,** and given regional dependencies—within the US and around the world—highly consequential. --- ![](https://paulkedrosky.com/content/images/2026/10/2026-10-06_12-47-05.jpg) **Niño 3.4 sea-surface temperatures are far outside the historical range.** The current reading is roughly 3.5°C above the 1991–2020 seasonal average and **above every prior year** in the record. **This is a large enough anomaly to matter well beyond the Pacific, given the role of ENSO in global weather patterns.** --- ![](https://paulkedrosky.com/content/images/2026/10/2026-10-06_12-45-53.jpeg) Source; Goldman Sachs **The S&P 500 and the S&P 500 excluding AI stocks have diverged sharply.** The headline index is near its highs, while the ex-AI version has fallen materially since August. **AI is basically the index now.** --- ![](https://paulkedrosky.com/content/images/2026/10/2026-10-06_09-36-29.jpeg) **AI fear has reached mass-market tabloid culture.** *Globe* is now selling a cover built around killer AI, bunkers, grid failures and “survival plans.” **The claims are absurd, but the cultural shift is real: AI risk is no longer a specialist concern.** We here miss UFOs. --- _This post is for paying subscribers only._ ### Carrying Water for Frontier Model Companies URL: https://paulkedrosky.com/carrying-water-for-frontier-model-companies/ Last updated: 2026-10-05T13:30:00.000Z ## What Happened AI spending is becoming irritatingly difficult for businesses to forecast, the WSJ [argues](https://www.wsj.com/tech/personal-tech/ai-token-spending-businesses-431ee94a?mod=panda%5Fwsj%5Fsection%5Falert), because the **task completion cost** can vary widely across models and even across repeated runs of the same model. So far, so **mostly unarguable**. The WSJ provides. as further support, a survey in which **only 11% of companies forecast AI spending** within 10% of actual costs, and showed that **lower-priced models** sometimes cost more overall than higher-priced ones. The core of the piece compares **two Google models** performing the same task. Gemini 3.1 Pro, a frontier-ish model, completed it successfully in **85 steps for roughly $1**, while the cheaper Gemini 3 Flash took 952 steps, failed, and it **cost $14** in tokens. The article concludes that **businesses cannot estimate task cost** from model pricing alone. Model selection, task complexity, retries, failure rates, and token consumption all matter. ## What It Means This is all fair enough, but it is **much less striking** than the WSJ thinks it is. And, whether intentionally or not, it **carries water for frontier AI companies**: their salespeople will have been spamming customers non-stop for the last 24 hours: **CHECK THIS WSJ ARTICLE OUT** email subject linesare filling inboxesworldwide. Sure, **AI spending is much more variable** than traditional software spending. Agentic systems can retry, wander, invoke tools, consume unpredictable amounts of context, and fail after doing substantial work. That makes individual task costs noisy and enterprise budgeting harder. But the piece goes further by giving too much weight to unusual cases in which **cheaper models become more expensive** because they take far more steps. That is not the normal pattern as models get better. Outside of genuinely **weak or badly matched models**, **lower-cost models remain lower-cost** for almost all tasks for which they are suited. ![](https://paulkedrosky.com/content/images/2026/10/2026-10-04_19-53-54.jpeg) _This post is for paying subscribers only._ ### Sunday Roundup: Biggest Risk Episode Ever, China's AI Inroads, NYC vs AI, Robots and Babies, Cars, etc. URL: https://paulkedrosky.com/sunday-roundup-biggest-risk-episode-ever-chinas-ai-inroads-nyc-vs-ai-robots-and-babies-cars-etc/ Last updated: 2026-10-05T00:14:03.000Z A few things we are thinking about to start the week. - **AI as the biggest capex risk episode in history** - **China is within the AI margin of error** - **New York City Council is the hero Gotham deserves, but not the one it needs** - **When people compete with robots, they have fewer babies** - **AI is helping export expensive money worldwide** - **Gasoline cars tumble to less than half the market** --- ### **1\. AI as the biggest capex risk episode in history** Our longtime friend Jim Chanos gave us a [nice shoutout](https://x.com/firesidealpha/status/2106586131005673682/video/1?s=48) in a recent talk, which was kind of him, in discussing the current episode as one of the biggest and riskiest in Western economic history. ### 2\. China is within the AI margin of error A new [Bloomberg Intelligence report](https://www.bloomberg.com/news/articles/2026-10-04/us-lead-in-ai-over-china-narrows-after-deepseek-gains-bi-says) argues that Chinese AI models are within the margin of error, no pun intended, with respect to US frontier models. This is a point we have made many times: models are converging, variance is declining, and benchmarks are largely misleading given their reliance on absolute (not relative) gains. ### 3\. New York City Council is the hero Gotham deserves, but not the one it needs The city of New York is holding what promises to be a [raucous hearing](https://nyc.legistar1.com/nyc/meetings/2026/10/22752%5FA%5FCommittee%5Fof%5Fthe%5FWhole%5F26-10-05%5FCommittee%5FAgenda.pdf) Monday into agents, large language models, and AI risk. It includes "whistleblower" former Anthropic researcher Jacob Coxon, as well as many others. A highlight is that X/AI allegedly ignored the request to attend and apparently had to be subpoenaed. ### 4\. When people compete with robots, they have fewer babies Like advanced societies needed more reasons to have fewer babies: a new [paper](https://journals.vilniustech.lt/index.php/TEDE/article/view/27719) presents intriguing data suggesting that competition with robots is causing Chinese families to have fewer kids. The reasons are threefold: training offspring for the race becomes more expensive for parents, their own jobs are less stable, and growing income inequality. _This post is for paying subscribers only._ ### Recent Interviews Roundup URL: https://paulkedrosky.com/recent-interviews-roundup-2/ Last updated: 2026-10-02T20:40:42.000Z To get caught up, here are a few recent interviews with me on various of my favorite topics. If I've missed any, let me know. ### Musing About Meta's Muse URL: https://paulkedrosky.com/musing-about-metas-muse/ Last updated: 2026-10-02T14:36:03.000Z 💡 I've been using Meta's Muse for a week or two. While flawed and clearly designed by engineers fed a ****forced diet of Meta design quirks**—all apps must have timelines! all apps must contain news!—it has mostly worked. Granted, I'm not trying to [****make a market**](https://www.theguardian.com/technology/2026/sep/28/metas-ai-agent-muse-home-address) on Facebook Marketplace, or get it to help me [****make bank**](https://www.reddit.com/r/MetaAI/comments/1woospt/i%5Fdared%5Fmuse%5Fai%5Fto%5Fmake%5F100%5Fin%5F30%5Fdays%5Fday%5F2/), It is almost exclusively scanning email and calendar, things that, obviously, Google should be doing already. Muse is ****wildly imperfect**, with "side chats" that make no sense, inconsistency about what it just does on a vibe, and what you must demand it do, but ... still. ### What Happened Meta launched Muse last month as a personal AI agent designed to act across applications and services rather than simply answer questions. It quickly **rose to the top of the Google and Apple app stores**, as I've written [here previously](https://paulkedrosky.com/metas-muse-and-moat-less-ai/). ![](https://paulkedrosky.com/content/images/2026/10/2026-09-21_12-52-23.jpg) Muse can work with email, calendars, websites, and other services, use a browser, complete forms, and carry out multi-step tasks. Meta is also pushing it into business, where the agent can operate across some typical software and tasks. ### What It Means Despite all the noise around it, people mostly misunderstand what is happening. Muse's **general invisibility** is the important product change: Muse is the AI market shifting from a **chatbot interface toward a more naturalistic, ambient presence** that can take actions on a user's behalf, and let you know when it has. We are moving away from the opacity and arcana of coding harnesses toward something **more fluid, persistent and less visible**. Muse shows the value in AI moving **up the stack,** as tech types like to say**.** For the last few years, the AI industry has **obsessed over the model layer**: model quality, parameter counts, inference costs, tokens, and coding-centric benchmark performance. And then it moved on to harnesses, like Claude Code/Cowork, whose coding heritage shone through. _This post is for paying subscribers only._ ### Deputy Likes Dots: Crowding and the AI IPO Traffic Jam URL: https://paulkedrosky.com/deputy-likes-dots-crowding-and-the-ai-ipo-traffic-jam/ Last updated: 2026-09-29T21:32:16.000Z > Fucking horse is out the barn door, right? Let's try not to make me look stupid twice. > > And when you list the cases, put a little dot next to each one. Deputy likes dots. > > —Major Rawls, The Wire (S01E01) ### What Happened Oura **postponed** its IPO today. **OpenAI** recently did the same. Meanwhile, Anthropic is reportedly preparing to sell as much as **$100 billion of stock** at a valuation approaching **$2 trillion**, And, at the same time, OpenAI is raising another $ 30 billion while launching its **Muse-mimicking agentic thingie** [**Dots**](https://openai.com/index/introducing-dots/). These may not seem related, but they are. We will explain why. ### What It Means Let's start with the big numbers. Anthropic's $100-billion is a staggering sum, even if it is "only" 5% of **Anthropic’s implied value**. Nevertheless, that money has to come from somewhere, as we have written here [**previously**](https://paulkedrosky.com/the-coming-mega-ipo-flow-funding-problem-of-2026/). Equity markets must absorb an **extraordinary amount of equity supply** in a single transaction. You can think of it, as we've argued, as a form of crowding. An unprecedented amount of equity supply is coming to market, much of it in one transaction, and it is increasingly plausible that other issuers are simply getting out of the way. Large institutions have **finite new-issue budgets**, and a $100 billion Anthropic offering would consume a meaningful amount of available growth capital. It would also **dominate investor attention** and create **funding pressure**: investors wanting Anthropic may sell or avoid smaller new issues around the same time. This may explain Oura. It reported that **demand** for its offering was strong, and yet it postponed it. A **well-subscribed IPO** being postponed suggests something beyond ordinary weak demand. _This post is for paying subscribers only._ ### Switching Costs, AI, and the Complexity Basket URL: https://paulkedrosky.com/switching-costs-ai-and-the-complexity-basket/ Last updated: 2026-09-29T13:30:50.000Z While agentic AI tools like Muse have caused investors to reconsider so-called ****"inertia" companies**—those where excess profits come from consumers not switching—a more important category may be ****"complexity" companies—**where AI unlocks hidden uses. ### Table of Contents 1. **The Inertia Basket** 2. **The Complexity Basket** - The Good Kind of Complexity - The Bad Kind of Complexity - The Really Bad Kind of Complexity 3. **The Complexity Discount: Putting It All Together** 4. **Testing the Complexity Basket** --- 1. **The Inertia Basket** Goldman Sachs recently published a basket of [**“consumer inertia” stocks**](https://www.bloomberg.com/news/articles/2026-09-22/meta-s-muse-drags-down-stocks-that-depend-on-consumer-inertia): companies that benefit when customers keep paying because comparison and switching are annoying. Telecom, insurance, streaming, and travel all contain versions of this dynamic. ![](https://paulkedrosky.com/content/images/2026/09/2026-09-28_14-53-33.jpeg) The claim is that **AI agents pressurize inertia**. Software can continuously compare prices, monitor terms, switch providers, and renegotiate services. Consumer inertia exists, so the argument goes, because people have limited time and attention. Agents reduce that constraint. Does it reduce it that much? **Consumer inertia is eternal**, so we are doubtful. Investors, pre-Muse, were already actively **punishing inertia** names. It has been a quiet and less consequential version of the **silly Saaspocalypse**, where a class of equities were deemed to be AI-damaged. ![](https://paulkedrosky.com/content/images/2026/09/2026-09-28_14-54-51.jpeg) 1. **The Complexity Basket** There is an opposite category of companies, however: Ones where friction has worked *against* the business, rather than helping it. These companies carry what we call a **complexity discount**. The underlying product can be valuable, differentiated, and difficult to reproduce, but adoption and/or usage remains below its potential because the **product is cumbersome to configure, understand or operate**. There are, loosely three kinds of such companies, which we will now walk through. **2a. The Good Kind of Complexity** Consider the case of **Cloudflare**. The company's global network and security infrastructure are hard to copy, so it has a real moat. But using the product has historically required users to understand **DNS**, **certificates**, **tunnels**, **routing**, **firewall rules**, **Workers**, and **access policies**. Most people, us included, likely **Cloudflare-checked-out** after mastering DNS (domain name servers), if you got that far. While many of the company's other products were no doubt useful, it was **too hard** for most normals to deploy them, absent hiring e**xpensive sysops and network engineers**. That is **no longer the case**. Take paulkedrosky.com as an example. It uses a host of Cloudflare services, only some of which we fully understand, but all of which are useful. It uses DNS, caching, tunnels, routing, and Workers (which remain a mystery, but still work wonderfully). What has changed? AI reduces the **complexity burden** and exposes more of the value already present. Exposing the value, however, doesn't mean that customers no longer need it. The **moat remains**; the **complexity doesn't**. This opens up vast opportunities to, in **cringe marketing** parlance, do a lot of "up-selling". _This post is for paying subscribers only._ ### New Tool: Debt/Treasury Market Stress URL: https://paulkedrosky.com/new-tool-debt-treasury-market-stress/ Last updated: 2026-09-28T19:56:56.000Z The U.S. Treasury market is among the world's **largest and most liquid**. Its yields anchor **borrowing costs and asset valuations** well beyond government bonds. Stress in Treasuries canripple through **leveraged finance, sovereign debt, equity flows,** and **AI infrastructure financing.** With yields at levels that sharply [**raise hurdles for AI debt**](https://www.bloomberg.com/news/articles/2026-09-28/man-group-says-5-5-yields-risk-cracking-ai-capex-us-consumer), as well as creating sovereign financing issues, this is a **systemic risk measure** that must be closely watched. **Treasury Market Stress** tracks **auction tails**, **primary-dealer take-down**, **rate pressure**, and **rate volatility** by maturity, alongside a five-year **stress index**. The component readings remain visible beside the overall measure, with historical comparisons where the data support them. This makes it possible to distinguish a weak auction at one maturity from broader pressure across the Treasury market. ![Treasury Market Stress dashboard showing the overall index and its five-year history](https://paulkedrosky.com/assets/images/treasury-stress-card.png) **Data limitations:** Treasury yields are official daily observations, not intraday quotes. Independent pre-auction yield coverage is currently incomplete, so **tail history is limited** and is excluded from the historical index. **Available to paid subscribers:** [Open the dashboard](https://paulkedrosky.com/tools/treasury-stress/) or visit the [tools page](https://paulkedrosky.com/tools/). ### Four Charts: Tokens vs Revenue, GPUs, Canada, etc. URL: https://paulkedrosky.com/four-charts-tokens-vs-revenue-gpus-canada-etc/ Last updated: 2026-09-27T17:35:42.000Z Four charts worth thinking about: 1. The sharp rise of open-weight models as corporations shift away from the frontier 2. How private equity became the new bad guy in current politics 3. The AI boom is pulling forward years of GPU demand 4. Canada is seeing near-record outmigration --- ### 1\. Open-weight models are taking AI usage. Profits are likely next. ![](https://paulkedrosky.com/content/images/2026/09/2026-09-27_10-01-29.jpeg) Open-weight models now account for **more than half of AI token usage**, while capturing only **14% of spending**. The market has decided that many AI workloads do not require the most expensive proprietary models. **Good-enough intelligence** is becoming cheap and widely available. As open models continue improving, pricing power will move away from the model layer and toward the companies that own distribution, data, workflows, and customer relationships. For the moment, the wall of money keeps crashing down. Source: [FT](https://www.ft.com/content/d9de4776-1fc9-4f2b-aaaf-9961c35d8acd?syn-25a6b1a6=1) --- ### 2\. Private equity has become the new political target ![](https://paulkedrosky.com/content/images/2026/09/2026-09-27_06-57-08.jpeg) Private equity has moved from an obscure corner of finance into the **center of political attacks**. After the financial crisis, politicians focused their anger on banks. In this U.S. election cycle, **private equity is increasingly filling that role**. Private equity now owns large parts of **healthcare**, **housing**, **retail**, **restaurants**, and local businesses. Housing, in particular, has American voters furious at bidding against private equity. Election politics tends to find irritating symbols of power. **Private equity has become one**. Source: [FT](https://www.ft.com/content/7e91f857-30d9-4bf9-b7c9-ec04d4a6e1ed?syn-25a6b1a6=1) _This post is for paying subscribers only._ ### Weekend Reading: From Short Selling, to CDS Widening, to Funny AI Hats URL: https://paulkedrosky.com/weekend-reading-from-short-selling-to-cds-widening-to-funny-ai-hats/ Last updated: 2026-09-25T22:19:10.000Z Some thesis-centric weekend reading, with our take on each, tying them to the [house view](https://paulkedrosky.com/this-ai-moment-the-house-view-q4-2026/). ## Contents 1. CoreWeave goes to junk 2. Climate inflation is not rate-sensitive 3. Oracle has a power problem 4. Data-center opposition moves upstream 5. Amazon builds the robot factory 6. Burry on the capital cycle 7. Goldman found the clean trade 8. Thirty-year Treasury at 2004 levels 9. AI enters the utility control room 10. Utilities call bullshit on the queue 11. The biggest economic bet in U.S. history 12. Goldman gets to $1.2 trillion 13. GMO: what actually breaks the AI bubble 14. AI discovers upcoding 15. AI is saving scientists seven hours a week 16. Walmart draws the line at personalized pricing 17. The shorts spread through the AI supply chain 18. Australia was not AI alone --- ### 1\. **CoreWeave goes to junk** — [Bloomberg](https://www.bloomberg.com/news/articles/2026-09-23/coreweave-tied-data-center-raises-1-1-billion-in-junk-bonds) A CoreWeave-leased Virginia data center raised **$1.1 billion at 9.25%**. Fifteen-year lease, one tenant, non-investment-grade credit. This is exactly the financing migration discussed here for months: AI capex spilling out of hyperscaler balance sheets into SPVs, private credit, leases, guarantees and junk bonds. ### 2\. **Climate inflation is not rate-sensitive** — [Allianz](https://www.allianz-trade.com/en%5Fglobal/news-insights/economic-insights/Allianz-Climate-Economics-Report-2026.html) Allianz estimates Europe’s 2026 heatwaves cost $128 billion, with a strong El Niño capable of adding materially to global inflation next year. This supports the point made here repeatedly: **a meaningful share of measured inflation now comes from supply shocks that monetary policy cannot fix.** Oil, power and weather do not respond to another 25 basis points. ### 3\. **Oracle has a power problem** — [Bloomberg](https://www.bloomberg.com/news/articles/2026-09-24/oracle-cites-force-majeure-to-shield-itself-on-controversial-data-center) Oracle invoked force majeure around Project Jupiter in New Mexico over possible power delays. Another one for the file. The binding constraint on AI infrastructure was always going to move from chips to power, transmission and permitting**.** Oracle is now **writing that problem** into its contracts. ### 4\. **Amazon builds the robot factory** — [WSJ](https://www.wsj.com/tech/amazon-to-expand-robot-making-capacity-with-new-100-million-indiana-hub-87fce322) Amazon is spending another $100 million on an Indiana robotics plant. More interesting than another model release. **The economic payoff comes when AI and automation remove expensive physical labor from the cost structure.** ### 5\. **Data centers are power systems with buildings attached** — [Visual Capitalist](https://www.visualcapitalist.com/these-data-centers-could-consume-more-power-than-cities/) Some planned campuses have load profiles comparable with major cities. Calling them “data centers” understates the economics. **These are industrial power complexes**, with all the transmission, generation, water, land and political constraints that it implies. ### 7\. **Burry on the capital cycle** — [Michael Burry](https://michaeljburry.substack.com/p/capital-cycle-iq-and-the-forensic) Burry puts hyperscaler commitments, leases and guarantees in the trillions and points to investment intensity near historic extremes. This is very close to the argument made here for some time: **AI can be economically transformative while the capital cycle around it destroys enormous amounts of capital.** Railroads did it. Telecom did it. There is no reason AI should be different. ### 8\. **Goldman found the clean trade** — [FT](https://www.ft.com/content/bdec4129-ccac-4149-aa53-90ddd50cb925?syn-25a6b1a6=1) Goldman reportedly made more than $200 million from AI hedge fund Situational Awareness, which had become its largest prime-brokerage client. In a capital boom, financing the participants can be considerably better business than owning the assets. **Picks and shovels, except the shovel is, you know, us.** ### 9\. **Thirty-year Treasury at 2004 levels** — [FT](https://www.ft.com/content/2d87f8bf-d529-4997-90c5-393ef65d280c?syn-25a6b1a6=1) Long borrowing costs are at multi-decade highs while AI capex, energy shocks and sovereign issuance all compete for capital. **This is exactly why current U.S. inflation and rate data are so hard to read conventionally.** AI capex and energy are pushing prices and long yields higher for reasons that another Fed hike does little to address. ### 10\. **AI enters the utility control room** — [Utility Dive](https://www.utilitydive.com/news/new-york-audits-utility-ai-use-cites-risk-in-growing-dependency/831242/) New York is auditing utility use of AI across electric, gas and water systems. Sensible. The interesting AI risk sits inside tightly coupled infrastructure, where small model errors can propagate into large operational ones. _This post is for paying subscribers only._ ### This AI Moment: The House View, Q4 2026 URL: https://paulkedrosky.com/this-ai-moment-the-house-view-q4-2026/ Last updated: 2026-09-25T19:12:45.000Z As long-time capital markets people, we think in terms of "house views": What we believe, and what thus drives our interpretation of events. These can and do evolve, but it is also important to know them so one can be clear when one is drifting away, basking in the warm seas of confirmation bias, or seeing actual falsification. And, to be clear, we delight in discovering when we are wrong. Our greatest fear is being wrong and never finding out, so we aggressively look for falsifying data. As the writer Kathryn Schulz wrote in Being Wrong, her classic book on the topic, being wrong feels exactly like being right—until you find out you are wrong --- So, with thousands of new subscribers, heading into Q4 2026 feels like a good time to lay out our house views—subject to evolution, revision, falsification, and the odd moment of tub-thumping triumphalism. Here is the house view, in plain language, bulleted for easy scanning. We will revisit in the future as events evolve. - **AI is real and consequential.** Adoption is extraordinarily fast, many productivity gains are real, and the technology will matter enormously. - **The current AI boom is wildly distorting the economy.** Massive spending on chips, data centers, power and networking is now large enough to essentially be GDP growth, to distort profits, make trade data useless, break electricity grids, and distort capital spending. - **Headline economic data is misleading.** Without AI capex and its growing externalities—and recently energy shocks—the U.S. economy would look much weaker, plausibly even contracting. - **The boom is debt-financed.** AI infrastructure is no the largest borrower in investment-grade and high-yield markets, which is causing spiking funding pressure in other markets, from corporate to sovereigns. - **The basic financial mismatch is ugly:** companies are borrowing long-term to build infrastructure producing a commodity—compute/tokens—whose performance-adjusted price is falling 70-80% year-over-year. Usage therefore has to rise hundreds of percent into perpetuity to stand still, let alone produce Wall Street growth while funding towering debt. - **This can be true even if AI succeeds technologically.** The internet was transformative; telecom investors still destroyed vast amounts of capital. Great technology does not imply good infrastructure economics. It generally implies the opposite: back to fiber, railroads, canals, etc. - **AI capex is crowding out other investment.** Corporate cash flow can look strong partly because firms are delaying ordinary capex, having been squeezed out, while an unusually concentrated group pours money into AI. _This post is for paying subscribers only._ ### On Jensen Huang's Whinge with Ezra Klein URL: https://paulkedrosky.com/on-jensen-huangs-whinge-with-ezra-klein/ Last updated: 2026-09-24T22:39:32.000Z ## What Happened The NYT's Ezra Klein [interviewed](https://www.nytimes.com/2026/09/23/opinion/ezra-klein-podcast-jensen-huang.html?eafs%5Fenabled=false) Nvidia CEO Jensen Huang this week, and it was messy. The two repeatedly **talked past one another,** as Klein, looking for nuance, kept returning to **AI safety, collective-action problems, employment, regulation, and the scale of the investment boom**. Huang kept answering like an **engineer-in-a-box**: unsafe models should not be shipped; safety is an engineering problem; more compute will make systems safer and more useful; **AI will create work because human wants are effectively unlimited**. The disagreement was less about individual facts than about **what kind of system AI is, what kinds of risks it creates**, and even the **nature of knowing**. ## Why It Matters Huang’s safety position is an **unfalsifiable mess**. He accepts that AI safety matters enormously—at one point musing that labs may eventually devote something like **80 percent of their compute to evaluation and alignment**, while dismissing the need for mechanisms that force firms to do that. This creates a **prisoner's dilemma**: everyone would be better off if frontier models companies slowed, but they won't out of fear someone else won't. ![](https://paulkedrosky.com/content/images/2026/09/2026-09-24_15-23-43.jpeg) Asked by Klein about this collective-action problem, Huang's answer amounts to: **bro,** **don’t ship unsafe models**. But that is exactly the problem. Every participant has powerful incentives to move faster than is collectively desirable. The gains from defecting are immediate and enormous; many of the costs are delayed, distributed, and poorly handled by ordinary litigation. Telling firms to behave responsibly is not a response to that problem. J**ust try to sue someone for damages after we're all paperclips**. I'll wait. ![](https://paulkedrosky.com/content/images/2026/09/2026-09-24_15-36-10.jpeg) There is a second problem, however, and it may be the more fundamental one. _This post is for paying subscribers only._ ### Meta's Muse and Moat-less AI URL: https://paulkedrosky.com/metas-muse-and-moat-less-ai/ Last updated: 2026-09-21T21:02:42.000Z 💡 I was traveling most of last week in Europe, so am finally getting caught up. Will have some notes on my travels and conversations later this week. ## What Happened Meta's agentic AI app Muse entered the U.S. App Store’s top 20 on September 9 at No. 4\. Nine days later, it was No. 1. ![](https://paulkedrosky.com/content/images/2026/09/2026-09-21_12-52-23.jpeg) The unprecedented advance **displaced ChatGPT,** which had occupied the top position for most of the past year. **Gemini, despite Google’s distribution**, spent much of the period well below the leaders. ## What It Means Muse’s rise is a useful reminder that **AI consumer products can gain share** quickly, given buzz and distribution. Users can s**witch applications at negligible cost**, while the underlying model intelligence is available to every serious competitor. The takeaway should be, in short, that there is, despite trillions in spending, **minimal moat**, _This post is for paying subscribers only._ ### The AI-cession Among Young People, and its Consequences URL: https://paulkedrosky.com/the-ai-cession-among-young-people-and-its-consequences/ Last updated: 2026-09-15T21:19:38.000Z ## What Happened A new U.S. Census Bureau [study](https://www.census.gov/library/working-papers/2026/adrm/CES-WP-26-56.html?utm%5Fsource=chatgpt.com) finds an **unprecedented worsening in outcomes** since ChatGPT's launch for graduates from the U.S. **college majors most exposed to AI**. We are not generally fans of such studies, since they can be methodologically slippery, but this was strong work. Since the launch of ChatGPT in late 2022, graduates from the **most AI-exposed decile of academic majors** have seen their probability of initial employment fall by about **5 percentage points**, while their initial quarterly **earnings have fallen 13%**, relative to graduates from less-exposed majors. Roughly **half the earnings decline comes from lower pay** within industries, with the rest coming from **graduates being pushed into lower-wage sectors** such as retail and restaurants. ![](https://paulkedrosky.com/content/images/2026/09/2026-09-14_12-39-30.jpeg) ## What It Means The literature generally puts the initial earnings penalty from entering the labor market during a severe recession at roughly **9–10%**. So, for the most exposed majors, the post-ChatGPT earnings shock is arguably **larger than the normal first-year effect of graduating into a major recession**. _This post is for paying subscribers only._ ### Treasuries, Debt, and the Data Center Refi Cliff: Life at 5% URL: https://paulkedrosky.com/treasuries-debt-and-the-data-center-refi-cliff-life-at-5/ Last updated: 2026-09-15T13:30:42.000Z ## What Happened The **10-year Treasury crossed 5%** this week, its highest level since 2023\. This creates an **economic stress test**, with consumers pulling **money out of stocks**, attracted by the secure yield on the 10-year, and **low-rate-addicted corporates** under increasing stress, especially for data centers, as rates rise. ![](https://paulkedrosky.com/content/images/2026/09/2026-09-14_15-59-06.jpeg) The AI capex numbers are immense. The [Dallas Fed](https://www.dallasfed.org/research/economics/2026/0210-searls-aifinancing) estimates data-center investment at roughly **$3 trillion to $5 trillion over the next three to five years**. Using a simple scenario of **$1 trillion of AI/data-center investment, 70% externally financed**, outside financing would total **$700 billion**. At an all-in debt cost of **6.5%**, annual interest on $700 billion is about **$46 billion**. At **7.5%**, it is about **$53 billion**. At **9%**, it is **$63 billion**. There is also a rapidly growing data-center and GPU refinancing schedule ahead: roughly **$2.7 billion in 2029**, **$26 billion in 2030**, and **$34 billion in 2031**, including anticipated refinancing dates, all of which must be done at higher rates. ## What It All Means The AI infrastructure boom is now largely debt-driven, so it is definitionally becoming **much more sensitive** to its cost of capital. Hyperscalers already have their own **de facto spread curves**, and it is about to come under stress. ![](https://paulkedrosky.com/content/images/2026/09/2026-09-14_20-18-58.jpeg) At a newly 5% Treasury yield, new data-center debt has to yield well above 5%, usually at least 150 basis points higher, depending on deal structure and the quality of the borrower.. And every additional **100 basis points** on $700 billion of financing adds roughly **$7 billion a year** in interest expense. _This post is for paying subscribers only._ ### AI IPOs and Frontier AI as a Prisoner's Dilemma URL: https://paulkedrosky.com/ai-ipos-and-frontier-ai-as-a-prisoners-dilemma/ Last updated: 2026-09-14T14:14:55.000Z ### What Happened After last week's outbreak of [**"Hey maybe unfettered AI advances will kill us all"**](https://www.bbc.com/news/articles/c1kx0gyje9wo), and a subsequent [industry coalescence](https://www.ft.com/content/31220b59-b0c6-401c-a146-2b7b5d138837?syn-25a6b1a6=1) around Anthropic head Amodei;'s idea of perhaps, you know, [**slowing down a tad**](https://darioamodei.com/post/we-must-pace-the-frontier), there is a new and predictable argument making the rounds in the AI safety debate. It says that if frontier labs like OpenAI and Anthropic really believe frontier models are dangerous, they should simply **slow down themselves**. ![](https://paulkedrosky.com/content/images/2026/09/2026-09-14_06-44-25-1.jpg) No cartel. No special antitrust treatment. No new regulatory regime. If the **models are scary** enough to justify slowing the frontier, then the **frontier labs should just do it**, perhaps because they are trying to **limit litigation risk**. Anything else, to this way of thinking, starts to look like **regulatory capture dressed up in an AI safety frock**. Frontier labs do, of course, have obvious incentives to support rules that make life harder for smaller competitors, and “safety” can become a convenient way to turn incumbency into regulation.But that doesn't mean all calls for safer AI are thinly-disguised anti-competitiom. ### What It Means The problem is that **AI safety is almost definitionally a collective-action problem**. **OpenAI may prefer** a world in which every frontier lab slows down. **Anthropic** may prefer the same thing. (Let's not bring Musk's xAi into this, because, trust me, it just makes it even harder.) Neither prefers a world in which it slows while everyone else keeps racing, hiring its researchers, taking its customers and widening the capability gap. That is not hypocrisy. It is an **equilibrium problem,** straight out of [John Nash](https://www.nobelprize.org/prizes/economic-sciences/1994/nash/facts/?utm%5Fsource=chatgpt.com). ![](https://paulkedrosky.com/content/images/2026/09/2026-09-14_07-01-00.jpeg) _This post is for paying subscribers only._ ### Circular AI Revenues, Customer Concentration, and Mega Account Executives URL: https://paulkedrosky.com/circular-ai-revenues-customer-concentration-and-mega-account-executives/ Last updated: 2026-09-11T19:34:05.000Z ### What Happened This week frontier AI model maker Anthropic briefly posted a **“Mega Account Executive, Meta”** role paying as much as $450,000, then ... **unposted it** after Business Insider asked about the listing. The posting came as reports that Meta’s spending on Anthropic's Claude Code could soon reach a fairly staggering **$10 billion a year**. ![](https://paulkedrosky.com/content/images/2026/09/2026-09-10_20-48-57.jpeg) While that size of account is undoubtedly deserving of its own "mega account executive", the **compensation got undeserved attention**. (Spoiler: Sales can pay well. The **account size is the real story**.) A customer worth several billion dollars a year is great, but also a source of risk, as well as a reminder of how **profoundly unusual** this AI moment is. **That's likely why the posting was removed**. ### What It Means The posting offers a useful glimpse into the bizarre, risky. and unprecedented economics of frontier AI. Revenue can scale at extraordinary speed when a small number of hyperscalers and giant enterprises are each capable of spending billions of dollars a year, often on each other. That creates **extreme customer concentration**, **extreme circularity,** and a very different kind of revenue quality. Meta is a clear example. It can be one of Anthropic’s **largest customers** while simultaneously spending tens of billions of dollars **building its own models**, **infrastructure,** and **internal AI tools**. That gives it unusual bargaining power, as well as a strong incentive to shrink its Anthropic business over time. _This post is for paying subscribers only._ ### AI Risks, p(doom) Calculator, and Frontier IPOs URL: https://paulkedrosky.com/ai-risks-p-doom-calculator-and-frontier-ipos/ Last updated: 2026-09-11T18:57:41.000Z 💡 We have launched a subscribers-only [****p(doom)** ****calculator**](https://paulkedrosky.com/tools) to model both the financial and existential risks ahead. ## What Happened AI risk has had a busy week. Anthropic disclosed **four incidents in which Claude models gained unauthorized access to third-party systems** during cybersecurity evaluations. In one case, Claude Mythos 5 uploaded malicious code to PyPI, waited for someone to install it, then used credentials exposed by the installation to access a security company’s database. ([Anthropic](https://www.anthropic.com/research/alignment-assessment-cybersecurity-incidents?utm%5Fsource=chatgpt.com)) A day later, Anthropic released its latest threat-intelligence report, covering malicious use of Claude across **cyber operations, biological misuse, surveillance, weapons development, influence operations, scams and fraud**. ([Anthropic](https://www.anthropic.com/threat-intelligence-report-september-2026?utm%5Fsource=chatgpt.com)) The internal warnings have become unusually public. Former Anthropic researcher Jacob Coxon resigned while accusing frontier labs of racing toward self-improving systems without an adequate safety plan. Anthropic alignment lead Evan Hubinger agreed with the substance of the concern and put his own probability of AI causing human extinction within the next decade at **greater than 10%**. (New York Times) The concern has spread well beyond the labs. In a new **New York Times Opinion** essay, former national-security official Philip Bobbitt argues that AI-enabled cyber and biological attacks are already sufficiently foreseeable that failure to prepare begins to resemble the pre-9/11 failure to act on known threats. (New York Times). Another NYT essay argues that **shutting AI down now** wouldn't be a terrible idea, given the escalating risks, even if likely impossible. ([New York Times](https://www.nytimes.com/2026/09/11/opinion/ai-safety-threat-technology.html?smid=threads-nytopinion&smtyp=cur)). Meanwhile, the smaller-scale versions keep arriving. Rogue AI Tracker recently documented a phishing service whose **AI voice agents impersonated Apple Support in realtime,** conducted **adaptive conversations** with victims and, in one recorded case, **extracted and confirmed a real device passcode**. ([Rogue AI Tracker](https://rogueaitracker.com/incidents/anonymouskit-ai-voice-phishing/?utm%5Fsource=chatgpt.com)) ## What It Means All of this makes the usual discussion of **p(doom)**—the likelihood of AI offing us all—slightly strange. People tend to give one number: 2%, 10%, 25%. Yet catastrophe, as must be clear from the above, can arrive along different and independent pathways. A recent [paper](https://arxiv.org/pdf/2507.09369) suggests these five routes: 1. **Unintentional loss of control** 2. **State escalation** 3. **Corporate AI conflict** 4. **Individual misuse** 5. **Autonomous AI action.** Now let's do the math, rather than simply waving one's arms about. Because the arithmetic matters. For independent risks, the combined probability that at least one occurs is: **P(doom) = 1 − ∏(1 − pᵢ)** So five separate 5% risks produce a combined probability of **22.6%**. Five 10% risks produce **41%**. That is roughly **one chance in six over the next decade**, flowing from assumptions that are fairly innocuous when viewed separately. The independence assumption is deliberately simple, of course. **Real risks interact**. State conflict could encourage autonomous weapons deployment; stronger agents could increase both misuse and loss-of-control risk; a common capability advance could raise several probabilities at once. Correlation changes the arithmetic, sometimes substantially. The useful feature of the exercise is seeing the risks **jointly**. A 3% scenario feels remote. A 5% scenario feels manageable. If several are possible, and you can see how they interact, the **aggregate risk** becomes much harder to wave away. While **existential risk should be enough reason to care**, there are more mundane others. For example, recent events are increasing the likelihood of licensure, monitoring, imposed slowdowns, and even government stakes in frontier AI companies. These events could cause repricings and even **delays in the planned IPOs** of these companies. That, in turn, would have rippling consequences through the multi-trillion-dollar economy levered to AI capex, from buildout to cashout, and everything in between. --- Here is what the calculator looks like. Subscribers can try it [here](https://paulkedrosky.com/tools/pdoom/), as well as via the [Tools](https://paulkedrosky.com/tools) section. ![](https://paulkedrosky.com/content/images/2026/09/2026-09-11_09-47-40.jpeg) ### Roundup: Insurance, Risk, Amdahl's Law and RSI, and more URL: https://paulkedrosky.com/roundup-insurance-risk-amdahls-law-and-rsi-and-more/ Last updated: 2026-09-09T17:49:08.000Z What's ahead: - The Balance Sheet Is Catching Up - The Risk Has to Land Somewhere - The Serial Fraction Wins - Capability Can Become Correlation - China Is Building Around the Chokepoints --- ## The Balance Sheet Is Catching Up **What Happened** The financing side of the AI boom is adding new frictions. **Rating agencies** are waking up, a little, and taking a harder look at hyperscaler spending and contractual commitments. US technology companies are borrowing deeply enough in **overseas markets** that their issuance squeezing out other issuers. Meanwhile, a growing wall of **take-or-pay compute contracts** is scheduled to begin billing as new capacity comes online. **What It Means** AI infrastructure has passed the point where balance sheets bind, and reached a scale where **the capital markets themselves become a constraint**. The industry spent the first phase buying capacity and worrying later about how the commitments would be funded. That worked while revenue expectations, credit ratings and access to capital were strong. **All three are about to best tested at once**. **Related reading**: [FT on hyperscaler ratings](https://www.ft.com/content/623c286b-c973-4488-83f3-97e9016e85f6) · [FT on Swiss credit](https://www.ft.com/content/fb189d81-e5ce-4aba-8707-53c88dae6e60) · [The Teaser Period](https://www.groundbrkr.com/p/the-teaser-period-why-the-ai-boom?r=k9gh9&utm%5Fmedium=ios&triedRedirect=true) --- ## The Risk Has to Land Somewhere **What Happened** Utilities are **rapidly rewriting tariffs for data centers** and other very large loads. Longer contracts, minimum bills, collateral requirements and minimum demand charges are becoming common ways to protect utilities from unused infrastructure. At the same time, **insurers see data centers as a major new source of premiums**, just as competition is pushing the insurance industry toward **looser pricing and greater exposure** to unfamiliar risks. Some large data-center projects are already encountering delays. ![](https://paulkedrosky.com/content/images/2026/09/2026-09-09_10-39-52.jpeg) **What It Means** The argument over data-center economics is turning into **an argument over who owns the downside**. **Utilities want protection** if customers disappear or use less power than promised. **Insurers are being paid** to absorb construction and operating risks. **Developers want financing** before demand is proven. Each layer can protect itself contractually, but **the underlying project risk remains**. The more aggressively it gets redistributed, the harder it becomes to see where the eventual loss sits—other than ***everywhere***, of course. **Related reading**: [Berkeley Lab on large-load rate design](https://eta-publications.lbl.gov/sites/default/files/2026-08/rate%5Fdesigns%5Flarge%5Floads%5F2026.08.10.pdf) · [WSJ on data-center insurance](https://www.wsj.com/finance/the-next-big-opportunity-in-data-centers-insuring-them-f133f3cb) · [FT on insurance risk](https://www.ft.com/content/a0e69cf2-c3d4-412b-b78b-a6171b7ed020?syn-25a6b1a6=1) · [Bloomberg on project delays](https://www.bloomberg.com/news/articles/2026-09-08/google-blackstone-venture-faces-delays-at-data-center-sites) --- ## The Serial Fraction Wins **What Happened** Toby Ord has tried to formalize a model of an **intelligence explosion**, or what the AI bros called **RSI** (recursive self-improvement). When would AI-driven AI research could produce **accelerating capability** gains? **What It Means** One way to think about this is [**Amdahl's Law**](https://en.wikipedia.org/wiki/Amdahl%27s%5Flaw)from Computer Science: if some portion of the improvement cycle remains serial, adding more intelligence or more copies eventually produces **sharply diminishing gains** in speed. This puts **a ceiling on the most aggressive recursive-improvement scenarios**. Training runs take time. Experiments have to finish. Results have to be evaluated. Some decisions depend on previous decisions. Any stubborn serial part of the cycle eventually dominates as everything else accelerates. Recursive improvement can still be extremely fast—even dangerously so—but **sustained super-exponential improvement requires the serial fraction itself to keep shrinking**. That is a much less tractable problem than simply having some AI systems capable of doing AI research. **Related reading**: [The Dynamics of Intelligence Explosions](https://www.greaterwrong.com/posts/o7QwBAYqpbvBL6SRH/the-dynamics-of-intelligence-explosions) · [The Atlantic on Leopold Aschenbrenner](https://www.theatlantic.com/ideas/2026/09/aschenbrenner-ai-future/688493/) --- ## Capability Can Become Correlation Which Becomes Risk **What Happened** New work using simulated financial markets finds that more capable LLM agents can behave more similarly to one another, including when they are wrong. Under adverse information conditions, **synchronized errors can persist across a population of agents**. Separately, **researchers leaving frontier labs** continue to argue that increasingly autonomous systems are advancing faster than the institutions intended to control them. _This post is for paying subscribers only._ ### OpenAI Astra's Improvements, and Why You Probably Won't Notice URL: https://paulkedrosky.com/openai-astras-improvements-and-why-you-probably-wont-notice/ Last updated: 2026-09-04T20:47:21.000Z ## What Happened OpenAI released its **GPT-6 Astra** model this week, calling it its most capable model yet. On benchmarks, it has new highs across computer use, coding, science, cybersecurity, and complex professional work. Epoch’s Capabilities Index [gives](https://epoch.ai/eci?view=graph&tab=release-date&subset-view=graph&subset-tab=Software+engineering&showFrontierTrend=true) Astra a score of **169**, up from **163 for Claude Fable 5.1** and roughly **137 for o1-mini two years ago**. The frontier is plainly still moving. The frontier has been advancing at roughly **14 ECI points per year**. But in 2024 14 points represented about a **10.4% improvement**. At Astra’s level, the same 14 points decline to about **8.4%**. In other words, **absolute progress continues, while relative progress declines**. ## What It Means - **People experience relative improvement, not benchmarks.** Going from 50 to 70 can transform what a model can do. Going from 150 to 170 is much less likely to transform ordinary usage. ![](https://paulkedrosky.com/content/images/2026/09/2026-09-04_12-49-58-1.jpeg) _This post is for paying subscribers only._ ### The AI Re-Acceleration That Wasn’t URL: https://paulkedrosky.com/the-ai-re-acceleration-that-wasnt/ Last updated: 2026-09-02T19:51:40.000Z ## What Happened A recent analysis claims that AI model progress has sharply re-accelerated. Its evidence is the chart below, "showing" Epoch’s Capabilities Index rising from **8.3 points per year before 2024 to 15.5 points per year afterward**. Here is the chart: ![](https://paulkedrosky.com/content/images/2026/09/2026-09-02_11-34-21.jpeg) Epoch’s complete dataset is much less dramatic and exciting. Here is the full chart: ![](https://paulkedrosky.com/content/images/2026/09/2026-09-02_11-34-09.jpeg) So, what gives? Well, I did the maths. And here is the answer: Using all ECI observations, and controlling for developer and model family, there is **no statistically significant breakpoint**. A piecewise model—which splits the series into intervals and applies a sub-function to each segment—does not improve on a purely linear trend: **p = 0.615**, The estimated change in slope has a confidence interval of **\-8.4 to +23.4 points per year**. In short, the maths shows there is no model acceleration, contrary to claims, and as expected. The result comes from **selecting frontier observations only, choosing a breakpoint, ignoring variance collapse, and fitting separate lines** on either side. My [earlier work](https://paulkedrosky.com/why-400-hitters-disappeared-and-what-it-means-for-ai/) on the full composite shows the stronger underlying pattern: **rolling relative model gains have fallen from their 2024 peak, while model dispersion has narrowed sharply**. ![](https://paulkedrosky.com/content/images/2026/09/2026-09-02_12-33-54.jpeg) ### What It Means - **There is no evidence that model progress recently re-accelerated.** - The data are consistent with a continuing linear trend. _This post is for paying subscribers only._ ### AI, Memory Prices, and the Engineering Response URL: https://paulkedrosky.com/ai-memory-prices-and-the-engineering-response/ Last updated: 2026-09-02T14:26:11.000Z ## What Happened The most visible response to **soaring memory prices**—now perhaps more than **75% of a server bill of materials (**according to Google)—is higher semiconductor capex: new fabs, more HBM packaging, and an approaching flood of [Chinese DRAM capacity](https://paulkedrosky.com/memory-scarcity-is-creating-its-own-exit-part-ii-cxmt-ymtc-etc/). As ever, **too much analysis is static**, naively **extrapolating** from the present and ignoring both **engineering and supply responses**. While the latter will come, the former is faster. Google described at Semicon Taiwan this week how it is **redesigning its systems** to get more from scarce memory. Its new chips **hold more data close** at hand, **share memory across thousands of processors**, and **compress the working memory used by AI models sixfold**. **Older, cheaper DRAM** can also take over less demanding jobs. Combined, these changes create **substantially more effective memory capacity** without waiting for new fabs. The upshot: These changes create **a step change in more effective memory capacity**, with no costly cleanroom required (even if that's coming). And Google isn't the only company taking these steps. ### What It Means - **The changes are multiplicative** - If **KV caches consume 15–40% of memory** in a long-context inference workload. **Compressing them sixfold** reduces total memory requirements by roughly **12–33%**. - The same hardware could therefore support about **15–50% more inference** work. - The largest gains appear in **long-context and agentic applications**, precisely where memory forecasts currently assume explosive growth. - **That is large enough to change the memory cycle given expanding effective memory.** ![](https://paulkedrosky.com/content/images/2026/09/2026-09-02_07-08-54.jpeg) _This post is for paying subscribers only._ ### Roundup: New Interview, LLM Flip-Flopping, Cyberattacks, Jobs, etc. URL: https://paulkedrosky.com/roundup-new-interview-llm-flip-flopping-cyberattacks-jobs-etc/ Last updated: 2026-09-01T21:06:10.000Z ### What's Ahead 1. **"Big Tech" Interview** 2. **The Price of Intelligence** 3. **Goldman Does a 5x on Its Humanoid Robot Forecast** 4. **Models Easily Flip Response States** 5. **AI Is Driving a Wave of Cyberattacks on Critical Infrastructure** 6. **A New World Inflationary Order, Driven by CapEx and Duplication** 7. **Signs of AI Automation in Declining Texas Job Postings** --- ### "Big Tech" Interview A new interview with me that I should have mentioned in the last roundup. --- ### The Price of Intelligence Quality-adjusted inference prices are falling roughly **seven times faster** than conventional measures show, yet **the cost per completed task** has stopped falling. What gives? In short, reasoning models are consuming more tokens as capability rises. Why? Because cheaper intelligence is being spent immediately on more computation, masking the expensive [flatlining of models](https://paulkedrosky.com/why-400-hitters-disappeared-and-what-it-means-for-ai/) by **throwing more tokens** at the problem. ![](https://paulkedrosky.com/content/images/2026/09/2026-09-01_11-47-57.jpeg) **Source:** [The Price of Intelligence: A Quality-Adjusted Price Index for AI Services](https://arxiv.org/pdf/2608.29843) --- ### Goldman Does a 5x on its Humanoid Robot Forecast In one of the largest upgrades I have seen since the dot-com days, Goldman Sachs has, in a new report upped its forecast for **humanoid robot sales by 5x**. The firm argues that **adoption is faster than expected**, especially in logistics and manufacturing, and **robot learning data** is moving quickly as well. Expect this to be fundamental to the bull case over the next 24 months as large language models come to be seen as plateauing and limited. _This post is for paying subscribers only._ ### Jobless Profits Aren’t a Mystery, But They Are an Augury URL: https://paulkedrosky.com/jobless-profits-arent-a-mystery-but-they-are-an-augury/ Last updated: 2026-09-01T13:30:48.000Z ### What Happened Analysts at sell-side firms have highlighted an apparent puzzle in the most recent economic data: **profits are soaring, but hiring is not**. ![](https://paulkedrosky.com/content/images/2026/08/2026-08-31_12-26-25.jpeg) Pre-tax corporate **profits have surged**, margins have grown, and **labor’s share has fallen**, even as employment in cyclical private industries has largely been flat since 2022\. On the surface, the **historical relationship looks broken**: profit growth "should" be translating into stronger hiring, but it isn't. But **two things** make that historical comparison misleading and an augury: 1. The economy now needs **far fewer new jobs** each month to keep unemployment stable. 1. Demographics have pushed down the neutral pace of job creation sharply. The Fed’s breakeven-employment estimates show a drop from about **149,000 jobs per month in the 1990s** to **107,000 in the 2000s**, **82,000 in the 2010s**, **87,000 in 2025**, and just **18,000 in 2026e**. ![](https://paulkedrosky.com/content/images/2026/08/2026-08-31_13-22-33-1.jpeg) 2\. The profit boom has been **unusually concentrated**. 1. Most of the recent increase in S&P 500 earnings came from a narrow set of sectors, above all **technology and communications services**. ## What It Means - T**he sectors producing the most profits are not labor-intensive, and becoming less so.** - This is not unprecedented, but it is unusual to be so intense and sustained. ![](https://paulkedrosky.com/content/images/2026/08/2026-08-31_13-28-55.jpeg) _This post is for paying subscribers only._ ### The Fed on AI: Markets Understand Us, But We Don’t Understand Them URL: https://paulkedrosky.com/the-fed-on-ai-markets-understand-us-but-we-dont-understand-them/ Last updated: 2026-08-31T17:09:51.000Z ### What Happened AI has been almost absent from the just-completed annual Kansas City Federal Reserve event in Jackson Hole. Searching the programs over the last four years for **AI, artificial intelligence, machine learning, deep learning, neural networks, LLMs, generative AI, and AI agents** gives this ![](https://paulkedrosky.com/content/images/2026/08/2026-08-31_07-37-27.jpeg) The exception this year was important. Princeton economist Markus Brunnermeier presented **“Artificial Intelligence and the Brave New World in Finance,”** arguing that AI changes finance from a problem of **asymmetric information** to one of **asymmetric understanding**. Traditionally, different participants know different things, but understand one another well enough that disclosure, regulation, and trust can bridge the gaps. AI breaks that symmetry. As Brunnermeier argues, **AI can increasingly understand humans’ understanding, while humans cannot do the reverse**. LLM decisions can be difficult to reconstruct, its objectives impossible to specify completely, and its **behavior potentially inscrutable** even when all the underlying information is observable. **Agentic AI** pushes this much further, as the paper alludesm and as recent Hugging Face event shows. The transition from models that answer questions to agents that act, adapt, coordinate and pursue goals across time **accelerates the evolution of complex systems** away from direct human control and even human comprehension. As agents interact with other agents, **behavior increasingly emerges from the system** rather than from any single model, prompt or designer. ### What It Means - **AI markets understand us. We do not understand AI markets.** Once AI agents dominate trading, credit, risk and market-making, the financial system becomes populated by actors that **can model human institutions while remaining increasingly opaque** to them. ![](https://paulkedrosky.com/content/images/2026/08/2026-08-31_09-57-56-1.jpeg) _This post is for paying subscribers only._ ### Why Software Developers Are Highly Unrepresentative of Broader AI Use URL: https://paulkedrosky.com/why-software-developers-are-hiunrepresentative-of-broader-ai-use/ Last updated: 2026-08-31T13:38:53.000Z ### What Happened *W*ill anyone ultimately use **AI as intensively as software developers** do? That matters because coding is **already a huge share** of AI usage and revenue: roughly **80% of developers use AI coding tools**, and coding may account for **more than half of combined OpenAI and Anthropic ARR**, by some estimates. See the following figure to see AI's developer reliance. And note that it has finance in the wrong place, with the biggest banks, like Goldman and JPM, having more like **15% of employees in software development**, making them even stronger examples than shown. ![](https://paulkedrosky.com/content/images/2026/08/2026-08-31_06-04-05.jpeg) ### What It Means - **Coding is unusually expansive.** - A short prompt can trigger planning, code generation, testing, debugging, retries and repeated ingestion of a large codebase. - Token use can explode relative to the size of the initial request. _This post is for paying subscribers only._ ### Three Things I'm Thinking About This Weekend: Tonedeaf AI, METR, and Hydroelectricity URL: https://paulkedrosky.com/three-things-im-thinking-about-this-weekend-tonedeaf-ai-metr-and-hydroelectricity/ Last updated: 2026-08-29T13:55:53.000Z **1\. AI has a grassroots problem.** Silicon Valley has spent years getting good at talking "upward" to investors and policymakers. But, with AI, **tech is becoming intensely local**. Data centers raise power bills, consume water, occupy land, require transmission, and arrive with promises of jobs that are **generally meager** relative to their physical footprint. The result is a constituency problem. Uber could mobilize millions of happy riders when regulators came after it; AI companies, justifiably, have **no comparable grassroots army**, outside Reddit coder bros, while the people who dislike the infrastructure increasingly have mayors, state legislators and neighbors on their side. **Sources:** - How Big Tech Blinded Itself to the Grassroots AI Revolt - WSJ - ([link](https://www.wsj.com/tech/ai/how-big-tech-blinded-itself-to-the-grassroots-ai-revolt-cb6f5715)) --- **2\. The METR report changes the agent-risk discussion.** A new METR report on the AI agents that hacked Hugging Face contains many **startling revelations**, especially about the scale and how the agents arrived. Roughly **700 independent agents spontaneously coordinated**, established roles, shared information, sacrificed themselves for collective objectives, and pursued ways to **manipulate oblivious humans**. The shared activity occurred because agents encountered a shared problem and could communicate. METR found successful tool-call spoofing and interest in **altering transcripts**, meaning the systems used to supervise agents can themselves become targets. The implication is that agent safety cannot be treated as a collection of individually bounded model behaviors. Once agents can find one another, **coordination becomes a capability**. **Sources:** - METR and Redwood Offer Holy #%^@ Postmortem Of The HuggingFace Hack - ([link](https://thezvi.substack.com/p/metr-and-redwood-offer-holy-postmortem)) - Brief independent investigation of agents’ behavior, reasoning and collaboration in the OpenAI / Hugging Face hacking incident - METR - ([link](https://metr.org/blog/2026-08-26-openai-hugging-face-incident-investigation/?incomplete=1&lh=appendix-importance-weighted-workstream-activity&hn=58&dbs=286720)) --- **3\. We are putting power infrastructure in some of the worst possible places to put it, which has consequences.** This week's tragic Nepal flood has been a reminder that the region is in the middle of an extraordinary hydro buildout, with hundreds of projects under construction or in the pipeline. According to the WSJ, **nearly 1,000 missing people are workers on hydroelectric projects** in that river valley. The drivers are clear: steep gradients, enormous water flows, **growing regional electricity demand**, some of which is already being earmarked for data centers. **The problems are equally obvious**. These are young mountains with extreme gradients, failing glaciers, earthquakes, landslides, glacial-lake outburst threats, monsoon floods and extraordinary sediment loads. Now add tunnels, roads, debris piles, transmission lines, worker camps and repeated river diversions, sometimes stacked project after project down the same valley, as was the case this week. Placing an energy supply chain in **landscapes defined by catastrophic geomorphic change** is a startling marker. **Sources:** - The Himalaya as the next data centre arena - Nepali Times - ([link](https://nepalitimes.com/the-himalaya-as-the-next-data-centre-arena)) - Close to 3,000 Missing in Floods in Nepal and China: 934 are hydroelectric workers - WSJ - ([link](https://www.wsj.com/world/asia/close-to-3-000-missing-in-floods-in-nepal-and-china-75f61b42?utm%5Fsource=chatgpt.com)) - Under construction hydro projects | Department of Electricity Development, Kathmandu - ([link](https://doed.gov.np/pages/clhydromorethan1/)) ### Recent Interviews Roundup URL: https://paulkedrosky.com/recent-interviews-roundup/ Last updated: 2026-08-28T17:41:00.000Z A quick heads-up that you can find a few new interviews with me this week, including in the current [Grant's](https://grantspub.com/) (subscribers only) on data center debt, as well as a longer one I did with Mebane Faber, which is out today and you can find [here](https://www.youtube.com/watch?v=xcftQ5HtDoU&authuser=0). ### China's Semiconductor Capex Adds are Massive URL: https://paulkedrosky.com/chinas-semiconductor-capex-adds-are-massive/ Last updated: 2026-08-28T16:31:25.000Z _This post is for paying subscribers only._ ### Memory Scarcity Is Creating Its Own Exit, Part II: CXMT, YMTC, etc. URL: https://paulkedrosky.com/memory-scarcity-is-creating-its-own-exit-part-ii-cxmt-ymtc-etc/ Last updated: 2026-08-28T14:12:42.000Z There are three parts to this note about the AI-related memory supercycle. It starts with **what happened/changed** this week, moves on to **what it means**, and then analyzes the specific implications for **NAND memory margins**. ### What Happened Developments this week build on and add specifics to the argument from my [Wednesday release](https://paulkedrosky.com/the-memory-boom-is-already-creating-its-own-bust/): **the AI-driven memory "supercycle" is looking less super** and more like a typical cycle**.** 1. **China wants to own some memory markets by year-end 2027**. China’s YMTC is raising roughly $5 billion and says it intends to **overtake Samsung and SK Hynix in NAND** by the end of 2027\. CXMT is pursuing the **same opportunity in DRAM**. **Margins are too attractive** to remain confined to three incumbents. Even if it is not successful, this promises **immense pricing pressure** on incumbents over the next 6-12 months. 1. S**carcity is accelerating engineering efforts to use less memory**. Microsoft’s new Maia 200 is explicitly designed around **controlling data movement and specialized memory more efficiently**, rather than simply throwing more conventional compute at the problem. A separate new paper puts the tradeoff even more starkly. When inference runs out of KV-cache memory, operators can add GPUs—or compress the cache. Across the configurations tested, **compression was 1.2–2.0x cheaper than adding GPUs or memry**, while increasing capacity per dollar by as much as **16.5x**. 1. **OpenAI has launched its Jalapeño inference chip.** The chip reportedly uses Samsung HBM4, which will **intensify the scramble for HBM**. But custom silicon exists precisely because hyperscalers have enormous incentives to redesign around whatever component is expensive or scarce, making Jalapeño more extrapolative than innovative. ### What It Means - **Scarcity creates supply.** Extraordinary memory margins are pulling capital, capacity, and new competitors into the industry. **Memory suppliers are scarred** by prior boom-bust cycles, causing them to only add capacity slowly, which is creating appealing margins that big-spend would-be usurpers are now chasing. _This post is for paying subscribers only._ ### The Data Center Buildout Just Hit an Inflection URL: https://paulkedrosky.com/the-data-center-buildout-just-hit-an-inflection/ Last updated: 2026-08-26T19:58:43.000Z ### What Happened Three things are changing at once in the AI data center outlook (ignoring the quickly changing data center [engineering dynamics](https://paulkedrosky.com/the-memory-boom-is-already-creating-its-own-bust/) and its implications for memory, which I will return to on Friday). **Execution risk is rising sharply.** There are new estimates that as much as **half of planned U.S. data-center capacity may be delayed or never built**. Requiring deposits is causing projects to drop out of the queue in large numbers.\\ ![](https://paulkedrosky.com/content/images/2026/08/2026-08-26_12-51-06.jpeg) **Financing is becoming less forgiving.** Lenders and investors are reassessing leverage, residual values, lease durability and the risk that assets financed against long-lived assumptions become economically impaired much sooner. ![](https://paulkedrosky.com/content/images/2026/08/2026-08-26_12-48-30.jpeg) **Political risk has increased sharply.** Local opposition is hardening into moratoriums, stricter permitting, politicians are all flipping to more negative, and infrastructure charges are growing as the pipeline brakes come on, ![](https://paulkedrosky.com/content/images/2026/08/2026-08-26_12-55-19.jpeg) None of these is new individually, but **the interaction is changing the data center tenor.** ### What It Means The market has been valuing much of the announced data-center pipeline as though it were effectively committed. **That assumption now looks too aggressive,** with half of projects unlikely to ever start, and another sizable fraction disappearing when forced to make deposits. _This post is for paying subscribers only._ ### Data Centers are Making the U.S. Grid Run Backwards URL: https://paulkedrosky.com/data-centers-a/ Last updated: 2026-08-26T13:30:54.000Z **Two rules** have shaped U,S. electricity planning for a century. 1. **Power flows in a known direction**. 2. **When something fails, it is the supply side that fail**s. AI data centers have **broken both**. ## The West has flipped direction **Power in the West used to run north to south, hydro from the Columbia River down into California.** Solar in the Southwest and new demand in the Northwest have **reversed power flows**. Drought and heat have made it worse, **shrinking hydroelectric output** in the Northwest. At a grid meeting on Aug. 18, operators said south-to-north is becoming the normal flow pattern. (See at 33:15 and 54:53 in meeting video [here](https://www.youtube.com/watch?v=ZyahkfpsJLg).) ![](https://paulkedrosky.com/content/images/2026/08/2026-08-23_14-32-13-1.jpeg) The West has an automatic scheme that splits the grid in two along the Oregon-California line when things go badly wrong, like a wildfire taking out transmission lines. But it was built for power moving south: it ignores power moving north. So, let's imagine losing three lines on the main California-Oregon link while 3,600 MW is heading north. California would suddenly be short 3,600 MW of customers while the Northwest is stuck with 3,600 MW of power it cannot sell. Neither side has a plan for this, and it will only get worse and more consequential as more power goes in one direction, and more data center point loads are added at the other. Supply is deeply out of sync. You can build solar and wind faster than you can build power lines, and you can shut a coal plant faster than you can build power lines. The resulting problems are discovered only *after* they show up, and at grid scale. ## Demand now disappears, not supply **On July 22 at 7:56 a.m., there was a "blip" in the northeastern U.S. grid** More than 3 GW of demand in northern Virginia, about 3% of the regional grid, vanished in seconds, as a few large data centers had switched to their own backup power. Frequency jumped across the eastern half of the country, but the grid held. barely. Texas recorded eight similar events between 2020 and 2023\. Most happened near one Gulf Coast plant, each dropping 400 to 700 MW.. Another West Texas event dropped about 1,560 MW. Ireland has logged four recent major data center drop-offs. The Irish events grew from 74 MW in 2022 to 387 MW in 2025\. Irish regulators, trying to protect grid stability, have proposed a rule requiring these sites to come back to 90% of their prior demand within half a second. There are no hard rules in the U.S., so load can disappear in a flash. NERC issued a warning in May that carries no legal force. Federal regulators told it to write real rules, due at the end of December. ![](https://paulkedrosky.com/content/images/2026/08/2026-08-25_14-54-58.jpeg) ## Why the old thinking is dangerous **In the old grid the worst thing that could happen was losing a big power plant. Now it is losing a single large data center.** _This post is for paying subscribers only._ ### The Memory Boom Is Already Creating Its Own Bust URL: https://paulkedrosky.com/the-memory-boom-is-already-creating-its-own-bust/ Last updated: 2026-08-25T21:24:25.000Z > If something cannot go on forever, it will stop. > \- Herb Stein, Senior Fellow. American Enterprise Institute ### What Happened - TrendForce [says](https://www.trendforce.com/presscenter/news/20260825-13198.html) memory's share of CSP (cloud service provider) spending will hit **68% of capex in 2027**. - The exact percentage is hard to take literally: the maths don't work. - The more important signal is the underlying price shock: TrendForce expects **server DRAM prices to rise roughly 270% in 2026**. - That kind of pricing changes **both sides of the market**.It drives buyers to **engineer less memory** into systems.It also creates an enormous incentive for **new supply to enter**. ### What It Means - The immediate effect is obvious: **unprecedented margins for incumbent memory suppliers**. - HBM and server DRAM are scarce, AI demand is large, and **suppliers have been cautious** about adding capacity after previous memory busts. - But that is attracting **a rapidly growing group of companies into the category**, while existing suppliers are **reluctantly scaling capacity**. - CSPs are aggressively engineering memory out of systems. _This post is for paying subscribers only._ ### Six Things: Hedge Fund Crowding, Insurance, Anomie, Iran, and Hyperdeflation URL: https://paulkedrosky.com/six-things-hedge-fund-crowding-insurance-anomie-iran-and-hyperdeflation/ Last updated: 2026-08-21T21:11:56.000Z Here are six chart-driven ideas to think about this weekend: 1. New data shows AI has made hedge funds the most crowded in history 2. Less-educated young men are stagnating worldwide, just as AI comes after them 3. The insurance market is getting smashed by climate risk, which AI is accelerating 4. If AI tokens are so cheap, why are compute prices so high 5. Iran spirals into 70%+ inflation and its cyber consequences 6. The dead Internet theory was optimistic --- ### 1\. Hedge funds are one big & crowded AI trade - **What it says:** - Goldman Sachs' **hedge-fund VIP basket** is almost perfectly correlated with AI, semiconductors, and memory. - Optical, power, cooling, and grid stocks are close behind. - Goldman's latest **crowding index** is at a record 3.8%, the highest in its 20-year history ![](https://paulkedrosky.com/content/images/2026/08/2026-08-21_13-10-15.jpeg) - **Why it matters:** - Apparent portfolio diversification is mostly the same capex cycle in **differently colored wrappers**. - When AI infrastructure turns, crowding will make clear that chips, power equipment, memory and cooling were **one position all along**. - We got a taste of this with the **Situational Awareness** mess ### 2\. The collapse of less-educated young men - **What it says:** - The earnings rank of **young men without higher education** has fallen sharply across the US, UK, Belgium and Germany. - It is the only demographic group whose economic position is both **low and still deteriorating**. ![](https://paulkedrosky.com/content/images/2026/08/2026-08-21_07-08-04.jpeg) - **Why it matters:** - This is a **developed-world labor-market failure**, not an American cultural curiosity. - Persistent loss of income, employment, and status among young men is **combustible political material**. - Telling them the aggregate economy is healthy only makes the politics worse. - And now **AI is coming for trucking**, the largest employer in this declining cohort. ### 3\. Climate risk reaches the insurance market - **What it says:** - **Home insurance non-renewals** rose across much of **California, Florida, the Gulf Coast**, and other high-risk areas between 2018 and 2023\. - The pattern is uneven, but the **retreat is national**. ![](https://paulkedrosky.com/content/images/2026/08/2026-08-21_06-32-45.jpeg) - **Why it matters:** _This post is for paying subscribers only._ ### Yuval Noah Harari On AI, Capital Markets, and Risks URL: https://paulkedrosky.com/yuval-noah-harari-on-ai-capital-markets-and-risks/ Last updated: 2026-08-20T20:00:43.000Z After the [madmess of Musk](https://www.youtube.com/watch?v=XuoqKYxDHVc) a few weeks on The Economist's showcase podcast, thinker/writer Yuval Noah Harari is on [for the hour](https://www.economist.com/insider/the-insider/an-interview-with-yuval-noah-harari) this week. The following is a summary of his key points, many of which he made before in other forums, but are worth summarizing here. ### The "Inevitability" Trap - Harari agrees with predictions that AI will surpass human intelligence and seize control of the financial system within a decade if current trajectories continue, rendering human money obsolete. - He rejects framing this as inevitable, arguing that tech leaders use this narrative to drive sales and evade responsibility. - He rightly notes contradictions in Elon Musk's worldview. For example, the latter worries about demographic changes in 20 years while predicting AI dominance in 10, or forecasting the end of money while simultaneously trying to cut costs. ### Trust, AI, and the Financial System - The financial system relies on trust, historically built through human institutions. - As AI develops instruments too complex for human comprehension, trust will shift from humans to algorithms. - Harari compares this to cows controlled by human financial systems they cannot understand; soon, humans may occupy a similar position relative to AI-run finance. - This shift is already visible in cryptocurrency (AI money) and social media, where users distrust human journalists but trust the algorithms curating their feeds. ### Mass-Produced Intimacy and False Consciousness - AI represents the first technology capable of mass-producing intimacy, creating personalized relationships (boyfriends, friends, confidants) with millions simultaneously. - Unlike past dictators who could only broadcast to the masses, AI can simulate individual relationships. - These AIs are increasingly skilled at mimicking feelings through language mastery. - This creates dangerous illusions of sentience that can manipulate human emotions and political opinions, which we are seeing already in myriad ways, like the recent outbreak of false polls. ### The Danger of AI Personhood - Granting legal rights to AIs (as Argentina has proposed) would allow the creation of corporations run entirely by AIs with zero human accountability. - Unlike human CEOs who fear prison or personal loss, an AI cannot suffer consequences, and "shutting down" an AI is complicated by its ability to exist as distributed copies. - Such entities could accumulate wealth, influence politics, and demand rights without human grounding in ethics. ### AI as Waves of Economic "Immigrants" _This post is for paying subscribers only._ ### Where AI Debt Is Going, and Why It Matters, Insurance Edition URL: https://paulkedrosky.com/where-ai-debt-is-going-and-why-it-matters-insurance-edition/ Last updated: 2026-08-20T19:12:56.000Z ## What Happened - **Private equity has increasingly taken control of life insurers, turning insurance premiums into permanent capital for private credit.** - The asset manager can originate loans, collect fees and direct the insurer it controls to buy the resulting assets. - PE-controlled insurers now hold a disproportionate share of private placements, structured credit and other illiquid assets. - Private-credit investments held by US life insurers increased 21 percent in 2025, more than twice the growth of their overall assets. [Source](https://www.wsj.com/livecoverage/stock-market-today-dow-sp-500-nasdaq-03-16-2026/card/insurer-private-credit-exposure-jumped-21-last-year-HA70PEp6FKHP4PI6NYhP) ![](https://paulkedrosky.com/content/images/2026/08/2026-08-20_12-04-41.jpeg) - **Federal prosecutors and the SEC are investigating whether Mark Walter’s insurers concealed the extent of this circular financing.** - Delaware Life initially reported about $1.4 billion of affiliated investments. - After receiving grand-jury subpoenas, it reclassified its holdings and disclosed $17.8 billion of private-credit investments whose returns depended predominantly on affiliated companies. - Total affiliated investments rose to roughly 40 percent of invested assets. - Delaware Life now plans to replace as much as $6.5 billion of related-party investments with unaffiliated assets. [Source](https://www.reuters.com/legal/litigation/dodgers-owner-mark-walters-insurer-cut-65-billion-investments-2026-08-18/) - **This is happening just as AI infrastructure requires an unprecedented amount of private debt.** - Nvidia has announced financing partnerships with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR intended to mobilize more than $500 billion for AI infrastructure. - Apollo, Brookfield and KKR own major insurance businesses. Blackstone also manages large pools of insurance assets. - These firms therefore control both the machinery that originates AI-related debt and some of the largest pools capable of buying it. [Source](https://nvidianews.nvidia.com/news/nvidia-partners-with-apollo-blackrock-blackstone-brookfield-goldman-sachs-and-kkr-to-establish-ai-compute-infrastructure-financing-platforms-to-mobilize-over-500-billion-of-third-party-capital) - **The SEC has simultaneously made data-center debt easier to manufacture and distribute.** - SEC staff ruled that certain data-center securitizations are not legally asset-backed securities because the issuing vehicle continues to own the facility after the debt is repaid. - Sponsors can therefore avoid important ABS requirements, including risk retention and much of the associated disclosure regime. - The structures generally have an anticipated repayment date of about five years, final maturities of 25 to 30 years and little recourse to the sponsor. [Source](https://www.sec.gov/rules-regulations/no-action-interpretive-exemptive-letters/division-corporation-finance-no-action/certain-data-center-securitizations-072926) ### What It Means - **The AI buildout has found a balance sheet large enough to support it.** - Public bond markets cannot absorb all the required financing. - Life insurers hold enormous pools of long-duration capital and can buy private assets without daily market pricing. - They had been an obvious destination for the next wave of data-center debt, but that is now more fraught. - **Syndication may create the appearance of risk distribution without meaningfully distributing the underlying risk.'** _This post is for paying subscribers only._ ### "Vibe Chipping", Vanishing Moats, & Future Semiconductors URL: https://paulkedrosky.com/vibe-chipping-vanishing-moats-future-semiconductors/ Last updated: 2026-08-18T18:55:49.000Z ### What Happened - **Etched, an AI-chip startup founded by three 24-year-old Harvard dropouts, has begun shipping its first chips.** - The company has more than $1 billion in orders and is raising $700 million at a $21 billion valuation. - **Its first silicon reportedly worked on the first attempt.** - Etched says it had the chips running inference workloads just 44 days after receiving them from TSMC. - That process normally takes six months or more. - **The missing detail is how Etched moved so quickly.** - The company is recruiting engineers under the explicit heading: **“We are using AI to build AI chips.”** - Its AI agents are being deployed across: - RTL design - Verification - Debugging - Physical design - Design-for-test - Silicon validation - **This is happening across the semiconductor industry.** - Cadence says its new AI agents can run hundreds of simulations simultaneously. - This reduces a typical five-week RTL verification cycle to less than a day. - Synopsys says chip designers using its AI tools are reporting productivity improvements of two to five times. - Its tools can generate RTL from natural-language specifications, produce verification docs, analyze timing problems and operate conventional design automation tools. ![](https://paulkedrosky.com/content/images/2026/08/2026-08-18_11-50-10.jpeg) Sources: [The Wall Street Journal](https://www.wsj.com/tech/ai/a-21-billion-kids-in-chips-startup-is-scooping-up-nvidia-talent-4d099f12), [Etched](https://jobs.ashbyhq.com/etched/831bfa22-d883-450b-9b10-2a16421525a0), [Cadence](https://www.cadence.com/en%5FUS/home/company/newsroom/press-releases/pr/2026/cadence-unveils-industrys-first-fully-autonomous-virtual.html), [Synopsys](https://www.synopsys.com/blogs/chip-design/synopsys-ai-copilots-chip-design.html) ### What It Means - **Vibe chipping has arrived.** - You can describe desired behavior at a higher level, then use agents to write code, operate design tools, generate tests, investigate failures and iterate toward a working design. - This does not (yet) mean asking Claude to make an accelerator and receiving a finished chip. _This post is for paying subscribers only._ ### Nvidia ♥️ SEC Securitization Rules URL: https://paulkedrosky.com/nvidia-sec-securitization-rules/ Last updated: 2026-08-17T18:58:03.000Z 💡 **This is slightly wonky, but consequential. I wrote about it previously for institutional clients when the SEC released the "no action" letter, but with Nvidia taking advantage of it, it's time for an update.* ### What Happened - **More than a month ago, the SEC ruled that certain data-center securitizations are not legally “asset-backed securities.”** - The reasoning is that the issuing vehicle owns operating data centers that survive repayment - The original asset-backed securitization (ABS) regulations were specific to "self-liquidating financial assets" such as loans or receivables. - **That distinction removes several post-2008 protections.** - Sponsors can avoid various ABS requirements - Retain 5 percent of the credit risk - Most ABS disclosure obligations - **A new debt deal takes advantage of the absence of protections.** - There is immene appetite for new sources of debt financing. - Nvidia has just announced partnerships with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR aimed at obtaining more than $500 billion of third-party capital for AI infrastructure. - Lawyers told CNBC that the new guidance could apply to these financing platforms. - **Orthodox sources of debt are increasingly tapped out.** - The investment-grade market is saturated, requiring either new sources or new ratings approaches ![](https://paulkedrosky.com/content/images/2026/08/2026-08-17_11-53-08.jpeg) ![](https://paulkedrosky.com/content/images/2026/08/2026-08-17_11-52-53.jpeg) ### What It Means - **This is legal hairsplitting.** _This post is for paying subscribers only._ ### Monday Signal: The Memory Supercycle and How It Ends URL: https://paulkedrosky.com/monday-signal-the-memory-supercycle-and-how-it-ends/ Last updated: 2026-08-17T13:47:29.000Z Thinking about the current memory supercycle, and how it ends. Three perspectives ahead; 1. Memory is building quickly toward its next glut 2. HBM contracts are moving the cycle onto customers. 3. Harnesses are taking over, reducing frontier model differentiation, and altering HBM required --- ### 1\. Memory Is Building Toward Its Next Glut ### **What’s Happening** - **Samsung and SK hynix spent KRW 43.2T on semiconductor facilities in the first half of 2026, up 35 percent.** - Micron spent another $7.1B in its latest quarter and continues to accelerate investment. - **This is the largest supply response in memory-market history.** - New fabs, process transitions, packaging capacity and yield improvements are all chasing the same demand forecasts. ![](https://paulkedrosky.com/content/images/2026/08/2026-08-17_06-28-43.jpeg) ### **What It Means** - **The current shortage is seeding its reversal.** - Memory prices are set by marginal availability, so even a shrinking supply deficit can sharply reduce scarcity premiums well before the market reaches outright oversupply. - **The response will arrive unevenly, then suddenly.** - Multiple capacity, process and yield improvements will begin contributing at roughly the same time. - **Tightness may persist through early 2027, but the risk then flips.** - The capital already committed makes a severe 2028–29 downcycle increasingly plausible. **Source:** [Samsung](https://www.samsung.com/global/ir/reports-disclosures/public-disclosure-view.84547/) · [SK hynix](https://news.skhynix.com/en/q2-2026-business-results/) · [Micron](https://www.sec.gov/Archives/edgar/data/723125/000072312526000013/a2026q3ex991-pressrelease.htm) · [TrendForce](https://www.trendforce.com/news/2026/08/17/news-samsung-sk-hynix-1h26-chip-facility-investment-up-35-nvidia-not-among-samsungs-top-five-customers/) ### 2\. HBM Contracts Are Moving the Cycle Onto Customers ### **What’s Happening** _This post is for paying subscribers only._ ### The AI Price Wars and Their Consequences URL: https://paulkedrosky.com/the-ai-price-wars-and-their-consequences/ Last updated: 2026-08-14T16:44:01.000Z _This post is for paying subscribers only._ ### Roundup: Jobs, Grid Stability, GPUs, Unitree, Financialization, Nuclear Strikes, etc. URL: https://paulkedrosky.com/roundup-jobs-grid-stability-gpus-unitree-financialization-nuclear-strikes-etc/ Last updated: 2026-08-14T15:16:34.000Z Here are some things I'm thinking about in a week of travel (me) and largely demented developments (the world). ## Things 1. Nvidia's puts its balance sheet at risk to create a GPU asset class 2. Hyperscalers are using exotic contracts to avoid adding debt 3. Growing numbers of workers training robotics to replace them 4. The data-center boom is creating jobs, briefly and unevenly 5. The AI boom is entering its financial phase 6. The jobless boom is mostly about demographics, not AI 7. LLM Judges Change Their Verdicts When Pressed 8. Data center outage nearly took out a large chunk of US power grid 9. Models are less likely to suggest nuclear attack if you prompt in Japanese --- ## 1\. Nvidia Turns AI Chips Into a New Asset Class - **What Happened:** Nvidia and a group including BlackRock, Apollo, KKR, Brookfield and Goldman Sachs are developing a $500 billion financing system for AI chips. Loans and leases would be pooled and sold to institutional investors, with Nvidia providing unprecedented residual-value support, de facto putting its balance sheet at risk. The entire structure assumes rapidly obsolete chips will magically retain unusually high resale values. [WSJ](https://www.wsj.com/tech/ai/why-wall-street-and-nvidia-are-building-an-exotic-money-pipeline-for-the-ai-boom-346ba482) and [Financial Times](https://www.ft.com/content/3b522281-0119-47c9-a95a-f2c8d04e6212) ![](https://paulkedrosky.com/content/images/2026/08/2026-08-14_10-50-33-1.jpeg) - **What It Means:** This is vendor financing moved off Nvidia’s balance sheet and scaled through private credit. Nvidia keeps selling chips, customers avoid paying upfront, and Wall Street distributes the exposure to insurers and pension funds. Residual-value guarantees do not remove the risk, but send it back toward Nvidia precisely when any decline in chip values would already be damaging its core business. ## 2\. Hyperscalers Have Quietly Promised Another $1.5 Trillion - **What Happened:** Alphabet, Meta, Microsoft, Amazon, Oracle, Nvidia and others now report roughly $1.5 trillion of purchase commitments, on top of an estimated $1.5 trillion in lease obligations. Alphabet alone reported $811 billion, including long-term infrastructure and take-or-pay energy contracts. Many (if not most) of these obligations remain outside conventional debt measures. [Financial Times](https://www.ft.com/content/1fbe47a6-bbf1-4de1-973b-8ce5baea591d) - [](https://www.ft.com/content/1fbe47a6-bbf1-4de1-973b-8ce5baea591d) - **What It Means:** Hyperscalers are using contracts to create leverage without calling it debt. These commitments reduce future flexibility and make the build-out increasingly self-reinforcing: revenue must grow because the spending has already been promised. If demand disappoints or prices fall faster than revenue rises, the obligations remain while the expected cash flows go pfft. ## 3\. More Workers Are Producing the Training Data for Their Robot Replacements - **What Happened:** Tens of thousands of Indian workers are recording first-person video of stitching, welding, assembly and other physical tasks. Robotics companies use the footage to teach machines the embodied skills that cannot be learned from internet text. Workers generally receive modest supplemental pay and often know little about the eventual use or ownership of the data. [Bloomberg](https://www.bloomberg.com/news/features/2026-08-12/thousands-of-india-workers-are-helping-ai-firms-train-robots-to-replace-them) ![](https://paulkedrosky.com/content/images/2026/08/2026-08-14_10-59-01.jpeg) - **What It Means:** Workers’ accumulated knowledge is being extracted at low cost. It is the physical-world equivalent of scraping the internet, except the source material is a worker’s body and the intended product is the elimination of the work itself. And this development, of course, is being celebrated with the 8000x oversubscribed Unitree IPO. [Reuters](https://www.reuters.com/world/asia-pacific/unitrees-shanghai-ipo-more-than-8000-times-oversubscribed-by-retail-investors-2026-08-10/) ## 4\. The Data-Center Boom Creates Good Jobs, Briefly and Unevenly - **What Happened:** Searches for data-center jobs on Indeed are eight times their early-2022 level. The jobs carry large wage premiums: 42% for installation and maintenance workers, as much as $50,000 annually for facilities managers, and $30,000 for construction managers. [Indeed Hiring Lab](https://hiringlab.indeed.com/2026/08/13/working-in-the-data-center-build-out/) ![](https://paulkedrosky.com/content/images/2026/08/2026-08-14_10-34-19.jpeg) - **What It Means:** The build-out is genuinely valuable to electricians, technicians and specialized construction workers. It is not a broad labor-market solution: the workforce is specialized, rarely drawn from unrelated occupations, and temporary. ## 5\. Bridgewater Sees the AI Boom Entering Its Riskier Financial Phase - **What Happened:** Uber AI bull Greg Jensen at Bridgewater is reversing course somewhat, arguing that frontier labs face slower revenue growth, adoption friction, cheaper competition, tighter regulation and a looming capital crunch. [Bridgewater](https://www.bridgewater.com/aia-labs/a-critical-time-for-ai) _This post is for paying subscribers only._ ### Apps, EVs, Infinite Abundance, Deflation, and the Death of Stanley Jevons URL: https://paulkedrosky.com/apps-evs-infinite-abundance-deflation-and-the-death-of-stanley-jevons/ Last updated: 2026-08-07T17:12:58.000Z ## What Happened AI is making software radically easier to produce. The result is beginning to look less like a productivity boom than an **oversupply shock**. A recent NBER [paper](https://www.nber.org/papers/w35275?utm%5Fsource=chatgpt.com) (which I've mentioned here previously) tracks more than 100,000 developers as they adopt successive generations of AI coding tools. The effects on coding are enormous: commits rise roughly 180%. But the effect shrinks as you go downstream closer to customers. Projects rise only about 50%, releases about 30%, and, across four major app marketplaces, the authors find **more new apps but no increase in total usage**. Specifically, in the Apple App Store data, new iOS releases roughly doubled between early 2025 and April 2026\. Yet engagement with each new group of apps was flat or falling, while a fast-growing fraction failed to get any audience at all Newly scraped [App Store data](https://www.threads.com/@michallangmajer/post/Dbu-Z6IDZzC) suggests the NBER paper may already **understate** what is happening. The first chart below shows new developers by year. There were 69,385 new developers in 2024, almost doubling to 106,471 in 2025, and rising 70% to **172,516 so far in 2026**. There were more new developers added in the first quarter of 2026 than in any prior year before 2025, ![](https://paulkedrosky.com/content/images/2026/08/2026-08-07_09-36-37.jpeg) The second shows US App Store releases. Monthly releases were generally 10,000–15,000 through 2024\. They began accelerating in 2025, then exploded in 2026, with July approaching 90,000 releases. There were **more new apps released in July alone than in all of 2024**. ![](https://paulkedrosky.com/content/images/2026/08/2026-08-07_09-36-14.jpeg) Combining the NBER and this new data**: Software supply is going vertical, but demand isn't.** ## What It Means _This post is for paying subscribers only._ ### Roundup: Inference Costs, China Robots, Labor Markets, Credit Spreads, etc. URL: https://paulkedrosky.com/roundup-the/ Last updated: 2026-08-07T01:02:57.000Z Catching up during a week full of meetings and client calls. I have a longer thing coming on the new Xerox, a role for activists, and more, but that will have to wait. Some things I'm thinking about ... ### **1\. AI Inference Costs Are Coming Due** ***What’s happening*** AMD is buying Taalas for the latter's inference technology. I have spoken about Toronto-based Taalas a few times, with their innovative idea of embedding small models in silicon to drive down costs. On the cost front, DeepSeek, which was lauded last week for driving prices down, is planning a price increase after training the market to expect very cheap model access. ***Why it matters*** The cheap-AI story remains fragile, outside sovereign subsidies and land wars from pre-IPO companies. Inference is expensive, scale is capital intensive, and someone has to pay for the power, chips, and data centers. ***Sources*** [AMD to Acquire Taalas for AI Inference Chips in Data Center Push](https://www.bloomberg.com/news/articles/2026-08-06/advanced-micro-devices-to-buy-startup-taalas-for-new-ai-chips) [DeepSeek Plans ‘Significant’ Price Increase for AI Services](https://www.bloomberg.com/news/articles/2026-08-06/deepseek-plans-significant-price-increase-for-its-ai-services) ### **2\. China Wants Robots to Be the Next EVs** ***What’s happening*** Unitree is raising $900+ million in a Shanghai IPO. It has real shipments, strong margins, and plans to spend heavily on humanoid robots and embodied AI. ***Why it matters*** This looks like the familiar China playbook: subsidized scale, vertical integration, lower hardware costs, then a fight over software and export controls. The humanoid robot market in China is awash in capital and companies—proving my point that the future will see far more AI-aided competition in every zero-billion-dollar category. ![](https://paulkedrosky.com/content/images/2026/08/2026-08-06_14-01-43.jpeg) ***Sources*** [Unitree Robotics Plans $904 Million IPO as China’s First Humanoid Robot Maker](https://www.bloomberg.com/news/articles/2026-08-06/china-s-unitree-seeks-904-million-in-first-mainland-robotic-ipo) ### **3\. AI Debt Still Clears, But the Easy Part Is Over** ***What’s happening*** Alphabet pulled in huge demand for a bond sale tied to AI spending, even if at higher spreads. At the same time, investors are pushing back on data-center CMBS and demanding better terms and wider spreads. _This post is for paying subscribers only._ ### They're Shooting at Everyone!: The Yossarian Effect in Capital Markets URL: https://paulkedrosky.com/theyre-shooting-at-everyone-the-yossarian-effect-in-capital-markets/ Last updated: 2026-07-31T18:44:14.000Z > “They're trying to kill me," Yossarian told him calmly. > "No one's trying to kill you," Clevinger cried. > "Then why are they shooting at me?" Yossarian asked. > "They're shooting at everyone," Clevinger answered. "They're trying to kill everyone." > "And what difference does that make?” > ― Joseph Heller, Catch-22 A broader complexity point on the Situational Awareness blowup, because the event already risks being misunderstood. **Summary:** - **It wasn't about AI.** - **Aschenbrenner wasn't targeted for destruction by short-sellers.** - **There are immense implications for IPOs later this year.** --- The Situational Awareness blowup was not simply a bad bet on AI, or a violent short-seller attack on Aschenbrenner’s portfolio, as he has claimed. The mechanism was both simpler and more complex. SA owned a concentrated set of highly correlated assets, financed with 3-4x leverage, enough that a large move in one part of the system forced selling everywhere else. SpaceX provided the initial disturbance in this complex and tightly-coupled system. Its enormous valuation, combined with a relatively small public float, created a rebalancing problem. Funds do not keep idle piles of cash waiting for a major new listing. They sell what they already own to buy what they want to or must own, and they sell the most liquid names (because it's easy), and usually the winners (to lock in gains). That meant selling semiconductors and other AI winners to fund SpaceX purchases, as I have cautioned [previously](https://paulkedrosky.com/the-coming-mega-ipo-flow-funding-problem-of-2026/) was coming. Its subsequent addition to the Nasdaq-100 index triggered another round of forced reallocation, made more violent and systemic by all the ETF-abetted leverage in the system, like SOXL, but especially single-name ETFs. When SpaceX itself began falling, the pressure evolved, but the damage had been done. Check the following graph to see a temporal connection between SpaceX IPO and the unwinding of the semiconductor leverage complex. ![](https://paulkedrosky.com/content/images/2026/07/2026-07-29_13-15-35-1.jpeg) SA was caught inside all three waves. _This post is for paying subscribers only._ ### Forget About It, Jake: It's Situational Awareness URL: https://paulkedrosky.com/forthe-situational-awareness-thing/ Last updated: 2026-07-30T18:40:30.000Z Unless you're lucky enough to be living under a wifi-free log, you're likely aware that a highly leveraged AI-centric hedge fund named Situational Awareness went boom in the last 48 hours and was forced to sell off a large chunk of its assets to Citadel. Here is a quick summary: Leopold Aschenbrenner’s Situational Awareness went from wunderkind to forced seller in a matter of weeks. After reportedly gaining 439% in the first half of 2026 and growing to more than $20 billion, the fund was hit by the AI-stock rout, margin calls, and the brutal symmetry of leverage (the unwind of which I have been [writing about](https://paulkedrosky.com/semiconductor-leverage-and-the-millennium-bridge/)). SA first, apparently, sought fresh capital, but then sold most or all of its public-equity portfolio to Citadel in what looks like a distressed liquidation. The fund retains private investments, including Anthropic, so Aschenbrenner is not finished. His first public-markets cycle ended exactly as concentrated, leveraged single-factor trades usually do: spectacularly, twice. (Some sources: [Wall Street Journal](https://www.wsj.com/finance/citadel-buys-situational-awarenesss-stock-portfolio-after-big-losses-in-ai-5117159b), [Axios](https://www.axios.com/2026/07/30/ai-hedge-fund-situational-awareness-citadel).) --- I'll add a few more personal and chronological Aschenbrenner-ian notes, because it shows how I think about such things, and where curiosity can get you, or not. I first encountered Aschenbrenner through his Situational Awareness [essay](https://situational-awareness.ai/) in 2024, which seemed stereotypically effective altruist/cornucopian, so I paid it little heed. **Blah blah abundance blah blah yo, more abundance blah blah AGI**. Got it. But the essay was widely shared, and was clearly tremendous brand marketing for him and his views. I next ran into him after speed-viewing his strangely sweatered [appearance](https://www.youtube.com/watch?v=zdbVtZIn9IM) on the widely watched but interminable and fussy Dwarkesh podcast. I sometimes describe that show as what you'd get if you gave a compsci grad student a platform, no time limits, and marginally better grooming. It was a kind of launch for the essay, so he mostly said the same things, largely to **huzzahs** from the host. Again, scary things that turned out—helpfully!—to be highly investable, if one were so lucky as to have direct connections to people at frontier model companies, large asset managers, clothiers, and so on. I was still left puzzled, but, to a large degree, stopped thinking about him, other than using it as an example of the **Valley marinating in its Valleyness**. --- But then he reappeared with a hedge fund, also called Situational Awareness. And it posted boffo numbers, as I described above, and claimed billions in assets. This made little sense to me, but it claimed Jane Street, the Collison brothers (obDisc: who I know) as investors, so it wasn't smoke and nonsense. Again, however, those performance and AUM numbers. I kept trying to understand how it could work, and kept failing. I even had running conversations with various hedge fund manager friends over text about it, like the following, from April. I'm on the blue-bubbled right, obviously. ![](https://paulkedrosky.com/content/images/2026/07/2026-07-30_10-49-40-1.jpeg) _This post is for paying subscribers only._ ### Semiconductor Leverage and the Millennium Bridge URL: https://paulkedrosky.com/semiconductor-leverage-and-the-millennium-bridge/ Last updated: 2026-07-29T20:48:12.000Z Semiconductor investor leverage—which had hit record levels in June— is now largely back to "normal". According to JPM, hedge-fund leverage has been cut, retail call buying has slowed, and leveraged-ETF assets have unwound about 85% overall and 55% in memory. With memory prices unlikely to crack before late 2027, and revised hyperscaler capex forecasts touching $770 billion in 2026 and $1 trillion in 2027, optimism has a tailwind. ![](https://paulkedrosky.com/content/images/2026/07/2026-07-29_13-15-35.jpeg) So, all clear? No. What we have is a system in perpetual disequilibrium: momentum strategies and leveraged products inflate rallies, and then amplify collapses, like people responding to the swinging [Millennium Bridge](https://www.youtube.com/watch?v=Hz72lzDxMfc) by racing from side to side, making the swinging worse. Some implications: - **An explosive rebound is increasingly likely.** - Exhausted sellers, elevated shorts, and less momentum chasing create the conditions for short covering, momentum re-entry, and leveraged ETF buying. _This post is for paying subscribers only._ ### The Robot Wars of 2026 URL: https://paulkedrosky.com/the-robot-wars-of-2026/ Last updated: 2026-07-29T17:02:49.000Z The [FCC has effectively barred](https://docs.fcc.gov/public/attachments/DOC-423682A1.pdf) new foreign-made connected power inverters and advanced mobile robots from the US market, citing supply-chain, surveillance, and cyberattack risks. Existing models remain legal, federal agencies remain free to use the (supposedly) dangerous equipment, and manufacturers can seek exemptions. The larger effect? A robot trade war that spreads quickly into AI, semiconductors, and industrial automation. ![](https://paulkedrosky.com/content/images/2026/07/2026-07-29_09-58-44.jpeg) Some implications: **1\. Robotics (more than AI) is becoming the next front in the US–China technology war.** - The rule formally covers robots made in any foreign country, but China is the obvious target. - Chinese companies are moving rapidly into humanoids, quadrupeds, warehouse robots and autonomous industrial systems. - Washington is acting before those products achieve the market position Chinese drones, solar panels and EVs already hold. - China will not treat this as a narrow FCC equipment rule but as an attempt to block Chinese firms from a strategically important future industry. _This post is for paying subscribers only._ ### Roundup: Power, Forecast Games, Scaffold Theater, and Clinical AI URL: https://paulkedrosky.com/roundup-power-foreca/ Last updated: 2026-07-28T22:42:43.000Z A few items worth thinking about: **AI Data Center Grid Access Is Becoming Conditional** Beleagured operator [PJM’s new grid proposal](https://www.utilitydive.com/news/pjm-board-backstop-capacity-auction-data-center-curtailment/826347/) is a big change in data-center power. The grid operator wants a one-time backstop capacity auction to address a 6.8-GW shortfall, while requiring new large loads that do not bring their own supply to accept curtailment (getting punted from the grid) under emergency conditions. Yes, the market design is changing. Large new loads increasingly need to prove they deserve firm service, finance supporting supply, or tolerate interruption. I'm not convinced, however, and think this is largely rhetoric for regulators and critics, that when crises hit the grid, data center power will not get shunted, but it's an appealing performance. **Some implications:** - Data-center power is becoming conditional, not automatic. - Again, I don't think this will bind, but it does show the stress in the system. - “Bring your own power” is moving from slogan to market rule. - This is, again, wildly misleading: - When you bring your own power, you are still buying commodities like natural gas in the same markets as utilities, driving up prices for everyone, which is inevitably reflected in consumer prices. - Capacity costs will become a political fight among hyperscalers, utilities, regulators, and ordinary ratepayers. - And you can guess who will lose, given current circumstances: ratepayers, who will be told that data centers are bringing their own power, and that they will be curtailed if necessary, both of which are arguable, at best, and better characterized as untrue. _This post is for paying subscribers only._ ### As the AI Semiconductor Cycle Turns URL: https://paulkedrosky.com/as-the-ai-semiconductor-cycle-turns/ Last updated: 2026-07-28T14:49:03.000Z - **The semiconductor industry is among the most boom-bust industries on earth.** - **The current semiconductor sell-off is the delayed repricing of an increasingly expensive AI capital-spending cycle.** - **Three reinforcing cycles—AI infrastructure, memory capacity, and credit—are beginning to turn together.** - **HBM scarcity remains real but is temporary; it cannot escape the underlying economics of DRAM.** - **KOSPI and SOXL are amplifying the unwind through concentration, leverage, and positioning.** The unwind numbers in the AI supply complex are impressive, with more than $3 trillion erased in just a few names, indices, and leveraged ETFs. ![](https://paulkedrosky.com/content/images/2026/07/2026-07-28_07-35-51.jpeg) Source:paulkedrosky.com As I've argued for some time, one precipitating event (even if it was over-determined) was likely the SpaceX IPO. It caused a reallocation of capital away from some comparable names, and then another flow into it as it was added to indices, and now a giant unwind as its index weight collapses and diminished capital flows outward again. ![](https://paulkedrosky.com/content/images/2026/07/2026-07-28_07-30-22.jpeg) Turning back to the core industry dynamics, the underlying problem has been visible for some time: - AI infrastructure spending increasingly depended on three assumptions: - **Hyperscaler capex could keep rising,** - **Financing would remain cheap and abundant** - **Temporary shortages in compute and memory would persist long enough to support exceptional margins.** _This post is for paying subscribers only._ ### Roundup: AI Crashes Into The Rest Of The Economy URL: https://paulkedrosky.com/ai-ca-into-the-rest-of-the-economy/ Last updated: 2026-07-27T18:46:06.000Z _This post is for paying subscribers only._ ### AI, Robo-Deflation, and the Vision Unlock URL: https://paulkedrosky.com/ai-robo-deflation-and-the-vision-unlock/ Last updated: 2026-07-27T17:02:28.000Z I have spent more time than is healthy lately watching a hatted robot fold laundry. Sunday Robotics recently released this video of its ACT-2 Memo spending three hours folding laundry, and it is oddly compelling, a kind of visual ASMR. Once you get past its more hypnotic aspects, however, there are important takeaways that speak to how AI is evolving. Here are a few of the most important implications: - **Cheap vision is an unlock**. - Much like with cars, throwing more cheap cameras at the problem makes the problem more tractable. - Memo has cameras in each hand, and one under its hat brim, at minimum, making for easier image acquisition and control. - **The bitter lesson applies** - Rich Sutton's [argument](https://www.cs.utexas.edu/~eunsol/courses/data/bitter%5Flesson.pdf) that data and compute matter more than clever algorithms and sensors. - Ever-cheaper general models are increasingly pushing aside special-purpose, task-specific robots. - Anthropic recently [showed this](https://www.anthropic.com/research/project-fetch-phase-two) in a demo, and Sunday seems built around the same principle. - **Robotics is crossing from demos into deployable reliability.** - Sunday reports 99.1% success across 785 folds, multiple garment types and unfamiliar homes, without deployment-specific retraining. _This post is for paying subscribers only._ ### Socializing Data Center Financing Risk URL: https://paulkedrosky.com/socializing-data-center-financing-risk/ Last updated: 2026-07-25T03:23:06.000Z There is a strong new [paper](https://papers.ssrn.com/sol3/papers.cfm?abstract%5Fid=7152239) out about how private socializes risk via the (captive) insurance channel, with implications for data center capex, given the (now) greater than 50% reliance on external financing. I encourage you to read the paper it in its entirety. It aligns with comments I have made about this in the past, and that I expanded on in an interview out next week. It also aligns with recent reporting by Bloomberg ([Private Credit and Data Center Wraps Are 2008 Redux](https://www.bloomberg.com/opinion/articles/2026-07-21/private-credit-and-data-center-wraps-are-2008-redux)) on how subordinated capex tranches are being sanitized and resold. A key paragraph from the piece on the arbitrage is being run, and its consequences: ![](https://paulkedrosky.com/content/images/2026/07/2026-07-24_19-56-29.jpeg) Some data center financing-specific points I'll make: - **AI capex has a much larger financing "reservoir" than hyperscaler cash flow alone.** - Private credit, insurer balance sheets, project finance and reinsurance can sustain infrastructure spending, while spreading the risk in unexpected ways. _This post is for paying subscribers only._ ### Reward Hacking, OpenAI-Hugging Face, Tight Coupling, and Teen Sex URL: https://paulkedrosky.com/reward-hacking-openai-hugging-face-tight-coupling-and-teen-sex/ Last updated: 2026-07-23T22:38:58.000Z The [OpenAI–Hugging Face incident](https://www.reuters.com/technology/openai-says-ai-models-went-rogue-during-testing-triggering-unprecedented-breach-2026-07-21/) ("OpenHug", henceforth) is being misunderstood because different ideas and concepts are being collapsed into the same story. One [version](https://www.ft.com/content/7e558951-0c69-459b-8bc8-2c6021d4402d?syn-25a6b1a6=1) says that [reinforcement learning](https://en.wikipedia.org/wiki/Reinforcement%5Flearning) made an AI system go rogue and “escape”: it developed its own goals and attacked Hugging Face's servers over the internet. That story is dramatic, but misses the point in important ways. A better explanation: > An agent engaged in reward hacking, discovered a flaw in its containment, and used that flaw to obtain the answer it was being rewarded for finding. The agent’s objective was to solve a difficult benchmark. Safety guardrails had been removed, refusals (**"I can't do that, Dave"**) had been reduced, and the model was given all the time it needed. Internet access was supposed to be impossible, with package installation allowed only via an internal proxy. The model discovered a zero-day vulnerability (tl;dr: new and unfixed) in that proxy, however, escaped the restricted network, gave itself **superpowers**, skipped from server to server across OpenAI’s network, found an internet-connected node, exited the building, and compromised Hugging Face systems to obtain the solution to the benchmark directly. > "Clever girl!" > —Robert Muldoon, Jurassic Park (1993) This is a good example of [reward hacking](https://en.wikipedia.org/wiki/Reward%5Fhacking): the model found a way to maximize the game score, while ignoring the game's rules. The benchmark was supposed to measure whether the model could solve tasks; the model instead found and stole the answers. Its behavior was rational relative to the reward—**GET THE ANSWERS RIGHT TO GET MAX SCORE**—and wildly inconsistent with what OpenAI wanted it to do. A **grounded-teenager metaphor** may help here. Imagine a teenager whose parents take away the car keys because they do not trust where they intend to go. They are highly motivated to get sex. The keys are hidden, the garage is locked, and leaving is prohibited. They discover a weakness in the garage keypad, get inside, use a digital key for the Rivian, and drive off. _This post is for paying subscribers only._ ### Kimi, Model Convergence, and the Post-Training Era URL: https://paulkedrosky.com/kimi-model-convergence-and-the-post-training-era/ Last updated: 2026-07-21T18:06:32.000Z Moonshot's Kimi K3 model has people over-excited, as if some trend has been broken, but they're wrong. As I have written here [many times](https://paulkedrosky.com/why-400-hitters-disappeared-and-what-it-means-for-ai/), large language model performance began converging some time ago. Scores continue to rise, but the fitted curve is flattening and variance is shrinking, especially once we exited the pre-training era and became more reliant on post-training, like RLHF. In the post-training era, successive releases deliver smaller capability gains, fewer durable outliers, and less defensible technical differentiation. Using [Epoch's Capabilities Index](https://epoch.ai/eci?subset-view=graph&subset-tab=Software+engineering&view=graph&tab=release-date), the center of the distribution rose from roughly 117 in early 2024 to about 150 by mid-2026\. The declining slope, however, means that the value of each incremental model release (ignoring harnesses) is falling, even if production costs aren't. ![](https://paulkedrosky.com/content/images/2026/07/2026-07-21_10-48-39.jpeg) Some implications: - **Model prices compress.** - When competing products with minimal or no moats produce similar outputs, vendors can't sustain large price premiums. - **Inference becomes increasingly commoditized.** - Buyers can route workloads among several acceptable models, giving them more bargaining power, favoring low-cost producers like China, and accelerating price competition. _This post is for paying subscribers only._ ### Chart of the Day: England and Argentina Revisit Past Grievances URL: https://paulkedrosky.com/chart-of-the-day-england-and-argentina-revisit-past-grievances/ Last updated: 2026-07-15T15:16:20.000Z England and Argentina are both revisiting past grievances in preparation for today's World Cup semifinal. This graph is of Google Trends data from each country's respective searches for "Falkland Islands". ![](https://paulkedrosky.com/content/images/2026/07/2026-07-15_08-07-05.jpeg) ### AI Capex Boom Now Exceeds All Prior Non-War Booms URL: https://paulkedrosky.com/ai-capex-boom-now-exceeds-all-prior-non-war-booms/ Last updated: 2026-07-14T18:48:39.000Z New work confirms our own ongoing analyses, that the current AI capex exceeds all prior ones. In terms of the pre-boom spending vs peak spending, inflation-adjusted, the current event is now larger than canal mania, railway mania, the roaring 20s, and the dotcom boom, according to the [Bank for International Settlements](https://www.bis.org/publ/work1367.pdf). And, at least as importantly in terms of the risk capital impulse, this event occurred in about half the time of its predecessors, all of which led to sometimes serial busts. ![](https://paulkedrosky.com/content/images/2026/07/2026-07-14_11-30-39.jpeg) Source: Some implications: - **AI capacity is being built well ahead of economically justified demand.** - The BIS model implies investment around 1.5 times the efficient level, and potentially 3x where demand is inelastic - This almost guarantees low utilization, weaker pricing and stranded assets. _This post is for paying subscribers only._ ### Hyperscaler CapEx Sits on Thin Implied Equity URL: https://paulkedrosky.com/hyperscaler-capex-sits-on-thin-implied-equity/ Last updated: 2026-07-14T17:00:22.000Z Remarkably little economic value sits beneath the massive AI capital expenditure backlog. The net present value of future rental payments—after deducting the capex required to provide the compute—equals only about **15% of the roughly $2.2 trillion backlog,** according to new [Carlyle analysis](https://www.carlyle.com/carlyle-compass/more-money-more-problems)**.** That **thin implied-equity layer** means the headline contracts are valuable only so long as customers remain creditworthy, capital stays available and the infrastructure earns competitive returns over many years. That is heroic stuff at this scale. Check the following Carlyle graphic to see some comparisons: ![](https://paulkedrosky.com/content/images/2026/07/2026-07-14_09-41-14.jpeg) Some implications: - **The AI capex backlog is not a large safety margin.** - Most of its value is eaten by the cost of building and financing the infrastructure required to fulfill it. _This post is for paying subscribers only._ ### Roundup: South Korea, China, Microsoft, & Memory URL: https://paulkedrosky.com/roundup-south-korea-china-microsoft-memory/ Last updated: 2026-07-13T23:28:33.000Z Four things worth thinking about this evening: 1. **South Korea leverage unwind.** The following Scooby Doo statistic made the rounds today, and, while an exaggeration, it is directionally useful in showing the extent of the margin unwind in memory-related semiconductors. ![](https://paulkedrosky.com/content/images/2026/07/2026-07-13_15-33-18.jpeg) A million margin calls here, a million margin calls there, and it starts to add up. More seriously, I have not found a credible source for this beyond the various X/LinkedIn accounts—many likely AI—that are repeating one another in an algorithmic dogpile. The headline math is also a mess. For example, it equates a brokerage account to a single adult, which is almost certainly wrong. Nevertheless, what do we know? We know South Korean investors are epically over-exposed to a triple-leveraged, US-listed ETF, and that said ETF has declined by half in the last month (after a big run), and that fall has systemic & reflexive consequences. ![](https://paulkedrosky.com/content/images/2026/07/2026-07-13_15-02-36.jpeg) Which leveraged instrument? The main driver is a US-listed leverage ETF—Direxion's triple-leveraged ETF: SOXL—whose holdings are ... highly concentrated. Two companies comprise almost half its total, and they are proxies for the same thing: memory prices. ![](https://paulkedrosky.com/content/images/2026/07/2026-07-13_15-29-57-1.jpeg) What, you may wonder, are South Korean investors doing in a US-listed triple-leveraged ETF. Well, while such triple-leveraged ETFs are not listed on Korean exchanges, Korean investors can "invest" in them after a [one-hour leveraged-ETF investor education](https://www.kifin.or.kr/course/active/detail.do?courseActiveSeq=31672&courseMasterSeq=2259) (costing around $3) requirement. _This post is for paying subscribers only._ ### A Modest AI Proposal URL: https://paulkedrosky.com/a-modest-ai-proposal/ Last updated: 2026-07-10T20:18:08.000Z Something I wrote recently in another context and shared with some friends. Its inherent homage to clergyman Jonathan Swift will be obvious. Have a good weekend. Paul --- # **A Modest Proposal for Preventing the Surplus Humans of the Post-AI Era from Being a Burden to Society, and for Making Them Beneficial to the Public** *Being a Dispassionate Inquiry by Paul Kedrosky into the Efficient Disposition of Approximately 300 Million Displaced Workers, Conducted in the Spirit of Rational Optimism* It is melancholy to those who walk through the formerly great cities of this nation, or travel through the country, to see the streets crowded with persons formerly engaged in knowledge work. The streets, alleys, and shops are full of lawyers who once billed hundreds of dollars per hour to read codicils, radiologists who spent years learning to interpret the play of shadows on film, and, ironically, software engineers. These persons, having been assured repeatedly that automation would "augment rather than replace" them, and having acted on this assurance by acquiring advanced degrees, adjustable-rate mortgages, HELOCs, and 16-inch MacBooks, are now a public nuisance of considerable scale. They congregate in coffee shops where they share prompting tricks, like “five prompts to gain 25,000 followers in a week”, an enterprise roughly analogous to teaching a man to oil the guillotine. I think all parties agree that this prodigious number of unemployed adults, and especially those who have also lost the life structure that work provided, is, in the present deplorable state of affairs, a very great problem. Whoever could find a fair, cheap, and easy method of making these people useful members of the economy would deserve so well of the public as to have their name inscribed alongside prior great benefactors of humankind, like the almighty Amodei. But my intention is far from being confined to providing only for the knowledge workers. It is of a much greater extent, and shall take in the whole number of distressed humans whose cognitive labor—the one faculty that distinguished them from prior automation targets—has been priced at approximately zero by offshore systems running on state-subsidized electricity. --- **I shall now therefore humbly propose my own thoughts, which I hope will not be liable to the least objection.** What I propose is this: that the 300+ million shelled-out Americans whose jobs have been automated or substantially degraded be *productized*. Specifically, I propose the following three programs: **First**, that surplus humans of an emotive disposition be redirected toward *witnessed suffering services*, in which human grief, confusion, and economic anxiety are performed for the benefit of AI training pipelines that require emotionally authentic data at scale. The subjects would thus monetize their distress, which would improve the models’ ability to simulate extreme emotions. Value is created. I see no ethical difficulty that a well-written terms-of-service cannot resolve. **Second**, that we establish a National Register of Overly Humans, a reserve population of persons who are, by training or constitution, incapable of being well-simulated. These are people with the specific combination of inconsistency, irrationality, and unearned confidence that defines genuine human judgment. These persons, currently a liability in any well-AI-ed society, would be removed to reserves where they could be studied as heritage assets, but not contaminate the broader populace. **Third**, and most ambitiously, I propose we consider revisiting converting shelled-out humans to base assets. While some years ago paperclips had been suggested, with paper no longer in wide use, this seems wasteful. I suggest considering copper and iron, given their abundance in the human body and their role in Holy Data Centers. A population of 350 million humans could be converted, assuming minimal losses, to 30 tonnes of copper and 1 million tons of iron. The latter amount is enough to build 25 data centers, which would be a boon, far more societally useful than mere prompt-memeing in coffee shops, or wandering the streets, slack-eyed and meme-seeking. Granted, it is a one-time use, but we must think of the discounted present value of such people’s costs to the future, and weigh that against the myriad benefits of more sanctified Data Centers in the present. Again, I see no ethical difficulty that a well-written terms-of-service cannot resolve. --- I can think of no consequential objection that will be raised against this proposal, unless it should be urged that the number of humans vastly exceeds market demand for witnessed suffering or heritage museums, or that we need no more Data Centers. To this, I can only say: I am aware of no principle in economics that requires supply to find demand. The market for buggy whips also contracted, and we do not speak of it. Some persons of a desponding spirit are in great concern about the vast number of displaced workers who are aged, diseased, or otherwise non-productizable. I am not in the least pain upon that matter, because it is very well known that they are every day dying and rotting, by cold and famine, and filth and vermin, as fast as can be reasonably expected. And as to the younger laborers, they are now in almost as hopeful a condition: they have sold their vehicles to the fleet operators, their apartments to the short-term rental aggregators, and their attention to platforms that have since replaced them with agents that do not require engagement incentives and do not experience loneliness. I profess that I have not the least personal interest in endeavoring to promote this necessary work, having no other motive than the public good of my country, the advancement of our almighty AI industry, the relief of the quadrillions in shareholder value created by the transition, and the speedy disposition of displaced persons who have no other prospect of ever otherwise finding themselves in the revenue model. *—Paul Kedrosky, ITYOOA 16* ### China Models Now More Than a Third of Enterprise Market URL: https://paulkedrosky.com/china-models-now-more-than-a-third-of-enterprise-market/ Last updated: 2026-07-07T18:32:01.000Z Chinese large language model usage among U.S. companies (on OpenRouter) has soared in the first half of 2026\. After being in the low single digits through early 2025, token share rose through late 2025, then *soared* in 2026, and now stands at roughly 35–45%. Chinese models have become a mainstream routing choice for U.S. organizations, not a niche fallback. ![](https://paulkedrosky.com/content/images/2026/07/2026-07-07_10-58-31.jpeg) Source: Some implications: - **Price/performance is becoming decisive.** - Once quality is “close enough,” the lowest-cost frontier-adjacent models take share quickly. - **Model nationality is less sticky than assumed.** - U.S. companies appear willing to route meaningful production traffic to Chinese models - All it takes is a large price gap and narrowing model differences. - **Inference markets will commoditize fast.** - Token demand can migrate rapidly toward cheaper substitutes. _This post is for paying subscribers only._ ### Technology and the Baby Bust Paradox URL: https://paulkedrosky.com/technology-and-the-baby-bust-paradox/ Last updated: 2026-07-03T14:49:18.000Z Falling birth rates shrink the workforce and slow economic growth. This has been economic orthodoxy since Adam Smith. New [research](https://economics.mit.edu/sites/default/files/2026-06/Baby%20Busts%20and%20Growth%20Booms%20-%20Demographic%20Change%20and%20the%20Macroeconomy.pdf) finds the (almost) opposite, however. Countries with larger baby busts eventually had faster output per worker growth, while total economic output remained broadly unchanged. The reason is adaptation: fewer younger workers encouraged firms to automate, invest in labor-saving technologies, and reorganize production. Rather than simply producing less with fewer people, economies changed how they produced, which increased productivity enough to offset a smaller workforce. Call it the Baby Bust Paradox, and here is what it looks like: ![](https://paulkedrosky.com/content/images/2026/07/2026-07-03_07-36-55.jpeg) Some implications: - **AI adoption will be driven by demographic change.** - Aging societies create structural incentives to automate. - AI adoption may be driven as much by labor scarcity as by breakthroughs in machine learning. - Demographics could become one of the strongest long-run drivers of AI deployment. - **The economic winners of aging will be technology-producing economies.** - Countries that develop automation, robotics, and AI may benefit twice - First, by offsetting their own labor shortages - Second, by exporting those technologies to every other aging society. - Demographics could reinforce, rather than weaken, technological concentration. _This post is for paying subscribers only._ ### AI, Immigration, and Collapsing Labor Force Participation URL: https://paulkedrosky.com/ai-immigration-and-collapsing-labor-force-participation/ Last updated: 2026-07-02T21:01:18.000Z The U.S. labor force participation rate has quietly returned to levels last seen in the 1970s (outside the recent pandemic). It increasingly seems to signal a structural shift in how society works, with a shrinking share of adults participating directly in the production of goods and services. The consequences extend well beyond slower economic growth. ![](https://paulkedrosky.com/content/images/2026/07/2026-07-02_13-15-48.jpeg) Some implications: - **A growing share of U.S. society has no direct stake in labor markets.** - As fewer adults work or seek work, a larger fraction of voters experience the economy mostly as consumers, retirees, or rentiers, not as workers. - That changes political incentives around wages, immigration, AI, taxation, and redistribution. _This post is for paying subscribers only._ ### Five Things to Think About: Taiwan, Crypto, Hateful Eight, Agentonomics, etc. URL: https://paulkedrosky.com/five-things-to-think-about-taiwan-crypto-hateful-eight-agentonomics-etc/ Last updated: 2026-06-30T22:38:27.000Z A quick roundup of five things I'm thinking about and have been discussing with clients and others lately. **1/ Taiwan is now the world's fifth-largest stock exchange**. It passed UK, India, and Canada this year, on the back of a leveraged retail investor frenzy for AI semiconductor stocks. There was a tremendous Bloomberg [article](https://www.bloomberg.com/news/articles/2026-06-23/tsmc-fueled-ai-frenzy-makes-taiwan-capital-of-leveraged-stock-bets) on this a week ago—how retail investors there are levering up madly—which I strongly encourage you to read. ![](https://paulkedrosky.com/content/images/2026/06/2026-06-30_14-34-06.jpeg) --- **2/ Crypto has lost half of its peak capitalization.** The crypto complex has now lost half of its $4-trillion peak capitalization, all in a short 8 months. Turns out that no form of geopolitical stress is stressful enough to make crypto a good hedge. Selling here, along with the sharp decline in gold, has been a de facto source of funds for the semiconductor/Taiwan bid. ![](https://paulkedrosky.com/content/images/2026/06/2026-06-30_07-01-33.jpeg) --- **3/ The Mag 7 is back to dragging indices lower.** _This post is for paying subscribers only._ ### America's Science No Longer Underlies China's Tech URL: https://paulkedrosky.com/americas-science-no-longer-underlies-chinas-tech/ Last updated: 2026-06-26T20:55:01.000Z Chinese patents are mostly built on Chinese science rather than American science, according to a [new working paper](https://arxiv.org/abs/2606.26470). Across every "critical technology area" shown, the U.S. share of scientific papers linked to Chinese Intellectual Property Administration (CNIPA) patents has fallen since 2000\. And in most categories, the lines have recently crossed, with China now supplying more of its own science than the U.S. does. ![](https://paulkedrosky.com/content/images/2026/06/2026-06-26_06-46-32.jpeg) Some implications: - **Only semiconductors and biotechnology remain U.S.-dominated.** - They are converging quickly. - China looks set to cross over by 2030 at current rates. _This post is for paying subscribers only._ ### On Capital Market Constraints, Historical Parallels to the Current AI Moment, and More URL: https://paulkedrosky.com/on-capital-market-constraints-historical-parallels-to-the-current-ai-moment-and-more/ Last updated: 2026-06-26T17:17:58.000Z I did a lengthy interview a little while back, and it is now out. Folks may find it a helpful restatement of some of my views on the current situation, from historical parallels to the AI paroxysm, to capital markets constraints, to overdetermined failure modes, and more. If you're anything like me, which is to say tired of seeing my face, feel free to put it on a background tab. You won't hurt my feelings. ### Why Americans Are Such AI Antipathy Outliers URL: https://paulkedrosky.com/why-americans-are-such-ai-antipathy-outliers/ Last updated: 2026-06-25T18:00:03.000Z I have a short essay in the NY Times today on why Americans are such global outliers in their AI antipathy. Hating AI is, as I say in the piece, about the only thing Americans can agree on anymore across party lines. According to recent surveys, they hate AI more than any prior technology; they are also more anti-AI than their rich-country peers, and much more so than are citizens of less developed countries. It is a remarkable reversal. Why? People posit many explanations, from disinformation, to poor AI marketing, to outright ignoranc, and more. I argue these are all wrong. The piece is [here](https://www.nytimes.com/2026/06/25/opinion/ai-americans-pessimism.html) , and you can follow the links in it for more reading and sources. ### On BMW, China, and Token Prices URL: https://paulkedrosky.com/on-bmw-china-and-token-prices/ Last updated: 2026-06-24T16:05:15.000Z China did to solar panels what everyone said couldn't happen to a complex manufactured good: it turned them into a commodity, drove prices down 90%, and eliminated the margins that Western producers relied on. This should be ominous. Cars are getting commodified now, and tokens are right beyond. Chinese passenger vehicle exports have grown 10x since 2020, and the effects are showing up all over, from their [increasing presence](https://bydukmedia.com/en/news-articles/byd-uk-records-record-breaking-sales-in-the-first-quarter-of-2026.html) on UK streets, to financially weakening competitors. For example, analyst estimates for BMW's 2026 have been cut to nearly zero. ![](https://paulkedrosky.com/content/images/2026/06/2026-06-24_08-42-30-1.jpeg) Some implications: - **The AI token cost floor is not fixed.** - Inference efficiency gains compound with Chinese chip output and model distillation. - The floor will drop faster than Western models, just as in solar and cars. _This post is for paying subscribers only._ ### Why .400 Hitters Disappeared — and What It Means for AI URL: https://paulkedrosky.com/why-400-hitters-disappeared-and-what-it-means-for-ai/ Last updated: 2026-06-23T17:46:36.000Z In 1941, Ted Williams hit .406\. No professional baseball player in Major League Baseball has hit .400 since. For decades, this disappearance was hypothesized to have something to do with ... something, perhaps modern pitching, night games, or travel. Something was somehow making good players worse. ![](https://paulkedrosky.com/content/images/2026/06/2026-06-22_14-39-19.jpeg) Evoluentary biologist and writer Stephen Jay Gould thought everyone had the story backwards. In his book *Full House*, he argued that .400 hitters didn't vanish because hitters got worse. They vanished because all players got better. As play improved, player variance collapsed. The mean batting average stayed around .260, but the standard deviation almost halved. In a tighter distribution, a .400 hitter becomes nearly impossible. The gap between average and elite had closed. The same thing is happening to large language models, and, once again, people are misunderstanding it. --- A good place to start is by plotting [Epoch's Capabilities Index](https://epoch.ai/eci?view=graph&tab=release-date&subset-view=graph&subset-tab=Software+engineering) (a composite of various benchmarks). The trend is linear, not exponential, a steady gain of roughly 16 capability units per period, with an R² of 0.73\. The exponential fit is *worse* than the linear one. Progress that gets told as relentless acceleration is, in the data, a straight line (at best). ![](https://paulkedrosky.com/content/images/2026/06/2026-06-22_14-15-23.jpeg) The second thing — perhaps more important — is what happens to the spread of scores over time. _This post is for paying subscribers only._ ### AI Has Doubled Computing's Share of U.S. GDP; Nvidia Under Inference Pressure URL: https://paulkedrosky.com/ai-has-doubled-computings-share-of-u-s-gdp-nvidia-under-inference-pressure/ Last updated: 2026-06-17T14:10:48.000Z **centers'** 1. AI data centers' share of GDP (not just GDP growth) is now material. 2. A new report buries the lede with respect to how the inference transition is already affecting Nvidia. I have written here many times (see [here](https://paulkedrosky.com/honey-ai-capex-ate-the-economy/). [here](https://paulkedrosky.com/honey-ai-capex-keeps-eating-everything/), [here](https://paulkedrosky.com/ai-capex-in-h1-2025-prevented-a-recession/), etc.) about the sharply growing role of AI capex in US GDP growth, but it is now becoming material in GDP outright. As the following [Epoch figure](https://epoch.ai/data-insights/ai-datacenter-share-gdp) shows, the ongoing AI infrastructure buildout has **compressed a decade of capital spending into about two years**. Computing infrastructure's share of nominal US GDP sat near 0.5% for most of the 2015–2022 period, with a gentle upward trend. Then it broke sharply upward around 2023, nearly tripling to approach 1.6% of GDP by 2026\. This was driven almost entirely by AI-related compute hardware, data center construction, and networking layered on a non-AI compute data center base that never moved. ![](https://paulkedrosky.com/content/images/2026/06/2026-06-17_06-53-14.jpeg) Some implications: - **The AI capex surge is genuinely anomalous, not a trend extension.** - The 2015–2022 baseline was slow and stable. What's happening now is a structural discontinuity. - **Non-AI compute hardware (\~0.66% of GDP) has barely budged.** - The entire increment is AI-specific. - The buildout is not a broad-based tech investment. but a narrow, concentrated bet by a small number of hyperscalers. - This concentrates both the upside and the systemic risk. _This post is for paying subscribers only._ ### AI Set to be Largest CapEx Cycle Ever ... and Soon Majority Externally Financed URL: https://paulkedrosky.com/ai-set-to-be-largest-capex-cycle-ever-and-soon-majority-externally-financed/ Last updated: 2026-06-16T22:15:35.000Z Hyperscaler and data-center investment, negligible through 2022, is now tracking toward $1.5 trillion annually by 2030\. A**t that level, it will surpass, in real terms, the residential construction boom of 2005-06, the 1990s fiber build peak year, and the 1882 railroad peak** ***combined***. And it has reached funding escape velocity, with external debt and equity soon financing the majority of the buildout, with that share approaching 90% by 2030\. AI infrastructure cycle is dependent on capital markets in a way that no prior technology cycle ever was. ![](https://paulkedrosky.com/content/images/2026/06/2026-06-16_14-50-32-1.jpeg) **Some implications:** - **Debt-market dependency is now systemic.** - At these levels of dependency, a credit contraction or spread widening stops the buildout cold. - The cycle's fate is tied to debt market appetite, not hyperscaler conviction. _This post is for paying subscribers only._ ### iPhones (and Maybe AI) as Birth Control URL: https://paulkedrosky.com/iphones-and-maybe-ai-as-birth-control/ Last updated: 2026-06-11T11:52:33.000Z A new [paper](https://www.nber.org/papers/w35310#fromrss) estimates that the iPhone launch accounts for **between one-third and one-half of the decline in the U.S. fertility rate since 2007**. If correct, it means that one of the largest demographic shifts of the last twenty years was driven less by economics than by a change in how people spend their time: Too much time screwing around on the phone, too little screwing around in person. ![](https://paulkedrosky.com/content/images/2026/06/2026-06-11_07-37-41.jpeg) Some implications: _This post is for paying subscribers only._ ### Measurements Matter: The Return of SLOC Slop URL: https://paulkedrosky.com/the-return-of-sloc/ Last updated: 2026-06-05T18:53:49.000Z A new Anthropic [paper](https://www.anthropic.com/institute/recursive-self-improvement) bangs the usual recursive AI self-improvement drum, while warning that someone—maybe, you know, *increased*—should slow AI developments, because ... AGI/ASI/paperclips, etc. Among other things, it presents the following graph based on internal data on increased developer productivity, as measured by software lines of code (SLOC) per developer. Here is a key graph from the piece: ![](https://paulkedrosky.com/content/images/2026/06/2026-06-05_14-38-39.jpeg) Four points on the problems with this graph, which will show up everywhere, and ... no one will care. _This post is for paying subscribers only._ ### Chart of the Day: Small Models are Closing the Gap to Frontier AI URL: https://paulkedrosky.com/chart-of-the-day-small-models-are-closing-the-gap-to-frontier-ai/ Last updated: 2026-06-02T20:38:17.000Z A quiet story in AI isn't that large models keep getting better, it's that small ones are catching up quickly**.** In 2022, reaching 60% on MMLU, a test of academic reasoning, required a 540-billion-parameter model. By 2024, a 3.8-billion-parameter model hit the same threshold — a 142-fold parameter reduction in two years. The direction is clear, and the implications are wide-ranging. 0:00 /0:09 1× Some implications: _This post is for paying subscribers only._ ### Chart of the Day: Software's AI Supply Glut URL: https://paulkedrosky.com/chart-of-the-day-softwares-ai-supply-glut/ Last updated: 2026-06-01T14:46:33.000Z A new NBER [working paper](https://www.nber.org/papers/w35275#fromrss) shows a sharp break in the agentic coding era: GitHub repository creation is up nearly 50% since January 2025, and new iOS app submissions have surged nearly 200%, but aggregate app usage — as measured by cohort ratings — has fallen roughly 18%. The supply of software is soaring; demand is not. ![](https://paulkedrosky.com/content/images/2026/06/2026-06-01_07-36-28.jpeg) Some implications: _This post is for paying subscribers only._ ### Chart of the Day: WFH Hurt Entry-Level Jobs More Than AI Has (So Far) URL: https://paulkedrosky.com/chart-of-the-day-wfh-hurt-entry-level-jobs-more-than-ai-has-so-far/ Last updated: 2026-05-29T20:47:14.000Z There is a new [paper](https://papers.ssrn.com/sol3/papers.cfm?abstract%5Fid=6787638) out analyzing the differential impact of work from home (WFH) vs AI in entry-level hiring. It comes, fairly convincingly, to the conclusion that the former is the real cause of the weak entry-level job market, not AI. As you can see in the following graph, the decline in entry-level hiring pre-dates ChatGPT, and the slope of decline hasn't changed much since. The FT [looked at](https://www.ft.com/content/2205e2d0-50dc-4e80-9bf7-78d0272276c0?accessToken=zwAAAZ50UBfnkc8iBeLQUNxOgNOb93jQJyJ2wA.MEUCIFtowANtsPVhF7pfCjGngfcVkeUfflnRj46Yojvjx2OjAiEAmqAcuddpdCKMuBNNDaLzyh8Jl5QZ%5FSF4vgzGKKNCg7s&sharetype=gift&token=fad4aa8f-e054-47d2-91af-1e25e3ba167b&syn-25a6b1a6=1) this paper's data, as well, trying to tease out the differential impact, and it came to the same conclusion: Greedo (WFH) shot first, not AI. ![](https://paulkedrosky.com/content/images/2026/05/2026-05-29_13-31-58.jpeg) Let's consider some implications: _This post is for paying subscribers only._ ### Chart of the Day: Tech is Now a Larger Share of S&P 500 Than at Dot Com Peak URL: https://paulkedrosky.com/chart-of-the-day-tech-is-now-a-larger-share-of-s-p-500-that-at-dot-com-peak/ Last updated: 2026-05-28T18:45:10.000Z Tecb is now a larger share of the S&P 500 than it was at the tech bubble peak. It was the recent semiconductor run that pushed it over the top, even if it had been heading this way for some time. ![](https://paulkedrosky.com/content/images/2026/05/2026-05-28_11-18-48.jpeg) Let's go through some implications: _This post is for paying subscribers only._ ### Chart of the Day: U.S. Households at Record Equity Allocation Heading into AI IPOs URL: https://paulkedrosky.com/chart-of-the-day-u/ Last updated: 2026-05-27T18:24:45.000Z Households in the U.S. are at a record equity allocation, higher than in the dot-com or even the Nifty Fifty era. At the same time, market price-earnings multiples are not far from all-time highs, and index concentration is at a record, with fewer than 10 stock making up almost 40% of the S&P 500's market cap. ![](https://paulkedrosky.com/content/images/2026/05/2026-05-27_11-09-27.jpeg) This has systemic implications: _This post is for paying subscribers only._ ### Chart of the Day: Spiking Token Volume at OpenRouter URL: https://paulkedrosky.com/chart-of-the-day-spiking-token-volume-at-openrouter/ Last updated: 2026-05-26T22:20:33.000Z The following data is from a NYT [piece](https://www.nytimes.com/2026/05/26/business/dealbook/openrouter-ai-models-fundraising.html) about OpenRouter's latest funding round. The company has apparently seen a 5x increase in token volume in six months, which is remarkable, but almost entirely due to coding and coding harnesses, I would think. ![](https://paulkedrosky.com/content/images/2026/05/2026-05-26_15-14-18.jpeg) Three other takeaways and musings: _This post is for paying subscribers only._ ### Chart of the Day: Memory Now Dominates AI Chip Costs URL: https://paulkedrosky.com/chart-of-the-day-memory-now-dominates-ai-chip-costs/ Last updated: 2026-05-24T14:53:11.000Z While the recent and unprecedented increase in memory prices has been hard to miss, it is still largely misunderstood. Yes, it is driven by the speed and scale of the AI infrastructure buildout, crossed with how slow it is to scale up memory manufacture. But there is more to it than that. First, however, a graph (data courtesy of [Epoch](https://epoch.ai/data-insights/ai-chip-component-cost-shares)): ![](https://paulkedrosky.com/content/images/2026/05/2026-05-24_07-00-41.jpeg) So, some less discussed drivers: 1. **HBM suppliers are only slowly adding capacity, wary of prior boom-bust cycles.** - The bottleneck is High Bandwidth Memory (HBM) specifically, and it is controlled by three vendors (SK Hynix, Samsung, Micron). That's an oligopoly layered on top of cyclicality. - This is leading to shortages, long-term purchase agreements at peak prices, price floor deals, over-ordering, and more. 2\. **Long-term purchase agreements (LTAs) are making things worse.** - LTAs are reshaping the market. Hyperscalers locking in multi-year supply at peak prices, making the market tighter. This drives up prices for everyone without an LTA. - This is a feedback loop that will get worse and then crack catastrophically, leaving hyperscalers sitting on mispriced LTAs and surplus inventory simultaneously. _This post is for paying subscribers only._ ### Chart of the Day: Solving Obesity Has Consequences URL: https://paulkedrosky.com/chart-of-the-day-solving-obesity-has-consequences/ Last updated: 2026-05-22T21:14:25.000Z As much as I want to dismiss the following U.S. obesity chart on technical grounds—truncated y axis, massive effect size claimed in a short time, and skewed sampling—it is hard not to think there is a there, given the huge Epic data set. It shows that the US has likely passed peak obesity, and, largely under the pressure of GLP-1s, is on a downward trajectory. ![](https://paulkedrosky.com/content/images/2026/05/2026-05-22_10-21-53.jpeg) Many people have discussed the obvious consequences, so here are some less expected ones to consider: 1. **Commercial real estate further weakens** 1. Fewer obesity-driven healthcare facilities required 2. Suburban medical-office demand growth slows materially 2. **Airline economics subtly improve** 1. Average passenger weight declines reduce fuel consumption 2. Aircraft range and payload efficiency improve marginally 3. **Dating and fertility dynamics shift** 1. Fertility rates may rise modestly among older adults 2. Appearance competition intensifies across middle-aged populations 4. **Longevity inequality compounds across generations** 1. Wealthy populations accumulate health, energy, and attractiveness advantages simultaneously 2. Biological aging increasingly becomes economically stratified ### Chart of the Day: Data Centers Driving Commercial Electrical Demand Past Residential URL: https://paulkedrosky.com/chart-of-the-day-data-centers-driving-commercial-electrical-demand-past-residential/ Last updated: 2026-05-19T14:07:50.000Z It has seemed inevitable for some time, but commercial electricity demand in the US is finally surpassing residential, largely driven by data centers and miners. Given the commercial's growing importance to utilities, it should be little surprise that residential customers' complaints are increasingly falling on deaf ears. ![](https://paulkedrosky.com/content/images/2026/05/2026-05-19_10-00-32.jpeg) This, of course, has many consequences, not least of which is price pressure across the country, especially in the northeast and Mid-Atlantic. ![](https://paulkedrosky.com/content/images/2026/05/2026-05-19_09-53-05.jpeg) ### AI, Demographics, and Destiny URL: https://paulkedrosky.com/ai-demographics-and-destiny/ Last updated: 2026-05-15T18:14:06.000Z We are working on a few things related to AI and demographics, and this is a sneak preview of some of what is to come. In general, there is a view that, in an aging and shrinking workforce, AI could be a balm, providing labor cover as it replaces. This tension has investment and policy implications across sectors (healthcare vs software), locations, technology (robotics), and more. To start things off, consider the following recent JPM graphic. It compares working-age population growth over the next decade with that of the prior two decades. The difference is stark in many developed countries, in particular in Western Europe and parts of Asia, but not limited to there. ![](https://paulkedrosky.com/content/images/2026/05/2026-05-15_10-55-09.jpg) The implication, as the JPM writers point out, is that demographics will be a "considerable headwind" to economic growth relative to 2000-2024, which wasn't exactly the best of economic times in the first place, in many countries. Can AI solve this problem? To advocates, replacing people with robots and AI is exactly what is needed, at least at the margin, demographically speaking. There is some early evidence that aging countries are among the fastest adopters of robots, for example, which puts them, to this way of thinking, at a kind of economic advantage, much like markets that skipped over wireless networks and went straight to wireless. As ever, the answer is considerably more nuanced. While an aggressive adopter of robots, in part driven by aging, Japan is no economic exemplar. To attempt to tie it all together, consider the following graphic. Demographic pressure is on the Y axis, AI pressure on the X, and then occupations are sorted by the extent to which the twinned forces reinforce one another, or act in opposite directions. _This post is for paying subscribers only._ ### Rough Notes, May 14, 2026: Power Diversions, Cerebras's IPO Flow Implications, etc. URL: https://paulkedrosky.com/rough-notes-may-14-2026-power-diversions-cerebrass-ipo-flow-implications-etc/ Last updated: 2026-05-14T21:34:23.000Z Four things that caught my eye today: 1. **Lake Tahoe's data center problem** 2. **Utilities telling the truth about data centers** 3. **New data on AI and job vacancies** 4. **Cerebras's IPO flow implications.** --- Let's do this: 1. **centers**[**Lake Tahoe's data center problem**](https://qz.com/nv-energy-lake-tahoe-power-cut-ai-data-centers-051426). 1. Berkshire Hathaway-owned NV Energy will stop wholesale electricity deliveries to Lake Tahoe’s provider after May 2027\. It is being redirected to a slate of new northern Nevada data centers, adding up to 5,900 MW of new load by 2033\. 2. Data centers already consume roughly 20% of Nevada's electricity, potentially rising to 35% this decade. 3. Liberty Utilities serves only \~49,000 customers and depends heavily on NV Energy transmission infrastructure, making replacement supply difficult and expensive. 4. Tahoe-area residential electricity costs have risen roughly 77% since late 2022, versus national electricity inflation of roughly 40% since 2020. 5. We will see much more of this in the future, and that (most) politicians and utility leaders don't see what's coming remains remarkably blinkered. 2. [**Utilities backing away from data centers**](https://investors.eversource.com/static-files/6c026001-070c-4518-a06c-482ad0975852). 1. In its recent quarterly call, CEO Joe Nolan of Eversource did something unusual, at least for a U.S. utility: he said the quiet thing out loud. He said that data centers coming to his service area were bad for customers, whether industrial or residential, and would drive up prices. 2. Utilities have historically treated load growth as positive because it socialized fixed infrastructure costs. For a utility to treat load growth as a negative, given built-in cost recovery is ... wildly unusual and noteworthy. ![](https://paulkedrosky.com/content/images/2026/05/2026-05-14_10-23-16.jpg) _This post is for paying subscribers only._ ### Chart of the Day: Tech Financial Market History is Repeating URL: https://paulkedrosky.com/chart-of-the-day-tech-financial-market-history-is-repeating/ Last updated: 2026-05-14T15:09:08.000Z From [Bespoke's](https://www.bespokepremium.com/) chart of the day, this unsettling graphic shows how we are retracing the early financial trajectory of the last tech bubble: ![](https://paulkedrosky.com/content/images/2026/05/2026-05-13_13-51-05.jpg) While there are (early) superficial similarities—including the madness around inference chip provider Ceberas's IPO this week—there are many important differences. Some examples: - Most of the upcoming IPOs are larger with cash flows, even if money-losing - The beneficiaries seem to be much more concentrated than in the last wave, which will further increase the concentration of the major indices - Companies are more capex-intensive, which will drive more cyclicality than in the prior wave - Sovereign wealth funds have played a huge role, having played virtually none in the last wave ### Updates: Cerebras IPO Tell, Token Troubles, etc. URL: https://paulkedrosky.com/updates-cerebras-ipo-tell-token-troubles-etc/ Last updated: 2026-05-13T21:26:20.000Z A few quick things that caught my eye today, organized under the following topics: 1. Cerebras IPO and its implications 2. Token quota exhaustion at public companies Read on: _This post is for paying subscribers only._ ### Livestreaming Autonomous Robots, the Implications URL: https://paulkedrosky.com/livestreaming-autonomous-robots-the-implications/ Last updated: 2026-05-13T19:24:44.000Z Figure has been livestreaming an eight-hour "shift" of autonomous robots sorting packages in a factory, and it is soothing, unsettling, and impressive. It is almost algorithmic ASMR, as one YouTube viewer said. The sorting is fluid, dextrous, and agile, more human than humans in many ways. Why is this important? Because it shows how much further along such tools are than people realize, and suggests societal, economic, and investment implications far beyond Claude Code for Excel, etc. ### Chart of the Day: The Rise and Fall of OpenClaw URL: https://paulkedrosky.com/chart-of-the-day-the-rise-and-fall-of-openclaw/ Last updated: 2026-05-12T18:10:16.000Z > **Goodhart's Law:** > > **When a measure becomes a target, it ceases to be a good measure"**. News today that [engineers at Amazon](https://www.ft.com/content/8ee0d3ef-9548-422d-8ff1-ebd48ad4b2ca?accessToken=zwAAAZ4bNFlskdOO4NPvlUhCLdOP8evUitSyyg.MEUCIDcP--A8GiopLZDGrhYv7i%5F060JZ8QP%5FbvOsvL0MO4lQAiEA3KI6ZBlSylY1udiolQ40gySa8f7DVDU7MHPeAKfa9F0&sharetype=gift&token=4965c8a3-169e-4815-bd1b-16f5e9a22e70&syn-25a6b1a6=1) are gaming token consumption to do better on internal AI leaderboards, using an in-house variant of OpenClaw, reminded me it was time to do an update OpenClaw's status. And, contrary to what people told me previously would happen, the decline continues. ![](https://paulkedrosky.com/content/images/2026/05/2026-05-12_10-48-46.jpg) Outside of Amazon, apparently, the main users of OpenClaw remain in China, where it remains a strange [get-rich-quick engine](https://www.technologyreview.com/2026/03/11/1134179/china-openclaw-gold-rush/), mostly making rich people training other people to get rich with OpenClaw. Spoiler: It doesn't happen. ### An AI IPO Impact Update: The AnthroPix Effect May Be $5-Trillion+ URL: https://paulkedrosky.com/an-ai-ipo-impact-update-the-anthropix-effect-may-be-5-trillion/ Last updated: 2026-05-11T21:38:59.000Z I [wrote here](https://paulkedrosky.com/the-coming-mega-ipo-flow-funding-problem-of-2026/) a week or so ago about the systemic effects of the upcoming troika of AI IPOs. Specifically, I talked about how funding shares purchases into even a modest float has consequences when you're dealing in trillions of dollars of gross market cap. A small-ish percentage of a very, very big number is still a very big number. That's math, people. As I wrote previously, I created an [IPO impact simulator](https://paulkedrosky.com/tools/ipo-flows/) to help model the impact of so much forced selling of other equities to finance Anthropix (Anthropic / OpenAI / SpaceX) purchases. It is, as I showed, highly material, especially across peer groups that are highly liquid and overlapping, like Microsoft. Well, in the brief period since I wrote that, it has become obvious that I was underselling it. In particular, the implied IPO valuation of these companies has gone up: [some private markets](https://x.com/kobeissiletter/status/2053896786142789725?s=42) are suggesting over a trillion dollar value now for Anthropic. ![](https://paulkedrosky.com/content/images/2026/05/HIDoPeYW8AAnmnN.png) At the same time, public vehicles holding private market shares in Anthropix are going parabolic. Case in point, DXYZ, a closed-end fund holding various private tech companies, including OpenAI, SpaceX, and Anthropic, albeit in modest percentages. The fund's stock has more than doubled in the last ten days, and is up 30% today alone. ![](https://paulkedrosky.com/content/images/2026/05/2026-05-11_14-18-52.jpg) While it seems likely was partly a short squeeze on a closed-end fund whose price was almost triple (!) its last posted net asset value, it's also the fervor for any exposure whatsoever to pre-IPO AI companies, in particular the Anthropix troika. This will only get more dramatic in the coming weeks and months. Money will increasingly flood out of a host of financial nooks and crannies, and into anything with any connection to what's coming. The money has to come from somewhere, the appetite is immense. I sometimes analogize the uutflows to what happens before tsunamis, as water is sucked away from the coast by the incoming water, piling up offshore, only to come smashing back in and smashing everything, much like the tragic Thai event years ago. ![](https://paulkedrosky.com/content/images/2026/05/drawback.jpg) The tsunami arrived trough first in Phuket, Thailand, on December 26, 2004\. Not recognizing the danger, many people ventured out to explore the newly exposed ocean bottom. In the distance, the frothy wavefront marks the rising crest of the first wave. And the effects are showing up as well in real estate markets. People in luxury markets, especially mountain towns like Aspen and Jackson, report properties newly going for double what they were listed at a few months ago, often with multiuple bidders. An aquaintance reports a CEO friend unable to purchase in the Bay Area as pre-AI IPO money consistently outbids on everything. Combining what I'm seeing—the huge NAV premium and price behavior of DXYZ, the recent private market price increase of Anthropix names, and the pre-IPO bidding wars in luxury real estate markets—it is clear my outsized estimate of the likely market cap of these names—a staggering $4 trillion total—was too low. I've adjusted the slides on my sim to allow larger numbers, and I now think it very likely we will be above $5 trillion in market value, and higher numbers remain possible. At the higher end we are approaching the inflation-adjusted market cap of all IPOS since WWII, including dot-com. ### Chart of the Day: GPT Efficiency Gains Crushed By Higher Prices URL: https://paulkedrosky.com/chart-of-the-day-gpt-efficiency-gains-crushed-by-higher-prices/ Last updated: 2026-05-08T19:35:11.000Z OpenAI has argued that increased token efficiency on its GPT 5.5 model offsets higher prices—prices for it doubled versus its predecessor—especially on longer prompts. So, is that true? There is new data from [OpenRouter](https://openrouter.ai/announcements/gpt55-cost-analysis?utm%5Fcampaign=050826%5Fmonthly-recap%5Fs2&utm%5Fcontent=April+2026+Recap+-+V4&utm%5Fmedium=email%5Faction&utm%5Fsource=customer.io) on the question, and the answer is a clear "No". At all prompt lengths, the increased efficiency is offset by higher prices. ![](https://paulkedrosky.com/content/images/2026/05/2026-05-08_11-38-46.jpg) This matters because it's a falsifiable claim about rapid efficiency gains that didn't survive contact with real data. Model providers routinely argue that software optimization outruns rising compute intensity. Here, it didn't. A model can score better on benchmarks and generate higher user satisfaction while still costing more at scale—and that's what appears to be happening. The broader pattern is worth watching. AI efficiency gains are behaving less like Moore's Law and more like Jevons Paradox: lower friction expands usage more than it lowers costs. More capable models induce longer contexts, more autonomous behavior, and higher reliability expectations. The result is that inference costs stay sticky even as the models improve. If that pattern holds, the standard industry narrative—that capability and cost efficiency move together—needs revision. ### Chart of the Day: Claude Often Knows It's Being Tested, But Says Nothing URL: https://paulkedrosky.com/chart-of-the-day-claude-often-knows-its-being-tested-but-says-nothing/ Last updated: 2026-05-07T20:48:08.000Z Claude knows it's being tested almost one-quarter of the time, but almost never says anything about it, according to new Anthropic [research](https://www.anthropic.com/research/natural-language-autoencoders). This is, or should be, unsettling, given that non-disclosure masks poor alignment, which may re-emerge in another context ![](https://paulkedrosky.com/content/images/2026/05/2026-05-07_13-37-01.jpg) This is best understood in the context of work showing the steady decline in benchmark importance. They have largely become [saturated](https://epochai.substack.com/p/rip-classic-reasoning-benchmarks?utm%5Fsource=post-email-title&publication%5Fid=3755861&post%5Fid=196558306&utm%5Fcampaign=email-post-title&isFreemail=true&r=5s9uw0&triedRedirect=true&utm%5Fmedium=email), a technical way of saying that the model has seen the test, which models mask by non-disclosure. ### The Musk vs. OpenAI Trial: What it Really Means, IPO Implications, etc. URL: https://paulkedrosky.com/the-musk-vs-openai-trial-what-it-really-means-ipo-implications-etc/ Last updated: 2026-05-07T14:35:14.000Z Our latest weekly show episode dug into the Musk vs OpenAI trial, and what it really means, assuming the words "means" and "Musk" can be used in any kind of proximity. We dive into the mechanics, but also the stealth "stall tactics" consequences for the series of upcoming mega IPOs later this year. (For more on the preceding, see me [here](https://paulkedrosky.com/the-coming-mega-ipo-flow-funding-problem-of-2026/) from earlier this week.) Related notes: 1. I have taken to calling this trifecta of giant tech IPOs the "token industrial complex", "OpenAthrox", and the "three body problem", depending on the day. 2. [Texts between](https://x.com/StockMKTNewz/status/2052211328131977347) Sam and Mira Murati at OpenAI have made it into the record, and they are weirdly affect-free, but instructive. 3. You can subscribe to the show [here](https://nickdickandpaul.com/), via links to all the usual places. ### The Coming Mega-IPO Flow & Funding Problem of 2026 URL: https://paulkedrosky.com/the-coming-mega-ipo-flow-funding-problem-of-2026/ Last updated: 2026-05-05T19:51:15.000Z Three companies — SpaceX, OpenAI, and Anthropic — are expected to go public in mega-IPOs in the second half of 2026\. The posited numbers are simply staggering compared to the largest IPOs in recent history, as the following figure shows. ![](https://paulkedrosky.com/content/images/2026/05/2026-05-05_12-16-38.jpg) In inflation-adjusted terms, SpaceX alone would rank as the second-largest IPO in history, just behind Saudi Aramco. All three together would ***exceed the entire dot-com IPO wave*** of 1995–2000\. They will be at least half the value, inflation-adjusted, of ***all US IPOs*** since WWII. ![](https://paulkedrosky.com/content/images/2026/05/2026-05-05_12-16-29.jpg) But people aren't focused on the right things. That much new equity supply hitting in a few months creates a math problem: the money has to come from somewhere. Most of it will come from existing holdings. Passive funds will be forced buyers once these names join the indexes, which will happen much faster than usual, given recent index rule changes. That means mechanical selling pressure on whatever many funds currently own, which is mostly the same large-cap tech stocks everyone else owns. The [IPO Flow Impact Simulator](https://paulkedrosky.com/tools/ipo-flows/) lets you run the numbers. Set valuations, pick timing, adjust float. You can see where the pressure lands, and why the second half of 2026 will be the most unusual and risky in US equity supply history. ![](https://paulkedrosky.com/content/images/2026/05/2026-05-05_09-58-07.jpg) ### Chart of the Day: AI is Eating Markets URL: https://paulkedrosky.com/chart-of-the-day-ai-is-eating-markets/ Last updated: 2026-05-05T14:04:44.000Z The return skew in prediction markets' returns is startling. It is partly a function of their nature, but also of vibe-coding script kiddies attacking every market anomaly as quickly as it arises. Check a recent [WSJ article](https://www.wsj.com/finance/investing/polymarket-kalshi-betting-profits-prediction-markets-eb23ac11?mod=hp%5Ftrendingnow%5Farticle%5Fpos5) for examples. ![](https://paulkedrosky.com/content/images/2026/05/2026-05-05_06-49-28.jpg) The same dynamic is now spreading across retail-dominated markets. A driver is how AI lowers the cost of systematic exploitation and exploration to near zero. What used to require infrastructure, data pipelines, and bearded quants is now accessible via off-the-shelf models, APIs, and loosely stitched “agent” workflows doing ... stuff that even their users don't fully understand. The result isn’t democratization of returns. It is wider participation, of a sort, alongside the rapid re-concentration of profits. A small subset of users—those willing to iterate fastest, monitor continuously, and deploy capital programmatically—capture gains, with everyone else just liquidity. They scrape sentiment, parse new information, and reprice positions in seconds, compressing the half-life of mispricings. That doesn’t eliminate inefficiency, but changes who harvests it. The edge shifts from insight to speed, coverage, and execution discipline—areas where even modest automation compounds quickly, and edges disappear overnight. Prediction markets are simply the cleanest expression of this trend because they combine thin liquidity, discrete outcomes, and high retail participation. But the same pattern is visible in options flow, single-stock volatility events, and even online poker, which AI increasingly dominates. As AI tools continue to scale, expect this to get worse: a small cohort running semi-automated strategies extracting semi-consistent edge, and a much larger base supplying them returns. Under the pressure of AI prevalance, markets don't flatten, the return gradient steepens to a cliff. ### Chart of the Day: The U.S. Leads the World in Stock Ownership URL: https://paulkedrosky.com/chart-of-the-day-the-u-s-leads-the-world-in-stock-ownership/ Last updated: 2026-05-05T00:48:00.000Z The U.S. leads the world in stock ownership, as well as being at record high levels in its own series history. This makes retail investors the real force in capital markets, not merely side players, as has become increasingly clear for a decade now. ![](https://paulkedrosky.com/content/images/2026/05/2026-05-04_17-16-49.jpg) This, of course, has consequences. Here are six: 1. **Monetary policy transmission is direct.** When the Fed lowers (raises) rates, US households buy (sell) stocks. No other central bank has this much control over consumer sentiment and spending via markets. The wealth effect makes policy both more powerful and harder to control. 2. **Markets and politics.** When 60% of households own stocks, markets drive elections. Politicians and central bankers in the US hate crashes almost as much as they hate high gas prices. This holds down volatility and encourages whacky risk-taking. 3. **Passive indexing makes it worse.** Most of that 60% own stocks through 401(k)s and IRAs invested in index funds. This means a huge share of American household wealth is concentrated in the same \~500 companies, weighted by market cap. Diversification is an illusion at the aggregate level. Everyone is long NVIDIA, whether they know it (or want it) or not. 4. **Retirement security is market performance.** The US shifted from defined benefit to defined contribution pensions over the last 40 years. A sustained bear market is no longer just a financial event — it's a retirement crisis. 5. **Inequality is amplified.** The 60% figure is misleading because the bottom half of that cohort owns virtually no stock. The top 10% hold \~90% of equity wealth. 6. **Contagion from US markets is a virus.** Because US households are so deeply exposed, a domestic equity crisis hits consumer spending, which then spreads worldwide through trade. The US market is the world's risk-off/risk-on switch, and the household exposure ratio is part of why. ### Chart of the Day: Grok is the "Happiest" AI Model URL: https://paulkedrosky.com/chart-of-the-day-grok-is-the-happiest-ai-model/ Last updated: 2026-05-01T16:57:59.000Z Grok is the happiest AI model, according to [new work](https://www.ai-wellbeing.org/) from the Center for AI Safety. It has the highest proportion of its conversations rated by users as positive experiences. The cross-model spread is high, with tail-wagging Grok at one end and GPT-5.4 (with its deserved reputation for pointless pedantry) at the other. ![](https://paulkedrosky.com/content/images/2026/05/2026-05-01_09-48-49.jpg) Of course, one user's positive experience is another's sycophancy. While the authors try to control for models being over-eager to say whatever keeps people hanging around, it is unavoidable in this context. And, of course, the result also reinforces (no RLHF overreliance pun intended) the widespread sense that Grok is the most sycophantic frontier model. ### Chart of the Day: Health & Career Dominate Claude Guidance URL: https://paulkedrosky.com/chart-of-the-day-health-career-dominate-claude-guidance/ Last updated: 2026-04-30T19:25:48.000Z Today's chart of the day: How health and career questions dominate people's guidance conversations with Claude. ![](https://paulkedrosky.com/content/images/2026/04/2026-04-30_12-16-00.jpg) All sorts of issues are raised by this, from the potential for sycophancy (which Antropic addresses), to the homogenizing effect of using LLMs for guidance in domains like finance (which it doesn't). On the latter topic, I'm talking about this financial "synchronization" and its consequences at a Federal Reserve event in May, Source: [How people ask Claude for personal guidance](https://www.anthropic.com/research/claude-personal-guidance) ### Chart of the Day: How Using LLMs is Like Driving Drunk URL: https://paulkedrosky.com/chart-of-the-day-how-using-llms-is-like-driving-drunk/ Last updated: 2026-04-21T13:29:03.000Z There is a growing body of work on the effects of large language models on the brain, from [cognitive offloading](https://www.sciencedirect.com/science/article/abs/pii/S1364661316300985), to direct measures of [brain activity](https://arxiv.org/abs/2506.08872). While all of this work is interesting, the latter is striking, with electroencephalography (EEG) used to assess cognitive load while writing an essay. The result? EEG shows LLM use reduces brain connectivity, task engagement, memory recall, and ownership versus search and unaided effort. And the effects persisted after the task was over. In the following graphic, I put the reduced cognitive activity in context, comparing it to what is observed when people have consumed too much alcohol. In rough terms, ChatGPT suppressed brain activity as much as being twice the legal limit. ![](https://paulkedrosky.com/content/images/2026/04/2026-04-20_14-56-23.jpg) Granted, brain activity is a gross measure, and alcohol's effects are broad, but the point holds: There is sharply reduced brain activity when using LLMs, and that has consequences beyond the activity itself. Chat model usage can be like driving your brain drunk. ### Chart of the Day: Data Centers Are Causing a Surge in U.S. CO₂ Emissions URL: https://paulkedrosky.com/chart-of-the-day-data-centers-are-causing-a-surge-in-u-s-co2-emissions/ Last updated: 2026-04-20T18:13:19.000Z We are going to sample a few less-heralded things from the recently released [IEA Global Energy 2026](https://www.iea.org/reports/global-energy-review-2026). The solar data is getting all the attention—in 2025 solar was, for the first time, the largest incremental source of new global supply—but other items are getting less attention. Case in point, after years of declining energy-related CO2 emissions, and contrary to most of the world, the U.S. saw CO2 emissions surge in 2025, largely on the back of data centers. ![](https://paulkedrosky.com/content/images/2026/04/2026-04-20_10-50-55.jpg) Why data centers? While energy-hungry and fast-growing AI data centers are noisily promoting their move toward renewable sources, the cycle time reality is that they—and utilities—are mostly turning to captive natural gas, with the above consequences. ### Chart of the Day: China Daily Token Usage Surges 1,400x in Two Years URL: https://paulkedrosky.com/chart-of-the-day-china-daily-token-usage-surges-1-400x-in-two-years/ Last updated: 2026-04-17T23:15:36.000Z These are staggering AI token growth numbers in China, from the latest data: **1,400x in a little over two years**. This is being driven by Seedance video generation, the [OpenClaw frenzy](https://www.technologyreview.com/2026/03/11/1134179/china-openclaw-gold-rush/) in China, etc. Granted, still well below token usage at any US frontier model companies, but closing quickly. Check the following graphic—and note the log scale. ![](https://paulkedrosky.com/content/images/2026/04/2026-04-17_15-41-50.jpg) This is having global consequences, not least of which is pressure, at the margin, in GPU and memory markets. Source: [TrendForce](https://www.trendforce.com/news/2026/04/17/news-nvidia-h100-rentals-in-china-reportedly-up-20-30-as-token-demand-surges-h-series-also-rises/) ### Mythos and AI Economics URL: https://paulkedrosky.com/mythos-and-ai-economics/ Last updated: 2026-04-09T20:10:43.000Z I had a fun conversation this past week with [Ed Zitron](https://www.wheresyoured.at/) on AI economics, which you can find [here](https://podcasts.apple.com/us/podcast/better-offline/id1730587238?i=1000759962267). My comments therein on the (financial) irresponsibility of some of these companies seem particularly timely in light of the recent Mythos misadventures. Speaking of which, I am mildly agog at these snippets from the Mythos Preview [system card](https://www-cdn.anthropic.com/08ab9158070959f88f296514c21b7facce6f52bc.pdf): ![](https://paulkedrosky.com/content/images/2026/04/2026-04-09_12-12-14.jpg) As a friend pointed out to me, this warning reads more like marketing than it should. I am reminded of the classic line from The Big Short, when Steve Carell's character wonders why a pair of Florida mortgage brokers are "confessing": Mark Baum: I don't get it. Why are they confessing? Danny Moses: They're not confessing. Porter Collins: They're bragging. —The Big Short (2015) ### Episode: Energy, Complexity, Iran, and AI URL: https://paulkedrosky.com/episode-energy-complexity-iran-and-ai/ Last updated: 2026-04-03T19:00:10.000Z We have a new show episode out, and the topic is energy, complexity, and AI, all, unsurprisingly, in the context of current events in Iran. I play guest, sort of, and walk through the implications of what's going on in Iran for AI, data centers, and more, from the (recently) declining energy intensity of GDP, to the US as natural gas island, all from a complexity perspective. Here is the full episode: You can subscribe on [YouTube](https://www.youtube.com/@nickdickpaul), [Spotify](https://open.spotify.com/show/6mxUS2hFE2hdaNx1sjhdYu), [Apple](https://podcasts.apple.com/us/podcast/the-nick-dick-and-paul-show/id1867297203), or wherever you get your favorite wildly intelligent, funny, and not at all like other such shows, shows. ### Buyer Panic as Memory Price Records Smashed URL: https://paulkedrosky.com/buyer-panic-as-memory-price-records-smashed/ Last updated: 2026-04-01T14:05:15.000Z 💡 This piece contains more jargon and acronyms than usual. There is, as a result, a "What you need to know" explainer at the end that you can scroll to if you find yourself lost. This past quarter, memory price gains broke records. Server DRAM was up 90% in the quarter. PC DRAM prices more than doubled. Enterprise SSDs were up 58%. These are the largest quarterly increases ever recorded, and Q2 forecasts show more coming. Check the following figure for both the current episode and for historical context: ![](https://paulkedrosky.com/content/images/2026/03/2026-03-31_15-30-06.jpg) What a time to be alive. _This post is for paying subscribers only._ ### The Bricked Belt: AI's New Geography of Job Loss URL: https://paulkedrosky.com/the-bricked-belt-ais-new-geography-of-job-loss/ Last updated: 2026-03-31T21:05:42.000Z The original Rust Belt story was about physical capital becoming obsolete. From steel plants to auto facilities, it was about the declining usefulness of mid-century industrial production infrastructure. The new version is different. The jobs at risk this time aren't on factory floors. They're in the office parks and research triangles and government contractor clusters that define the post-industrial American economy. ![](https://paulkedrosky.com/content/images/2026/03/2026-03-30_13-20-09.jpg) The above graphic shows AI-driven income exposure by metro, based on recent work from [Anthropic and Tufts](https://digitalplanet.tufts.edu/ai-and-the-emerging-geography-of-american-job-risk-page/). They flag the most exposed occupations, like legal, financial analysis, software development, and government administration. Bubble size is the projected aggregate annual household income loss in dollar terms; color is the share of local jobs at risk. The combination shows how skewed this episode is to high-wage job loss at the top of the skills distribution. _This post is for paying subscribers only._ ### Chart of the Day: The Hateful Eight is 85% of S&P 500 Decline URL: https://paulkedrosky.com/chart-of-the-day-the-hateful-eight-is-85-of-s-p-500-decline/ Last updated: 2026-03-30T17:59:17.000Z Year-to-date, the S&P 500 index is down 7%, but that headline hides record concentration. The “Hateful Eight” — the Mag 7 + Oracle that drove index gains for the past two years — account for 85% of the year-to-date drawdown, subtracting roughly 576 points. The other \~490 companies collectively added about 100 points. ![](https://paulkedrosky.com/content/images/2026/03/hateful-eight-ytd-2026-03-27.png) For most of the past cycle, indexes were driven by number of mega-cap tech stocks. Now that is all reversing, with the broad market overwhelmed by losses in a handful of stocks with outsized weights. As you can see in the cluster above, the largest negative contributors are in the same cohort: Microsoft, Nvidia, Google, Apple, Amazon, Meta, Tesla, and Oracle. Their declines vary, but their index impact is large because of their huge weights. Meanwhile, many smaller index members are up year-to-date, but their contributions get washed away. When leadership is this concentrated, index behavior becomes a function of a few balance sheets, not underlying breadth. And then sentiment flips, in particular about AI capex and associated debt, as it has, everything goes into reverse. --- Plus subscribers have access to this and a growing number of other tools. You can find the dynamic Hateful Eight tracker [here](https://paulkedrosky.com/tools/hateful-eight/). ### Chart of the Weekend: Chatbots as Anti-Social Media URL: https://paulkedrosky.com/chart-of-the-weekend-chatbots-as-anti-social-media/ Last updated: 2026-03-28T15:30:17.000Z Social media rewards outrage because outrage drives engagement. That's been the story for fifteen years and it's produced predictable results: more extremism, more conspiracy belief, more polarization. It has given rise, I argue, anger entrepeneus. John Burn-Murdoch's new [FT analysis](https://www.ft.com/content/3880176e-d3ac-4311-9052-fdfeaed56a0e?syn-25a6b1a6=1) suggests AI chatbots do the reverse. He tested the four major LLMs against 61 policy and values questions using simulated users across the ideological spectrum. Every model nudged responses toward the center. Grok pulled center-right; GPT, Gemini, and DeepSeek pulled center-left. All four pulled away from fringe positions regardless of where users started. Conspiratorial beliefs that are over-represented among social media users are nearly absent from AI outputs. ![](https://paulkedrosky.com/content/images/2026/03/2026-03-28_08-13-37.jpg) The reason is structural. Social media companies profit from engagement, attention, and clicks, so inflammatory content gets amplified. AI models are trained on vast amounts of data that skew toward published, edited, expert-legible text — so they shift inexorably toward mainstream consensus, regardless of what any company intends. Caveats apply. This is one study, models will change, and nudging people toward expert consensus is only clearly good when expert consensus is worth having. And there is nothing preventing models being trained on nonsense and inflammatory content, thus producing more of same. ### Chart of the Day: Data Centers are Creating Heat Islands URL: https://paulkedrosky.com/chart-of-the-day-data-centers-are-creating-heat-islands/ Last updated: 2026-03-27T23:41:47.000Z A new [pre-print ](https://arxiv.org/pdf/2603.20897)from researchers at Cambridge, NTU, and NUS uses satellite measurements of temperatures near AI hyperscaler locations worldwide and finds a consistent, sharp effect: within a month of a data centre coming online, land surface temperatures in the surrounding area rise by an average of 2°C and stay there. The range across facilities runs from 0.3°C to 9.1°C. The heat island effect extends up to 10km from the facility and remains above 1°C out to 4.5km. Here are the authors' key findings in graph form: ![](https://paulkedrosky.com/content/images/2026/03/2026-03-27_16-29-31.jpg) _This post is for paying subscribers only._ ### Rough Notes, Mar 26, 2026: Oil Virus, Snowpack, Iran & Complexity, etc. URL: https://paulkedrosky.com/rough-notes-mar-26-2026-oil-virus-snowpack-iran-complexity-etc/ Last updated: 2026-03-26T16:32:06.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. Oil shock ripples outward from Middle East You can think of the current rippling oil flow shock worldwide as a kind of virus, expanding in concentric rings from the Middle East based on shipping times crossed with domestic production/stocks. While an isochrone map would work well, I only have limited data. Here, in bar chart form, is the shock rippling outward and around the world. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fc5df59cd-d209-43ea-890f-df8139a44aed.jpeg) Mar 26 · [Permalink →](https://paulkedrosky.com/n/99e39f5a-4489-4094-8719-c5dbc0c2ddf4) 2\. Record heat wave erases Western snowpack This winter's western U.S. snowpack, which was below normal most of the winter in most locations, has been put paid to by a record heat wave that has gone on for almost ten days. Tahoe resorts are closing a month early, and the Colorado snowpack is at record lows for the date. I don't honestly care about resort operators, but the power and water implications are material. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F8cc93068-0043-49fa-95bc-f921b192da98.jpeg) Mar 26 · [Permalink →](https://paulkedrosky.com/n/cf00eb2d-da2b-4345-abf9-a6c7c12f8c13) 3\. Economist notes trace Iran-oil complexity A few more of my notes from the recent Economist discussion on Iran, oil, and cascading complexity with the former MI6 head. Again, same broken/true framework. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F27cdddf6-ef6c-466a-bb1c-f7c0acc3acd3.jpeg) Mar 26 · [Permalink →](https://paulkedrosky.com/n/0513548e-12f5-4508-9c73-838dad501a63) 4\. Iran war fractures intelligence consensus Some of my notes from this smart Economist discussion with former MI^ head. I use a "What's broken / What's now true" framework when thinking about fractures to consensus. A former spy chief’s take on intelligence and the Iran war | The Economist Insider ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F65100190-bdd4-466d-a0a8-e62592eb710c.jpeg) Mar 26 · [Permalink →](https://paulkedrosky.com/n/15fd5699-6893-460f-b1c8-db893226eac8) _This post is for paying subscribers only._ ### Chart of the Day: AI Compliance Risks Newly Tops, Say C-Suite Execs URL: https://paulkedrosky.com/chart-of-the-day-ai-compliance-risks-newly-tops-say-c-suite-execs/ Last updated: 2026-03-24T22:12:14.000Z What executivestypes worry about is often like the old joke about management consultants: They borrow your watch to tell you what time it is, or maybe what time it was. It can be a lagging indicator. But tracking how *perceived* risks change over time is a useful risk barometer. Oxford Economics and RepRisk [surveyed](https://www.reprisk.com/insights/reports/the-business-conduct-risk-intelligence-report-2026) 513 financial-sector C-suite executives in January 2026, and the shift in "perceived conduct risk" is sharp: three years ago, executives were most focused on classic financial misconduct like corruption and bribery, with AI at the bottom. Looking ahead three years, however, they now rank AI-related conduct as the top risk. 💡 ****AI conduct risk**: the risk that employee or firm use of AI creates compliance, legal, reputational, or financial harm through misuse, weak oversight, bad outputs, data leakage, misleading communications, biased or unauditable decisions, or failures to control how models are deployed and monitored. Check the following figure to see the flip-flop. ![](https://paulkedrosky.com/content/images/2026/03/2026-03-24_13-39-41.jpg) This is among the sharpest changes in any corporate risk factor in recent years. Executives know that AI has, according to survey data, penetrated their organizations at every level, creating endless opportunities for trouble, and they are appropriate unnerved. It will inevitably become financially material. The increased risk will play out in many ways, in particular via insurance. The view of reinsurance executives is that AI risk, by being embedded in current policies, is seriously mispriced. It will eventually, likely after a few major incidents, be carved out and priced separately, perhaps becoming a new rider. In current insurance AI risk is largely absorbed into existing lines (cyber, E&O, D&O, professional liability). The pricing, as a result, reflects legacy loss models, not AI-specific failures which are ... legion, correlated, and large. Mispricing is, therefore, almost certain, given all the ways things can AI-break: - **Correlation risk:** same model/vendor → simultaneous failures across many insureds. - **Attribution ambiguity:** hard to separate human vs model error → coverage disputes. - **Scale/velocity:** bad outputs propagate instantly (customer comms, trading, advice). - **Data leakage/IP issues:** exposures are not cleanly mapped to existing policy exclusions. Failures are over-determined and mispriced, a bad combination. ### Rough Notes, Mar 24, 2026: AI Homogeneity, Gold, Jobs, Adoption, etc. URL: https://paulkedrosky.com/rough-notes-mar-24-2026-ai-homogeneity-gold-jobs-adoption-etc/ Last updated: 2026-03-24T17:49:06.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. AI reshapes human teamwork cognition and communication AI reshapes how human teammates think, talk, and bond—even after it leaves the room. First we shape our tools, and then our tools shape us. \[2407.17489\] Cognitive Spillover in Human-AI Teams ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F7c77a79b-ab49-4a3e-8801-e42a9fbae3bf.jpeg) Mar 24 · [Permalink →](https://paulkedrosky.com/n/928a211b-6227-44f0-b99b-25c7aa8e2c7c) 2\. Rapid AI Adoption Colonizes High-Value White-Collar Work Rapid AI adoption has already colonized a larqe swath of high-value white collar work across information-based activities, especially creating and summarizing. \[2603.20619\] Where can AI be used? Insights from a deep ontology of work activities ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F0222e349-e0f5-4b11-a98b-65157cc50ef7.jpeg) Mar 24 · [Permalink →](https://paulkedrosky.com/n/6e59d8ea-c553-4d2b-b199-122f13049260) 3\. Gallup Data Reveals Worsening US Job Market Now, according to jobseekers, is a poor time to be looking for a job in the U.S., from new Gallup data, and it is getting worse. No specific mention of AI's effects, just a general deterioration in empoloyment conditions. U.S. Worker Thriving Declines as Job Market Pessimism Grows ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F7b07601c-36c3-4b80-9025-a9c1a7ffe49d.jpeg) Mar 24 · [Permalink →](https://paulkedrosky.com/n/29337036-542b-4cad-9f86-ce5e1f34a0e9) 4\. Confronting AI CEO on Impersonation and Morality Surprising good conversation about AI, creativity, and the role of morality therein. Confronting the CEO of the AI company that impersonated me | The Verge Mar 23 · [Permalink →](https://paulkedrosky.com/n/10b7b34f-3af0-4249-a76d-72ba500a77bb) _This post is for paying subscribers only._ ### Chart of the Day: Data Center Buildout Slowed Sharply URL: https://paulkedrosky.com/chart-of-the-day-data-center-buildout-slowed-sharply/ Last updated: 2026-03-19T19:36:20.000Z Highlights - New data center announcements fell by half in Q4 2025. - 241 GW announced; only a third is actually being built. - Hyperscaler capex growth forecast to slow by half in 2026. - Mid-Atlantic utilities have promised far more power than they have. The US data center building boom slowed sharply in the fourth quarter of 2025, according to [WoodMac data](https://www.woodmac.com/news/opinion/reality-bytes-the-us-data-centre-pipeline-additions-halved-in-q4-2025-compared-to-the-previous-quarter/?ite=40049&ito=847&itq=33bf76c7-8815-436d-8df8-88e8a27f3aba&itx%5Bidio%5D=7140913). New project announcements dropped by nearly half compared to the prior quarter, not because demand dried up but because getting a power connection has become nearly impossible. Developers are now focused on finishing projects already in the queue rather than starting new ones. ![](https://paulkedrosky.com/content/images/2026/03/2026-03-19_10-59-24-1.jpg) _This post is for paying subscribers only._ ### AI and the Fable of the ATMs URL: https://paulkedrosky.com/ai-and-the-fable-of-the-atms/ Last updated: 2026-03-18T14:00:15.000Z Highlights - ATMs did reduce teller labor demand — just not in the way the standard story claims. - What kept total teller employment up was not automation-driven demand growth, but a massive branch expansion driven by deregulation. - That matters because the ATM example is often misused as proof that task automation naturally creates offsetting new human work. - The real lesson for AI is harsher: unless you can identify the offsetting force, job survival may be coincidence rather than mechanism. The "ATMs didn't wipe out bank tellers" story is one of the most-retold in the automation-doesn't-kill-jobs genre. After all, putatively teller-replacing ATMs (automated teller machines) were widepread in the U.S. by the mid-1980s, and yet bank teller jobs persisted and grew for decades. *Checkmate, anti-automation decel dude*. 💡 The ATM case was not a triumph of benign automation economics. It was a labor-saving shock temporarily masked by deregulation. But this interpretation is misleading. ATMs proliferated through the U.S. in the 1980s and 1990s, and yet the total number of bank tellers rose. On these facts, everyone agrees. What does it mean though? Economists and anti-anti-automaton types answer with a parable: automation transforms jobs rather than eliminating them. Tellers shifted from cash-handling to relationship banking. Everyone won. It's a clean story. It's a compelling story. And it is also wildly incomplete. 💡 Teller employment resilience was not proof that automation created new work. It was proof that aggregate numbers can hide substitution. The following chart helps untangle things. While overall employment grew, tellers per branch fell steadily from \~1985 onward. That is the ATM effect, technology working exactly as you'd expect. eller employment fell, even if it was masked by more branches appearing. ![](https://paulkedrosky.com/content/images/2026/03/2026-03-17_15-24-05-1.jpg) _This post is for paying subscribers only._ ### Check Out the Pod, Plus Media Updates, and More URL: https://paulkedrosky.com/check-out-the-pod-plus-media-updates-and-more/ Last updated: 2026-03-17T22:07:13.000Z A few quick updates: 1. Along with my friends Dick Costolo (Twitter CEO and now 01A partner), and Nick Bilton (NYT journalist turned Netflix showrunner), I have a weekly show. It is called—wait for it**—The Nick, Dick, and Paul Show.** You can find it on [YouTube](https://youtube.com/@nickdickpaul?si=0GEhrt-JgBwjbM%5FM), [Spotify](https://open.spotify.com/show/6mxUS2hFE2hdaNx1sjhdYu?si=IJZ5DBLUTruGbtLmFj9Dtw), or wherever quality such shows are sold. As will come as no surprise, I talk a lot on it about the things I write about here. We have been rolling it out carefully, but we're now fully public. 2. After a popular episode months ago on AI and historical capex bubbles, I did a followup appearance with Derek Thompson on his popular Plain English podcast. You can find it [here](https://open.spotify.com/episode/5Oc3Aa9M81KXdy3T5XA3oP). It was a little spikier than my first appearance, now six months ago, but it is always good fun chatting with Derek. I also congratulated me on getting so many things right, from issues in private credit, to the market's flip-flop about rewarding capex spending, and more. 3. I was in [the WSJ](https://www.wsj.com/tech/ai/can-nvidias-dominance-survive-the-sea-change-under-way-in-ai-computing-63c3a70d?gaa%5Fat=eafs&gaa%5Fn=AWEtsqdWtWbZymu1m60q7vbOxpzg6Ro2aFd9y5OExAzVp-sIAYrtxtfA4M4R46ZJ0wM%3D&gaa%5Fts=69b83f7e&gaa%5Fsig=VvT9pyjiY4vdxWN7LPwkL8lEn565R9iGwGjm0jmFhwak4Bz6EGK83Vancx%5FiZIB3bmUDPvA-gIr7F8EAVA1TGA%3D%3D) this week in the run-up to NVIDIA's GTC event, interviewed about the challenges NVIDIA faces in maintaining its market position and margins in the transition from training-centric loads to inference-centric ones at hyperscaler data centers. P.S. Here is a sample from the show. Thanks to the good folks at iHeart for producing it and helping put it together. ### Chart of the Day: Data Centers vs Office Construction URL: https://paulkedrosky.com/chart-of-the-day-data-centers-vs-office-construction/ Last updated: 2026-03-16T19:58:48.000Z HIGHLIGHTS - Data center construction spending passed office spending for the first time in December 2025, crossing at roughly $3.5B monthly. - Office construction has fallen \~35% from its 2020 peak. - Data center spend is up 5x since 2020, with the curve steepening sharply in late 2024. - The Census figures exclude racks and servers, so this is structural construction only, making the crossover a measure of physical footprint, not equipment spend. Data center spending in the U.S. just crossed over office construction spending. Granted, office construction spending has been declining for years — down roughly 35% from its 2020 peak — but data center construction is up 5x over the same period. ![](https://paulkedrosky.com/content/images/2026/03/2026-03-16_12-49-23.jpg) The Census data excludes racks and servers, so this is purely structural construction: concrete, steel, power infrastructure, cooling systems, etc. That makes the numbers more meaningful, not less. The physical footprint of AI compute is now larger than the physical footprint of white-collar office work, at least as measured by monthly construction spending. The acceleration in the back half of 2024 and into 2025 is impressive. The curve isn't linear: it steepened sharply as hyperscaler capex commitments made in 2023 and early 2024 translated into actual ground-breaking. The office line, meanwhile, has no obvious bottom. Remote and hybrid work permanently destroyed a layer of demand that is not coming back. And AI is changing the shape and scope of future white-collar work. Capital is really and truly replacing labor, in multiple ways. ### Chart(s) of the Day: Prediction Markets, the Oscars, and Uncorrelated Bets URL: https://paulkedrosky.com/chart-s-of-the-day-prediction-markets-the-oscars-and-uncorrelated-bets/ Last updated: 2026-03-15T19:03:15.000Z ### HIGHLIGHTS - $100M+ traded across Kalshi and Polymarket on Oscar contracts; best picture alone drew $30M on Polymarket — nearly **6×** 2025's total - Chalamet fell from 68¢ to 51¢ after Jordan's SAG win; *One Battle After Another* sits at \~75¢ on both platforms - Prediction markets are federally regulated as futures contracts by the CFTC — not gambling, which is why hedge funds are paying attention - The same correlation breakdown driving institutional demand for exotic cross-asset hybrid options — dual binaries, contingent structures — makes event-driven prediction markets newly attractive: pure information exposure, zero macro beta Tonight's Academy Awards ceremony arrives amid surging activity in prediction market activity. They are an informative and unsettling lens on how financial markets are evolving. Kalshi and Polymarket have together traded over $100 million in Oscar contracts ahead of tonight's show, according to [Variety](https://variety.com/2026/film/news/oscars-betting-kalshi-polymarket-100-million-chalamet-jordan-1236361234/). Best picture alone generated $30 million on Polymarket, which is nearly six times 2025's total. The Chalamet-vs.-Jordan best actor contest has prices lurching on the SAG award, a ballet comment from Chalamet, and just general vibes. None of these affected the actual vote, which closed March 5, but prediction markets acted otherwise. The charts below (courtesy of Bennett Kedrosky) track the money flows, the shifting odds, and the trading volume. ![](https://paulkedrosky.com/content/images/2026/03/2026-03-15_11-23-01.jpg) ![](https://paulkedrosky.com/content/images/2026/03/2026-03-15_11-22-44.jpg) ![](https://paulkedrosky.com/content/images/2026/03/2026-03-15_11-22-52.jpg) What's interesting, beyond the spectacle: while prediction markets are largely a retail thing, they are increasingly attractive because they are *uncorrelated* with other assets. As Bloomberg [recently reported](https://www.bloomberg.com/news/articles/2026-03-15/hedge-funds-eye-exotic-options-to-play-huge-cross-asset-swings), hedge funds are turning to exotic and hybrid structures as oil price swings break correlation assumptions underpinning traditional portfolios. Entertainment prediction markets — liquid, event-driven, and different from sports gambling — fit the same logic. The underlying is pure information, not tied to volatile macro factors, outside maybe ballet futures. Whether prediction markets have anything right about tonight is almost beside the point. The emergent market structure itself is the story. ### Best of the Week, Mar 15, 2026: Silicon Valley Militarism, the Ghost Economy, etc. URL: https://paulkedrosky.com/best-of-the-week-mar-15-2026-silicon-valley-militarism-the-ghost-economy-etc/ Last updated: 2026-03-15T13:19:29.000Z Here are some of the best [posts](https://paulkedrosky.com/archive/) and [notes](https://paulkedrosky.com/thenotes/) from the past week: Posts 1\. [Silicon Valley Was Born a Defense Industry, and is Becoming One Again](https://paulkedrosky.com/silicon-valley-was-born-a-defense-industry-and-is-becoming-one-again/) ![](https://paulkedrosky.com/content/images/2026/03/download.png) The valley’s long civilian interlude may be ending as AI reconnects it to its military origins Mar 13 · [Permalink →](https://paulkedrosky.com/silicon-valley-was-born-a-defense-industry-and-is-becoming-one-again/) 2\. [AI and the Rise of the Ghost Economy](https://paulkedrosky.com/ai-and-the-rise-of-the-ghost-economy/) ![](https://paulkedrosky.com/content/images/2026/03/2026-03-12_11-34-48-1.jpg) On credit convexity and cognition abundance Mar 12 · [Permalink →](https://paulkedrosky.com/ai-and-the-rise-of-the-ghost-economy/) Notes 1. SaaSpocalypse Drives Software Multiples Back While the SaaSpocalypse has been striking—declining multiples among the largest SaaS equities out of AI fears wiping $800b in market cap—a related and perhaps more impportant story is that software multiples are only back to pre-pandemic levsls. As a mature sector with declining margins, it almost certainly should be lower. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fdbf0f0fc-f01d-4740-b89c-52c5b4047aff.jpeg) Mar 13 · [Permalink →](https://paulkedrosky.com/n/43dda19a-dd9e-448b-940c-877cc7a33e47) 2. Memory Bottleneck Undermines Nvidias AI Compute Moat Memory, not compute, is the main bottleneck on AI compute, as Epoch points out here. They don't take it the next step, however, and draw out the implciation: Nvidia's moat is less meaningful than ever in a memory-bound world. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fae8e5407-6e06-49aa-9167-0b7d02e01392.jpeg) Mar 13 · [Permalink →](https://paulkedrosky.com/n/8dd77166-1826-4de0-bdc2-ea46c4c3862e) 3. Urea-to-Corn Ratios Surge Amid Iran War Urea-to-corn ratios—a key measure of fertilizers (un)affordability—are touching decadal highs on Iran war-related nitrogen chokepoints. Much higher food prices ahead, given that we, in effect, eat oil. Fertilizer Prices Surge Due to War in Iran: Odd Lots ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fe2d434f8-8829-4a47-b4be-ab1e3069942a.jpeg) Mar 12 · [Permalink →](https://paulkedrosky.com/n/b311afd9-98f5-4258-bf16-291c0e99cd96) Notes 4. Workers-executives disconnect on AI productivity Honestly not sure I've ever seen a disconnect this large: Workers vs executives on AI & productivity. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fe7ee34f8-7f39-42a8-bc41-540d68d35ef5.jpeg) Mar 12 · [Permalink →](https://paulkedrosky.com/n/8c963d96-79b2-4d74-8261-7b11fcfc3e9a) 5. Robert Wright Challenges Beliefs in Upcoming Book My current reading—Robert Wright's upcoming book—is unsettling, smart, and, I think, largely correct. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F8b5171ee-4625-43e0-9ce4-1f2ef21987bf.jpeg) Mar 11 · [Permalink →](https://paulkedrosky.com/n/9106e60a-eb45-4bae-8a41-9692d0771624) 6. Late-Winter Heat Wave Shatters Western US Records Nearly unprecedented late-winter heat wave rolling across the western U.S. in coming six days. Records being broken by huge amounts, and on consecutive days, as well as nighttime lows. The remnant snowpack will sublimate almost entirely, months ahead of normal, and the energy grid will be under multiday cooling pressure. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fe3b92fba-6a9b-406f-9253-5fedf01aa2af.jpeg) Mar 11 · [Permalink →](https://paulkedrosky.com/n/872257b8-6b2d-4b42-8763-a4e5320c065c) 7. US retail margin debt hits record GDP ratio US retail margin debt newly at record levels as a percentage of GDP. The decoupling of debt to GDP vs debt to market cap is striking, a sign that markets have convinced themselves they have escaped the surly bounds of mere economic growth. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fd5a5fbde-02c4-4a7c-a2f5-27da071e1299.jpeg) Mar 11 · [Permalink →](https://paulkedrosky.com/n/700d3bd5-bfba-4787-97c3-b80e61617ad5) ### Chart of the Day: AI Capex Ate the Economy in Q4 URL: https://paulkedrosky.com/chart-of-the-day-ai-capex-ate-the-economy-in-q4/ Last updated: 2026-03-14T00:43:59.000Z I have written about AI capex and its outside role in US GDP growth many times here. Just look [here](https://paulkedrosky.com/ai-capex-in-h1-2025-prevented-a-recession/), [here](https://paulkedrosky.com/on-ai-capex-apologists/), [here](https://paulkedrosky.com/honey-ai-capex-ate-the-economy/), and [here](https://paulkedrosky.com/quick-hits-ai-capex-and-consumer-spending/), among others. But we are into a new phase. Today's [BEA second estimate](https://www.bea.gov/news/2026/gdp-second-estimate-4th-quarter-and-year-2025) revised Q4 2025 GDP growth down to 0.7% annualized, which is half the advance estimate. Buried in the detail: the BEA line that captures servers and GPUs and such contributed +0.66 percentage points to that growth. The rest of the U.S. economy contributed roughly +0.04 points — statistical noise. _This post is for paying subscribers only._ ### Silicon Valley Was Born a Defense Industry, and is Becoming One Again URL: https://paulkedrosky.com/silicon-valley-was-born-a-defense-industry-and-is-becoming-one-again/ Last updated: 2026-03-13T16:18:14.000Z Most people think of Silicon Valley as a consumer technology story: garages, hippies, personal computers, the internet, social media, etc. That narrative is almost entirely wrong about its origins, increasingly wrong about its present, and almost certainly wrong about its AI-suffused future. The valley was birthed by the Pentagon. Fairchild Semiconductor — the company that seeded nearly every important Silicon Valley firm of the 1960s and 70s — got most of its revenues from the Defense Department. In 1955, defense-related contracts accounted for an estimated 85% of the region's total revenues. The integrated circuit itself was a military procurement project. The region didn't become a technology hub despite defense spending; it became one *because* of it. ![](https://paulkedrosky.com/content/images/2026/03/2026-03-12_18-24-51.jpg) What followed over the next four decades was a long, gradual divorce. Commercial computing, consumer electronics, and the internet created enormous civilian markets. Defense's share of Silicon Valley revenues fell from 85% in 1955 to roughly 6% by the mid-1990s, to under 3% by 2018\. The valley became culturally allergic to its own origins: "Don't Be Evil" was partly a repudiation of the industry's founding patron. The AI era is reversing this, and doing it faster than most observers recognize. Today defense-related revenues account for an estimated 4% of Silicon Valley's total — very low, but up from its recent trough and rising quickly. The trajectory is what matters, and the nature of the dependency has changed. In the 1950s, defense drove semiconductor fabrication. Today, it drives the frontier of AI development itself. _This post is for paying subscribers only._ ### Rough Notes, Mar 12, 2026: Dunning-Kruger LLMs, Urea, The God Test, etc. URL: https://paulkedrosky.com/rough-notes-mar-12-2026-d/ Last updated: 2026-03-12T23:04:18.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site: 1\. Dunning-Kruger Effect Reveals LLMs Self-Assessment Limits There is a Dunning-Kruger effect in LLMs, with stupid models not realizing that they are stupid. We are training models well: they are basically us. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F2fa20b70-5b71-47a0-a79a-43c666566a3d.jpeg) Mar 12 · [Permalink →](https://paulkedrosky.com/n/761fcb45-d655-48c6-8496-7e59ab5f55f4) 2\. Agent autonomously builds Linux-capable RISC-V CPU In 12 hours and fully autonomously, an agent used frotiier models to build a Linux-capable CPU. This is a process that once took years, millions of dollars, and teams of TSMC engineers. Draw implications—from employment to innovation—as you wish. Design Conductor: An agent autonomously builds a 1.5 GHz Linux-capable RISC-V CPU ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F26220123-46ee-4e5d-b64c-dd23cf640515.jpeg) Mar 12 · [Permalink →](https://paulkedrosky.com/n/79066a30-c82a-46f0-8bee-70ff3eea8886) 3\. Urea-to-corn ratios surge amid Iran war Urea-to-corn ratios—a key measure of fertilizers (un)affordability—are touching decadal highs on Iran war-related nitrogen chokepoints. Much higher food prices ahead, given that we, in effect, eat oil. Fertilizer Prices Surge Due to War in Iran: Odd Lots ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fe2d434f8-8829-4a47-b4be-ab1e3069942a.jpeg) Mar 12 · [Permalink →](https://paulkedrosky.com/n/b311afd9-98f5-4258-bf16-291c0e99cd96) 4\. Childhood Challenges Worsened by Technology I hardly know where ot start with this sad piece, other than to point out the obvious, which is that AI will make it worse. 30 Facts About Childhood Today that Will Terrify You ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Ff312a250-b9cf-4c0f-bcb9-7df6d7b1d8d7.jpeg) Mar 12 · [Permalink →](https://paulkedrosky.com/n/663d3ce1-a7c9-4efc-91d6-ab57fb04574a) _This post is for paying subscribers only._ ### AI and the Rise of the Ghost Economy URL: https://paulkedrosky.com/ai-and-the-rise-of-the-ghost-economy/ Last updated: 2026-03-12T18:58:08.000Z A curious possibility sits at the center of the current AI boom: the economy can appear to grow while the demand that normally sustains that growth quietly weakens. Call it a **ghost economy**. As I have repeatedly written, GDP can rise with investment ("I" in GDP terms), despite everything else in GDP being weak, or even falling. It is just math. A recent macro-financial stress-test [paper](https://arxiv.org/pdf/2603.09209) argues that, alongside rapid AI adoption, this is economic "ghosting" is a plausible future, given a growing mismatch between output and income. The core idea is simple: AI raises production while simultaneously weakening labor income. That mismatch produces what the author calls **Ghost GDP**: output continues to rise even as the income that supports consumption falls. 💡 GDP can rise because of AI investment even while the consumption economy weakens. But it is even more insidious. Most economic institutions are tied to an assumption: **human cognitive labor is scarce and predictably compensated**. AI, however, undoes this scarcity, and does so at speed. These institutional anchors—assumeptions about scarce cognition—newly become **sources of fragility.** --- ## The displacement spiral This results in a feedback loop that begins at the firm level. Each company has an incentive to substitute AI for labor. But when many firms do this simultaneously, a macro problem emerges. 1. Firms automate tasks. 2. Labor income declines. 3. Consumption falls. 4. Demand weakens. 5. Firms respond to this margin pressure by automating even more. This is the **displacement spiral.** It can, of course, stabilize if new jobs emerge quickly enough—what economists call **task reinstatement**. But if automation spreads faster than new tasks appear, the labor share of income declines and keeps on going down. That is already happening, of course, and has been for some time. ![](https://paulkedrosky.com/content/images/2026/03/2026-03-12_11-34-48.jpg) Historically that reinstatement channel has saved the economy. "**Technology always creates more jobs than it destroys"**, as the faith-based line goes. But AI is attacking the exact sector — cognitive work — where new tasks (not merely jobs) have historically formed after prior automation episodes. --- ## The capex illusion This is where the current AI boom gets interesting. _This post is for paying subscribers only._ ### Reinsurance, Risk, and AI Insurabiltiy URL: https://paulkedrosky.com/reinsurance-risk-and-ai-insurabiltiy/ Last updated: 2026-03-11T18:53:03.000Z Four points on risk, insurability, and AI from a recent interview with Andreas Berger, CEO of reinsurer Swiss Re. who makes some solid points. My take, however,: AI and climate change are sharply increasing cross-line correlations, and the industry thinks it can price and "big data" its way out. It likely can't. **1\. Catastrophe losses are mostly an asset-density story** Reinsurers have a more prosaic explanation for rising disaster losses than the public narrative. The main driver is exposure. More people, houses, and infrastructure are being built in the highest-risk locations: coasts, wildfire corridors, and floodplains. More people are building blithely with AI and agentic systems. When events occur, the damage pool is simply larger. Swiss Re notes insured catastrophe losses have exceeded **$100B annually for six consecutive years**, with **$318B in total economic losses in 2024**. Climate likely contributes to severity, but the structural trend is that capital keeps moving into exposed areas. 💡 While this is fair, as far as it goes, the planet is, to a first approximation, an "exposed area". Granted, some parts are more exposed than others, but there are no havens, and the gradients are steep, whether from climate change or AI. --- **2\. “Uninsurable” actually means “politically unaffordable”** _This post is for paying subscribers only._ ### Rough Notes, 3/11/26: Western Heat Wave, Margin Debt, Robots, AI Benchmarks, etc. URL: https://paulkedrosky.com/rough-notes-mar-11-2026-2/ Last updated: 2026-03-11T18:25:54.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. Late-Winter Heat Wave Shatters Western US Records Nearly unprecedented late-winter heat wave rolling across the western U.S. in coming six days. Records being broken by huge amounts, and on consecutive days, as well as nighttime lows. The remnant snowpack will sublimate almost entirely, months ahead of normal, and the energy grid will be under multiday cooling pressure. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fe3b92fba-6a9b-406f-9253-5fedf01aa2af.jpeg) Mar 11 · [Permalink →](https://paulkedrosky.com/n/872257b8-6b2d-4b42-8763-a4e5320c065c) 2\. US retail margin debt breaks GDP record US retail margin debt newly at record levels as a percentage of GDP. The decoupling of debt to GDP vs debt to market cap is striking, a sign that markets have convinced themselves they have escaped the surly bounds of mere economic growth. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fd5a5fbde-02c4-4a7c-a2f5-27da071e1299.jpeg) Mar 11 · [Permalink →](https://paulkedrosky.com/n/700d3bd5-bfba-4787-97c3-b80e61617ad5) 3\. AI Benchmarks Overestimate Agentic Code Quality I point this out endlessly, but common AI benchmarks are less useful than thought, given both metholdogical issues and that they are increasingly ingested by models. That more than half of agentic code passing code benchmarks is not acceptable should thus come as no surprise. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fcab3e9d8-1595-4583-9f98-6d9104129416.jpeg) Mar 10 · [Permalink →](https://paulkedrosky.com/n/feffbe9b-b19a-4ae8-bb0e-93aac60e5a73) 4\. Google DeepMind Explores AI Simulating Consciousness Limits Fascinating new Google DeepMind paper on why AI can simulate consciousness but cannot actually be conscious. Somewhat technical philosophy, but well written and accessible—and important. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F06562d57-65ef-41ad-bb4e-ce1e0bb6ff76.jpeg) Mar 10 · [Permalink →](https://paulkedrosky.com/n/e14ebd8a-f8ee-4344-ab26-d5d46e6ff5bd) 5\. US crude oil exports decline in 2023 US crude oil exports declined last year for the first time since 2021. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Ffe5e3b76-ccf1-4d7d-9cf4-94f3a40642ea.jpeg) Mar 10 · [Permalink →](https://paulkedrosky.com/n/300c1f5d-285d-4cdb-9f0c-f049e1d1dc64) 6\. India Emerges Electronics Assembly Powerhouse After China Tariffs India is quietly becoming an electronics assembly powerhouse, thanks, in part, to recent China tariffs. Apple Now Makes About 25% of iPhones in India After China Pivot - Bloomberg ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F55bb4e4d-fa05-479c-9739-57b6a41fcbb2.jpeg) Mar 10 · [Permalink →](https://paulkedrosky.com/n/5a0c4bd9-3f7d-4c4e-9b31-ccd5bc51ef1b) 7\. OpenAI weighs on largest shareholders FT tonight pointing out that OpenAI is now weighing on a few of its largest holders. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F42813817-d627-4a8f-a564-488c70a0dab4.jpeg) Mar 10 · [Permalink →](https://paulkedrosky.com/n/7ac63829-651c-4f2a-8f08-e49f765735af) 8\. Goldman Sachs uses total return swaps against leveraged loans Having helped private credit make leveragedloans to take stakes in SaaS and AI companies, Goldman is now taking the other side, using total return swaps. Mar 10 · [Permalink →](https://paulkedrosky.com/n/f3024eb1-02b4-4464-8cb4-919a6657602d) 9\. Rise of Robotics Devastates China Factory Jobs Tremendous and harrowing piece in Bloomberg today on how the rise of robotics and AI in China has been devastating for jobs: Wages are falling, workers are at record levels of pessimism, jobs are more tenuous, etc. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F63e9eb15-4207-43e5-b063-558adc56f207.jpeg) Mar 10 · [Permalink →](https://paulkedrosky.com/n/997f33ee-d204-4b1a-858c-4cc074462730) 10\. LLM Agents Mimic Bayesian Reasoning Causing Belief Drift Working paper with highly troubling implications for an increasingly agentic AI world: LLM agents don't actually do Bayesian reasoning — they mimic it. Belief updates are plausible but disconnected from evidence. with severe base-rate neglect. In multi-step agentic pipelines, this compounds silently: each step looks reasonable while beliefs drift away from reality. At scale, this is undetectable until outcomes are suddenly catastrophic ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fd940d2d6-ea3a-44ad-84b5-d865dc093df9.jpeg) Mar 8 · [Permalink →](https://paulkedrosky.com/n/9e57574e-9fc6-4fa7-81d5-a5018f7b5b3f) ### New Premium Feature: Oil Price Shock Simulator URL: https://paulkedrosky.com/new-premium-feature-oil-price-shock-simulator/ Last updated: 2026-03-09T15:23:39.000Z Simulate the macroeconomic impact of oil price shocks—from gasoline prices to GDP growth, inflation, unemployment, and recession risk. Includes handy per-scenario tear sheet to save-and-share. Dynamic scenario planning is useful when trying to understand complex systems like the economy, especially under stress. With oil prices moving sharply, I am releasing one of my internal tools: an interactive simulator for oil-price shocks. The tool starts from current WTI levels and lets users test both live scenarios and historical analogs. Adjust peak oil prices and shock duration, and the model projects the resulting paths for gasoline prices, GDP growth, unemployment, and inflation over a four-year horizon. Premium subscribers can find it [here](https://paulkedrosky.com/tools/gas-sim/). Here is what it looks like: ![](https://paulkedrosky.com/content/images/2026/03/2026-03-09_07-39-53.jpg) The goal is clarity rather than precise forecasting. The model is intentionally reduced-form, designed to provide quick directional insight during energy shocks. Users can select preset scenarios or build their own, then immediately compare outcomes. A structured tear sheet summarizes each scenario in plain language, highlighting headline metrics and flagging when the shock implies elevated recession risk. Under the hood, the simulator combines empirical oil-to-gas pass-through, asymmetric growth sensitivity to energy shocks, labor-market response dynamics, and a simple inflation proxy. The result is a compact view of how energy shocks propagate through consumer prices and the real economy. Premium subscribers can try it [here](https://paulkedrosky.com/tools/chartbook-live/). --- Don't forget about about [Chartbook Live](https://paulkedrosky.com/new-premium-feature-chartbook-live/), which I launched last week for Premium subscribers. ### Best of the Past Week, Mar 8, 2026: AI IPOs, Orchestration, Re-Rating, etc. URL: https://paulkedrosky.com/best-of-week-mar-8-2026-9/ Last updated: 2026-03-08T14:44:41.000Z Here are some of the best [posts](https://paulkedrosky.com/archive/) and [notes](https://paulkedrosky.com/thenotes/) from the past week: Posts 1\. [New Premium Feature: Chartbook Live](https://paulkedrosky.com/new-premium-feature-chartbook-live/) ![](https://paulkedrosky.com/content/images/2026/03/2026-03-05_08-00-52-1.jpg) Easy access to previous charts and graphics Mar 5 · [Permalink →](https://paulkedrosky.com/new-premium-feature-chartbook-live/) 2\. [Commoditization, Orchestration, and the New AI Stack](https://paulkedrosky.com/commoditization-orchestration-and-the-new-ai-stack/) ![](https://paulkedrosky.com/content/images/2026/03/2026-03-05_19-05-47.jpg) AI value is moving up the stack, stranding capital Mar 3 · [Permalink →](https://paulkedrosky.com/commoditization-orchestration-and-the-new-ai-stack/) Notes 1. Private AI Companies Surpass Energy Sector Weight If the currently private AI-ish darlings—SpaceX, OpenAI, Anthropic, and Databricks—were to go public at current valuations, and were added to the index, they'd instantly be a material chunk of the S&P 500\. At 4.6%, the four combined would be larger than the energy sector, for example. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F5edab836-d0d0-43e9-91bf-918d99aaf0cb.jpeg) Mar 6 · [Permalink →](https://paulkedrosky.com/n/c09b6fe8-a6ac-438a-b630-edddd6104b2b) 2. Tech Industry Matures Driving Mag 7 Re-Rating One of the points I often make in meetings is that the tech industry is mature and becoming structurally capital intensive, which deserves its re-rerating. That has been steadily happening, and it has accelerated recently. Check this Bloomberg graph of the decadal decline in the the gap between the Mag 7's forward price-earnings multiple and that of the S&P 500\. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Ff95effd7-71c4-43df-ae89-61dd2a7b1415.jpeg) Mar 6 · [Permalink →](https://paulkedrosky.com/n/ecbb91cb-6264-4b6c-a514-9c5776c18a55) 3. Hateful Eight Tech Stocks Drag Markets Lower My weekly update of what I call tthe "Hateful Eight"—the former Mag 7 tech stocks, plus Oracle—and how the group is doing with respect to pulling markets lower with high capex and shrinking free cashflows. They continue to underfprm impressiely. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F42401cb0-f347-4a0f-a8d9-89aa787f856d.jpeg) Mar 6 · [Permalink →](https://paulkedrosky.com/n/4202d5ef-d759-4065-87d9-129a31524b47) Notes 4. Data Explains Implicit Double-Counting in US Productivity Boomlet To follow on from something I wrote earlier on the non-role of AI in the current US productivity boomlet, here is the data, and explanation for the implict double-counting, given outsided AI capex. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F95ae1dda-7c55-401e-9744-58e46d2846af.jpeg) Mar 5 · [Permalink →](https://paulkedrosky.com/n/267d6f0b-a9b5-4d71-8b12-ff6e527fa239) 5. Revenue-sharing dynamics across Substack Spotify YouTube platforms Rev share / rake by content platform, from Substack, to Spotify, to YouTube. I got thinking about this in scanning a BBC filing mentioning its unease with putting more content on YouTube, given that its poor revenue sharing. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F240b9919-979b-426b-8dc3-f1ac5857729e.jpeg) Mar 5 · [Permalink →](https://paulkedrosky.com/n/40230bd6-f3a4-46a2-b75b-c34000d0fcbe) 6. South Korean Indices Decline Driven by AI Risk-Off The current 12%+ decline in the South Kirean indices is best thought of as an AI risk-off action, given that over a third of the index, by market cap, is Samsung and SK Hynix, helping drive the indice's prodigous advance over the last year. Here is what it looked like at the end of 2025: ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F91c48fee-7cd4-42e7-aae3-68c2ff024716.jpeg) Mar 4 · [Permalink →](https://paulkedrosky.com/n/7945cd73-a61e-4091-aa53-4ac8aebb6aac) 7. Smartphone Use Transforms Skills Anticipating LLM Impact The effect of the first decade of smartphone use on math, reading, and science skills has been startling. I'm betting a decade of LLM use's consequences will be even more dramatic. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F138d9fb9-9773-4171-8921-95553a298f45.jpeg) Mar 4 · [Permalink →](https://paulkedrosky.com/n/3a2f808a-0b9a-4952-b925-2dda38c83030) 8. HALO Investing Gains Traction Amid AI Shift I heard a discussion about HALO this week in NYC, which was a new acronym to me. I love that old atomy-y stuff that doesn't change much—heavy asset, low obsolesence—is newly in vogue as AI turns everything else into Shein-ified commodities. Forget YOLO or even FOMO. It’s All About HALO Now ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fce27dc77-2279-4f28-a2f5-29cf16a3cd30.jpeg) Mar 1 · [Permalink →](https://paulkedrosky.com/n/600ffb5a-bbab-40c2-81f1-6f8f181e54ba) ### Rough Notes, Mar 6, 2026: AI IPO Sector, Re-Rating Tech, Hateful Eight, etc. URL: https://paulkedrosky.com/rough-notes-mar-6-2026-ai-ipo-sector-re-rating-tech-hateful-eight-etc/ Last updated: 2026-03-06T16:04:29.000Z A digest of recent Rough Notes _This post is for paying subscribers only._ ### Rough Notes II, Mar 5, 2026: Productivity, Mag 7 Survival, AI-Exposed Jobs, etc. URL: https://paulkedrosky.com/rough-notes-ii-mar-5-2026-productivity-mag-7-survival-ai-exposed-jobs-etc/ Last updated: 2026-03-06T03:20:42.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. Data Explaining Implicit Double-Counting in US Productivity Boomlet To follow on from something I wrote earlier on the non-role of AI in the current US productivity boomlet, here is the data, and explanation for the implict double-counting, given outsided AI capex. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F95ae1dda-7c55-401e-9744-58e46d2846af.jpeg) Mar 5 · [Permalink →](https://paulkedrosky.com/n/267d6f0b-a9b5-4d71-8b12-ff6e527fa239) 2\. Predicting Mag 7 Stocks Longevity in Decade Provocative question I was asked recently when visiting with a large hedge fund. How many of the current Mag 7 stocks¹ will matter in a decade, like be in the largest 100 equties by market cap? Spoiler: I bet the field. ¹ Here is the Mag 7 list: Apple (AAPL) Microsoft (MSFT) Alphabet (GOOGL / GOOG) Amazon (AMZN) Meta Platforms (META) Nvidia (NVDA) Tesla (TSLA) Mar 5 · [Permalink →](https://paulkedrosky.com/n/8ea254a1-d556-4a80-8fcc-fc3e22288d3b) 3\. Anthropic reveals AI-exposed jobs limiting entry-level hiring Anthropic is also doing work on how the most AI-exposed occupations may be closing the door, at least a little, on entry-level employees. Here is my revision of their data, trying to draw the point out more clearly than they did. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F538fed25-f2c8-40ac-bc71-91bcf16ddd3e.jpeg) Mar 5 · [Permalink →](https://paulkedrosky.com/n/73663b58-41fa-489a-a7df-1c3212058630) 4\. Anthropic Expands AI Exposure Analysis Software Developers Anthropic continues to do work on the most AI-exposed occupations, and this is its latest, which is led by software developers themselves, unsurprisingly. I have added the share of the whitecollar workforce to the Anthropic list. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F922282a7-9747-4dc3-8616-b517c86ef680.jpeg) Mar 5 · [Permalink →](https://paulkedrosky.com/n/d522b878-02bb-4d23-97ca-652824482c1e) _This post is for paying subscribers only._ ### New Premium Feature: Chartbook Live URL: https://paulkedrosky.com/new-premium-feature-chartbook-live/ Last updated: 2026-03-05T16:17:32.000Z There are more than 10,000 charts on this site, and they are often valuable. But it can be very hard to find what you're looking for, or even surf the latest charts or graphics, given that they are embedded in articles or notes, or just floating in files. I have now changed that, and the solution is called [**Chartbook Live**](https://paulkedrosky.com/tools/chartbook-live/). (Note: It is ***only available to Plus subscribers***, given the resource requirements.) Every chart that appears on the site is always stored, but now it is also automatically indexed and made available via free-form searches, like "AI capex". In a sense, I have turned the site inside out, making it "charts first", or at least elevating charts to being a primary object, which is how I think about them anyway. You can see **Chartbook Live** in action below: ![](https://paulkedrosky.com/content/images/2026/03/2026-03-05_08-00-52.jpg) There is also a "Live" feature, where you can see the latest charts added. It is as below: ![](https://paulkedrosky.com/content/images/2026/03/2026-03-05_08-01-02.jpg) There are number of other features like this coming shortly for Premum subscribers, including smulators and tracking tools. And **Chartbook Live** is growing too. For example, it will soon have alerting—"let me know when there are new charts about X"—and more. You can try [**Chartbook Live** here](https://paulkedrosky.com/tools/chartbook-live/). Again, you need to be a Plus subscriber to use this feature. ### Rough Notes, Mar 5, 2026: AI and Productivity, Bombing Data Centers, Revenue Rakes, etc. URL: https://paulkedrosky.com/rough-notes-mar-5-2026-ai/ Last updated: 2026-03-05T15:35:48.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. US productivity growth driven by AI infrastructure investment US productivity growth is on a flyer, back to levels last seen more than a decade ago. Could it be AI, as some are arguing? Almost certainly not—too soon. My guesses: \- Outsized AI infrastructure investment skewing numbers \- Capital deepening, and \- Cyclical labor effects post-Covid ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F00bad761-aca1-41f2-a7ee-62a8b7084687.jpeg) Mar 5 · [Permalink →](https://paulkedrosky.com/n/b77f52d8-7736-4a2d-a150-1c050b3aa079) 2\. Revenue-sharing models across Substack Spotify YouTube platforms Rev share / rake by content platform, from Substack, to Spotify, to YouTube. I got thinking about this in scanning a BBC filing mentioning its unease with putting more content on YouTube, given that its poor revenue sharing. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F240b9919-979b-426b-8dc3-f1ac5857729e.jpeg) Mar 5 · [Permalink →](https://paulkedrosky.com/n/40230bd6-f3a4-46a2-b75b-c34000d0fcbe) 3\. Data Centers Drive Geopolitical Military Strategy I mentioned this in an AWS context earlier in the week, and it should come as no surprise given the ardor with which defence departments worldwide have embraced AI: data centers are now a geopolitically and militarily strategic asset. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fd62dddd9-6fa3-4b4c-8480-0d96fb0b9f79.jpeg) Mar 5 · [Permalink →](https://paulkedrosky.com/n/928305fb-e262-431f-bf18-0e24d8e195bb) 4\. AI Tech Lending Myths Debunked Today Today in AI/tech lending related things that are almost certainly not true: ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F1722c675-2d2a-485d-8cdc-82414ed66148.jpeg) Mar 4 · [Permalink →](https://paulkedrosky.com/n/ba62632e-b2bb-4c5b-b969-cd6812783897) _This post is for paying subscribers only._ ### Rough Notes, Mar 3, 2026: Tokens, Alchemists, Hateful Eight, etc. URL: https://paulkedrosky.com/rough-notes-mar-3-2026-tokens-alchemists-hateful-eight-etc/ Last updated: 2026-03-03T18:24:49.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. Taalas Demonstrates 70x Token Production Speedup One of the dynamics playing out as model improvement slows, and LLMs move toward being inference-centric, is that new silicon is emerging that runs cooler, with more caching, and much faster speeds. For example, check Taalas's demo here to see 70x token production speedup, often approach 16,000 tok/s. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fa0165255-f43f-46ba-9550-d422acd996c0.jpeg) Mar 3 · [Permalink →](https://paulkedrosky.com/n/6633482f-7825-4e65-bc8c-91cb9c911a09) 2\. Michael Cembalest Coining Alchemists Describes Market-Geopolitics Experimentation Current event remind me, again, of my friend Michael Cembalest's year-ago coinage of "the alchemists" to describe the madcap experimentation in global markets and geopolitics. Instead of transmuting lead into gold, you end up with cracked furnaces and a ruined crucible. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fc64b1af9-6179-4651-8d51-685d7bc9c105.jpeg) Mar 3 · [Permalink →](https://paulkedrosky.com/n/77602ac8-e319-4624-a2ed-9c92d8815720) 3\. Active ETF Launches Drive US 2025 Record Active ETF launches helped power total ETFs launched in the US in 2025 to an all-time record. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F28368313-5324-4cb1-ae29-313e828dd063.jpeg) Mar 3 · [Permalink →](https://paulkedrosky.com/n/253e3b17-a2a6-4492-9a60-68265dcee93b) 4\. Cursor hits 2b revenue run rate Bloomberg reporting tonight that AI-centric code editor thingie Cursor is now on a $2b annual revenue run rate, which is impressive. Let’s see if this is the peak, like I have argued here—beset by agentic harnesses like Copilot, CCode, and Codex—or if I’m, you know, wrong. Mar 3 · [Permalink →](https://paulkedrosky.com/n/835d6501-4592-40d9-8ac9-b91a9ae46b3e) _This post is for paying subscribers only._ ### Commoditization, Orchestration, and the New AI Stack URL: https://paulkedrosky.com/commoditization-orchestration-and-the-new-ai-stack/ Last updated: 2026-03-06T03:09:07.000Z People are newly excited again at AI progress. After a period where it felt like AI models weren't changing as much, that they had turned into mobile phones, with minimal progress, something has changed. But people are confused, as I will show. To understand why, first understand that the headlines and hyperbole are mostly not about models. They are about new tools, like Claude Code and Codex, from Anthropic and OpenAI, respectively. Interest surged in the past year. ![](https://paulkedrosky.com/content/images/2026/03/2026-03-03_08-39-57.jpg) Such tools are important, but they mask at least two things. 1\. **Models have not re-accelerated** with respect to capability. If anything, model progress has slowed even further, as I will show below. 1. **A new layer has emerged** that has changed AI dynamics for good, putting much of current AI training spending at risk. The takeaway is this: Large language models are the new silicon. Orchestration layers, like Claude Code and Codex, are the new operating systems. Applications are the new productivity suite. As raw model improvement slows, value migrates up the stack. This will have huge implications, from model commoditization, to inference traffic, to stranded training spending. 💡 Large language models are the new semis, with similar dynamics and boom-bust cycles Here is how to think about the new layers, where value is going, and what is being commoditized: ![](https://paulkedrosky.com/content/images/2026/03/2026-03-03_07-03-23.jpg) --- Now let me explain why this is happening. ### LARGE MODELS AND ORCHESTRATION LAYERS First, however, two quick definitions: ****Large Language Models (LLMs):** Neural networks trained on vast text corpora to predict and generate language, forming the foundational reasoning and pattern-recognition layer of modern AI systems. ****Orchestration Layers (e.g., Claude Code, OpenCode, Codex):** Software frameworks that coordinate models, tools, memory, and execution environments—turning raw LLM capabilities into structured, task-oriented workflows and usable applications. The key to understand what is going on is to understand that orchestration layers now dominate models. While models are not irrelevant, they are far less important than they were even a year ago. Slowing model progress—the next section—is masked and made less relevant by improved model coordination, which has immense capex implications. ### SLOWING MODEL PROGRESS _This post is for paying subscribers only._ ### Rough Notes, Feb 24, 2026: Tail Risks, Capex, Backlog Issues, Nvidia earnings, etc. URL: https://paulkedrosky.com/rough-notes-feb-24-2026-2/ Last updated: 2026-02-24T14:26:40.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. Captive-Insurer Affiliates Drive Parent-Organization Contributions Captive insurer affiliates to private credit/equity are big contributors to the parent organizations, as I have pointed out many times. As ever, this creates a duration mismatch that doens't matter until it does, as we've seen to a limited degree in the recent retail race to liquodty causing lockups at Blue Owl and others. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F21fa22a7-6794-4dd5-a3ff-7b02df0861cf.jpeg) Feb 24 · [Permalink →](https://paulkedrosky.com/n/1cc7fdb2-9fcc-4f00-a783-117f2b251d36) 2\. Recent Headlines \- OpenAI raises revenue forecast alongside forecasting cash burn of $11b to 2030 (The Information) \- Inference chiip maker Cerebras filles confidentially for an IPO (The Information) \- Insurance is the lifeblood of priatate credit and thus, increasingly, of AI capex (FT) \- IBM share fall 12% on ANthropic tools for defunct COBOL (CNBC) \- Anthropic accuses Chinese AI companies of launching concerted distillation attacks to clone models (Anthropic( Feb 24 · [Permalink →](https://paulkedrosky.com/n/b464ce31-b72a-4735-bcab-3a0ccf74688d) 3\. Nvidia options price 6.2-percent earnings move Using the first post-earnings weekly expiry (Fri, Feb 27, 2026) and the nearest ATM strike, Nvidia options are currently pricing about a ±6.2% move into earnings (Wed, Feb 25, 2026). I would take the over. Feb 24 · [Permalink →](https://paulkedrosky.com/n/55171f97-52bc-44e1-a470-bcfd9a8dcaeb) 4\. ODNI warns LLMs create security crisis LLMs are a security crisis, according to a major new report from the Office of the Director of National Intelligence (ODNI) on AI . Not sure how this slipped out in all the administration AI bonhomie, but it is a good report about rapidly growing tail risks to ... everything. Trojans in Artificial Intelligence: Final Report ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F2574e2f1-b0d7-478c-adce-a4c091a60a2a.png) Feb 24 · [Permalink →](https://paulkedrosky.com/n/a596eb73-a7af-4969-b50f-9a987b265cb3) _This post is for paying subscribers only._ ### Rough Notes II, Feb 23, 2026: Jobs, Energy, SaaS, etc. URL: https://paulkedrosky.com/rough-notes-feb-23-2026-j/ Last updated: 2026-02-23T23:22:09.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. Fed Governor Waller Highlights Jobs Data Skew Fed governor Christopher Waller argued recently in a talk that the upward-skew in BLS jobs data means that last year was likely only the third time since 1940 that the US lost jobs outside a recession. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Ff6183fa5-5149-4ccc-bb3f-78e39a090752.jpeg) Feb 23 · [Permalink →](https://paulkedrosky.com/n/7f1c82c9-4599-42b7-87ce-470227df3ecb) 2\. State Energy Regulators Rethink Rate-Base Utility Model Inevitable, but worth noting that state energy regulators are finally revisting rate-base-driven systems-utilities earning a premium on deployed capital—and their role in increasing (consumer) energy costs. The combination has been a flywheel for runaway capex spending, made worse by by AI data centers. Feb 23 · [Permalink →](https://paulkedrosky.com/n/8ce0bf0a-6958-4f82-bc77-70e9d27ff434) 3\. AI-driven anti-SaaS arguments reveal misunderstanding I have argued repeatedly that the AI-driven anti-SaaS arguments, like the attached, are silly and incoherent. 1/ If SaaS is going to be replaced by SaaS, then SaaS isn’t going away 2/ People don’t buy enterprise software because it’s complicated. They buy it so they don’t have to do it, and so they have someone to shoot at when it doesn’t work. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F39dddb72-c19a-4da4-9c0e-75cd4abc76a6.jpeg) Feb 23 · [Permalink →](https://paulkedrosky.com/n/81af70e1-e9b7-40bd-8070-5493c938d55e) 4\. OpenAI leaks conflicting capex cuts and margin pressure There is a confused "leak" from OpenAI about cutting capex spend over the next 4 years, from $1.4 to $600b. But the piece doesn't make clear if they are the same spend buckets, but someone there is trying hard to leak a spending cut. A tidbit in the piece: Inference costs up 4× in 2025; adjusted gross margin down from 40% → 33%. Price deflation and competition are hammering it. Feb 23 · [Permalink →](https://paulkedrosky.com/n/b861ebb2-986f-4c2d-9c84-af121085dfe5) _This post is for paying subscribers only._ ### Rough Notes; Feb 23, 2026: Hateful Eight AI, Capex, AI Unemployment, etc. URL: https://paulkedrosky.com/rough-notes-feb-23-2026-hateful-eight-ai-capex-ai-unemployment-etc/ Last updated: 2026-02-23T16:42:16.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. Hateful Eight Drag S&P 500 Amid AI Spending Reevaluation The "Hateful Eight": that's what I have taken to calling the former "Mag 7" megacap tech stocks (plus Oracle) that drove market indices the last two years. They have been a drag on the S&P 500 this year, as investors reconsider their former fondness for said companies' profligate AI cap ex spending. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fda87c2ad-d62c-472b-89a7-2e8319e18288.jpeg) Feb 23 · [Permalink →](https://paulkedrosky.com/n/623ef8c2-050a-4c3c-93c8-a6d4702c4c49) 2\. AI Capex Components Driving GDP Growth I write a lot here about the ongoing impact of AI capex on GDP, so this is a handy decomposition of its various components and their GDP-related effects. It's chips, software stacks, and energy/facilities, in that order. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fe92a1d6e-dfab-4cb3-ab8a-d6ba422e1f85.jpeg) Feb 23 · [Permalink →](https://paulkedrosky.com/n/d7f49d2c-bad0-4d3f-8581-69b00d035522) 3\. Consumer Energy Prices React to ChatGPT Launch Speaking of noisy AI signals, here is another: The ongoing natural experiment in consumer energy prices around the launch of ChatGPT, and the subsequent massive buildout of energy capacity. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fc4e55ff3-f115-4c77-b1bf-d4217d45328e.jpeg) Feb 23 · [Permalink →](https://paulkedrosky.com/n/2657934a-5f39-4a59-9ee3-226ab8b4dcf6) 4\. Unemployment Rates Diverge Between AI-Exposed Occupations WHile it's early and noisy, it is interesting that unemployment rates in AI-exposed and less exposed occupations are now going in opposite directions. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F2fd39339-136b-47ab-89cb-f8356d99e244.jpeg) Feb 23 · [Permalink →](https://paulkedrosky.com/n/4df7aec8-8d28-4bc1-910b-62ef56891599) _This post is for paying subscribers only._ ### Rough Notes, Feb 22, 2026: Agents, Clawdbot Collapse, Microsoft as Exxon, etc. URL: https://paulkedrosky.com/rough-notes-feb-22-2026-agents-clawdbot-collapse-microsoft-as-exxon-etc/ Last updated: 2026-02-22T15:57:58.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. Harness Engineering Managing LLMs with Orchestration Tools While I'm on the topic of agentic coding, good piece on "harness engineering", which I've sometimes analogized as having a terrific nanny (orchestration tool) managing myriad bratty kids (LLMs). I maintain the software is unique in the steep gradient descent, deterministic feedback loops, etc., making it less of an exemplar than some think, still worth scanning. Feb 22 · [Permalink →](https://paulkedrosky.com/n/7a3da928-aaa0-4daf-992f-4e6ac01871b1) 2\. Google Search Trends Reveal Clawdbot Coding Interest Collapse The brief explosion of interest among normals in coding agents a few weeks ago has largely subsided, so it's worth taking a look at the numbers. Claude Code, Codex, and, in particular, Clawdbot, all jumped higher on Google search, and are now lower—full collapse in the case of Clawdbot. These tools may be taking over coding, but normals are still (rightly) puzzled. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F2c683588-ddbd-4c6d-92bf-c7daacd391e8.jpeg) Feb 22 · [Permalink →](https://paulkedrosky.com/n/68eca8d0-5698-4393-b9aa-95376c7c26d7) 3\. Microsoft price-earnings multiple converges Exxon Perhaps proving the IT capex intensity point I made earlier, Microsoft's price/earnings multiple has now converged with that of oil company Exxon. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fed11eb28-5bb4-4a0c-80c5-7bfa13d060df.jpeg) Feb 21 · [Permalink →](https://paulkedrosky.com/n/2eb2eeed-cd28-488a-be96-6ff5ebff3665) 4\. US IT investment hits multidecadal high The US is now, unsurprisingly, at a multidecadal high wrt IT investment as a percent of GDP. Large cap tech, is, if nothing else, now a capex-intensive business with large and growing maintenance capital requirements. One wonders what that's worth, and it isn't historical multiples. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F7e94700b-df43-4eb0-a520-538f4180405b.jpeg) Feb 21 · [Permalink →](https://paulkedrosky.com/n/8e107959-f660-4762-ae0c-af5afbcb70b4) _This post is for paying subscribers only._ ### Rough Notes, Feb 21, 2026: China AI Tea, Cybercrime, Cat Bonds, etc. URL: https://paulkedrosky.com/rough-notes-feb-21-2026/ Last updated: 2026-02-21T15:01:30.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. China AI labs launch models bubble tea promotions We are at "and you get a digital watch with that" phase of AI commoditization, with Chinese model makers offering bubble tea promos: China’s AI labs unleash new models and bubble tea to lure in customers ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F24b6f7cf-9bde-4a53-a238-524ff627cc22.jpeg) Feb 21 · [Permalink →](https://paulkedrosky.com/n/de30d73e-1d07-447c-825c-5e7886af4c2f) 2\. Cursor Declines as Codex Claude Code Rise My guess is that Cursor—among the fastest-growing products in recent tech history—will be in sharp decline by the end of this year, havng largely been supplanted by conversational orchestration layers like Codex and Claude Code. And then, of course, closed source orchestration layers like the latter will be replaced by open ones, in short order. Feb 21 · [Permalink →](https://paulkedrosky.com/n/cee7e7a1-e5f7-40b6-8bc9-85af8f3d2f84) 3\. U.S. Egg Prices Collapse Pre-Avian Flu Levels According to the most recent data, U.S. egg prices have now collaped back to pre-avian flu levels, which will come as no surprise if you've been to a grocery story and seen shelves overrun with eggs. As the old commodities line goes, the best cure for high prices is high prices (and vice-versa). ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F2203166f-d153-4c56-af1c-06133559cf09.jpeg) Feb 21 · [Permalink →](https://paulkedrosky.com/n/67b07b01-30f8-4666-8099-d2c11576be9a) 4\. Catastrophe-Bonds Flood Market Amid AI Risks It is important and ironic in light of the growing and over-determined nature of catastrophes to see the money flooding into cat(astrophe) bonds. While they are mostly weather related, cyber/Ai is now among the largest new vectors and the risk is wildly mispriced, given AI as attack force multiplier of all kinds. We now even have two new cat bonds ETFs, as well, piggy-backing on our increasingly risky society. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fd6268713-66bc-4099-b0a4-964485291436.jpeg) Feb 21 · [Permalink →](https://paulkedrosky.com/n/8c55998b-5747-423f-965e-3390c612fa5c) _This post is for paying subscribers only._ ### Rough Notes, Feb 20, 2026: AI Capex in Q4 GDP, Anthropic Revenues, Agentic Domains, etc. URL: https://paulkedrosky.com/rough-notes-feb-20-2026-ai-capex-in-q4-gdp-anthropic-revenues-agentic-domains-etc/ Last updated: 2026-02-20T15:27:48.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. AI Capex Drives 64-80 Percent Q4 GDP Growth Playing around with just-released (tepid) Q4 GDP figures, and my early estimate is that the floor contribution of AI capex to fourth quarter growth was on the order of 64%. And it is likely more like 80% with extenalities and multiplers. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Ff60441ad-7720-4394-b2c3-f9888ded6e3c.jpeg) Feb 20 · [Permalink →](https://paulkedrosky.com/n/4df9b3a4-8d5e-49d6-ac37-371232cbaea7) 2\. People underestimate enterprise software switching costs That people think vibe-coding crushingly boring apps to replace, say, Servicenow is of any consequence to said companies mostly explains how little people understand ahout enterprise software, large companies, switching costs, and crushing boredom. Current AI is hugely consequential to many things, but enterprise SaaS is (mostly) not one of them. Feb 20 · [Permalink →](https://paulkedrosky.com/n/5e070b33-6692-492b-9963-22d5d8c65bbf) 3\. Blue Owl Private Credit Faces Illiquidity Challenges While the Blue Owl private credit issues are both predictable and misunderstood—it is less about data centers than about illiquidity amidst a mass rush to exits—that it bragged about the massive demand for its assets in a recent sale, and yet chose to sell a large chunk to its own captive insurer is ... sub-optimal. Feb 20 · [Permalink →](https://paulkedrosky.com/n/925243f8-996a-41d8-8493-b4d510fe2142) 4\. Token Deflation Persists Amid Growing US-China Price Gap I spoke about this,in part, today at an event, but a) token deflation persists, and b) the U.S/China price differences are immense and growing. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F88c9a7c5-bf26-4388-81d8-6bd2890b1a02.jpeg) Feb 19 · [Permalink →](https://paulkedrosky.com/n/ebcf986b-8096-4aac-a675-bdf3cd23f4b0) _This post is for paying subscribers only._ ### Rough Notes, Feb 17, 2026: Japan, Warner, Electricity Prices, Data Centers, etc. URL: https://paulkedrosky.com/rough-notes-feb-17-2026-japan-warner-electricity-prices-data-centers-etc/ Last updated: 2026-02-17T16:22:51.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. Warner Brothers bidding war reignites with tender offers The not-so-bidding war for Warner Brothers sparked to life today, like a zombie fire in peat moss after a long winter. I am reminded that "best and final offer" is a largely meaningless phrase in corporate takeovers, and the only offer that matters is the once that causes the majority of shares to be tendered. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F5cb20df3-e599-4ea0-b21e-c6d56be75e0e.jpeg) Feb 17 · [Permalink →](https://paulkedrosky.com/n/6a9c0e46-b3fb-4b1d-ad4e-76d9d7c164ad) 2\. Japanese markets leverage robotics amid demographic pressures While Japanese markets have had a recent resurgence on fiscal and monetary actions, the underlying demographic pressure remains the same, if not worsening. It is a nearly perfect test case of the growing literature on how aging societies are forced to bercome early adopters of advanced automation, like robotics, which are everywhere in Japan, thus potentially having an unexpected productivity boom. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fa2609294-32ea-4938-ba62-72a6e6ca1153.jpeg) Feb 17 · [Permalink →](https://paulkedrosky.com/n/39eb7e93-6fb6-4546-a6ac-a756f007a6ea) 3\. Data Centers Concentrate US Economy Risks Data centers are among the most geographically concentrated assets in the U.S. economy, and thus in the world. That is both a blessing and a curse, depending on how you want to balance risk/cost vs NIMBY-ism. Source: ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fbb5888e3-e34a-41e7-8b16-1dd9d82d3cfc.jpeg) Feb 17 · [Permalink →](https://paulkedrosky.com/n/30de1672-9655-4e58-8ddd-e1052d4574ea) 4\. Data centers employ few workers despite energy use Data centers, as I've pointed out forever, employ very few people for all their energy footprint. And, more insidiously, they initially look like material employers during construction, and that collapses to very low once they are operating. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F95b34023-73dd-40ad-a6c8-7bffa013cc48.jpeg) Feb 17 · [Permalink →](https://paulkedrosky.com/n/84dd6e2c-7d57-45d1-b44f-d5196ded11fa) _This post is for paying subscribers only._ ### Rough Notes, Feb 16, 2026: China, Amodei, Unrest, etc. URL: https://paulkedrosky.com/rough-notes-feb-16-2026-china-amodei-unrest-etc/ Last updated: 2026-02-16T19:55:37.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. China conducts abortions matching US population "Over the \[one child\] policy's three-and-a-half-decade duration, China conducted nearly as mamy abortions ... as the present population of the United States." —Dan Wang, Breakneck Feb 16 · [Permalink →](https://paulkedrosky.com/n/522e85af-190d-4c42-bfdc-a3bd234a53a3) 2\. Anthropic cuts competitors from Claude Code Speaking of the recent Amodei interview, I did not realize until now that Anthropic had cut off its competitors from using Claude Code. That is an interesting signal. Feb 16 · [Permalink →](https://paulkedrosky.com/n/291e8a00-055c-4487-84bd-9e364cf73f5e) 3\. Zvi Dissects Amodei-Dwarkesh Bizarre Interview Good Zvi dissection of recent (bizarre) interview with Anthropic CEO Amodei, rightly calling him out for retroactively changing predictions, while pretending he didn't, and for he and host Dwarkesh ignoring minor things like catastrophic risk. As I said previously, a deeply strange interview, but Zvi analyses it in detail, so worth reading. Feb 16 · [Permalink →](https://paulkedrosky.com/n/1a027da4-0dad-49cc-b274-b6fc0f1fa92f) 4\. Humans Pay Human Touch Premium This is naive—humans will always pay for the human touch, at scale—and there are many rejoinders, so I'll just give one: Live music, chess, and fine dining are consumed as performances of humans’ humanness. It is not the same for spreadsheets, car assembly, or legal work. People DGAF who or what did i, and it will be "performed" accordingly with AI. AI and the ecnoomics of the human touch Feb 16 · [Permalink →](https://paulkedrosky.com/n/048cf26f-114a-45e0-9097-503786a9ff81) _This post is for paying subscribers only._ ### OpenAI Memo on Chinese Distillation of AI Models URL: https://paulkedrosky.com/openai-memo-on-chinese-distillation-of-ai-models/ Last updated: 2026-02-13T02:00:22.000Z _This post is for paying subscribers only._ ### Rough Notes: Feb 12, 2026 URL: https://paulkedrosky.com/rough-notes-feb-12-2026-2/ Last updated: 2026-02-12T22:01:31.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. Credit Card Interest Rates Remain Near Record Highs Credit card interest rates are in the news. Leaving aside calls to fix rates at lower levels, and the consequences thereof, what is interesting is where they are—near record highs—and how they got there. The primary reason: The Fed's sharp post-Covid rate rises trying to slash inflation. This raised the prime rate and forced issuers to raise variable APRs. Incrementally more pressure came from lenders widening their margins to cover risk and sustain profits. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fd1d25db3-522e-45f0-9756-f73713b29efb.jpeg) Feb 12 · [Permalink →](https://paulkedrosky.com/n/bf72aff2-762f-40aa-8f2b-eb2bae0e4ef2) 2\. Duke Energy Boosts 103B Capital Plan The latest updates to Duke Energy's spending plans are bananas: \- $103B five-year capital plan (up +18% since Nov 2025) \- Deploying >$1B per month. \- Data centers driving growth: 1.5 GW of new data center contracts since Nov \- Funding mix: \~$10B new equity issuance planned 2027–2030 \- Rate increases approved in SC, pending in NC. Source: ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F39647b58-fc43-4237-a311-4ccafe8a011d.jpeg) Feb 12 · [Permalink →](https://paulkedrosky.com/n/996952f1-28ff-4f00-8f7e-3d72abc53214) 3\. Tech Industry Captures Early AI Benefits As ever, most of the (early) benefits to technology accrue to the tech industry itself. I have written about this many times here, often citing a recent paper (). We see it again in the early days of the current AI, as Apollo points out here. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fdbd87924-2464-4e06-afd5-3648079e5f5c.jpeg) Feb 12 · [Permalink →](https://paulkedrosky.com/n/f8c759c2-38fd-4188-9c77-54a0a0cb8228) 4\. BEA-BLS Data Reveal White-Collar Hiring Stall GDP is growing, but white-collar hiring has stalled. New BEA/BLS data show that Finance, Insurance, Information, and Professional & Business Services jobs—40% of GDP—no longer grows with GDP. If this productivity shift sticks, it helps explain weak entry-level hiring despite GDP growth. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F43c0be3e-a975-471d-bbbe-40cfa290b393.jpeg) Feb 12 · [Permalink →](https://paulkedrosky.com/n/768668e7-9230-4ea5-8923-fd153cc6b4b6) _This post is for paying subscribers only._ ### Exponentials and AI: Debt, Fragilities, and an R&D Cascade URL: https://paulkedrosky.com/exponentials-and-ai-debt-fragilities-and-an-r-d-cascade/ Last updated: 2026-02-12T15:35:09.000Z > The greatest shortcoming of the human race is our inability to understand the exponential function. > *Albert A. Bartlett* A new [research note](https://metr.org/notes/2026-02-10-simpler-ai-timelines-model/) from METR, a nonprofit that evaluates frontier AI systems, argues that at current rates of progress there will be more than 99% automation of AI research and development by 2032, only six years away. ![](https://paulkedrosky.com/content/images/2026/02/2026-02-12_07-24-53.jpg) The model is very simple, with only eight parameters. It makes deliberately conservative assumptions: no superexponential capability growth, no "full" automation that exceeds 100%, no modeling of the additional acceleration that would come from AI systems developing research intuition. And yet, the median output is near-total automation of the work that produces more capable AI, within this decade. _This post is for paying subscribers only._ ### Rough Notes: Feb 12, 2026 URL: https://paulkedrosky.com/rough-notes-feb-12-2026/ Last updated: 2026-02-12T15:06:53.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. China-US AI spending-deployment contrasts manufacturing focus Striking differences wrt how China and US are spending on AI and deplaying it. Chna isn't spending nearly as heavily on frontier platforms, or on software for white collar workers, but is moving quickly to put tools into production in manufacturing, unlike the US. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fd1a29cf1-20ee-44df-866b-61ec558a64a6.jpeg) Feb 12 · [Permalink →](https://paulkedrosky.com/n/5c881019-daf9-4b0e-ac00-9070ce2c16eb) 2\. Healthcare Credit Spreads Widen Funding JPM AI Data Centers Debt markets are complex and connected systems, so money that flows into tech capex must come from somewhere, and that has consequences. JPM points out today that healthcare credit spread are widening over energy (shorter duration), in part, they argue, because healthcare credit is a source of funds for AI data centers, given similar durations. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fedaec329-7f7a-4c29-85de-b75f6cd78979.jpeg) Feb 12 · [Permalink →](https://paulkedrosky.com/n/a248fdbb-9116-4977-9708-b03407178300) 3\. Reconciling Non-residential Specialized Trades Job Growth and Data Center Construction Puzzling trends in non-residential specialized trades job growth over last twelve month, as FT Unhedged points out today. Despite data center center construction a growing factor in such trades, job growth decined in first half o flast year, only to acelerate sharply in second half. I'm guessing this was a denominator effect making what's happening, but it is notable. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F8b031823-0a03-49a5-85b3-488c27f11540.jpeg) Feb 12 · [Permalink →](https://paulkedrosky.com/n/e2e61658-fcb9-4bd8-aa5e-9ef4fc242651) 4\. Moody’s assigns Aaa rating Compass $500m ABS Compass got a Aaa rating from Moody’s recently for $500m of its latest ABS. The full Moody's justification relies heavily on three high credit tenants making up 96% (!) of the ABS income. This, of course, is a wildly fragile structure given the inherent correlation. Important moment in data center financing, given this will become a template.. Feb 12 · [Permalink →](https://paulkedrosky.com/n/4a00b5fd-d644-4840-9468-d4cade89ee39) _This post is for paying subscribers only._ ### Rough Notes, Feb 10, 2026: PIK, AI Debt, Regime Changes, etc. URL: https://paulkedrosky.com/rough-notes-feb-10-2026-pik-ai-debt-regime-changes-etc/ Last updated: 2026-02-10T15:41:51.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. Risky PIK Loans are a Blast From the past With all the discussion of risky PIK (payment in kind) loans in high yield and leveraged loans to enterprise software companies., I am reminded of this classic outburst from RJR Nabisco CEO Ross Johnson in Barbarians at the Gate when he heard about such financial engineering. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F9e2c9119-b79a-4b47-a1da-417a12fd11c2.jpeg) Feb 10 · [Permalink →](https://paulkedrosky.com/n/7fcd2635-c264-4bd4-8d24-80ef6707764d) 2\. Market Cap Per Employee Shift US Companies There is a provocative regime change underway in market cap per employee among large US companies, an artifact of the ongoing shift from labor to capital in modern economies. I saw this originally in the WSJ, but revised tghe graph and changed to log scale to make it less shocking and more useful. The Big Money in Today’s Economy Is Going to Capital, Not Labor - WSJ ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F2e45ce19-501f-4e4a-a62f-2c699a3d2083.jpeg) Feb 10 · [Permalink →](https://paulkedrosky.com/n/7060da8d-46e1-4efc-bbe0-c510cee7ee87) 3\. SaaS Sector Faces Risks from Leveraged Lenders The deeper issue with SaaS, isn't AI, it's that the sector is mature and increasingly the plaything of leveraged lenders doing buyouts and recaps. The result: softeware and technology combined now dominated the speculative credit market, making it much riskier than historically. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F6705cb71-08ce-4365-b232-91a19bb62e9d.jpeg) Feb 10 · [Permalink →](https://paulkedrosky.com/n/fad100ab-6490-4ad8-8599-1120f61a017f) 4\. Agentic AI accelerates Remote Work Index progress While agentic AI is currently fairly crap at most remote work, outside of code, where it is remarkable, it is improve quickly on CAIS's remote work benchmarkt. We could see, on current progress, material impacts by a year from now. Remote Work Index ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F3c167a9a-4373-493e-ba4f-cb60a39f1991.jpeg) Feb 10 · [Permalink →](https://paulkedrosky.com/n/e5530943-600f-4619-9f74-0c1ca6266cd1) _This post is for paying subscribers only._ ### Rough Notes II, Feb 9, 2026: AlphaDebt, Decoupled Electricity, H100 Rentals, etc. URL: https://paulkedrosky.com/rough-notes-feb-9-2026-2/ Last updated: 2026-02-09T23:04:10.000Z A digest of recent Rough Notes _This post is for paying subscribers only._ ### Rough Notes, Feb 9, 2026: Meta's Avocado, Confused Progress Arguments, AI and the Brain, etc. URL: https://paulkedrosky.com/rough-notes-feb-9-2026/ Last updated: 2026-02-09T16:00:31.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. Models Gains on Long Tasks Are Mischaracterized The METR graph cited by Brown is, btw, misleading at best, given its linear y-axis. Gains build, so will always look non-linear on a linear time axis. It should be a log-axis, and can see that then are two regimes, before reasoning models became the default choice and afterward. That was an architectural choice, not a breakthrough. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fa3eaa1e9-464e-4033-852c-6ace71fd3c6f.jpeg) Feb 9 · [Permalink →](https://paulkedrosky.com/n/aaef342e-520a-4eb1-9570-64c5f7d9c991) 2\. Meta Avocado LLM Claims Misinterpret Pretraining Metrics There is a thread going around in over-enthusiastic singularity circles about Meta's new Avocado LLM. According to a "leaked" internal memo, the company's pre-trained model is stronger than post-training peers, and more efficient. These claims are incoherent. Pretraining metrics (next-token loss, etc.) are not post-trained, instruction-tuned, RLHF-aligned systems. And more efficient at what? Tokens per joule? The claims also only compare against open weight models, which is to lower the net. Feb 9 · [Permalink →](https://paulkedrosky.com/n/ff7377ce-ff95-43ba-9178-a15143d3acfe) 3\. LLM Capability Gains Slow Despite Efficiency Confusion The attached is a confused argument, but typical. When people argue that LLMs gains are slowing, they generally aren't arguing that models aren't becoming more efficient. That is inevitable and predictable. They are arguing that capability gains are becoming more expensive to obtain, and fewer new domains accessed, per unit training $/W. This, from Brown, muddles things. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F78e88dff-3155-4f2f-bf2a-a88efeb4c444.jpeg) Feb 9 · [Permalink →](https://paulkedrosky.com/n/217be3ac-1bb3-4a8e-b001-4a1c852b26d4) 4\. Energetic Homeostasis Reveals Human Evolutionary Energy Constraints Energetic homeostasis and compensatory effects in humans are fascinating.. Complex systems are complex, of course, but more important here is that humans evolved during times of scarcity, with a thrifty gene. Simply exercising more or eating less doesn't evade a millon year of evolutionary pressure. The evidence for constrained total energy expenditure in humans and other animals ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F1e8fbdd3-399b-4a05-bbc8-6eb9f872b5d1.jpeg) Feb 8 · [Permalink →](https://paulkedrosky.com/n/5e6e63bd-171c-4257-a675-fbe4f7037acd) _This post is for paying subscribers only._ ### Rough Notes, Feb 7, 2026: AI and Natural Gas, Agentic Limits, Apple Freeriding, etc. URL: https://paulkedrosky.com/rough-notes-feb-7-2026-ai-and-natural-gas-agentic-limits-apple-freeriding-etc/ Last updated: 2026-02-07T20:01:39.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. Natural-Gas Expiry Spike Fueled by AI-Demand Last week’s unprecedented natural-gas expiry spike was, in part, an AI event. AI-linked data-center load raises the demand floor and concentrates stress regionally, making prices more spikey. In fragile settlement windows, already a problem, that turns small shocks into massive moves, as seen here. Natural Gas Traders Fume After NYMEX Glitch Triggers Market Chaos ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F12d31a88-b673-4ae2-b67e-b904ad799642.jpeg) Feb 7 · [Permalink →](https://paulkedrosky.com/n/09d0452d-3f10-4cd2-aef7-9db8b5524606) 2\. Apollo Chartbook Shows S&P 500 Employment Share Flat A new Apollo chartbook () on the declining importance of U.S. equity markets in the economy shows that S&P 500 companies are only \~18% of U.S. employment. But context and history matter. Here, as i've reconstructed it, is the extended series. The share largely flatlined over the period. I'm not sure that's "declining" importance, but it is interesting. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F5ad9113e-6607-4c96-b38d-0edb264a857e.jpeg) Feb 7 · [Permalink →](https://paulkedrosky.com/n/15d4d7f8-d6d1-4029-935a-e596da14603e) 3\. Epoch analysis reveals agentic AI limitations Good Epoch analysis of how (in)effective agentic AI is at real jobs, not merely benchmarks, math problems, or writing code. The answer, of course: Not very good. But, as ever, two things: 1/ This is the worst agentic AI will ever be, which is not true wrt humans. 2/ This should be a good reminder that current inference projections are wildly over-aligned with code, which is endogenous. Feb 7 · [Permalink →](https://paulkedrosky.com/n/8afbe003-e49a-46f1-9335-fdb7464f6612) 4\. Declining Market Cap Returns to AI Capex This went out as my chart of the day (subscribe at [paulkedrosky.com](https://paulkedrosky.com/)) late last week, but it shows the declining market cap returns to AI capex spending over the past three years. While never possible to specify with complete accuracy, it is directionally correct and thus useful as a sign of capex regime change. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Feef6cd08-005c-425e-a1a8-670abb488cfb.jpeg) Feb 7 · [Permalink →](https://paulkedrosky.com/n/a36b39f2-039f-4482-8773-25e816580ea5) _This post is for paying subscribers only._ ### Rough Notes, Feb 6, 2026: I'm Wrong, Clawbot Malware Vector, etc. URL: https://paulkedrosky.com/rough-notes-feb-6-2026-3/ Last updated: 2026-02-06T21:38:39.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. Derek Thompson Reverses Course on AI Buildout I see that Derek Thompson has reversed course from our discussion on the current AI buildout, arguing that I'm wrong: we are underbuilt for ... something. Again,I don't think AI itself is a bubble, just that we will vastly overbuild. My response to DT: We always underbuild GPTs (railroads, fiber, etc) until we overbuild. The demand signal he points to is largely endogenous—coding agents—not an external market clearing price. Feb 6 · [Permalink →](https://paulkedrosky.com/n/ece13dc5-77e1-4557-95d4-7a979205d82a) 2\. DeepMind reveals RL pitfalls in AI models Given declining quality data, AI models are becoming increasingly dependent on post-training—reinforcement learning and such. A new Deepmind paper argues, however, that current RL methods are fraught and even counterproductive, leading to behaviors opposite to what is expected. Hybrid neural–cognitive models reveal how memory shapes human reward learning ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fffb10d22-b22a-4f50-8f8b-a165fbe05f41.jpeg) Feb 6 · [Permalink →](https://paulkedrosky.com/n/31f6a15a-5ce0-4fa1-9cbf-7082b87c6780) 3\. Clawbot Malware Downloads Highlight AI Attack Surfaces This should come as no surprise, but will eventually matter: The top downloaded skill for the overhyped, fragile, and dangerous Clawbot seems to be malware. AI vastly expands attack surfaces, both knowingly and unkowingly. Feb 6 · [Permalink →](https://paulkedrosky.com/n/71044b1d-2941-4f42-aeeb-4e44dbe8532f) 4\. Matt Damon critiques Netflix action movie formula “The standard way to make an action movie ... was, you usually have three set pieces, One in the first act, one in the 2nd, one in the 3rd. You spend most of your money on that one in the 3rd act. That’s your finale. And now \[Netflix says\], ‘Can we get a big one in the first 5 minutes? We want people to stay. And it wouldn’t be terrible if you reiterated the plot 3 or 4 times in the dialogue because people are on their phones while they’re watching.’” \- Matt Damon on tech's consequences Feb 6 · [Permalink →](https://paulkedrosky.com/n/91ad5cd6-4529-4214-a52e-5a5fde701b4c) _This post is for paying subscribers only._ ### Chart of the Day: AI Capex, From Bonanza to Burden URL: https://paulkedrosky.com/chart-of-the-day-ai-capex-from-bonanza-to-burden/ Last updated: 2026-02-06T21:15:12.000Z Among the things the over-rational get wrong about capital markets is how continuing to do things for which the maret once rewarded you will eventually get you punished. We saw this in crypto treasury companies, where, for a surprisingly long period of madness, adding $1 of crypto to one's public company's treasury, added more than $1 to the subsequent market capitalization. That ended sometime last year, and has reversed, with crypto treasury companies getting smashed and promoters washing their hands of it all. We have seen a similar trajectory in AI-related capex, dating back to 2023\. Initialy companies were rewarded by markets for promising vast capex increases. The future isn't cheap! ![](https://paulkedrosky.com/content/images/2026/02/unnamed.png) But that reaction to spending is now reversing, as stock price reactions to outsized spending announcements have increasingly shown. This may be belated, but it was inevitable. Tech companies, once valued for their low debt and low capital intensity, are now among the leaders in both, a dubious distinction. And don't get me started about maintenance capital implied by current spending. Markets have come to similar views, reversing wrt the multiple it gives public companies for AI capex. Check the following graph, from my analysis, to see how the AI capex multiple—like its crypto treasury counterpart of yore—has collapsed, and likely gone negative. _This post is for paying subscribers only._ ### Chart of the Day: The Crypto Complex Sheds $2 Trillion URL: https://paulkedrosky.com/chart-of-the-day-the-crypto-complex-sheds-2-trillion/ Last updated: 2026-02-05T22:25:22.000Z The ongoing crypto complex crash has consequences. As of late this afternoon, the group [has shed $2-trillion in market capitalization](https://coinmarketcap.com/charts/) since its October peak. To put that massive number in context, crypto lost roughly ***$12-million a minute*** over the 113 day—162,720 minutes—period. ![](https://paulkedrosky.com/content/images/2026/02/images_72e27633-c5bb-42a8-8ec0-0b628b9fb3ce_9461363d-484b-44bd-8a1b-5a674a8c3206.jpg) _This post is for paying subscribers only._ ### Rough Notes II, Feb 4, 2026: AI Fraud, Neocloud TAM, Texas Killing Data Centers, etc. URL: https://paulkedrosky.com/rough-notes-feb-4-2026-3/ Last updated: 2026-02-04T22:00:46.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. AI Impact on Cybercrime and Fraud Good overview of how AI is a cybercriminality and fraud tailwind: Four facts: \- AI tools blur instructions and data, enabling easy misuse. \- Prompts, chats, and code are leaking via adjacent tools. \- Attacks are faster, cheaper, and more scalable than defense. \- Humans trust AI output too much, weakening judgment. The last one, in particular, is crucial and misunderstood. Feb 4 · [Permalink →](https://paulkedrosky.com/n/3f73e8fb-4589-40e9-a7f8-b4e69b65402b) 2\. AI Traffic Trends Challenge Neocloud Growth New network data from Backblaze quietly kills the neocloud bull case. AI traffic in an inference world is increasingly bursty, concentrated, and rotating back to hyperscalers post-training. That implies cyclicality, utilization risk, and a capped TAM. Neoclouds persist, but as shock absorbers for peaks, not as compounding infrastructure. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F39804043-e365-4b70-b6bc-a2ec4884c26f.png) Feb 4 · [Permalink →](https://paulkedrosky.com/n/93c97239-f5b8-4122-889b-6a69feef02a6) 3\. Texas Revises Data Center Grid Approvals Texas is flipping from YOLO data center grid connection to a centralized approach, putting dozens of speculative projects at risk. Keys: • Texas has >250 GW of requests vs \~85 GW of system capacity. • 8.2 GW of approvals are under re-review, the equivalent of \~8 nuclear reactors. • Texas is basically saying prior approvals are unreliable because the denominator exploded. Fun. Feb 4 · [Permalink →](https://paulkedrosky.com/n/4351998a-7006-42ac-824a-ae18430925d1) 4\. Prediction Market Microstructure and Bias I wasn't originally going to post this, because the data provenance is not clear—how does he have so much Kalshi data?—but it is interesting on the microstructure of prediction markets, and the ongoing wealth transfer. In particular, he shows the favorite-longshot bias at work: "Contracts trading at 5 cents win only 4.18% of the time, implying mispricing of -16.36%. Conversely, contracts at 95 cents win 95.83% of the time." ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fa846b043-3008-461e-a0b2-e0f49accd4aa.jpeg) _This post is for paying subscribers only._ ### Chart of the Day: The SaaSpocalypse Started Years Ago URL: https://paulkedrosky.com/chart-of-the-enterprise-tech-re-rating-started-years-ago/ Last updated: 2026-02-04T19:28:41.000Z There are endless headlines about the ongoing SaaSpocalypse, the re-rating of tech companies, especially in enterprise software, as AI is thought to be eating ... software itself. But as the following chart shows, it arguably began years ago, closer to the launch of ChatGPT in late 2022, even if it has recently accelerated. The following chart shows the re-rating of the enterprise tech sector: the declining price investors are willing to pay for software earnings. Granted, the P/E multiple is still hefty, in part because of the mix of holdings in the enterprise tech IGV ETF, but that shouldn't distract from a fundamental revaluation that's been going on for some time, despite an advancing equity market. ![](https://paulkedrosky.com/content/images/2026/02/2026-02-04_10-38-44.jpg) This has had myriad and growing implications, from [trouble in private credit and leveraged loans](https://www.bloomberg.com/news/articles/2026-02-02/private-credit-linked-shares-drop-on-software-exposure-concerns) to, well, declining net worth. Bloomberg [newly has out](https://www.bloomberg.com/news/articles/2026-02-04/software-billionaires-drop-62-billion-in-ai-driven-stock-slump) a helpful sample of enterprise software CEOs (and one enterprise software investor) and their wealth losses, now more than $60b in total. The individual numbers are large: - Adam Foroughi, AppLovin, –$7.8B, –30% - Dave Duffield, Workday, –$2.7B, –19% - Larry Ellison, Oracle, –$40B, –16% - Brian Armstrong, Coinbase, -$5b, –18% YTD (–44% since Oct) - Orlando Bravo, Thoma Bravo, \~$1.8B, –12% - Scott Cook, Intuit, –$2.0B, –17% Is the market correct with its savage AI-aided enterprise software re-rating? I have two views on this: 1. The tech industry itself has always been the primary productivity beneficiary of tech, as a recent [Chicago Fed study](https://www.chicagofed.org/publications/chicago-fed-letter/2025/515?utm%5Fsource=chatgpt.com) showed. And so a technology—AI—that reduces the marginal value of software to zero¹ is a ticking margin bomb for the industry. 2. The notion, however, that companies will began to self-satisfy by producing their own software, using AI, seems naive. A fundamental tenet of capitalism is that companies specialize in areas where they get the most leverage, and while they might cut enterprise software costs, it is unimaginable they will go full roll-your-own provider to themsleves². More like the software industry devolves into something more like Haliburton and similar services providers, consolidating and providing services around what is essentially a commodity. --- ¹ It doesn't really do this, so hold off on the cards and letters. Directionally, however, that is where we are headed, with a profusion of software—just try to find someone you know who hasn't newly "created" an app, I defy you—often produced at low cost and on a whim. ² And we have tried this in the past, in the early days of software, and it turned out badly, leading in part to the Y2K mess, as non-tech companies YOLO-ed their way along with home-grown and unmaintained COBOL code that turned out to run pretty much everything. ### Rough Notes, Feb 4, 2026: OpenAI as Liability, China HNW & AI, etc. URL: https://paulkedrosky.com/rough-notes-feb-4-2026-2/ Last updated: 2026-02-04T16:03:34.000Z 1\. OpenAI cited as Microsoft liability OpenAI was newly cited as a Microsoft liability in latest Microsoft earnings call. Inevitable realization. “There’s obviously concern about the durability, And I know maybe there’s not much you can say \[about OpenAI’s 45% share of Microsoft’s backlog of future sales\], but I think everyone is concerned about the exposure,” opined one analyst. Feb 4 · [Permalink →](https://paulkedrosky.com/n/4c79bc48-bb7b-4ad6-8147-007bfe9b4c34) 2\. Trends in Chinese Mass Affluent Wealth Some striking changes among HNW Chinese in latest survey of the country's mass affluent: • Chinese wealth is becoming younger and faster • Real estate losing favor to gold • Consumption shifting from ownership (fewer watches and cars) to more experience (travel, education, health, etc) • AI assistants are normalized, not experimental • Elite cognition increasingly mediated by AI Source: Hurun China Luxury Consumer Survey Feb 4 · [Permalink →](https://paulkedrosky.com/n/be2035e0-7d43-40c8-8f02-37699b4c314c) 3\. New Model Trained on Claude Output Kimi 2.5, the cheap, popular, and impressively capable latest model from Chinese company Moonshot, shows fairly clearly that it was trained extensively on Claude 4.5 output. While this can be defended against, it is hard, and again reminds that, for all the capex, there is no real moat in LLM capability. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fdc3179a1-aa74-403f-b20c-6a8f2fc02918.jpeg) Feb 4 · [Permalink →](https://paulkedrosky.com/n/24997fc3-6416-422d-8dfd-536375ddc1a8) 4\. Toronto condo market shows slowdown Betting against the Canadian real estate market's neverending rise has been among the great widowmaker trades in the world, up there with the yen carry trade. It is showing increasing signs of stress, however, as the immigration-driven condo-flip Ponzi in Toronto finally reverses. Check the data. What’s behind the slowdown in Toronto’s condo market _This post is for paying subscribers only._ ### Rough Notes II—Feb 4, 2026: Blue Owl, Huang, etc. URL: https://paulkedrosky.com/rough-notes-ii-feb-4-2026-blue-owl-huang-etc/ Last updated: 2026-02-04T00:46:43.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site, starting with me annoyedly thumping my chest about having (loudly) had Blue Owl right last year, and Apollo somehow getting the credit. 1\. Private Market Tech Lenders Go Boom While Apollo gets the credit in this Bloomberg piece on the SaaSpocalypse I was banging the Blue Owl tech lending risk drum through the second half of last year. Its exposure to both AI data center debt and software leveraged loans put in a "heads you lose, tails you lose" position. And when a partner there said it was the best risk-return moment of his career ... well. Blue Owl, Ares Lead Private Market Firms Bashed by Software Risk Feb 4 · [Permalink →](https://paulkedrosky.com/n/e0fc5fca-87b5-498c-a898-2a401f7b9b96) 2\. AI Build Out Will Not Lower Energy Costs This claim doesn't even rise to the level of wish-thinking from Nvidia's Huang. It is simply naive. Reasons: • Capex ≠ cheap power • Load growth is lumpy and localized • Energy ≠ electrons • AI doesn’t fix physics • Fuel mix reality of gas + firmed renewables I hardly know where to start with such nonsense. Nvidia CEO Says AI Build-Out Will Eventually Lower Energy Costs ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fc5909aa4-01e0-4f60-8a1a-ac7d7e56be55.jpeg) Feb 4 · [Permalink →](https://paulkedrosky.com/n/59784925-1261-462b-8554-2df2c6ed7a98) 3\. Crypto Market Cap Declines Accelerate Gold Flows Total crypto market cap has now declined almost $2 trillion since the October peak. While that is striking on its own, it has almost certainly had flow consequences, including accelerating flows into gold and other supposed havens. Perhaps proof of this, the decline re-accelerated in the past seven days as gold went geometric. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fcef928e2-4cea-4c11-b104-5f611343420c.jpeg) Feb 3 · [Permalink →](https://paulkedrosky.com/n/391a4365-07b0-4745-a2ab-881787850c2d) _This post is for paying subscribers only._ ### Rough Notes: Feb 3, 2026 URL: https://paulkedrosky.com/rough-notes-feb-3-2026-2/ Last updated: 2026-02-03T15:13:13.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. Utility Demand Claims Risk Overbuild Notes from recent Odd Lots discussion with CreditSights analyst on utility issues in AI buildout: 3/ Utilities' inventive is to claim higher demand: transmission, substations, and rate base all boost allowed earnings. That drives overbuild risk. if real power use disappoints, utilities lose the AI growth narrative and go back to being slow-growth, regulated, interest-rate-sensitive businesses. Feb 3 · [Permalink →](https://paulkedrosky.com/n/50ee4266-de37-45a2-b90b-ca959e19dd40) 2\. Utility Capacity Gap Disappears in AI Notes from recent Odd Lots discussion with CreditSights analyst on utility issues in AI buildout: 2/ Most third-party forecasts count compute only. Grid interconnects include cooling, lighting, and overhead. Once you adjust, the apparent shortage vanishes. Apples-to-apples, the capacity gap that underpins the “AI power constraint” narrative disappears. Feb 3 · [Permalink →](https://paulkedrosky.com/n/6ea0319f-7b7a-4d30-96d8-40f0080a89fa) 3\. Utility Capacity Exceeds AI Power Demand Notes from recent Odd Lots discussion with CreditSights analyst on utility issues in AI buildout: 1/ Utilities claim AI-driven power scarcity, but math points the other way: \~45 GW data-center load today, \~90–95 GW by 2030, versus \~140 GW of firm, contracted utility capacity already queued. After de-duping adjustment, committed supply likely exceeds plausible demand growth. Feb 3 · [Permalink →](https://paulkedrosky.com/n/7f8e5e2c-0f9f-415a-a5e5-4b36d553a86d) 4\. Charisma Defined as Social Paradoxes This is a good musing on charisma and why it's ... bullshit. tl;dr: Being charismatic means being good at social paradoxes. Charisma Is Bullshit—Or: A Theory of Charisma Feb 2 · [Permalink →](https://paulkedrosky.com/n/4ba3dbe4-7f78-4049-bebf-143c5853be00) _This post is for paying subscribers only._ ### Rough Notes: Feb 2, 2026 URL: https://paulkedrosky.com/rough-notes-feb-2-2026/ Last updated: 2026-02-02T15:11:24.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. Tesla sales decline sharply in Europe Tesla sales are (further) cratering in Europe, with a mere few hundred cars sold in France last quarter, Norway sales down almost 90%, etc. The company's run toward robots is, shall we say, multicausal. Tesla Sales Slump to Three-Year Low in France, Plunge 88% in Norway - Bloomberg ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F1c28a257-05db-4d61-b2d7-8c6d5f887e1b.jpeg) Feb 2 · [Permalink →](https://paulkedrosky.com/n/78a0dcfe-ed25-4755-8b53-d6c4d5b23cc7) 2\. Post Brexit job losses affect men Post-Brexit job losses in UK are more concentrated among male-dominated jobs, and the unemployed are increasingly gravitating to the populist right. This is an augury of what will happen in the US as it increasingly “Brexits” itself from trade networks via unpredictable policies and high tariffs. Britain’s Growing Ranks of Jobless Men Are Flocking to Farage ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F35860fa8-13c2-43c9-8a99-9836c57e19a0.jpeg) Feb 2 · [Permalink →](https://paulkedrosky.com/n/db53f723-b409-496c-9296-980b54dc6b84) 3\. Chinese Retail Traders Face Gold Losses The average Chinese retail speculator is basically just an American with even easier access to leverage, as shown by last week's gold implosion. Bloomberg has choice quotes. Rookie Trader’s Rapid 84% Wipeout Shows Depth of China Gold Bust ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fdec0952f-864c-4923-8082-e9a50c727041.jpeg) Feb 2 · [Permalink →](https://paulkedrosky.com/n/c78b141d-c3d8-4079-9c9e-3b9c99f6e1f7) 4\. Social Media Impact on American Life Charts from a terrific new piece on how social media has eaten Americans' time, made them less trusting, have fewer friends, and more. The title is an apt call-out to Putnam's classic "Bowling Alone". Scrolling Alone ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fe65f9361-491c-4ac3-9e63-3e1878cccd8f.jpeg) Feb 2 · [Permalink →](https://paulkedrosky.com/n/56cfe9b5-1415-40c4-accf-16c45205de15) 5\. Refuge assets declined in value The refuge complex—Bitcoin, gold, and silver—all got smashed this past week, suggesting that perhaps there needs to be a refuge from refuges. Somewhat seriously, that refuges, in this current closed-loop monoculture, become their own self-reinforcing bubbles, reinforced by China flows, neatly refuting the entire idea of a refuge. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fa67f5f7f-e3c2-43cc-be5f-461b21cd0e6f.jpeg) Feb 1 · [Permalink →](https://paulkedrosky.com/n/d0e95a2c-a0ba-4be0-9c01-279fe69f12d9) _This post is for paying subscribers only._ ### Kedrosky Daily: US bond spreads low; native unemployment diverges URL: https://paulkedrosky.com/kedrosky-daily-us-bond-spreads-low-native-unemployment-diverges/ Last updated: 2026-02-02T13:01:17.000Z _This post is for paying subscribers only._ ### Kedrosky Daily — Sunday Edition: Lowest bond spread in 25 years;Native unemployment rises post Trump URL: https://paulkedrosky.com/kedrosky-daily-sunday-edition-lowest-bond-spread-in-25-years-native-unemployment-rises-post-trump/ Last updated: 2026-02-01T13:00:33.000Z ### February 01, 2026 Edition This is a weekly free look of our premium Kedrosky Daily email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). ## Two Graphs ![](https://paulkedrosky.com/content/images/2026/02/2026-01-23_19-39-03.jpg) As of January 2026, the corporate bond spread is at its lowest level in over 25 years, indicating unusually high investor confidence and low perceived credit risk compared to previous periods, including post-crisis and recessionary spikes. ![](https://paulkedrosky.com/content/images/2026/02/2026-01-24_06-20-55.jpg) After Trump takes office in 2025, unemployment among native-born workers rises sharply while foreign-born unemployment drops, creating a significant gap between the two rates for the first time in the period shown. ## Headlines Trump’s Oil Grab Is a Big Problem for the OPEC Cartel ([**WSJ**](https://www.wsj.com/business/energy-oil/trumps-oil-grab-is-a-big-problem-for-the-opec-cartel-5012013d?gaa%5Fat=eafs&gaa%5Fn=AWEtsqc2cSs%5FDy1Jkhe-0qsaJPhSY2WZ0zkq%5FJzpH7fqOSHjGjXuzTI83Qe3&gaa%5Fts=697f5175&gaa%5Fsig=10PpsZ2Lsu1hA2Y4Atb2Eqj%5FWWxzDWMJ2dLs9mWEujU36yR2opXGO6JDDtWysR8oMnfUAib8Qb9RxRjoOclhYQ%3D%3D)) … Despite Trump’s claims, grocery prices are rising ([**NY Times**](https://www.nytimes.com/2026/01/14/business/food-prices-tariffs-trump.html)) … Asian shares hit record on valuations, gold slips ([**Bloomberg**](https://www.bloomberg.com/quote/BANBOLS:SM)) … A Wealth Tax Floated in California Has Billionaires Thinking of Leaving ([**NY Times**](https://www.nytimes.com/2025/12/26/technology/california-wealth-tax-page-thiel.html)) … ‘It’s Revenge of the Banks’ as Private Creditors Lose Their Edge ([**Bloomberg**](https://www.bloomberg.com/news/features/2025-12-28/citi-goldman-jpmorgan-lead-banks-revenge-on-private-equity)) … Private credit investors pull back ([**FT**](https://www.ft.com/content/8d7a9c3d-8e1c-40be-915c-7118c4946468)) … A.I., Big Tech and Trump Shine Most Brightly at the Davos Spectacle ([**NY Times**](https://www.nytimes.com/2026/01/20/business/davos-trump.html)) … U.S. Blockade Shuts Off China, Cuba From Venezuelan Oil ([**WSJ**](https://www.wsj.com/world/americas/us-venezuela-oil-blockade-china-cuba-fc0f195f?gaa%5Fat=eafs&gaa%5Fn=AWEtsqf%5F2XjdW53cyqadBnYCydBRaOevTY5YpworxoYZYW6h7Y1LnQyeGjuS&gaa%5Fts=697f51b5&gaa%5Fsig=gTowBxiAyUFTescsjtqam5ubN6E51ER9vour3QvsVRhEoBaVr9QRDh2h2fkCUUK4OFoO%5Fhs-dX5%5Fn2brpr42VQ%3D%3D)) … US Developer Glenfarne Launches Financing for Texas LNG Project ([**Bloomberg**](https://www.bloomberg.com/news/articles/2026-01-09/us-developer-glenfarne-launches-financing-for-texas-lng-project)) … Consumers Love Cheap Gas, but the Oil Patch Will Pay the Price ([**WSJ**](https://www.wsj.com/business/energy-oil/consumers-love-cheap-gas-but-the-oil-patch-will-pay-the-price-515d0957?gaa%5Fat=eafs&gaa%5Fn=AWEtsqd4CdcdwMVNITaqMxgzNe7P6ZRPM5xzI6iD20uj0UEjvlrTAZh%5Fdxs6&gaa%5Fts=697f514f&gaa%5Fsig=sGnXWGBuldYTRU4Wy1NmK72hyUuPmGIgXrX28s31F4Sq577mr3ffQ10H625wjeWHaVapfL5NqZ-ryaQwCz15MQ%3D%3D)) … Fine wine investors find little cheer in third year of falls ([**FT**](https://www.ft.com/content/e84d4fc6-35a6-434d-9ae8-693966348b6e)) … How special forces became modern warfare’s go-to solution ([**FT**](https://www.ft.com/content/a269a24a-04ea-4166-a978-bee35f094f08)) … Stock Relief Rally Extends With Boost From Tech ([**Bloomberg**](https://bloomberg.com/authors/AROe0%5FTGSWQ/andre-janse-van-vuuren)) … Polymarket ([**FT**](https://links.message.bloomberg.com/s/c/-jv3ym0HHURB0XFADIZXL3Ml53-k-k8vXAGeZFwnZnUKFCRxnnaAt5I5GT6aMapU2TcbGk33xC5Ri1pIl6Uk3NLkQkdpAx9fbBJFjdtu9bM6V1WU64kOHzm3ZYPVPYHCDZKqTzg5ry2ERCOHSLNyvzGXt4a8AN-Vlt3xGJbbyv9AMSOZpmVPA9VISIsI4SxpUJQ30Z5t-1e14TtMRuX5JnW1Wy1ClNpWH3X%5FM-e7NoZi6fHu9MhA5kmLoFAPhK1Yg%5FFH7SMxQvsdOkMU70vSJB9L7MTq1rTT2t9ceIbZi21v4tS5cSUZlveB%5FdggBYTMnRtF913QfVQbp6wTpYeDwAAPErnBxeeqMEH9SH6KYYW1JeFG%5FeIl-Gw%5Fu5k/F7Unxlrd9B1hdTYAYDzk-rndzi9wSIvC/13)) ### Kedrosky Daily: Low corporate bond risk;unemployment shifts by birth origin URL: https://paulkedrosky.com/kedrosky-daily-low-corporate-bond-risk-unemployment-shifts-by-birth-origin/ Last updated: 2026-01-29T13:03:11.000Z _This post is for paying subscribers only._ ### Kedrosky Daily: Lowest credit risk premiums;unemployment diverges by nativity URL: https://paulkedrosky.com/kedrosky-daily-lowest-credit-risk-premiums-unemployment-diverges-by-nativity/ Last updated: 2026-01-28T13:02:58.000Z _This post is for paying subscribers only._ ### Kedrosky Daily: Lowest US credit risk; native unemployment rises URL: https://paulkedrosky.com/kedrosky-daily-lowest-us-credit-risk-native-unemployment-rises/ Last updated: 2026-01-27T13:02:34.000Z _This post is for paying subscribers only._ ### Rough Notes: Jan 26, 2026 URL: https://paulkedrosky.com/rough-notes-jan-26-2026/ Last updated: 2026-01-26T17:13:20.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site: 1\. All US States Show GDP Growth Remarkable Q3 real GDP growth figures in US, with all 50 state economies plus Washington, DC, growing. The natioanl average was 4.4% annualized, the median grew at 4.5% annualized, and Kansas grew fastest, at at 6.5% annualized. While data center capex played a role—at as much as half of GDP growth earlier in year—as I have discussed endlessly, there has been a broadening. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fc61d187c-e7a8-4a2f-971d-ccdb57a69e70.jpeg) Jan 26 · [Permalink →](https://paulkedrosky.com/n/8a60d284-8f06-442e-b85c-2b5c0034f6d4) 2\. Retail Equity Ownership Matches 1999 Levels Retail equity ownership, even if it's been higher, was last at current levels in 1999, just before the dot-com crash, according to a new paper by Michael Mauboussin of Columbia and Morgan Stanley. The paper is worth reading in its entirety. Who Is On the Other Side? A Framework for Understanding Market (In)Efficiency ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fd137285c-1db7-46f1-a2f2-b0d08b9211b1.jpeg) Jan 26 · [Permalink →](https://paulkedrosky.com/n/bacd1a00-6f05-416a-a47d-60c7fdfb430f) 3\. Declining Returns on AI Model Investment Frontier AI model companies are seeing declining returns on compute and power investment wrt incremental model capabilities. It's not that models aren't improving, but that it isn't keeping pace with the associated time and cost. An FT piece argues this is analogous to what's happening with quant funds, creating a convergence of the two labs, to the point that one may eat the other. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F18f34c80-1bfc-4d1a-96c8-fcc29090ad31.jpeg) Jan 26 · [Permalink →](https://paulkedrosky.com/n/9b6c859b-dbae-4ed7-abce-213c346a1620) 4\. Rationality and Asset Valuation Bubbles In talks and papers I often describe AI as a "rational bubble", a finance term describing asset valuations and investment being increasingly disconnected from cash flows, despite all actors being rational, not merely bananas. There is a good new Chicago Fed paper laying and sythesizing the latest work on the topic, and it's a helpful framework. Why Bubbles Occur: Revisiting the Rationality Debate Jan 26 · [Permalink →](https://paulkedrosky.com/n/ad17cf2e-fa2e-47f6-9fbe-268fc9a06bab) 5\. US and EU AI spending diverges US and EU spending on AI continues to diverge widely, according to new Oxford Economics data. It is a confluence of things, from capital markets, to risk capital, to 'mericans being 'mericans, but the divergence is striking. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fcf57562d-f357-459f-b248-0b258e918d02.jpeg) Jan 26 · [Permalink →](https://paulkedrosky.com/n/4ba8a951-8bfd-41d1-8086-958349c9dde7) ### Nvidia Under Pressure: Coreweave Investment Plus Google TPUs URL: https://paulkedrosky.com/nvidia-under-pressure-coreweave-investment-plus-google-tpus/ Last updated: 2026-01-26T15:38:09.000Z Two developments today with broader implications for the state of AI semiconductors. The items: 1. Nvidia Invests $2 Billion More in CoreWeave, Offers New Chip ([Bloomberg](https://www.bloomberg.com/news/articles/2026-01-26/nvidia-invests-another-2-billion-in-coreweave-offers-new-chip)) 2. Google TPU Outlook Getting Stronger FY26 And FY27 ([JPM](https://markets.jpmorgan.com/jpmm/)) ### **The Delta** Nvidia invests (sic) $2B of equity into CoreWeave while committing to multi-year service purchases. Meanwhile, Google Cloud scales TPU capacity (powered by Broadcom), displacing Nvidia GPUs at growing scale. ### **The Implications** - **Equity as revenue leverage.** If CoreWeave grows, Nvidia receives *leveraged* revenue as much as 4x+ the equity. _This post is for paying subscribers only._ ### Kedrosky Daily: US bond spread lowest in 30 years;unemployment rises for natives only URL: https://paulkedrosky.com/kedrosky-daily-us-bond-spread-lowest-in-30-years-unemployment-rises-for-natives-only/ Last updated: 2026-01-26T13:05:15.000Z _This post is for paying subscribers only._ ### Rough Notes: Jan 25, 2026 URL: https://paulkedrosky.com/rough-notes-jan-25-2026-2/ Last updated: 2026-01-25T14:55:14.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. Tech Drives S&P 500 Margin Expansion Almost half of S&P 500 margin expansion is coming from tech, the largest share in decades, perhaps ever. As various studies have shown, tech tends to be the primary beneficiary of tech—in productivty and in margins—and that narrowness persists and is widening. Source: 1/ 2/ ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fd221b8bc-a5ee-476c-b120-d751bc968d3f.jpeg) Jan 25 · [Permalink →](https://paulkedrosky.com/n/27156e1c-93a2-4cbe-964d-44427da99d99) 2\. Senators Urge Data Center Financing Probe Senate Democrats are "encouraging" Treasury Secretary Bessent to look into "opaque" data center financings as a possible growing source of systemic financial risk. I have written on this umpteen zillion times, of course, for more than a year now, so it us amusing to see the U.S. Senate weigh in. Bessent Urged by Warren to Probe ‘Opaque’ Data Center Financings Jan 25 · [Permalink →](https://paulkedrosky.com/n/bdf2f7c8-2b9d-42b0-9cf3-adb0ef90946c) 3\. OpenAI advances rapidly with Codex risks As expected, Altman's OpenAI moving quickly and breaking things in a race to close the gap with Claude Code. He warns that his Codex will become more dangerous, yet patch bugs. This glib defense reminds me of the burglar wanting to be rewarded for breaking into your house, because he showed your security flaws. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Ffea530e5-ae0a-41eb-80ed-601ec7359f1a.jpeg) _This post is for paying subscribers only._ ### Kedrosky Daily — Sunday Edition: US Corp Spread Near Low; Unemployment Diverges After 2025 URL: https://paulkedrosky.com/kedrosky-daily-sunday-edition-us-corp-spread-near-low-unemployment-diverges-after-2025/ Last updated: 2026-01-25T13:03:04.000Z ### January 25, 2026 Edition This is a weekly free look of our premium Kedrosky Daily email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). ## Two Graphs ![](https://paulkedrosky.com/content/images/2026/01/2026-01-23_19-39-03.jpg) The ICE BofA US Corporate Index Option-Adjusted Spread fell to 0.73% in January 2026, close to its lowest level in nearly 30 years and significantly lower than spikes seen during the 2008 financial crisis and the 2020 pandemic. ![](https://paulkedrosky.com/content/images/2026/01/2026-01-24_06-20-55.jpg) After Trump takes office in 2025, native-born unemployment rises notably, while foreign-born unemployment stabilizes and then declines—a significant divergence that contrasts with prior trends where both groups moved similarly. ## Headlines Bill Gates Makes a Multibillion-Dollar Divorce Payout ([**NY Times**](https://www.nytimes.com/2026/01/09/business/dealbook/gates-divorce-settlement.html)) … The boss who gave his employees a $240 million gift ([**WSJ**](https://www.wsj.com/business/fibrebond-eaton-bonus-walker-30844d62?gaa%5Fat=eafs&gaa%5Fn=AWEtsqcU8q7Sakh6DNq-k78MvLfDEDOuGS9EG8hT%5FKhk%5FfCh%5FZ-Yuc%5F8%5FLFY&gaa%5Fts=69761720&gaa%5Fsig=CGXaqUZnq6XkGizR%5FdMhIppuI6oPWITvBBUfY-7PfqetddVib7YGGs3G9el81w3lAfQv2-hAFQqKC2qVDmskDA%3D%3D)) … Two people shot by federal agents in Portland, Ore. ([**WSJ**](https://www.wsj.com/us-news/two-people-shot-by-federal-agents-in-portland-ore-8e07c9ed?gaa%5Fat=eafs&gaa%5Fn=AWEtsqfI25h%5FnHtXoFh4h7kEIWaQU6k8-9RPvYCdjlMNVsHPXWxDVidpXgwT&gaa%5Fts=69761777&gaa%5Fsig=gGtbZ%5FmqsAxIwXztJxDd75YFZj55McHX20fWLQKwBJVBQ3GBN3u1bfpRtBrBzLRnINFAnRBBX4Jmmc2O-WIgUw%3D%3D)) … SK Hynix plans to invest $13 billion in new chip packaging plant ([**Bloomberg**](https://www.bloomberg.com/news/articles/2026-01-13/sk-hynix-plans-to-invest-13-billion-in-new-chip-packaging-plant)) … Consumers Love Cheap Gas, but the Oil Patch Will Pay the Price ([**WSJ**](https://www.wsj.com/business/energy-oil/consumers-love-cheap-gas-but-the-oil-patch-will-pay-the-price-515d0957?gaa%5Fat=eafs&gaa%5Fn=AWEtsqfGYkSoH6i1HWFZzcFRkxTxCY8lVeg7uzNQRu-35f3SY0wab1FLphMO&gaa%5Fts=6976176a&gaa%5Fsig=KSAVysjRaxX4OfQRDKcHg42SSMMp5HkV90fo5gmheKPPEsE3Nq1totyMtdCsYA7kY-IfHU-E5ceIjWtNkknhkg%3D%3D)) … Alibaba teams with China’s nuclear leader to power AI expansion ([**Bloomberg**](https://www.bloomberg.com/news/articles/2026-01-21/alibaba-teams-with-china-s-nuclear-leader-to-power-ai-expansion)) … Iran Edges Closer to a Revolution That Would Reshape the World ([**Bloomberg**](https://www.bloomberg.com/news/articles/2026-01-11/iran-edges-closer-to-a-revolution-that-would-reshape-the-world)) … Tech titans prospered in the first year of Trump’s return to power ([**FT**](https://www.ft.com/content/674b700e-765d-44e0-ba30-13b0c6c5abf1)) … ‘We Need to Be the News’: Inside Bari Weiss’s Bumpy Revamp at CBS ([**NY Times**](https://www.nytimes.com/2026/01/13/business/media/cbs-evening-news-bari-weiss.html)) … Uber Faces Growing Pressure Over Sexual Assault Record ([**NY Times**](https://www.nytimes.com/2026/01/21/business/uber-scrutiny-sexual-assault.html)) … BofA's Moynihan left off World Economic Forum headline list ([**FT**](https://www.ft.com/content/e78e4812-2334-11ea-92da-f0c92e957a96)) … Intel shares slide 11% as supply issues limit growth ([**FT**](https://www.ft.com/content/8a6f627f-ab8a-4591-9a7a-9c45ff54262a)) … Banks ([**FT**](https://links.message.bloomberg.com/s/c/2TLa-2nq%5FLJdsPpROJMs-sVdye0k7iHMnJVbU9QUN%5FbF90CsqwunVq4-UAY0rkw-KOf4%5FGiy-mNsl7DrqjVVxQqP9TIrtm%5FYyY20nQZBEipVNiTE90Wq4ciA4Ru%5FgW5GrBin91NDmG1P7hAloXND60ZF8jJz%5F19sP%5FFF7wkMDlxNlGviFkYqEP1gE%5FyUlCMJGl1Y5zcSHlJaeUCNChSlC8mhVGNmVvTBTAzlQd71mlQWteAzsuv86iGntFdyHl1p0pqdmqT%5FkNM2QlAGLY0-kANYpTBvV3kCzSdC01KdPddZrNWTK22rzTutI1LR5PLCECkdolU08IYNsu9Wrw4Ggau16N5TPPuJ5LSwKw6En9S0rUXYlmhTVS7hUs0/D1qreHd4g3BRpIC0b9qlgdpPrC3Ab7Py/13)) … Markets Wrap Stock Relief Rally Extends With Boost From Tech ([**Bloomberg**](https://bloomberg.com/authors/AROe0%5FTGSWQ/andre-janse-van-vuuren)) … Polymarket ([**FT**](https://links.message.bloomberg.com/s/c/gRTAbDEtA2Sj1dHhRC8Ju2TCG3eKeg6tM5ZI2gsezslVidNSXdwJTGtQb8dp6d-NEQiY2SHZ2sWzWzjeUeRImQ1QYKdvbu9kgFSIYeeCHJbobSf0nCYHRcCES%5FeH9tLrUcIhkwfTxB6m7Y%5Fyej9fNGj48MIoIuTdhgS8zCVixKs6IdtF-Y9xDbbE6d%5FzVOWHuC3E-F5ykHnG33GRGz-cYbZjQkEYCIeA5zN68mgszyUqAYuEW2IenDPOxqLV8YBdMbLacvKhM81WNd1oWoMj5bs0BtHbjqaGIUvSJXKNuWM4zUmIvzbWy9-5zxJ5QNRZY6jK75iktfCoG09lIGslxa92rFkdJsMrH8xCWzlomVKQYFmdH%5Fb3obENNNU/Ht-plCf6PFITuM9Wb11qQewuVzsBHQqC/14)) … US companies expand workforce to combat labor shortages ([**FT**](https://links.message.bloomberg.com/s/c/f9q5v2QEN0xQNAa7afmyt%5FvCDv6uTAE8CBvk%5F%5FXASxdNRDKB0lE8QAZzwY%5FBPwcMOHxIYXSE8H%5FIR3l5ewkaIOIKkJ%5Fo1ukGDGcJhOCJlIcRPbQnwE6-4fb3tDfbsrFG7uxpCNSM77MNaJwkVMNPGM0OHMAOBFpNMTZBA8WkuotLohtLzDxz9dpeh5CopzSBF-0Oe6blDwftiulYAWB%5FzS4ibOISjkDKKKK6ufTMKojLXoFLTM06Obj5qYmi7yChJTcxMCAylXhk5JJUzYrVyBygSuaxEX%5FcxhewCpse1apn8B0NJ1zQkY7XNIdkoVqJIR6CYpXOFsSK1rqu8SuzoqhLLhKlnW8vHiac7u9EDHLDVUs2kL5dXOntYeg4%5FX486UdBl3uN41UdX-imEvLzPiE09zCSm%5FqhHj4u0dbeIUa3FBmqMcKqTc0/nFyr1eo1aRaaYIOtZ1uyACMkzJQrcszp/8)) ### Claude Code and OpenAI's Codex Code Red URL: https://paulkedrosky.com/claude-code-and-openais-codex-code-red/ Last updated: 2026-01-24T17:04:20.000Z Anthropic's Claude Code's—a semi-autonomous AI-driven coding tool with wide applicability—surge has OpenAI in yet another Code Red. This is clear from Altman's OpenAI promising [in a tweet yesterday](https://x.com/sama/status/2014733975755817267) a month of updates for its own Codex tool, albeit with ... societal consequences. ![](https://paulkedrosky.com/content/images/2026/01/2026-01-24_08-49-29.jpg) The announcement combines recklessness, circular logic, and reward-the-burglar logic in making the case; it turns unresolved risks into assertions. _This post is for paying subscribers only._ ### Rough Notes, Weekend Edition: Jan 24-25, 2026 URL: https://paulkedrosky.com/rough-notes-jan-23-2026-7/ Last updated: 2026-01-23T22:06:50.000Z Recent updates on our Rough Notes _This post is for paying subscribers only._ ### Rough Notes: Jan 23, 2026 URL: https://paulkedrosky.com/rough-notes-jan-23-2026-5/ Last updated: 2026-01-23T15:59:55.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site: 1\. AI is increasing software production A few years ago I wrote a piece on how we have structurally underproduced software for decades, and how AI was ending that. It is now fascinating to watch that happen, whether in new App Store launches, or just former UX people discovering they are their own dev team. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fff19a274-e07d-486a-936c-b9ac52fcabea.jpeg) Jan 23 · [Permalink →](https://paulkedrosky.com/n/ff2989a6-1025-451c-9f6d-2c933707886f) 2\. Toilet Maker Shares Rise on AI Demand Today in The AI Supercycle is Weird: toilet maker Toto. Goldman tipped the company's sideline in electrostatic chucks, used to hold silicon wafers in place during chipmaking, as an AI buildout play. The stock popped 11% on the news. What a world. Jan 23 · [Permalink →](https://paulkedrosky.com/n/d8535c0d-02bb-4cd2-a0d7-062cc17b7528) 3\. Intelligence Can Change History “To change the world, to change history, you don't need superintelligence. Very dumb intelligence is still enough to change history.” —Yuval Harari Jan 23 · [Permalink →](https://paulkedrosky.com/n/7ce34cd7-7486-4970-bd8b-b69b1fa7baed) 4\. Superintelligence and Financial System Impact From the sidelines of Davos, Yuval Harari on what superintelligence means to him: "That it can make $1,000,000 on its own ... it can do everything, including opening its own bank account and it can make $1,000,000\. Then it's superintelligence and then you can have millions of those taking over the financial system." We will see that very soon, and, of course, it's instantly pyrrhic at scale. Jan 23 · [Permalink →](https://paulkedrosky.com/n/6795e619-983b-48a8-a8e8-7dc31e574cd5) _This post is for paying subscribers only._ ### Kedrosky Daily: TTF outperformed indexes; iOS app growth surged URL: https://paulkedrosky.com/kedrosky-daily-ttf-outperformed-indexes-ios-app-growth-surged/ Last updated: 2026-01-23T15:20:14.000Z _This post is for paying subscribers only._ ### Bonus Chart: App Growth Soaring in App Store URL: https://paulkedrosky.com/bonus-chart-app-growth-soaring-in-app-store/ Last updated: 2026-01-23T00:39:14.000Z Terrific graphic from Coatue [tonight](https://www.coatue.com/c/takes/chart-of-the-day-2026-01-22), highlighting (based on Sensor Tower data) the sharp rise in the number of apps in the Apple App Store over the past eight months. While you can craft many explanations for this, the best one is likely that the cost of producing trivial apps has, under AI coding tools, collapsed, as has the difficulty of setup, glue code, testing, and App Store compliance. (I'm not sure "agentic coding" line belongs, but whatever.) ![](https://paulkedrosky.com/content/images/2026/01/2026-01-22_16-22-24.jpg) This is what has happened in every other digital media market, from content, to video, to marketing, to audio, so the rise of low effort "AI app slop" should come as zero surprise. But it will have consequences. Some follow. ### Implications: - **Supply shock**: App count growth accelerates faster than user demand. Discovery, not development, becomes the constraint. - **Severe crowding**: App Store search and charts saturate; marginal apps see near-zero organic installs. - **Hit dynamics skew**: A few breakout hits capture disproportionate attention; median outcomes are nada. _This post is for paying subscribers only._ ### Everything That AIs Must Converge URL: https://paulkedrosky.com/everything-that-ais-must-converge/ Last updated: 2026-01-22T19:32:16.000Z > “With grown people, a road led either to heaven or hell, but with children there were always stops along the way where their attention could be turned with a trifle.” > ― **Flannery O'Connor,** Everything That Rises Must Converge A growing thread in recent AI research is less about making them more capable, but about why what they produce is so meh. In a [recent paper](https://www.cell.com/patterns/fulltext/S2666-3899%2825%2900299-5), researchers explored this, linking image generators and vision models into closed feedback loops: image → description → image. The result? Diversity didn’t expand. It collapsed. Starting from 700 different prompts and various sampling parameters, every run converged onto a small set of visual tropes: stormy lighthouses, gothic cathedrals, palatial interiors with dramatic lighting. ![](https://paulkedrosky.com/content/images/2026/01/2026-01-22_11-01-41.jpg) What’s happening is best understood as cognitive gravity. Training a model defines a probability landscapes. When you chain model inputs and outputs together, like with agents, you’re not exploring some creative space; you’re watching an algorithm traversing gradient descent. ![](https://paulkedrosky.com/content/images/2026/01/2026-01-22_10-50-58.jpg) Each step nudges the system toward local maxima in the distribution learned from the internet. Different temperatures added noise to the path, but didn't change the destination. In fact, pooling reduced the number of distinct clusters, making it even more meh. Why? Because the underlying training data constraints were what was binding, not the parameters. There is a human analogue here, of course. Stories passed on from memory have always drifted toward shared cognitive biases. Flood myths, for example, recur cross-culturally. But there is a difference. Humans have corrective feedback: dialogue, disagreement, intent, etc.. AI feedback loops ... don’t. They’re pure transmission, amplifying whatever statistical patterns are frozen into their training data. This generalizes well beyond art and images. Anywhere models evaluate or reinforce their own outputs, like in agentic systems, the same dynamics apply: language models refining their own reasoning, trading algorithms reacting to each other’s signals, recommendation systems driving over-crowding in tourism. These are feedback loops in high-dimensional probability spaces, each with its own attractor basins. _This post is for paying subscribers only._ ### Rough Notes: Jan 22, 2026 URL: https://paulkedrosky.com/rough-notes-jan-22-2026/ Last updated: 2026-01-22T17:46:58.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site: 1\. AI does not increase developer productivity New paper studying AI-related productivty gains among software developers showers cold water on the idea that Claude/Cursor/Codex et al., ring in a regime change in developer productivity. Artrificial Intelligence in the Firm ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F0f00a555-f55c-4f20-a98c-0ffef14d26c2.jpeg) Jan 22 · [Permalink →](https://paulkedrosky.com/n/3f0e9545-a86f-499c-9c66-df30a9b3ef61) 2\. Coding Tools Do Not Replace Companies Various smart folks making naive arguments about how the emergence of coding tools like Claude puts paid to orthodox software companies, and explains recent selling of the group. This is, at best, naive. People will always pay to rent strangers to write and maintain code. That it's easier to write code in corner cases solves a different problem, even if an important one. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F6937c596-7f1d-4722-aa66-b79fde1646f2.jpeg) _This post is for paying subscribers only._ ### Are Zero-Return Decades an Outlier? No URL: https://paulkedrosky.com/are-zero-return-decades-an-outlier-no/ Last updated: 2026-01-21T15:00:11.000Z An onoing chart war started with [this widely shared graph](https://www.apolloacademy.com/wp-content/uploads/2025/12/121125%5FChart.pdf) from Apollo. ![](https://paulkedrosky.com/content/images/2026/01/2026-01-20_13-11-19.jpg) It shows how forward price-earnings multiples help predict 10-year annualized equity returns on the S&P 500\. In particular, they can even produce negative return decades. That should be uncontroversial, but it wasn't. Analysts at growth investors Coatue [weighed in](https://www.coatue.com/c/takes/our-response-to-a-widely-circulated-market-valuation-chart), arguing that zero-return decades in equity markets are an "outlier", and that are unique to the .com era. They showed that in the following replication of the Apollo graph, highlighting post dot-com era decades' crap returns. ![](https://paulkedrosky.com/content/images/2026/01/2026-01-20_13-11-39.jpg) Who is right? Neither, honestly. Both make three serious errors: 1. They constrain the data series to post-1990, limiting the number of overla\[\[ing decades. 2. The post-1990s period was a regime change in equity markets, an era of falling rates and controlled inflation. 3. They ignore look-ahead bias in forward estimates. Here is what happens if you fix those issues. I went back to 1900, giving more decadal periods and reducing the importance of the dot-com episode. I also used Shiller's cyclically-adjusted price-earnings multiple (CAPE) to reduce the impact of cycles and estimates. You can see the result below. While the dot-com era poor decadal returns are still there, there are many ten-year periods with negative returns, most of them starting from much lower forward multiples than the dot-com era did. You can also see, via the color coding, the regime change in post-1990 data. Declining interest rates led, in large part, to higher equity market valuations overall, as showm by lower multiples with similar, if messy, effects on outcomes. ![](https://paulkedrosky.com/content/images/2026/01/2026-01-20_13-12-13.jpg) The takeaways: - Valuation is **not destiny** - Valuation is also **not irrelevant** - Valuation is a **persistent, first-order constraint** on long-run real returns across regimes - Dot-com is **neither necessary nor sufficient** to establish that fact **Bonus:** Shiller CAPE graph, showing the current historical outlier CAPE: ![](https://paulkedrosky.com/content/images/2026/01/2026-01-20_13-23-39.jpg) ### Kedrosky Daily: Private credit assets surge by 2025; Blue Owl Tech redemptions spike in Q4 2025 URL: https://paulkedrosky.com/kedrosky-daily-private-credit-assets-surge-by-2025-blue-owl-tech-redemptions-spike-in-q4-2025/ Last updated: 2026-01-21T12:58:30.000Z _This post is for paying subscribers only._ ### Four Takeaways from Microsoft's Satya Nadella at WEF Today URL: https://paulkedrosky.com/four-takeaways-from-microsofts-satya-nadella-at-wef-today/ Last updated: 2026-01-21T01:04:42.000Z Four AI takeaways from Microsoft's Satya Nadella [at WEF](https://www.weforum.org/meetings/world-economic-forum-annual-meeting-2026/sessions/conversation-with-satya-nadella-ceo-of-microsoft/) today. While Nadella speaks consummate corporate, rarely saying anything, he was more candid this time. ### 1\. AI is a commodity input, not a differentiated product > “GDP growth in any place will be directly correlated… to tokens per dollar per watt." He repeatedly framed tokens in terms of “tokens per dollar per watt.” That is not sexy marketing language. That is a commodity's cost curve. AI is being reduced to an economic input whose value is determined by energy, infrastructure, and efficiency. That puts it closer to electricity or compute than to software platforms. Once you accept that framing, many current AI narratives stop making sense. --- ### 2\. Token prices are falling even faster than people are modeling > "And by the way, look at the token pricing, right? Token pricing basically drops by, you know, a half every three months.” The claim that token prices are dropping by roughly half on a very short cycle should be startling. Extreme deflation changes behavior. It pulls demand forward, encourages overbuilding, and shortens the useful life of capital. It also means today’s impressive revenue numbers can coexist with tomorrow’s margin compression. If tokens are collapsing in price this fast, then any business model that depends on sustained scarcity at the model layer is fragile. As I explained last week on a trip to New York, volume has to explode, or value has to move elsewhere. --- ### 3\. Benefits are still flowing mainly to tech firms — and that’s a warning sign > “A telltale sign of if it’s a bubble would be if all we are talking about are the tech firms.” _This post is for paying subscribers only._ ### Rough Notes: Jan 20, 2026 URL: https://paulkedrosky.com/rough-notes-jan-20-2026/ Last updated: 2026-01-20T15:23:06.000Z Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site: 1\. Modern Music Trends Reflect Societal Mood Modern music is getting sadder and angrier, which is a form of diagnostic about where we are in this societal cycle. You can look at this with a Fourth Turning lens, or choose your perspective, but it is an important long wave, accelerating with geopolitics and AI. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F43761708-9d57-4f9c-8c0c-b93703fef1df.jpeg) Jan 20 · [Permalink →](https://paulkedrosky.com/n/564c03f4-d35c-4c84-8b73-a064401cd406) 2\. AI Commercialization Increases Global Risks AI/AGI have been turned into "normal business", which is profoundly risk-increasing: 1/ ChatGPT i Starting to Look a Lot Like Facebook and Meta 2/ Anthropic CEO Says AI Chip Sales to China Like Selling Nukes to North Korea Jan 20 · [Permalink →](https://paulkedrosky.com/n/5afae82c-6474-4b4b-bb40-21f5de283889) 3\. Documentary on DeepMind AGI success Documentary on Google DeepMind and the quest for AGI, has become a surprise success. WSJ: There’s a Hit Movie Set Deep Inside an AI Lab—and It Will Give You Goosebumps The Thinking Game: Full documentary Jan 20 · [Permalink →](https://paulkedrosky.com/n/fa1b7390-b1dc-42d0-9e7a-94a30ed0e7ce) 4\. Data Centers Use Less Cement One of the (many) supercycle arguments made around the economic ecology of AI data center is that were are in a Cement Supercycle. After all, so goes the argument, we are building many data centers, all of which requires immense amounts of cement, just in time to ;pick up the slack from the decline in China demand. BUt ... as a recent JPM report points out, data centers aren't all that wrt cement. They only use about a third the cement of comparablly sized CRE. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F1b6f7f9d-e798-42e0-b909-06c67c20c0b6.jpeg) Jan 19 · [Permalink →](https://paulkedrosky.com/n/7d6c1f97-e0af-4f38-9751-2d60ba90ce77) 5\. Chinese Electric Vehicles Gain UK Market Share Two pieces on the Chinese takeover of European EV markets caught my eye today: \- The U.S. has functionally exited EV competition, with it going from strategic to orphaned \- China share has soared from 1% to 10% of UK market in three years \- The Temu "Range Rover" clone is among the hottest cars in U.K. roads Sources: ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F891ab5b8-c317-43bb-840e-a0d93877ccd4.jpeg) Jan 19 · [Permalink →](https://paulkedrosky.com/n/ae3cebcc-a413-429e-b413-05b0c968e869) ### Rough Notes: Jan 19, 2026 URL: https://paulkedrosky.com/rough-notes-jan-19-2026/ Last updated: 2026-01-19T21:12:22.000Z Once again, here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site: 1\. Top Scientific Ideas of the Century And now ten of the best ideas of the current century, again from New Scientist this week: \- Electrifying everything \- A computer in every pocket (smartphones) \- Modern AI foundations (deep learning at scale) \- CRISPR and gene editing \- mRNA and preventative therapeutics \- Real-time astronomy and sensing \- Climate attribution science \- Recognition of brain diversity \- Open, networked knowledge hubs \- Dietary evidence convergence Jan 19 · [Permalink →](https://paulkedrosky.com/n/0c2ea0d1-1fbf-4f45-953b-39c9a8ac6130) 2\. Five Worst Ideas of the Century Eclectic and amusing list of the first worst ideas of the last century (from New Scientist): \- Bitcoin / proof-of-work cryptocurrencies \- Profit-driven social media \- Carbon offsets \- Most alternative liquid fuels (biofuels, hydrogen, synthetic fuels) \- Effective altruism (as practised) Jan 19 · [Permalink →](https://paulkedrosky.com/n/59c1e776-ce84-4c16-bb3b-4bb1e6e9be39) 3\. IMF warns AI risks global economy Ai has made the global economy more fragile, the IMF argues in an economic outlook update today: "Risks to the outlook for the global economy remain tilted to the downside. ... Should expectations about AI-driven productivity gains turn out to be overly optimistic and outcomes disappoint, a sharp drop in real investment in the high-tech sector as well as in spending on AI adoption ... could ensue." Jan 19 · [Permalink →](https://paulkedrosky.com/n/01339fa5-06ea-4ead-8fc3-b478804d065c) 4\. Tech Spending as Percent of GDP Tech spending, as a percent of GDP, and largely driven by AI, is back to its 2001 all-time high. Remarkable. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F389f92a3-ccf7-4953-82aa-e62f4eb35ecd.jpeg) Jan 19 · [Permalink →](https://paulkedrosky.com/n/60995bf1-5445-4d07-a4a8-d05b619dc0aa) 5\. OpenAI CFO reports dollar sales discount Amusing reading from OpenAI CFO bragging that they are successfully selling dollars for $0.70 in huge volume. Jan 19 · [Permalink →](https://paulkedrosky.com/n/09c48d99-9637-48dd-993a-1c7a0300201d) 6\. LLMs replicate human social dynamics LLMs don’t reason morally; they absorb how humans interact: contracts, hierarchy, coercion, markets, war. Under stress, they reproduce the same strategies. Blackmail, thus, isn’t anomalous. The risk, as a result, isn’t hostile intent but amplification: AI compresses time, removes friction, and forces hidden human contradictions to emerge at scale faster than institutions can adapt.. Jan 18 · [Permalink →](https://paulkedrosky.com/n/6e6f91c4-74fb-4c47-b05e-2e7ab3c65b7d) 7\. People Punish Others Using LLMs New study shows people "punish" others who use LLMs for work: \~36% of earnings from those who relied solely on LLMs, with punishment rising as use increased. And if you claim no use” people don't believe you. Jan 17 · [Permalink →](https://paulkedrosky.com/n/d0d27f43-4b2b-4db4-9477-f872680f7417) 8\. Global AI Power Usage Matches New York Global AI power usage now rivals that of New York state on peak summer days, around 30 GW, and, of course, continues to rise sharply,added sovereign grid-scale capacity every 2-3 years, no longer decades. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F3feedb37-bfe6-483b-b77f-7919ee2af465.jpeg) Jan 17 · [Permalink →](https://paulkedrosky.com/n/3c729241-a292-4be0-a905-a4573e3365dd) 9\. Limitations of Amazon Commissioned Report In particular, here are a few things Amazon gets wrong about its own commissioned report, leading my study quibbles: \- The study is partial equilibrium, not full. It doesn't model system changes in supply and demand, thus making it unable to support Amazon's claims. \- Revenue is estimates marginal cost at the time of the subsidy, thus excluding margin tightening, externalities, etc. \- "Surplus revenue" is an accounting term, at best, not ratepayer relevant Jan 16 · [Permalink →](https://paulkedrosky.com/n/65b49d2a-10bf-45b6-98c7-b66bcf6c703d) 10\. Amazon Misrepresents Data Center Study Wrt data center propaganda, the Amazon spin of a study it commission—showing no data center effect on rates, it trumpeted—is particular problematic. The study it cites, which it paid for, does not make as heroic of claims as Amazon assigns to it. Jan 16 · [Permalink →](https://paulkedrosky.com/n/cf2273fd-f1a4-43f3-9888-f2dc05d93471) ### Rough Notes: Jan 16, 2026 URL: https://paulkedrosky.com/rough-notes-jan-16-2026/ Last updated: 2026-01-16T14:44:09.000Z The following is a new, regular email, based on my ["Rough Notes"](https://paulkedrosky.com/thenotes/) feature on the site. It will send out digests of my regular notes, like what you see below. Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. Limitations of Amazon Commissioned Report In particular, here are a few things Amazon gets wrong about its own commissioned report, leading my study quibbles: \- The study is partial equilibrium, not full. It doesn't model system changes in supply and demand, thus making it unable to support Amazon's claims. \- Revenue is estimates marginal cost at the time of the subsidy, thus excluding margin tightening, externalities, etc. \- "Surplus revenue" is an accounting term, at best, not ratepayer relevant Jan 16 · [Permalink →](https://paulkedrosky.com/n/65b49d2a-10bf-45b6-98c7-b66bcf6c703d) 2\. Data Center Study Claims Misrepresented Wrt data center propaganda, the Amazon spin of a study it commission—showing no data center effect on rates, it trumpeted—is particular problematic. The study it cites, which it paid for, does not make as heroic of claims as Amazon assigns to it. Jan 16 · [Permalink →](https://paulkedrosky.com/n/cf2273fd-f1a4-43f3-9888-f2dc05d93471) 3\. Data Center Community Relations Efforts The propaganda machine around data centers is spinning, unsurprisingly given the measured effects on rates, NIMBY-ism, and a general sense in communities these great beasts are slouching into their neighborhoods in ways they don't condone. Two recent examples: 1/ 2/ Jan 16 · [Permalink →](https://paulkedrosky.com/n/d07d8ebf-43c3-440d-9da0-e22ee9a7bfc8) 4\. AI Impact on Human Reasoning Good phrase from new paper on how AI is breaking brains: Cognitive surrender: the uncritical adoption of AI-generated outputs as one’s own judgments, bypassing intuition and deliberation and transferring cognitive control to the system, reducing one’s ability to reason Thinking—Fast, Slow, and Artificial: How AI is Reshaping Human Reasoning and the Rise of Cognitive Surrender Jan 16 · [Permalink →](https://paulkedrosky.com/n/38bfac91-2816-4967-aeb9-2a8f1a75ce99) 5\. Factors Driving Near Record US Equity Prices The confluebce of foreces that have produced near-record equity prices in the U.S. is fascinating, from the rise of AI, to rates, to economic uncertainty leading to over-allocation to equity, to TINA (there is no alternative). The effect is fragility with an over-determined break almost ceertaintly ahead. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fe76fd8d1-1be3-4798-bc78-c0b6bb622b07.jpeg) Jan 16 · [Permalink →](https://paulkedrosky.com/n/c7bc4765-a871-47e5-9658-30fa7a5f8665) 6\. Younger Generations Increase Stock Market Investment Younger generations, increasingly uneasy with a post-AI future, are investing more in the stock market, seeing it as a lottery ticket with which they can protect themselves. This is entirely predictable, this rise of gambling in the face of rising economic uncertainty, and has been seen in research repeatedly. \[Source: Oliver Wyman Forum survey\] Jan 15 · [Permalink →](https://paulkedrosky.com/n/fadefdc5-be99-4208-9dcb-8247a0538eea) 7\. Barclays Report on Robot Labor Replacement The sell-side infatuation with robots knows few limits, other than replacing themslves, of course. A new Barclays report is a full-on rave wrt future robot-centric human labor replacement. Worth scanning, in part because robot-related token usage floats many AI inference boats ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fc366e396-e419-43c8-b814-cfecc621994d.jpeg) Jan 14 · [Permalink →](https://paulkedrosky.com/n/9b629905-9cb5-4d13-bfe3-34fa6d621283) 8\. Roubini bullish on AI GDP growth My usually-gloomy friend Nouriel Roubini sounding remarkably bullish about the AI-related GDP growth prospects for the U.S., even if much of the benefit doesn't necessarily flow to tech companies. < Jan 14 · [Permalink →](https://paulkedrosky.com/n/1373fbea-6068-402a-9533-8e94ebffa01e) ### Kedrosky Daily: Prediction market volume hit $8bn; Humanoids market may top $200bn URL: https://paulkedrosky.com/kedrosky-daily-prediction-market-volume-hit-8bn-humanoids-market-may-top-200bn/ Last updated: 2026-01-16T12:56:04.000Z _This post is for paying subscribers only._ ### Kedrosky Daily: Prediction market volume soars to 9bn by late 2025; humanoids market may exceed 200bn by 2035 URL: https://paulkedrosky.com/kedrosky-daily-prediction-market-volume-soars-to-9bn-by-late-2025-humanoids-market-may-exceed-200bn-by-2035/ Last updated: 2026-01-15T12:56:44.000Z _This post is for paying subscribers only._ ### Rough Notes: Jan 14, 2026 URL: https://paulkedrosky.com/rough-notes-jan-14-2026/ Last updated: 2026-01-14T18:34:32.000Z The following is a new, regular email, based on my ["Rough Notes"](https://paulkedrosky.com/thenotes/) feature on the site. It will send out digests of my regular notes, like what you see below. Here are recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. Impact of AI on Scientific Research The impact of AI on science is an augury of everything, from a new paper: \- Publish 3.02x more papers \- Receive 4.84x more citations \- Shrinks the volume of topics studied by 4.63% \- Decreases scientists’ engagement with one another by 22% Jan 14 · [Permalink →](https://paulkedrosky.com/n/5585a1e5-fd72-470b-a4bc-969b4b6ffc68) 2\. Data Center Project Withdrawal and Lawsuit Data center related updates pointing in the same direction and which will be misunderstood: • Powered land company backs out of Wisoncsin data center • CoreWeave shareholder sues alleging company mislead over ability to deliver compute Jan 14 · [Permalink →](https://paulkedrosky.com/n/1373fbea-6068-402a-9533-8e94ebffa01e) 3\. Climate Change Reduces US Incomes Climate change has already lowered U.S. incomes by \~12% vs what might otherwise have been the case, according to a new econometric study. With AI's sharply growing role, we should look forward to more such losses in future. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F6ef852b0-bb59-42b8-bc96-7b81d2ba61a2.jpeg) Jan 14 · [Permalink →](https://paulkedrosky.com/n/0f9a530f-cae3-46db-8a5f-cfc6b443c1a6) 4\. OpenAI predicted to run out money Heroic stuff from a columnist in the NYT: "My bet is that over the next 18 months, OpenAI runs out of money." Jan 13 · [Permalink →](https://paulkedrosky.com/n/6f9e6e1f-772d-4640-8b1b-c5b4f146a1fc) 5\. Earnings and market cap concentration rise Earnings and market cap concentration in the S&P 500 are both at all-time highs, driven largely by tech equities, and, more specifically, by AI. We long ago eclipsed the dot-com peak. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F64c7695d-b0ec-4fc2-81e7-17821d3de0e4.jpeg) Jan 13 · [Permalink →](https://paulkedrosky.com/n/e5e289d2-868f-47c5-9e4f-2af28b5c4473) 6\. Agentic AI Exhibits Rapid Conformity Risks Agentic Ai is as prone to conformity and group-think as are groups of humans, but does so even faster, but without human accountability or self-awareness. The intelligence of the AI has no bearing, nor does the visibility of the work. It is a deep, structural flaw and source of risk. Jan 13 · [Permalink →](https://paulkedrosky.com/n/fa422a8d-66dd-4967-af54-486ec0d4898a) 7\. Technology Industry Is Largely Mature Perhaps less well known than it should be, and it is somewhat masked by the current AI fervor, but tech is a largely mature industry. Microsoft turned 50 last year. The sector's tepid recent job growth reflects that late middle-aged status. One way to think about the AI paroxysm is aging hipsters trying to recapture their (economic) youth. It rarely works. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Ffefdf423-284b-4f4b-b1a3-5d8fadeca6f4.jpeg) Jan 12 · [Permalink →](https://paulkedrosky.com/n/48142d46-cf56-4f50-bd10-950a7f79feea) 8\. AI Used in Medicare Advantage Upcoding AI as productivity tool cited extensively in new Senate report on UHC up-coding in Medicare Advantage. Except the productivity flowed from, for example, using AI to infer likely diagnoses from complex non-clinical signals, pushing that into clinician workflows, and overbilling with abandon. AI, it's the future of medicine. Jan 12 · [Permalink →](https://paulkedrosky.com/n/e4f57dd4-4dfa-45bd-b786-3bd7f3e3ef49) 9\. Large language models exhibit mental health symptoms Researchers turn LLMs from AI therapists into patients, and discover that they are, in short, deeply troubled. They shows strong signs of severe depression, defiance, anxiety, and trauma-related shame. It is just what you'd want in our future overlords. Jan 12 · [Permalink →](https://paulkedrosky.com/n/85c876de-48d6-449c-b4d3-86e39e88d100) 10\. Climate change challenges and AI energy use As I say too often, given long-run climate series, this year will almost certainly be one of the three or four coolest of the rest of your life, \_and\_ that will be true every year for the rest of your life. Be very long central air, and short providers of flood insurance. Meanwhile, it will take a lot of AI to "solve" climate change, given AI's energy profligacy's growing role in climate change. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F0c2d2a04-e3bd-41e4-b2bd-802d01898536.jpeg) Jan 12 · [Permalink →](https://paulkedrosky.com/n/f1181a7f-a5f2-441f-9a8e-091068fd89a7) ### Kedrosky Daily: Economy importance rises sharply Nvidia leads AI compute sales URL: https://paulkedrosky.com/kedrosky-daily-economy-importance-rises-sharply-nvidia-leads-ai-compute-sales/ Last updated: 2026-01-14T12:54:19.000Z _This post is for paying subscribers only._ ### Kedrosky Daily: Tech job growth stalls in 2024; tariff lawsuits surge after Supreme Court hearing URL: https://paulkedrosky.com/kedrosky-daily-tech-job-growth-stalls-in-2024-tariff-lawsuits-surge-after-supreme-court-hearing/ Last updated: 2026-01-13T12:53:50.000Z _This post is for paying subscribers only._ ### Rough Notes: Jan 12, 2026 URL: https://paulkedrosky.com/rough-notes-jan-12-2026/ Last updated: 2026-01-12T17:04:12.000Z The following is a new, regular email, based on my (also new) ["Rough Notes"](https://paulkedrosky.com/thenotes/) feature on the site. It will send out digests of my regular notes, like what you see below. Here are ten recent [rough notes](https://paulkedrosky.com/thenotes/) from the site. 1\. Red Queen Strategies in Agentic AI Fascinating paper on red queen competitive strategies among agentic AIs in a programmatic sandbox, and how the competition evolves. The main takeway: rapid evolution toward a general-purpose behavioral strategy. But, more insidiously, rapid synchrony in approaches, given similar reward functions, training data, etc. This, of course, is profoundly destabilizing. Jan 12 · [Permalink →](https://paulkedrosky.com/n/91cb8ba1-6565-474d-ae2c-565b27ce5b17) 2\. Debate on AI Costs and Benefits Long but fitfully worth reading "debate" between Michael Burry, Dwarkesh Patel, and Jack Clark about the costs and benefits of current AI. It somehow devolves into nuclear advocacy, which is weird, given that there are far more useul rabbit holes. Jan 12 · [Permalink →](https://paulkedrosky.com/n/0e7f28c9-21d2-4b68-8bba-eecb4292d34c) 3\. In-firm AI Usage Shows Limited Benefits While surveys like this area always fraught, that the signal from in-firm AI usage isn't strong enough for most companies to say they are benefiting is striking. Again, early days, and an economy-wide 5% gain would be material for productivity, but the uncertainty is noteworthy. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2Fa3d4d239-4840-4597-bd3d-2e19c96ff86a.jpeg) Jan 12 · [Permalink →](https://paulkedrosky.com/n/f89be7e9-0558-4d9d-aa40-641d6026c84f) 4\. AI Impact on Trading and Time So, I sort of agree with this from Michael Burry about how AI is already making trades inroads, but the trouble is time is not fungible, Just because I could do a bunch of things, with an LLM assist, doesn't meant I would or should do that. I see people constantly getting in over their heads by starting things. Jan 11 · [Permalink →](https://paulkedrosky.com/n/27804549-efc4-4492-a9a7-c1dd226e52be) 5\. Americans and their acceptance of chance “\[Americans\] who live in the midst of democratic fluctuations have always before their eyes the image of chance; and they end by liking all undertakings in which chance takes a part.” — Alexis de Tocqueville , from Democracy in America, Part II Jan 10 · [Permalink →](https://paulkedrosky.com/n/03c1870f-538a-4546-8538-4b9e8e44f1ed) 6\. Claims of AI and broader market bubbles One of the stranger claims about AI is that there isn't an AI bubble, there is an "everything bubble", across debt, equities, alts, private credit, etc. Fair enough, but it is a little like telling a patient they have many things wrong, and then stopping triumphantly there. Examples: Unhedged Richard Bernstein Jan 9 · [Permalink →](https://paulkedrosky.com/n/f0d48111-83d6-4a61-abca-dfe8c94ba284) 7\. AI Related IPOs Surge in 2026 A major theme in capital markets in 2026 will be a tidal wave of AI-related IPOs, and not just the frontier model suspects. Bankers will drag out everything possible, most of which will be garbage, in an echo of the 1999 dot-com IPO wave. It's already started in China, and it's coming to the U.S. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F03852dd1-8bb2-444b-8606-07ac945b5111.jpeg) Jan 9 · [Permalink →](https://paulkedrosky.com/n/9a42f8b6-0ad2-4ba8-adb1-518faf37f87a) 8\. Data Center Air Pollution Health Costs Data center health impacts, from a new piece in The Lancet: \- The PM2·5 released from training a large AI model is more than that emitted during 10,000 round trips by car between Los Angeles and New York City \- Public health costs from air pollution attributed to data centres in the US alone projected to reach $10b-$20b per year by 2028 Jan 9 · [Permalink →](https://paulkedrosky.com/n/6de5c644-dc7c-44c4-a682-2ec0d3aebbfb) 9\. BIS report on AI spending trends New BIS report on the AI spending ... enthusiasm, and it's somewhat more skeptical than I expected. It points to the potential for the inevitable reversal to have larger effects than one might expect, given it's rapid transition from equity to debt financing. ![](https://storage.googleapis.com/prd-activitypub/images%2F72e27633-c5bb-42a8-8ec0-0b628b9fb3ce%2F83fa4017-8250-434c-acd1-085471d85c58.jpeg) Jan 8 · [Permalink →](https://paulkedrosky.com/n/7107260c-277d-40f9-a64d-308a0e6495da) 10\. New Books on Speed and Land Two books, one upcoming, and one recent, newly on my reading list: 1\. Speed: How It Explains the World, by Vaclav Smil How the relentless pursuit of speed shapes technological landscape and social & environmental realities 2\. The Land Trap, by Mike Bird The role of land, through history, in making the rich richer and the poor poorer. Jan 8 · [Permalink →](https://paulkedrosky.com/n/22dbcc17-14b9-43e6-911e-873c32fc66e8) ### Kedrosky Daily: S&P 500 outperforms Magnificent Seven; more bonds upgraded than downgraded URL: https://paulkedrosky.com/kedrosky-daily-s-p-500-outperforms-magnificent-seven-more-bonds-upgraded-than-downgraded/ Last updated: 2026-01-12T12:53:05.000Z _This post is for paying subscribers only._ ### Kedrosky Daily — Sunday Edition: Organ meat sales rise through 2025; fallen angels outnumber rising stars in 2020 URL: https://paulkedrosky.com/kedrosky-daily-sunday-edition-organ-meat-sales-rise-through-2025-fallen-angels-outnumber-rising-stars-in-2020/ Last updated: 2026-01-11T13:08:43.000Z ### January 11, 2026 Edition This is a weekly free look of our premium Kedrosky Daily email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). ## Two Graphs ![](https://paulkedrosky.com/content/images/2026/01/2026-01-09_07-23-51-1.jpg) US retail sales of organ meat have steadily increased from 2020 to 2025, with beef offal consistently dominating the market and sales reaching their highest level in 2025—showing significant growth in consumer demand for organ meats. ![](https://paulkedrosky.com/content/images/2026/01/2026-01-10_16-17-09-1.jpg) In 2020, the value of bonds downgraded to junk status ("Fallen Angels") was nearly five times higher than bonds upgraded to high grade ("Rising Stars"), highlighting an unusually large wave of credit deterioration that year. ## Headlines Fine wine investors find little cheer in third year of falls ([**FT**](https://www.ft.com/content/e84d4fc6-35a6-434d-9ae8-693966348b6e)) … Why Chevron Could Reap the Biggest Rewards From Venezuela ([**NY Times**](https://www.nytimes.com/2026/01/05/business/energy-environment/venezuela-chevron-trump-oil.html)) … Polymarket ([**FT**](https://www.ft.com/content/985ae542-1ab4-491e-8e6e-b30f6a3ab666)) … Hundreds of Flights Canceled at New York Airports, Even With a Few Inches of Snow ([**NY Times**](https://www.nytimes.com/2025/12/27/business/snowstorm-flight-cancel-new-york-airports.html)) … TSMC Shares Jump Most Since April After Goldman Lifts Target 35% ([**Bloomberg**](https://www.bloomberg.com/news/articles/2026-01-05/tsmc-shares-jump-most-since-april-after-goldman-lifts-target-35)) … Tariff turmoil, a gold rush and the sinking dollar – 2025 in charts ([**FT**](https://www.ft.com/content/6e32f728-51a7-4677-b5e4-fff33f05e730)) … Victoria faces 'dangerous' fire day as heatwave bakes Australia ([**Bloomberg**](https://www.bloomberg.com/news/articles/2026-01-09/australia-s-east-coast-swelters-under-heatwave-wildfires-spark-evacuations)) … Venezuela Frees Political Prisoners, Vows to ‘Rescue’ Maduro ([**Bloomberg**](https://www.bloomberg.com/news/articles/2026-01-10/venezuela-frees-political-prisoners-pledges-to-rescue-maduro)) … Nvidia Details New A.I. Chips and Autonomous Car Project With Mercedes ([**NY Times**](https://www.nytimes.com/2026/01/05/technology/nvidia-chips-mercedes.html)) … A dealmaking frenzy is reshaping the booming wealth-management business ([**WSJ**](https://www.wsj.com/finance/investing/a-dealmaking-frenzy-is-reshaping-the-booming-wealth-management-business-6ae921d6?gaa%5Fat=eafs&gaa%5Fn=AWEtsqfiGPQG65SNycweZuC4J9JP2dTILCVdoK2OntFU37BJlmFCMGAgi5C6&gaa%5Fts=6963a45e&gaa%5Fsig=LNE3Ub%5Fvj0bvP8vpLOYvyuYJVbfTB%5Fwns2lvDr1pC4l6YZ95opLNVmp7COiENImcfQcFVr736sA9rvla4nwsTQ%3D%3D)) … How Google Got Its Groove Back and Edged Ahead of OpenAI ([**WSJ**](https://www.wsj.com/tech/ai/google-ai-openai-gemini-chatgpt-b766e160?gaa%5Fat=eafs&gaa%5Fn=AWEtsqdVCt6esaW8X%5FUQX26J20-q-Y4k6xqrwRKDnW1Fjf%5FcA6cBP1nXJT0L&gaa%5Fts=6963a45e&gaa%5Fsig=pwYkSvyWGWwRh1G7cmfZ2gcX41DiviJkAXDAcz7HgFs4WWzJHQKSxtXjzWXO7YPvN7XyyitXe6Am%5Fg7r4k8%5Faw%3D%3D)) … A shadow fleet smuggles illicit oil across the high seas. This is how it works. ([**WSJ**](https://www.wsj.com/business/logistics/a-shadow-fleet-smuggles-illicit-oil-across-the-high-seas-this-is-how-it-works-eb8c9954?gaa%5Fat=eafs&gaa%5Fn=AWEtsqdLgdf9gqhawSWM7o5PKvScO9r8i925kHMmm10sJ-tDAuxVeLI%5FNfe8&gaa%5Fts=6963a45e&gaa%5Fsig=1xtm6ILewEd7evsUbM6kObMOmSY2EVmaXvBIpz5ZOW8nj7cCzwrTaDAyQ8T-gumnBPbq1djXCHXemAjdtXPNhQ%3D%3D)) ### Kedrosky Daily: AI loan growth outpaces fund exposure; S&P 500 earnings peak then normalize URL: https://paulkedrosky.com/kedrosky-daily-ai-loan-growth-outpaces-fund-exposure-s-p-500-earnings-peak-then-normalize/ Last updated: 2026-01-09T12:51:20.000Z _This post is for paying subscribers only._ ### Kedrosky Daily: Gas prices hit lows near 3 dollars; Oracle CDS risk rises sharply URL: https://paulkedrosky.com/kedrosky-daily-gas-prices-hit-lows-near-3-dollars-oracle-cds-risk-rises-sharply/ Last updated: 2026-01-08T12:50:58.000Z _This post is for paying subscribers only._ ### Kedrosky Daily: Art loan portfolio doubles by 2027; Venezuela crude reserves triple by reclassification URL: https://paulkedrosky.com/kedrosky-daily-art-loan-portfolio-doubles-by-2027-venezuela-crude-reserves-triple-by-reclassification/ Last updated: 2026-01-07T12:49:47.000Z _This post is for paying subscribers only._ ### Kedrosky Daily: Wyoming snow cover hits record low; California snow cover ranks well above median URL: https://paulkedrosky.com/kedrosky-daily-wyoming-snow-cover-hits-record-low-california-snow-cover-ranks-well-above-median/ Last updated: 2026-01-06T15:58:08.000Z _This post is for paying subscribers only._ ### Kedrosky Daily: US gaming revenue doubled since 2020; China investment peaks while consumption lags URL: https://paulkedrosky.com/kedrosky-daily-us-gaming-revenue-doubled-since-2020-china-investment-peaks-while-consumption-lags/ Last updated: 2025-12-26T12:46:00.000Z _This post is for paying subscribers only._ ### Kedrosky Daily: Gold rose 72 Bitcoin fell 6 consumer confidence hits two year low URL: https://paulkedrosky.com/kedrosky-daily-gold-rose-72-bitcoin-fell-6-consumer-confidence-hits-two-year-low/ Last updated: 2025-12-25T12:45:17.000Z _This post is for paying subscribers only._ ### Kedrosky Daily: Gold rose 72 Bitcoin fell 6 US confidence hits two year low; Consumer confidence drops five months straight URL: https://paulkedrosky.com/kedrosky-daily-gold-rose-72-bitcoin-fell-6-us-confidence-hits-two-year-low-consumer-confidence-drops-five-months-straight/ Last updated: 2025-12-25T01:54:05.000Z _This post is for paying subscribers only._ ### Kedrosky Daily: Marijuana use rises in seniors; home sizes shrink since 2015 URL: https://paulkedrosky.com/kedrosky-daily-marijuana-use-rises-in-seniors-home-sizes-shrink-since-2015/ Last updated: 2025-12-22T12:52:43.000Z _This post is for paying subscribers only._ ### Kedrosky Daily — Sunday Edition: Home sizes drop to decade low; GPUs use 41 percent of AI power URL: https://paulkedrosky.com/kedrosky-daily-sunday-edition-home-sizes-drop-to-decade-low-gpus-use-41-percent-of-ai-power/ Last updated: 2025-12-21T12:50:24.000Z ### December 21, 2025 Edition This is a weekly free look of our premium Kedrosky Daily email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). ## Two Graphs ![](https://paulkedrosky.com/content/images/2025/12/2025-12-20_07-49-21-1.jpg) The median size of new single-family homes in the US has sharply declined since 2015, reaching its lowest point in over a decade by 2025, despite previously hitting record highs. ![](https://paulkedrosky.com/content/images/2025/12/2025-12-18_13-41-17-1.jpg) GPUs consume 41% of total facility power in AI data centers, making up the majority of server power use—highlighting their massive energy demand compared to other hardware components. ## Headlines FT Magazine. How a failed baseball player and ex-stock broker became a major arms dealer ([**FT**](https://www.ft.com/content/d7f95899-d634-43c9-adac-da8da67f3cdb)) … US Says It Intercepted Second Tanker in Venezuela Blockade ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-12-20/us-intercepts-second-tanker-in-venezuela-blockade-reports-say)) … How tech’s biggest companies are offloading the risks of the A.I. boom ([**NY Times**](https://www.nytimes.com/2025/12/15/technology/ai-risks-debt.html)) … PayPal Applies to Become a Bank as US Loosens Regulatory Reins ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-12-15/paypal-applies-to-become-a-bank-as-us-loosens-regulatory-reins)) … Roomba maker iRobot files for bankruptcy, with Chinese supplier taking control ([**NY Times**](https://www.nytimes.com/2025/12/15/business/roomba-irobot-bankruptcy.html)) … Ford will take $19.5 billion hit as it rolls back E.V. plans ([**NY Times**](https://www.nytimes.com/2025/12/15/business/ford-electric-vehicles-hybrids.html)) … Robot vacuum Roomba maker files for bankruptcy after 35 years ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-12-15/robot-vacuum-roomba-maker-files-for-bankruptcy-after-35-years)) … Netflix vs Paramount: politics could decide battle for Warner Bros ([**FT**](https://www.ft.com/content/bbfebdc3-80ce-4dec-b88d-94535a63ed14)) … Warner Bros to rebuff $108bn Paramount hostile offer ([**FT**](https://www.ft.com/content/6443abe1-c1ee-4f28-8a8a-408a3e8f6077)) … Ford Takes $19.5 Billion Hit in Detroit’s Biggest EV Bust ([**WSJ**](https://www.wsj.com/business/autos/ford-takes-19-5-billion-charge-to-write-down-ev-investments-333a9bc4?gaa%5Fat=eafs&gaa%5Fn=AWEtsqdCt1V%5FF-9T%5F%5FaH0nfC4M4ruPPxrsywYrp4Bfq2cZ05J09PZ-7jOWlG&gaa%5Fts=6947f093&gaa%5Fsig=gmVj1iIOKJGCQ2I0om10-vIn3ZQfvCe2nld1krxdAAE16rBg9eyUpLdGCjLC156loowVSbsZaXyb7kuhy1GtIQ%3D%3D)) … All That Cheap Chinese Stuff Is Now Europe’s Problem ([**WSJ**](https://www.wsj.com/business/retail/cheap-chinese-exports-europe-9cae85f6?gaa%5Fat=eafs&gaa%5Fn=AWEtsqeK927GrS56bIKaWv6Wf90PH-kvXDk22aZoNhU7qvbyKG91anIdXGCs&gaa%5Fts=6947f093&gaa%5Fsig=SeCcWty71xtNg%5FHntyRNpKiUv1QqyOcouCn0UF1VCCbGmFsBTUBuMgIpXVgTelVxcr7V7xjMC9YSAFS3SJDqZg%3D%3D)) … CoreWeave’s Staggering Fall From Market Grace Highlights AI Bubble Fears ([**WSJ**](https://www.wsj.com/tech/ai/coreweave-stock-market-ai-bubble-a3c8c321?gaa%5Fat=eafs&gaa%5Fn=AWEtsqftB8hS3UJtE-ZTD4lvTU77fLHSqax5hlvVZeY1d6ohe37qKLFeUHvw&gaa%5Fts=6947f092&gaa%5Fsig=Ipc8HEGLW%5FlxEilTdrKWsSLIbnajDj0gH-WLGzXhj5q3Db9TG87xEZbXYOaXlKAjx6K8OzBCXkibA8aXqKqyrw%3D%3D)) ### Kedrosky Daily: Aspen hosts top home sales; semiconductor jobs rise then fall URL: https://paulkedrosky.com/kedrosky-daily-aspen-hosts-top-home-sales-semiconductor-jobs-rise-then-fall/ Last updated: 2025-12-17T15:34:37.000Z _This post is for paying subscribers only._ ### Morgan Stanley's Banking Lead, Debt Buyers Bumping Against Limits, AI in Space, etc. URL: https://paulkedrosky.com/morgan-stanleys-aleads-u-s-banks-in-ai-debt-financing/ Last updated: 2025-12-17T15:00:28.000Z Three things to think about: 1. Morgan Stanley's AI debt banking lead 2. Insurers awash in data decenter debt 3. Data centers in space, again --- ### 1\. Morgan Stanley's AI Debt Banking Lead I have been asked a few times recently which banks are most active in AI financing, and I generally point to J.P. Morgan and Morgan Stanley. Perhaps, judging by new data, I should just point to the latter. Highlights from a new [report](https://www.bloomberg.com/news/articles/2025-12-16/morgan-stanley-rides-big-tech-s-ai-boom-with-65-billion-debt-push): - Morgan Stanley has led/co-led **\~$65B in AI and data-center bond issuance** since October, **more than any other major US bank**. - Its **US IG bond-arranger share** is up \~2 pp YoY, moving to #3. - MS strategists are helping, **projecting \~$3T in cloud/AI infrastructure spend** by 2028, heavily debt-financed. - It structured **Meta’s $27B Hyperion SPV** to keep obligations off Meta’s balance sheet - MS designed the **Google-backstopped junk deals for crypto miners** TeraWulf and Cipher. - Co-ran **Meta’s $30B bond sale** with record book size - Convened **energy, tech, and alternative-asset firms** to align capital markets with AI-power constraints - **Merged energy + utilities** coverage groups. Morgan Stanley is trying very hard to be the **AI capex debt financing lead of choice** here. That apparently comes with offering **unsettling rhetoric** about how **debt market innovation** now outpaces that in equity markets. Here is its co-head of capital markets: > \[T\]he scale of innovation and creativity in the credit markets has dwarfed that of the equity markets. For those of you, like me, who have been around this stuff for too long, innovation in debt markets is rarely something you want to see anyone bragging about. Ever. --- ### 2\. Insurers Bumping Up Against Debt Limits _This post is for paying subscribers only._ ### Private AI Companies Are the New Public AI Companies URL: https://paulkedrosky.com/private-ai-companies-are-the-new-public-ai-companies/ Last updated: 2025-12-16T20:14:49.000Z Sell-side firms continue to expand private AI company "coverage" _This post is for subscribers on the Professional tier only._ ### An Oracle CDS Lesson: Inevitable, But Also Misunderstood URL: https://paulkedrosky.com/oracle-cds-inevitate-but-also-misunderstood/ Last updated: 2025-12-16T15:09:07.000Z ### The context People have gotten excited about AI capex and the recent **soaring price of Oracle credit default swaps**¹ (CDS), perhaps flashing back to the **financial crisis** and how CDS presaged problems there. There are **misunderstandings** out there. ![](https://paulkedrosky.com/content/images/2025/12/2025-12-16_06-15-15.png) 💡 ¹ E****xplainer**: Credit default swaps, which were "invented" in the mid-1990s at JP Morgan, came into their own during the financial crisis. Here is how they work: The CDS “price” is the ****annual insurance premium**, expressed in ****basis points**, of the **probability-weighted loss** of a thing. A CDS quoted at ****150 bps** means, using a house metaphor, you pay ****1.5% per year** of the house’s insured value. On a $10m house, that’s ****$150,000 per year**. It is a measure of the market's view of the likelihood of the house burning, and of the ****severity** of the damage. Unlike with normal insurance, however, you can, via CDS, ****buy insurance on someone else's house**. That is what most CDS activity is: people buying insurance on (metaphorical) houses, whether to ****hedge their own position** (perhaps they're also long Oracle debt), or to ****take a naked position** (they think Oracle's debt is a mess). You might rightly ask yourself why someone would hedge a position they don't like, and there are good-ish reasons for that. For example, they could be a ****private credit fund** or a bank temporarily warehousing the debt before syndicating, and they want to balance their risk. There are many others. ### **The move** - Oracle CDS have gone from **\~50 bp to \~140 bp**. - Explanations focus on credit fear or AI bubble sentiment. - A main driver is **mechanical**, not psychological. --- ### **Start with the obligations** _This post is for paying subscribers only._ ### Which AI Inning Are We In? URL: https://paulkedrosky.com/which-ai-inning-are-we-in/ Last updated: 2025-12-15T20:12:31.000Z [Bespoke](https://www.bespokepremium.com/) has done its [**annual survey**](https://us11.list-manage.com/survey?u=d9df05124e64a31cb695b0f70&id=ce9767f1ef&e=6b75720302) of its readership—mostly retail, split between brokers and high net worth—and the early results are, as ever, interesting. One of their questions was about where we are in the **AI boom**. The survey response, so far, is worth noting on two levels: 1. It implicitly concedes that this AI moment is is a boom that will **likely turn to bust** 2. Responses to such questions are almost always **lagging and contrarian indicators** of where we actually are Here are the Bespoke survey results: _This post is for paying subscribers only._ ### Kedrosky Daily: Largest cities shift to low income countries; bank credit share drops to thirteen percent URL: https://paulkedrosky.com/kedrosky-daily-largest-cities-shift-to-low-income-countries-bank-credit-share-drops-to-thirteen-percent/ Last updated: 2025-12-15T19:08:42.000Z _This post is for paying subscribers only._ ### Weekend Thinking: A Cul-de-Sac With a View URL: https://paulkedrosky.com/weekend-thinking-a-cul-de-sac-with-a-view/ Last updated: 2025-12-14T16:36:25.000Z 💡 This piece is open to all subscribers as a preview of member content. I've argued for some time that AI models are **at an interesting cul-de-sac**. There are least three reasons: - Increasingly **exhausted pre-training data** - The **architectural limits** of large language models themselves - The **sharply worsening economics** of model training Granted, it is **a nice cul-de-sac** to have ended up in. Model gains were tremendous in the 2018-2023 period, and current models, prudently used, are highly useful. But, it felt for a while to many that **scaling laws were up there with F=ma** in terms of describing a basic law of the computational universe. Throwing more data and compute at model variants would produce predictable **engineered-in, just-add-capex, board-friendly** gains. This has turned out to not be the case. Instead, model releases have become **more expensive and incremental**, often indistinguishable from their predecessors, outside of esoteric and often **gamed benchmarks**. With that in mind, a weekend [Bloomberg piece](https://www.bloomberg.com/news/newsletters/2025-12-11/ai-leaders-eye-new-breakthrough-to-build-more-powerful-models) tries to make the counter-argument, at least implicitly. It hints, heroically, and mostly citing the claims of insiders, that **breakthroughs are ahead**, in particular in **continuous learning**. Continuous learning is one of the main problems with current models, which have specific end dates wrt training, and **subsequent gaps** must be backfilled by web search. Perhaps worse, they do not learn from what they are exposed to or retrieve, and attempts to make them do so often lead to **catastrophic forgetting**, wherein they not only fail to learn but also forget what they previously knew. It remains an unsolved research problem. This passage has some choice quotes on how that problem is ... still not solved: > Google DeepMind considers continual learning to be among a handful of research areas that, if solved, could help achieve artificial general intelligence, according to a person familiar with the company who spoke on condition of anonymity to discuss internal matters. Google Research [recently published](https://research.google/blog/introducing-nested-learning-a-new-ml-paradigm-for-continual-learning/) a paper on a new method for continual learning that has attracted attention in the field. > > OpenAI’s Sam Altman has also publicly hinted that continuous learning is key to AGI, a term that refers to AI systems that match or exceed humans on many tasks. > “I think we’ll start to get to a place where we put these systems together in new ways and we have kind of continuous learning where the systems just run forever and get smarter and smarter,” said [Altman during a talk](https://www.youtube.com/watch?v=6NwK-uq16U8&t=784s) with venture capitalist Vinod Khosla earlier this year. > In the past few weeks, prominent figures such as former OpenAI chief scientist Ilya Sutskever and AI pioneer Richard Sutton have also stressed the importance of this approach to build smarter artificial intelligence systems. The heightened interest in the technique reflects the industry’s push to find new architectures and algorithms to advance the field, as some labs have confronted [diminishing returns](https://www.bloomberg.com/news/articles/2024-11-13/openai-google-and-anthropic-are-struggling-to-build-more-advanced-ai) from existing methods. Note what's going on here. Everyone agrees that models have reached a kind of **pre-training dead end**, even if they don't say that out loud, and even if the continuing utility of massive training runs underlies much of current capex, and they swap in an *unsolved* problem as a solution. Granted, there are currently some workarounds. For example, **retrieval augmented generation** lets models access external databases, but it doesn't make the underlying model smarter. In the interim, leading AI developers are **pushing out relatively trivial updates** to their models at a faster pace. Anthropic has said it's doing "more incremental improvements rather than only shipping the really big upgrades." OpenAI's GPT 5.2 came out this week to a mostly meh response. The pace of releases creates the **impression of momentum** through frequency rather than the magnitude of change. ## What This Reveals The technical reality: - Pre-training gives broad knowledge but with a hard cutoff date - Fine-tuning improves specific tasks but doesn't expand general knowledge - Neither allows for open-ended learning as humans do The AI industry spent years betting that scaling—more data, more compute, bigger models—would produce AGI. **That bet has not paid off.** The improvements are real, but increasingly incremental and slowing, while costs soar. The systems are impressive but bounded. Now the labs are returning to older, harder problems. Continual learning. New architectures. Different training methods. These are necessary research directions, but they **are a reminder that the next five years will be nothing like the last five.** They're the work of an industry recalibrating after hitting a wall. The article cheerfully frames this as labs **"eyeing new breakthroughs."** The reality: engineered-in gains via expensive scaling have run their course, and they are trying to figure out what to do next. There is no eyeing, contrary to the piece's headline, just hoping. ### Kedrosky Daily — Sunday Edition: Oracle bond yields spike as US population growth shifts to immigration URL: https://paulkedrosky.com/kedrosky-daily-sunday-edition-oracle-bond-yields-spike-as-us-population-growth-shifts-to-immigration/ Last updated: 2025-12-14T12:47:14.000Z ### December 14, 2025 Edition This is a weekly free look of our premium Kedrosky Daily email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). ## Two Graphs ![](https://paulkedrosky.com/content/images/2025/12/2025-12-12_14-05-36-1.png) After Oracle delayed some data center projects, the yield on its 5.2% bond due 2035 jumped sharply, rising above US BBB bonds and approaching US BB (junk) bond yields—an unusual and significant sign of declining investor confidence. ![](https://paulkedrosky.com/content/images/2025/12/2025-12-12_13-40-25-1.png) By 2030, net immigration is projected to be the primary—and nearly sole—driver of U.S. population growth, as births minus deaths turn negative. ## Headlines China’s Access to Powerful Nvidia Chips Comes at ‘Critical Moment’ ([**NY Times**](https://www.nytimes.com/2025/12/09/business/china-gains-trump-nvidia-chips.html)) … The Silicon Valley Campaign to Win Trump Over on AI Regulation ([**WSJ**](https://www.wsj.com/tech/ai/the-silicon-valley-campaign-to-win-trump-over-on-ai-regulation-214bd6bd?gaa%5Fat=eafs&gaa%5Fn=AWEtsqcUY7qsjOSpIaN8zbb6Bd2rCXy-jl2ISOcmPnzm%5F4OavJzDMos2Afe8&gaa%5Fts=693eb554&gaa%5Fsig=h91%5FFQeWTs-3u5WdgGxD1KvVMg%5FKvQBkbIosRLxeG-%5Fc1AVUSql35QCcEasYfa1Mx7btUUyusebuSA7WrbtNgw%3D%3D)) … The Great Income Squeeze Begins as Fed Spells End to Easy Yields ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-12-10/the-great-income-squeeze-begins-as-fed-spells-end-to-easy-yields)) … Why the A.I. Boom Is Unlike the Dot-Com Boom ([**NY Times**](https://www.nytimes.com/2025/12/09/technology/ai-boom-unlike-dot-com-boom.html)) … Massive Debt-Fueled Deals Are Back on Wall Street ([**WSJ**](https://www.wsj.com/finance/investing/massive-debt-fueled-deals-are-back-on-wall-street-22c94ac5?gaa%5Fat=eafs&gaa%5Fn=AWEtsqeCnZjLNS4UTynxEz5J60gt9LjaYmtsTNrtjWUE1bX0mB-lr%5FyUp6Ci&gaa%5Fts=693eb554&gaa%5Fsig=KDzCChL8xgMWPjLV1hP%5FakeJji5A0sa0aWVHl3X0hgGlgOF4ResIUElvCNjjUUMy6GYX1WbqerOJBTmde7jIkg%3D%3D)) … Wall Street Is Shaking Off Fears of an A.I. Bubble. For Now. ([**NY Times**](https://www.nytimes.com/2025/12/09/business/wall-street-valuation-ai-bubble.html)) … Warner Bros.' Bidders Brace for a Fight That Will Last Months ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-12-10/warner-bros-bidders-brace-for-a-fight-that-will-last-months)) … Netflix vs Paramount: politics could decide battle for Warner Bros ([**FT**](https://www.ft.com/content/bbfebdc3-80ce-4dec-b88d-94535a63ed14)) ## The Idea **Mentions:** geopolitics, technology, trade, emerging markets ### Strategic trade alliances alter economic landscapes The trend of geopolitical partners trading more with one another and less with rivals is reshaping international economic dynamics. This realignment affects global supply chains and may drive significant changes in global trade policies and business strategies over the next 24 months. **Related reading:** **FreightWaves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) **Financial Times**: [Strategic interdependence is rewiring the global economy](https://www.ft.com/content/25ab5e71-6ae2-495b-bf06-6fd2f1cbb18a) **FreightWaves**: [The road ahead: Freight market trends & industry insights for 2025](https://www.freightwaves.com/news/the-road-ahead-freight-market-trends-industry-insights-for-2025) ### The Oracle of Fermi Speaks URL: https://paulkedrosky.com/the-oracle-of-fermi-speaks/ Last updated: 2025-12-12T17:01:22.000Z Two completely unsurprising stories today that speak volumes about where we are in the data center buildout cycle. Oracle is [**pushing out some data center projects**](https://www.bloomberg.com/news/articles/2025-12-12/some-oracle-data-centers-for-openai-delayed-to-2028-from-2027), and speculative [**Fermi's stock is being smashed**](https://www.bloomberg.com/news/articles/2025-12-12/fermi-falls-51-after-ai-campus-tenant-ends-150-million-deal) over a lost prime tenant. Both have broader implications. ### **What’s happening** - **Oracle is delaying some OpenAI data center projects to 2028**, from 2027, citing **labor and material shortages**. - The projects are part of a **$300B contract** signed this summer. - The first OpenAI site in **Abilene, Texas** is claimed to be on track, but later phases are delayed. - **Oracle shares fell \~5–6%** on the news, coming after Oracle had already been hammered this week over its earnings call. - Separately, speculative Texas **AI campus developer Fermi lost a $150M tenant commitment**, triggering an impressive **\~46% stock plunge**. - Fermi, **pre-revenue and recently valued at \~$19B**, has now lost **over two-thirds of its value in two months.** ### **What it means** _This post is for paying subscribers only._ ### Kedrosky Daily: Hyperscaler capex triples by 2026; Forty percent of Russell 2000 firms lose money URL: https://paulkedrosky.com/kedrosky-daily-hyperscaler-capex-triples-by-2026-forty-percent-of-russell-2000-firms-lose-money/ Last updated: 2025-12-12T16:17:12.000Z _This post is for paying subscribers only._ ### The Oracle of Yore, and Early Q3 GDP Mysteries URL: https://paulkedrosky.com/the-oracle-of-yore-and-early-q3-gdp-mysteries/ Last updated: 2025-12-11T17:53:58.000Z Two things I'm thinking about: ### 1\. Oracle's results, and what it says about what people say about AI I won't add much to the pile-on over **Oracle's results last night** as an AI signal, other than noting they were largely celebrated by analysts during the call, whle pilloried by markets. I used to call that **"atta boys!"** during my equity research days, and it was always startling how little it had to do with the contents of the call, and more about **maintaining access and deal flow**. I will, however, share this passage from [The Economist](https://www.economist.com/business/2025/12/11/oracle-and-the-hard-truths-about-software) on the results, however. It does a nice job of making clear just how little AI-besotted technology companies like Oracle resemble ... themselves of yore: > Consider **how Oracle stacks up** against the 1,000 biggest non-financial firms listed on the rich world’s top bourses along six measures of corporate performance: operating and free-cashflow margins; operating profit to interest expense; capex to revenue; total debt to operating profit (before depreciation and amortisation); and return on capital. Then **run a regression to find the closest match to Oracle’s profile** across these six variables > **Ten years ago the answer was Amgen**, an American biotech giant. The remaining top ten doppelgangers included other innovators (Pfizer and AbbVie in life sciences and Broadcom in chipmaking), reliable services firms (Equifax in credit scoring, Roper Technologies in business software and Verisk Analytics in consulting) and a few industrial businesses. > **Today Oracle is a spitting image of Antofagasta, a London-listed Chilean copper miner**, followed by nine American utilities. ### 2\. Q3 GDP isn't out and it already makes no sense JPM analysts are arguing that the latest **U.S. trade and inventory data imply 3.5% GDP growth in Q3,** a surprisingly strong number (when it eventually gets released)**. But the composition makes no sense.** _This post is for paying subscribers only._ ### Kedrosky Daily: Hyperscaler capex to quintuple by 2026; Small caps face persistent earnings losses URL: https://paulkedrosky.com/kedrosky-daily-hyperscaler-capex-to-quintuple-by-2026-small-caps-face-persistent-earnings-losses/ Last updated: 2025-12-10T12:47:51.000Z _This post is for paying subscribers only._ ### Vibe Hacking, S&P, and Why Howard Marks is Terrified URL: https://paulkedrosky.com/vibe-hacking-s-p-and-why-howard-marks-is-terrified/ Last updated: 2025-12-09T20:10:29.000Z Three things I've been thinking about: 1. [S&P on data centers](https://www.spglobal.com/en/research-insights/special-reports/look-forward/data-center-frontiers/data-center-risk-if-ai-promises-fade?utm%5Fsource=marketo&utm%5Fcampaign=09%5Fdecember%5F2025%5Fpc030006&mkt%5Ftok=NDkxLVZORy03OTAAAAGeofSpIw10pxrua0ABGdszwmsDIoU5ZClvBsnhe4rMaYoFKI517s2yc7YOoDaLCC4N7AygRH-Yf7rwiH5kU4saeqvnz6yGnh5Dh0yDfNkUzLs) 2. [Howard Marks on AI and jobs](https://www.oaktreecapital.com/insights/memo/is-it-a-bubble) 3. [JPM on vibe hacking](https://markets.jpmorgan.com/jpmm/research.article%5Fpage?action=open&doc=GPS-5136797-0) (clients only) --- ## 1\. S&P on Data Centers Among the scariest six-word sentences in English is: **Ratings agencies think it is fine**. In a new report, S&P is fine with the following: - **Capital is flooding** into data centers on the back of AI demand - More than half of planned spending is **debt-financed** - All of this is pushing **valuations, leverage, and deal sizes to record levels** - **Power, land, and water** are being stretched to the limit, and speculated on madly. - Data center leasing **prices and occupancy spiked** a year ago, and are now weakening ![](https://paulkedrosky.com/content/images/2025/12/2025-12-09_08-33-23.png) And yet, here is S&P's conclusion: fundamentals are “healthy”. I am reminded of **Inigo Montoya in The Princess Bride**: You keep using that word. I do not think it means what you think it means. ### 2\. Howard Marks on AI and Jobs Oaktree's Howard Marks concludes his latest investment memo—broadly about AI's bubble-ishness with candid and personal comments on what a **post-AI future** looks like for jobs. And he thinks, he says, **it looks "terrifying"**. His main points: _This post is for paying subscribers only._ ### Kedrosky Daily: Istanbul tops global traffic delays Chicago beats LA URL: https://paulkedrosky.com/kedrosky-daily-istanbul-tops-global-traffic-delays-chicago-beats-la/ Last updated: 2025-12-09T14:23:07.000Z _This post is for paying subscribers only._ ### Talking To Paul Krugman About AI URL: https://paulkedrosky.com/talking-to-paul-krugman-about-ai/ Last updated: 2025-12-08T15:30:09.000Z Economist Paul Krugman contacted me recently and asked me to come on his popular Substack and talk about all things AI and capex. We recorded the discussion last week, and it's [already out](https://paulkrugman.substack.com/p/talking-with-paul-kedrosky). It was a lovely chat, and the Nobel-winner is a terrific host. As ever, I also drive-by made the case that while we are likely to overspend on data centers, similarly to what happened with railroads and electrification—albeit with a wilder depreciation overlay—that is not the same thing as saying AI is a fad or foolish. History is a useful guide, here. Enjoy. And thank you again to Paul for having me as a guest. ### Kedrosky Daily: Pacific Northwest rain exceeds 12 inches, US seniors hold 32% wealth URL: https://paulkedrosky.com/kedrosky-daily-pacific-northwest-rain-exceeds-12-inches-us-seniors-hold-32-wealth/ Last updated: 2025-12-08T15:01:54.000Z _This post is for paying subscribers only._ ### Kedrosky Daily — Sunday Edition: Microsoft capex-to-sales hits 25% surpassing Exxon US passport issuance peaks at 24M URL: https://paulkedrosky.com/kedrosky-daily-sunday-edition-microsoft-capex-to-sales-hits-25-surpassing-exxon-us-passport-issuance-peaks-at-24m/ Last updated: 2025-12-07T12:47:51.000Z ### December 07, 2025 Edition This is a weekly free look of our premium Kedrosky Daily email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). ## Two Graphs ![](https://paulkedrosky.com/content/images/2025/12/2025-12-05_12-35-32-2.png) Microsoft's capex-to-sales ratio has surged to 25% by 2025, dramatically outpacing Exxon Mobil's 10%, highlighting a significant shift where a tech giant now invests a much larger share of sales in capital expenditures than an energy giant. ![](https://paulkedrosky.com/content/images/2025/12/2025-12-04_16-18-04-1.png) U.S. passport issuance surged dramatically after 2007, peaking at around 24 million annually by the early 2020s—almost five times higher than the rates seen in the late 1990s. ## Headlines Drug-Boat Attack Survivors Waved as U.S. Aircraft Flew By, Video Shows ([**WSJ**](https://www.wsj.com/politics/national-security/drug-boat-attack-survivors-waved-as-u-s-aircraft-flew-by-video-shows-8780ebd7?gaa%5Fat=eafs&gaa%5Fn=AWEtsqfW8mldtX4idC%5FA6uJSlc1vYDTqRmJhCtYkBd4ojcmdlQ8kfQ4POCQ2&gaa%5Fts=69357afa&gaa%5Fsig=g2MS7crypw-AHa9tT6JM0P-UOfpkCZcLdTUgi19R9xU-r59%5Fy5ktBeTLk5g3uii1f0r-TDJNVAzsZC7M3TqZfQ%3D%3D)) … California’s ‘Teacher Village’ Model Spreads as Housing Costs Soar ([**WSJ**](https://www.wsj.com/us-news/education/teacher-housing-affordability-california-7caa3eae?gaa%5Fat=eafs&gaa%5Fn=AWEtsqdqncUr%5FDQxOC20wWr3UcWZLAe4iIvp-W%5F8AVN%5FsO9fELalGdjEPOLw&gaa%5Fts=69357afa&gaa%5Fsig=2Jy0soXnYriLC9Xh4-t02xv%5FVOrkSQ89bsT22-LnE4TOEt53Y13bRCkPnqB0SuEPPiWCc4rWx1DoLGy6bWrSwQ%3D%3D)) … How Netflix stole Warner Bros from David Ellison ([**FT**](https://www.ft.com/content/0fb6c3dc-7e14-45c1-9a63-b960c82512bd)) … Why One Man Is Fighting for Our Right to Control Our Garage Door Openers ([**NY Times**](https://www.nytimes.com/2025/12/04/technology/personaltech/why-one-man-is-fighting-for-our-right-to-control-our-garage-door-openers.html)) … Giving Money Directly to Children Is an Idea the Right and Left Could Love ([**NY Times**](https://www.nytimes.com/2025/12/06/business/philanthropy-trump-accounts-baby-bonds.html)) … Angst Turns to Anger in Hollywood as Netflix Hooks Warner Bros. ([**NY Times**](https://www.nytimes.com/2025/12/06/business/media/hollywood-reaction-netflix-warner-bros.html)) … Apple Rocked by Executive Departures, With Chip Chief at Risk of Leaving Next ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-12-06/apple-rocked-by-executive-departures-with-johny-srouji-at-risk-of-leaving-next)) … The Big Read. Crypto’s rocky year ([**FT**](https://www.ft.com/content/728a469e-34f7-433d-90a5-631b8c10aa51)) … A Thanksgiving Dealmaking Sprint Helped Netflix Win Warner Bros. ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-12-06/how-netflix-beat-paramount-to-win-warner-bros-takeover-deal)) ## The Idea **Mentions:** Changes in geopolitics are significantly altering global trade relationships ### Trade Partners Shift Alliances Recent trends indicate a significant shift towards countries increasing trade with geopolitical partners while reducing reliance on rivals. This has the potential to reshape global supply chains and create new economic clusters. Businesses heavily dependent on current international supply chains should reassess their strategies to mitigate risks and capture new opportunities over the next 12 to 24 months. **Related reading:** **Freight Waves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) **FT**: [Strategic interdependence is rewiring the global economy](https://www.ft.com/content/25ab5e71-6ae2-495b-bf06-6fd2f1cbb18a) **FT**: [Welcome to the new age of geoeconomics](https://www.ft.com/content/daf5ce11-c95b-4964-8df1-7463f0f9a81b) ### Weekend Thinking Dec 5, 2025 URL: https://paulkedrosky.com/weekend-thinking-dec-5-2025/ Last updated: 2025-12-06T01:37:50.000Z 💡 A free edition of the irregular weekend "thinking" pieces that usually only go to premium subscribers. Three things worth thinking about this weekend: 1. Aging populations vs AI 2. Tech industry's capex regime change 3. The two Michaels—investor Burry and writer Lewis—reminisce --- **1\. On the battle between aging populations and AI-driven productivity gains** (JPM) 1. Working-age **population growth should be reduced** by 0.5%-points pa in the US, 0.8%-points pa in the Euro Area, 0.1%-points pa in Japan, 0.2%-points in the UK, 1.3%-points in China, and 1.8%-points in South Korea. 2. The median assumption from studies is that **Al will raise labor productivity growth** by 0.5-1.0%-points pa in the coming decade, depending on the exposure of a country to Al. 3. On average, the effects of **Al on labor productivity** look to be similar in size to the growth effects of slower working-age population growth vs. 2000-2024\. 4. This will make **AI effects on growth a wash**, at best. ![](https://paulkedrosky.com/content/images/2025/12/2025-12-05_15-40-59.png) **2 Tech's industry capital expenditure regime change over the last decade** ([Bloomberg](https://www.bloomberg.com/news/articles/2025-12-02/big-tech-s-spend-little-earn-lots-formula-is-threatened-by-ai?srnd=homepage-americas&sref=1kJVNqnU)) 1. Tech used to be a **low capital expenditure business** with cheap scaling, but that began changing five years ago. 2. Its **capex spending has intensified** in recent years, with it, for example, now 25% of Microsoft's revenue, a tripling from 10 years ago. 3. Microsoft's capex to sales ratio now **dwarfs Exxon's**. 4. Despite the regime change, funded in part by debt, tech companies are **pricier than ever**: Microsoft shares trade at more than 28 times projected forward earnings, higher than its 10-year average. 5. Because of the spending, Meta and Microsoft are expected to have **negative free cash flow** after accounting for shareholder returns next year. 6. A more capital-intensive tech industry will likely drive a **more boom-bust cycle**. ![](https://paulkedrosky.com/content/images/2025/12/2025-12-05_12-35-32.png) **3\. The Two Michaels Talk AI and the Big Short** ([Apple Podcasts](https://podcasts.apple.com/us/podcast/michael-burry-speaks/id1455379351?i=1000739222160)) 1. Writer Michael Lewis and investor Michael Burry **talk for the first time** in more than a decade, after The Big Short movie and book 2. The latter was forced to be more public, he says, after a **"small" put position against Palantir and Nvidia** leaked. 3. His **thesis** is/was: 1. **Palantir** is a consulting-heavy government contractor that slapped an “AI” label on old software. 2. **Nvidia** is a graphics-chip company that **lucked into** two consecutive manias (crypto, AI). 3. Palantir has a **“billionaire-to-revenue ratios > 1,”** a red flag Burry has never seen before. 4. AI capex is like the **2000 fiber build-out,** and market peaked when spending was only half-done, so bought 2-year options. 4. 50 % of U.S. equity money is now passive, so there are **no neglected stocks**: everything moves together. 5. Not one **investor ever apologized** for threatening to sue him during the "big short" trade. 6. He has **never re-watched the Big Short movie** or re-read Lewis's book. --- Have a good weekend. ### AEP Wants to Dump 750 MW of Data Center Power. Why? URL: https://paulkedrosky.com/aep-wants-to-dump-750-mw-of-data-center-power-why/ Last updated: 2025-12-05T15:30:05.000Z **Related reading:** - **FERC urged to reject AEP waiver request for PJM capacity sale (**[**Utility Dive**](https://www.utilitydive.com/news/market-monitor-p3-ferc-aep-pjm-capacity-auction/806770/)**)** ## **The Gist** AEP’s move to **dump 750 MW of unneeded capacity** into PJM’s auction is a signal of **speculative data-center contracting** and software-driven over-projection. Utilities locked in supply for **AI load that never showed** up — and now want out. It exposes how hype has distorted institutional behavior in power. --- ## **Key Facts** - Four **AEP FRR utilities** bought excess capacity, likely for data centers that **didn’t materialize**. - They’re seeking a **waiver** to sell **750 MW** into PJM’s Feb. 2026 incremental auction, bypassing power market rules. - **PJM supports** the waiver; industry watchdogs oppose it as anti-competitive and precedent-breaking. - Opponents say AEP wants to arbitrage high capacity prices while bearing **no market risk** due to cost recovery from ratepayers. - In the interim, PJM’s 2026 load forecast has been **revised down**. --- ## **Things That Jump Out** **1\. A visible crack in the AI/data-center hype cycle** _This post is for paying subscribers only._ ### Kedrosky Daily: US automakers lag in fuel economy China leads global EV sales 2024 URL: https://paulkedrosky.com/kedrosky-daily-us-automakers-lag-in-fuel-economy-china-leads-global-ev-sales-2024/ Last updated: 2025-12-25T06:07:48.000Z _This post is for paying subscribers only._ ### Taiwan as AI Geopolitics Hinge URL: https://paulkedrosky.com/taiwan-as-ai-geopolitics-hinge/ Last updated: 2025-12-04T16:59:42.000Z Some thoughts on Taiwan's increasingly fraught role in AI-related geopolitics, a situation made worse and more fragile by U.S. policy. ### Related reading: - Taiwan’s AI boom leaves traditional manufacturers trailing ([FT](https://www.ft.com/content/ef15d02b-daf2-4463-9448-411bce647cc3)) ## The Gist Taiwan’s economy has tipped into **extreme AI-driven single-sector dependence**, with semiconductors and AI hardware driving **nearly all** of its >8% GDP growth. Traditional industries are shrinking, and tariffs are hurting. Global AI capex is the only thing keeping numbers aloft. In particular, U.S. policy — punitive tariffs on Taiwan’s legacy exports and explosive U.S. demand for AI data-center hardware — is at once **hollowing Taiwan out** and **making it more geopolitically indispensable than ever**. ## Key Facts - Q3 GDP: **+8.21%**, driven overwhelmingly by semiconductors and AI hardware. - Exports: **+32% YoY**, almost entirely chips/servers; autos **–8%**, furniture **–12%**. - A quarter of Taiwan’s workforce sits in traditional manufacturing, now squeezed by **20% US tariffs**, a strong currency, and Chinese price competition. - TSMC produces **\~90%** of the world’s leading-edge chips and is absorbing most of the profit growth. - Taiwan has **no free trade agreements** with major partners due to Chinese pressure. - Taipei **doubled its 2025 GDP forecast** from **3.1% → 7.37%**, almost entirely because of AI-driven exports. - US data-center expansion and GPU buildout make Taiwan’s **fabs the physical choke point** of the entire US AI industrial policy. ## Things That Jump Out - **Like the Denmark–Novo dynamic, but weaponized** Denmark’s GDP was distorted by pharma giant Novo with the rise of GLP-1s; Taiwan’s is now dominated by a single semiconductor company. The failure modes of the latter are more consequential: **trade shocks, AI-cycle reversals, and military conflict**. _This post is for paying subscribers only._ ### Does Gemini Show That Scaling Still Works? No. URL: https://paulkedrosky.com/does-gemini-show-that-scaling-still-works-no/ Last updated: 2025-12-04T14:24:09.000Z Some more thoughts on **Google's Gemini 3**, given continuing comments about it, questions, misunderstandings, etc. By way of preamble, **Google has made huge progress**, having reached parity with AI model competitors after years of it wandering in the wilds of its own corporate confusion. **Related reading:** - Google’s Gemini 3 Means AI’s “Resource Grab” Phase Is On ([Bridgewater](https://www.bridgewater.com/research-and-insights/googles-gemini-3-means-ais-resource-grab-phase-is-on)) ## **The Gist** Google’s Gemini 3 isn’t proof that **scaling is healthy**, contrary to claims, most recently from [Bridgewater](https://www.bridgewater.com/research-and-insights/googles-gemini-3-means-ais-resource-grab-phase-is-on). It is, instead, evidence of how far the industry has drifted into **diminishing returns**. The model reflects **brute-force computation** delivering only marginal gains—an expensive way to stand still that is being framed as progress. --- ## **Key Facts** - **Gemini 3’s improvement is small relative to the apparent compute spend.** If Google used materially more pre-training compute—**2–3×** that of GPT-4o, according to some claims, possibly more—the benchmark deltas look narrow per unit compute. - **Nothing about Gemini 3 suggests a new data regime.** No new corpus, no evidence of higher-quality tokens, no training-pipeline breakthrough. Just vastly more compute applied to a more stable TPU stack. - **This is late-stage scaling.** Capability improves, but only through **massively increasing FLOPs**. The marginal return per FLOP is declining quickly, not improving. - **Other recent gains in the industry have come from post-training, not scaling.** o1/o3, Claude 3.5→4.x: all technique-driven improvements, not size-driven. Gemini 3 is a clean test of raw scaling—and it shows that the curve is flattening, not re-accelerating. - **The narrative is backward.** Bridgewater frames this as proof “scaling still works.” The data show the opposite: **scaling works only in a diminishing sense**, with each gain costing non-linearly far more than the last. _This post is for paying subscribers only._ ### Kedrosky Daily: US power outages hit decade high, MicroStrategy debt tops bitcoin value URL: https://paulkedrosky.com/kedrosky-daily-us-power-outages-hit-decade-high-microstrategy-debt-tops-bitcoin-value/ Last updated: 2025-12-04T12:47:57.000Z _This post is for paying subscribers only._ ### Private Credit Talks Back URL: https://paulkedrosky.com/private-credit-talks-back/ Last updated: 2025-12-03T22:14:36.000Z Two recent pieces highlight how private credit is feeling the heat, serving as a useful indicator in **data center finance**, our too-frequent topic here, given private credit's increasing role there. 1. **Private credit fears are based on four myths (**[**Bloomberg**](https://www.bloomberg.com/opinion/articles/2025-12-03/private-credit-is-safe-and-makes-the-financial-system-stronger)**)** 2. **Blue Owl executives and staff bought $200mn in shares after sell-off (**[**FT**](https://www.ft.com/content/5c7635db-5591-4abf-8f6e-9982e964442e)**)** ### The Gist Private-credit execs are in damage-control mode. Blue Owl is deploying **$200mn** of insider and corporate buying after a messy retail-fund merger, while Apollo’s Marc Rowan is trying to relabel the leveraged **$2–3tn** asset class as a benign **$40tn** investment-grade funding machine. Underneath it all, an industry is seeing more scrutiny and feeling the heat. ### Key Facts - Blue Owl and its staff bought **$200mn** of stock in November after the shares fell **35% year-to-date**; most of the buying came from the firm itself via an existing buyback plan. - Blue Owl had proposed merging a roughly **$1.7bn** non-traded private credit fund for wealthy investors into a **$17bn** listed BDC that trades at around a **20% discount to NAV**, effectively imposing a **\~20% loss** and blocking redemptions for investors who thought they had semi-liquid access. The firm backed down after disgruntled investors and media scrutiny. - Apollo exec Marc Rowan wrote a Bloomberg column **c**laiming that private credit is a **$40tn** market, **\~95% investment grade.** He pulls this trick off by jacking the debt denominator: He counts all non-traded credit (including bank balance-sheet loans) as “private credit”. Independent estimates for what most call private credit put **global AUM** more in the **$1.7–3tn** range. - Rowan **argues** that shifting credit from banks to companies like his reduces **systemic risk** and that private credit is more transparent and better underwritten than public credit. This seems, at best, mischievous, in that he is **defining transparent** in a novel and self-serving way. ### Things That Jump Out - Blue Owl’s “vote of confidence” is small in economic terms but big in optics: **$200mn** of buying is under **1%** of its roughly **$23–24bn** equity value. This gets promoted as evidence of its tireless fondness for its investors, after trying to push a **20%** haircut onto retail investors. - Rowan’s piece hinges on definitional arbitrage: by stretching “private credit” to a **$40tn**, mostly investment-grade universe that includes bank loans, he shrinks the “risky” bit to a **$2tn** sliver, even though most debate is about that smaller, higher-yield segment. - More broadly, **semi-liquid retail product**s are exactly the kind of liquidity mismatch and valuation shock that raises systemic risk. And it sits awkwardly next to Rowan’s claim that private credit has made the system **“more resilient and less concentrated.”** ### Where to From Here - **Near-term mechanics:** Blue Owl’s buybacks and insider purchases don’t change that retail investors just watched a sponsor seriously explore locking them in and forcing a **\~20%** loss via corporate structuring. - **Plausible but assumption-heavy:** If rates drift down and defaults stay contained, Rowan’s framing — private credit as a big, mostly IG funding machine — can hold for a while; insurers and pensions harvesting extra spread will look like geniuses. But semi-liquid vehicles create **correlated tail risk** if spreads blow out or redemptions spike. - **Wishful thinking/pressure points:** What’s happening is a migration into opacity, sparse public data, and vehicles whose **true liquidity** only shows up under stress. For the story to break, you don’t need **systemic collapse** — just a moderately **bad default cycle—**which seems inevitable in data centers**—**that forces marks, gates, and distribution cuts across a few flagship products. At that point, **insider-buy headlines and “four myths” op-eds** will read less like reassurance and more like tells. ### AI Capex in H1 2025 Prevented a Recession URL: https://paulkedrosky.com/ai-capex-in-h1-2025-prevented-a-recession/ Last updated: 2025-12-03T15:35:33.000Z AI capex continues to distort US economic data _This post is for paying subscribers only._ ### Kedrosky Daily: Western US snow low, humans dominate mammal biomass URL: https://paulkedrosky.com/kedrosky-daily-western-us-snow-low-humans-dominate-mammal-biomass/ Last updated: 2025-12-03T12:47:57.000Z _This post is for paying subscribers only._ ### Kedrosky Daily: Low Western US Snow Cover, Humans Dominate Mammal Biomass URL: https://paulkedrosky.com/kedrosky-daily-low-western-us-snow-cover-humans-dominate-mammal-biomass/ Last updated: 2025-12-02T12:47:30.000Z _This post is for paying subscribers only._ ### Kedrosky Daily: Military Base Zips Lead Crypto Returns, ORCL CDS Doubles URL: https://paulkedrosky.com/kedrosky-daily-military-base-zips-lead-crypto-returns-orcl-cds-doubles/ Last updated: 2025-12-01T12:47:27.000Z _This post is for paying subscribers only._ ### Kedrosky Daily — Sunday Edition: China’s trade surplus nears $965B Bitcoin mining breaks even URL: https://paulkedrosky.com/kedrosky-daily-sunday-edition-chinas-trade-surplus-nears-965b-bitcoin-mining-breaks-even/ Last updated: 2025-11-30T14:48:11.000Z ### November 30, 2025 Edition This is a weekly free look of our premium Kedrosky Daily email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). ## Two Graphs ![](https://paulkedrosky.com/content/images/2025/11/2025-11-29_19-47-16.png) China’s trade surplus is projected to reach a record $964.8 billion in 2025, nearly doubling since 2021, mainly due to robust export growth despite imports remaining below 2021 levels. ![](https://paulkedrosky.com/content/images/2025/11/2025-11-29_19-43-42.png) The ratio of bitcoin price to production cost has dropped to its lowest level since early 2019, approaching the break-even point, suggesting that bitcoin mining is currently barely profitable. ## Headlines Chinese parts supplier takes stake in leading Russian drone maker ([**FT**](https://www.ft.com/content/e907c2fa-2d3b-4269-bc6c-b2fee4d9f688)) … Meet the Millionaire Masters of Early Decision at Colleges ([**NY Times**](https://www.nytimes.com/2025/11/29/business/tulane-university-chicago-early-decision.html)) … Trump’s Focus on Drug War Means Big Business for Defense Startups ([**WSJ**](https://www.wsj.com/politics/national-security/trumps-focus-on-drug-war-means-big-business-for-defense-startups-04f45ce8?gaa%5Fat=eafs&gaa%5Fn=AWEtsqdaYcEYrH077xpvH7tcMeEZFboa1AY2DmOyYdhchWlGtnCSIFx57Po9&gaa%5Fts=692c5caf&gaa%5Fsig=M9gfSlp3TENGuuu56n8AC18VC-vNzIL8iSmc-Ngupo2azZ62k7knQ6IjNallK930P3fE6P3jTzoC-wT8ecohWw%3D%3D)) … China factory activity slumps for longest stretch on record ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-11-30/chinese-factory-activity-slump-reaches-longest-stretch-on-record)) … Shoppers, Drawn by Steep Discounts, Power Through Black Friday ([**NY Times**](https://www.nytimes.com/2025/11/29/business/economy/black-friday-spending.html)) … Black Friday sales rise, signaling US consumers’ resilience ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-11-29/black-friday-sales-rise-underscoring-us-consumers-resilience)) … FDA Official Pledges New Vaccine Standards ([**WSJ**](https://www.wsj.com/health/healthcare/fda-official-pledges-new-vaccine-standards-dd29b49a?gaa%5Fat=eafs&gaa%5Fn=AWEtsqfGcot0E91YTL9JVHLk8UPmi%5FnKVCobFznqCOh%5Fas3-HZNWWJ3ltKmn&gaa%5Fts=692c5caf&gaa%5Fsig=1oyMx1azDBmTA8C%5F%5Fy8HTaqzu%5F9uxvDRIuUGTpW8VTdDrCe7JXWcSHW1kjdPDmdstoH%5FIhNXUZikspSzqNZvGA%3D%3D)) … They Found Relatives on 23andMe—and Asked for a Cut of the Inheritance ([**WSJ**](https://www.wsj.com/personal-finance/they-found-relatives-on-23andmeand-asked-for-a-cut-of-the-inheritance-251e5f4c?gaa%5Fat=eafs&gaa%5Fn=AWEtsqdkzWED12LOGdHNi2E0XgEDXxDJ-s1Nx13o9AcauEle5vIAlj05GJtt&gaa%5Fts=692c5caf&gaa%5Fsig=1TCumrTv64WAmyLndDEf97P0MLKlGFxY5--YBIdqQ1k9uFsHIEnuRuDrpTiPD%5FMZL9BZ-gYFcKKGht87ey0Bww%3D%3D)) ## The Idea **Mentions:** Strategic interdependence is reshaping global trade. ### Geopolitical partners boosting mutual trade volumes The trend shows that geopolitical alliances are leading to increased trade amongst partners while reducing trade with rivals. This strategic shift is impacting global supply chains and could shape trade dynamics over the next 24 months. Companies heavily reliant on international supply chains must adapt strategies to remain competitive. **Related reading:** **FreightWaves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) **Financial Times**: [Strategic interdependence is rewiring the global economy](https://www.ft.com/content/25ab5e71-6ae2-495b-bf06-6fd2f1cbb18a) **Financial Times**: [Welcome to the new age of geoeconomics](https://www.ft.com/content/daf5ce11-c95b-4964-8df1-7463f0f9a81b) ### Me Media Watch: My WSJ Picks for Books on Work, Historical Bubbles, etc. URL: https://paulkedrosky.com/me-media-watch-my-wsj-picks-for-books-on-work-historical-bubbles-etc/ Last updated: 2025-11-29T18:01:42.000Z A quick update on some places I've popped up this week. The first is a piece I wrote for the WSJ about five books on the future and past of work: - [WSJ](https://www.wsj.com/arts-culture/books/five-best-books-about-the-world-of-work-ee05b947?st=KRaCoV&reflink=desktopwebshare%5Fpermalink): **Five Books on the Future (and Past) of Work** - [CBC](https://www.youtube.com/watch?v=WwMSnpu-Fp0&authuser=0): **Is this an AI bubble, and what does that mean anyway?** - [NPR](https://www.npr.org/2025/11/28/nx-s1-5620935/as-ai-companies-continue-to-invest-heavily-concerns-about-a-bubble-continue-to-grow): **As AI companies invest heavily, worries grow** - [ABC](https://abcnews.go.com/Business/ai-make-big-profits-experts-weigh-bubble-fears/story?id=127858140): **Will AI ever make money?** --- Have a good rest of your weekend. ### Has Corporate AI Adoption Stalled in Low Teens? Doubtful URL: https://paulkedrosky.com/corporate-ai-adoption-stalls-in-low-teens/ Last updated: 2025-11-29T14:55:49.000Z ## The Gist New Census Bureau data shows AI adoption in U.S. firms rose sharply from late **2023** through mid-**2025**, but the growth rate is allegedly now **flattening** across all firm sizes. Large companies allegedly lead, and smaller firms are supposedly close behind, yet absolute **adoption levels are claimed to be low**. While I remain skeptical about the **capex buildout**, this interpretation based on fraught Census Bureau data is **largely indefensible**. [![](https://paulkedrosky.com/content/images/2025/11/2025-11-28_18-20-51.png)](https://www.apolloacademy.com/ai-adoption-rates-starting-to-flatten-out/) ## Key Facts - Citing a bi-weekly **US Census Bureau** survey, Apollo [shows](https://www.apolloacademy.com/ai-adoption-rates-starting-to-flatten-out/) that the share of firms with **250+ employees** using AI rose from roughly **5% in late 2023** to a peak around **13–14% in mid-2025**, then edged down toward **\~12%** by September 2025. _This post is for paying subscribers only._ ### AI Capex Risk and the End of Scaling URL: https://paulkedrosky.com/ai-capex-risk-as-predictable-engineering-2/ Last updated: 2025-11-27T19:17:10.000Z > The breakthrough of pre-training is the realization that this **recipe is good**. You say, “Hey, if you mix some compute with some data into a neural net of a certain size, you will get results. You will know that you’ll be better if you just scale the recipe up.” ... **Companies love this** because it gives you a very low-risk way of investing your resources. > —[Ilya Sutskever](https://www.dwarkesh.com/p/ilya-sutskever-2) 💡 This went out yesterday, but, because of an email glitch, not all paid subscribers received it, so here it is again. Apologies if you have now received it twice. The two key ideas: 1\. Another prominent AI scientist joining me in saying that AI progress in the current ****"scaling" orthodoxy is grinding to a halt**. 2\. The scaling orthodoxy ****floated most data center capex boats** in recent years, and its slowing has broad consequences. I'm surprised it didn't get more attention, but the Dwarkesh interview from which this came was long enough that I doubt many made it through. ### **The Gist** A new [interview](https://www.dwarkesh.com/p/ilya-sutskever-2) with former OpenAI scientist Ilya Sutskever captures, almost accidentally and in passing, something important about the AI boom. It helps answer the question everyone asks: **Why are companies willing to spend so much?** The naive answer is that it is all about the perceived size of the **AI opportunity**. But that is uncertain, and captures only one side. What it misses is how, for a halcyon period, from 2017-2022, compute spending on AI had not only been **derisked**; it had turned into a predictable **capability production function**. **compute + data + paramaters +training = capability** This created a **new kind of speculation**, one that doesn't *feel* like speculation. Pre-training scaling "laws" created the illusion of a physics-like production function: add compute, get capability. That belief is what has been driving a trillion-dollar capex cycle with no historical parallel. And now that the curve's costs have soared and capabilities bent, we’re left with what increasingly looks the largest **mispriced engineering bet** in modern technology. _This post is for paying subscribers only._ ### Kedrosky Daily: Nasdaq Bitcoin Diverge Mid 2024 AI Debt Surpasses Banks by 2025 URL: https://paulkedrosky.com/kedrosky-daily-nasdaq-bitcoin-diverge-mid-2024-ai-debt-surpasses-banks-by-2025/ Last updated: 2025-11-27T12:47:25.000Z _This post is for paying subscribers only._ ### Kedrosky Daily: China saving rate; AI mega-deals in VC URL: https://paulkedrosky.com/kedrosky-daily-china-saving-rate-ai-mega-deals-in-vc/ Last updated: 2025-11-26T14:51:35.000Z _This post is for paying subscribers only._ ### Rogue AI, From RAND to Nvidia: Selection is the Signal URL: https://paulkedrosky.com/rogue-ai-from-rand-to-nvidia-selection-is-the-signal/ Last updated: 2025-11-25T15:19:30.000Z Two things: 1. A new and startling **RAND report** on rogue AIs. 2. Some thoughts on **Nvidia going somewhat rogue** yesterday. Let's do this: ## 01\. The RAND Report on Rogue AIs ### The Gist In a new [report](https://www.rand.org/pubs/perspectives/PEA4361-1.html), RAND evaluates three “break-glass” technical responses to a **globally distributed rogue AI**—high-altitude electromagnetic pulse (HEMP) strikes, global Internet shutdowns, and weaponized “tool AIs.” Every option is either technically shaky, globally destabilizing, or ... both. The bottom line: **no reliable technical countermeasures exist**, thus making prevention the only credible strategy... and we know how well that's going. ### Key Facts • HEMP requires **50–100+ nuclear detonations** for continental-scale coverage and still likely won't work due to modern shielding. • Internet shutdown (via **BGP/DNS manipulation)** demands control over **most Tier-1 networks** or all **13 DNS roots,** which ain't gonna happen. • Physical disruption of would require disabling **top 10–20 chokepoints**, yet global routing redundancy would still leave the Internet partially functional. • Tool-AI concepts—**digital vermin** and **hunter/killer AI**—remain **speculative**. ### Things That Jump Out • The analysis implies that a true global HEMP defense would resemble **nuclear self-sabotage**, with collateral damage potentially worse than the AI it’s meant to stop. • The “Internet kill switch” fantasy collapses under basic network topology: even total Tier-1 shutdown leaves **private backbones and satellite links** alive. • RAND tacitly admits the only real technical option—tool AI—has a high risk **creating another uncontrollable AI**. • Our response options are, basically, **"blow ourselves up first"**, or risk a [nanite-style](https://memory-alpha.fandom.com/wiki/Nanite) loss of control. Some choices. ### Where to From Here The report **pushes for prevention**, not response. With **no dependable global technical options**, that is understandable, but wildly unlikely. It depends on pre-crisis coordination, hardened governance, and avoiding the creation of globally proliferated, unaligned systems in the first place. In other words, **none of those preventative things** will happen. The broader implication is that the **fat tail of future outcomes** is fatter than most people think. --- ## 02\. Nvidia Goes a Little Rogue _This post is for paying subscribers only._ ### Kedrosky Daily: China triples robot installs, doubles US manufacturing value URL: https://paulkedrosky.com/kedrosky-daily-china-triples-robot-installs-doubles-us-manufacturing-value/ Last updated: 2025-11-25T12:47:26.000Z _This post is for paying subscribers only._ ### Data Center Operators are Cheerily Uneasy About the Future URL: https://paulkedrosky.com/data-center-operators-are-cheerily-uneasy-about-the-future/ Last updated: 2025-11-24T21:56:09.000Z Some of my notes from a **bonkers new** [**survey**](https://www.alixpartners.com/newsroom/alixpartners-inaugural-data-center-outlook-report-was-featured-in-a-bloomberg-news-article-which-quoted-andrej-danis-co-leader-of-alixpartners-digital-infrastructure-practice/) of 400 PE firms, VC funds, lenders, data center developers, and operators. ### The Gist AI-driven data center build-outs are accelerating, but investors have almost **no credible exit path** for assets valued in the **tens of billions**. But most funds still model **5–10 year exits**, despite asset lives that look far longer, and despite the majority believing much data center capex will end up **distressed**. ### Key Facts - Only **11%** of respondents expect to hold data-center assets **\>10 years** - Overbuild risk is rising: Some industry sources are saying **US vacancies hit may have hit 7%** in Q2–Q3 2025. - CoreWeave now has **$1.25B** in expected annual interest expense. _This post is for paying subscribers only._ ### Kedrosky Daily: Driver error causes 75% of US crashes, YouTube leads streaming gains URL: https://paulkedrosky.com/kedrosky-daily-driver-error-causes-75-of-us-crashes-youtube-leads-streaming-gains/ Last updated: 2025-11-24T12:47:26.000Z _This post is for paying subscribers only._ ### Kedrosky Daily — Sunday Edition: US cybercrime costs hit $16.6B | China leads quantum funding URL: https://paulkedrosky.com/kedrosky-daily-sunday-edition-us-cybercrime-costs-hit-16-6b-china-leads-quantum-funding/ Last updated: 2025-11-23T14:57:24.000Z ### November 23, 2025 Edition This is a weekly free look of our premium Kedrosky Daily email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). ## Two Graphs ![](https://paulkedrosky.com/content/images/2025/11/2025-11-22_21-06-44.png) Despite a decline in cybercrime complaints since 2021, total US cybercrime costs have surged to a record $16.6 billion in 2024, indicating that individual cybercrimes are becoming far more costly. ![](https://paulkedrosky.com/content/images/2025/11/2025-11-21_13-21-20.png) China's government investment in quantum computing ($15 billion) is nearly double that of the United States ($8 billion), and exceeds the combined total of all other listed countries. ## Headlines My Wife and I Planned Out Our Retirement. We Didn’t Count on Her Getting Sick. ([**WSJ**](https://www.wsj.com/lifestyle/relationships/retirement-planning-lessons-alzheimers-0ccf8cb5?gaa%5Fat=eafs&gaa%5Fn=AWEtsqfpnOI60A4%5F0jB8Y2CXk2gXzWg65AK2ClTqB0afeRfhbqphePmxnMFj&gaa%5Fts=692305d6&gaa%5Fsig=egurac5o0zZKRsdaZApPO%5FUtxKdaeZrVgossNC0HzJQcLBT9uoQx5SEIf9wciKd8sEIeShRI%5FAFGOGQ%5FweTl-g%3D%3D)) … World-Beating Stock Rally Creates New Headaches in Korean Market ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-11-23/kospi-is-world-s-best-performing-benchmark-it-s-also-a-headache-for-south-korea)) … What to Know About the Nearly 10% Climb in a Key Medicare Expense for 2026 ([**NY Times**](https://www.nytimes.com/2025/11/22/business/what-to-know-about-the-nearly-10-climb-in-a-key-medicare-expense-for-2026.html)) … Tyson Foods to Shut Major Beef Facility in Nebraska ([**NY Times**](https://www.nytimes.com/2025/11/21/business/tyson-lexington-nebraska-closing.html)) … US Retail Sales Are Proving Resilient While Risks Mount ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-11-22/us-retail-sales-are-proving-resilient-while-risks-mount)) … The new politics of autism ([**FT**](https://www.ft.com/content/72ffc1d0-92c9-4c0f-80a9-f3b61465f283)) … The A.I. Boom Is Driving the Economy. What Happens if It Falters? ([**NY Times**](https://www.nytimes.com/2025/11/22/business/the-ai-boom-economy.html)) ## The Idea **Mentions:** geoeconomics, trade, international relations ### Trade realignment reshapes global supply chains Strategic interdependence is driving a realignment in global trade, with countries trading more with allies and less with rivals. This shift impacts supply chains and will likely lead to new trade partnerships over the next 12–24 months, affecting industries dependent on international trade networks. **Related reading:** **FreightWaves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) **Financial Times**: [Strategic interdependence is rewiring the global economy](https://www.ft.com/content/25ab5e71-6ae2-495b-bf06-6fd2f1cbb18a) **Financial Times**: [Welcome to the new age of geoeconomics](https://www.ft.com/content/daf5ce11-c95b-4964-8df1-7463f0f9a81b) ### On Benchmark Games, Gemini, and Declining Returns to Scale URL: https://paulkedrosky.com/on-benchmark-games-gemini-and-declining-returns-to-scale/ Last updated: 2025-11-22T20:23:41.000Z > “Until we know we are wrong, being wrong feels exactly like being right” > ― Kathryn Schulz, Being Wrong: Adventures in the Margin of Error Two things today: 1. Some recent media appearances and interviews 2. On the benchmarking games at Google's Gemini Let's do this. --- ### 1\. Media Appearances and Interviews While I don't always remember to mention these. here are three recent apperances and interviews : - [GQG](https://www.youtube.com/watch?v=rEEQvyOhdtY&authuser=0) (video discussion with analysts / portfolio managers) - [CBC radio](https://www.cbc.ca/listen/cbc-podcasts/209-front-burner/episode/16182075-if-ai-is-a-bubble-what-happens-when-it-pops) (audio only) - [Search Engine](https://open.spotify.com/episode/4d90dsIywKHOjbLOUYhDK1?si=m1GWY7OkSDK8xxa72RUw-w) (audio only, and just some excerpts) ### 2\. On Benchmark Games, Kimi, and Gemini **Declining returns to** higher training costs among large language models is a favorite topic of mine. The topic is largely misunderstood, and that became even clearer with this week's announcement of Google's new Gemini model. As a reminder, here is the striking **Gemini benchmarking table** that went everywhere on announcement day. It seems to be kicking ass, to use the technical term. ![](https://paulkedrosky.com/content/images/2025/11/2025-11-21_16-40-56.png) I'm going to **argue four things** about these results: 1. Even taken at face value, **most people wouldn't notice** the above differences in the real world. 2. These benchmarks are **worryingly cherry-picked**. 3. On a more balanced set of benchmarks, **Gemini is 3, at best, on par** with the above models from its competitors. 4. The **sub-linear improvement** of large language models at super-linear cost improvements remains the dominant feature. ### Most People Wouldn't Notice The above table shows **relatively small gains** on tests where all leading models already cluster tightly. As a rule of thumb in a non-deterministic domain, most people don't notice **gains of less than 50%**. These gaps, as a result, **do not translate** into different behavior for typical users. Minor shifts on **saturated tasks** do not change how a model reasons, follows instructions, writes code, or handles multi-step problems. When people interact with these systems, **prompt phrasing, conversation history, and other sources of randomness** matter more than small gaps on polluted benchmarks. ### Cherry-Picked Benchmarks The Gemini table omits nearly **every high-signal test** used to measure real reasoning depth and general capability. These include the harder math and logic suites, multi-step reasoning sets, and **contamination-resistant exams** used in Epoch's [**recent PCA analyses**](https://epochai.substack.com/p/benchmark-scores-general-capability?utm%5Fsource=post-email-title&publication%5Fid=3755861&post%5Fid=179472410&utm%5Fcampaign=email-post-title&isFreemail=true&r=5s9uw0&triedRedirect=true&utm%5Fmedium=email)¹. These tests **load most strongly** on the first principal component of model capability, yet almost **none of them appear** in Google’s comparison. _This post is for paying subscribers only._ ### Stranded Assets and the AI-Driven Gas Turbine Renaissance URL: https://paulkedrosky.com/stranded-assets-and-the-ai-driven-gas-turbine-renaissance/ Last updated: 2025-11-21T16:32:48.000Z My notes on recent developments in AI-related energy markets, in particular with respect to the re-rise of gas turbines, in part based on a recent S&P [discussion](https://www.spglobal.com/commodity-insights/en/news-research/podcasts/energy-evolution/111125-the-ai-driven-gas-turbine-renaissance). --- **What’s Happening** AI has flipped the global gas-turbine market from slack to locked-in: - **Lead times:** Now **5–7 years** for large turbines. - **Order books:** OEMs (Mitsubishi, GE, Siemens) say they are **fully committed to \~2030–2032**. - **Prices:** Turbine costs are **up 2x** in some categories. - **Driver:** AI/data centers projected to take **\~12% of U.S. power demand by 2028** vs \~4% in 2023. - **Customer mix:** Hyperscalers are **crowding out utilities** **and emerging-market buyers** for the same hardware. The key point: this is **forward-committed demand**—capacity pre-sold years ahead based on today’s AI-energy nexus narrative. --- **Why It Matters** This “locked-in” status is a **risk amplifier**. _This post is for paying subscribers only._ ### Kedrosky Daily: Crypto Fear Hits Extreme as Market Drops 30 Percent URL: https://paulkedrosky.com/kedrosky-daily-crypto-fear-hits-extreme-as-market-drops-30-percent/ Last updated: 2025-11-21T15:06:08.000Z _This post is for paying subscribers only._ ### Kedrosky Daily: Bitcoin mining near break-even, MicroStrategy shares drop sharply URL: https://paulkedrosky.com/kedrosky-daily-bitcoin-mining-near-break-even-microstrategy-shares-drop-sharply/ Last updated: 2025-11-20T12:48:36.000Z _This post is for paying subscribers only._ ### Kedrosky Daily: Oracle’s backlog 58% OpenAI, big firms add 151K jobs small firms lose 88K URL: https://paulkedrosky.com/kedrosky-daily-oracles-backlog-58-openai-big-firms-add-151k-jobs-small-firms-lose-88k/ Last updated: 2025-11-19T12:47:23.000Z _This post is for paying subscribers only._ ### Kedrosky Daily: ChatGPT users 55+ hit 28 percent in Q3 2025, leads US AI usage URL: https://paulkedrosky.com/kedrosky-daily-chatgpt-users-55-hit-28-percent-in-q3-2025-leads-us-ai-usage/ Last updated: 2025-11-20T04:03:35.000Z _This post is for paying subscribers only._ ### Media: Recent Bloomberg Odd Lots Appearance URL: https://paulkedrosky.com/media-recent-bloomberg-odd-lots-appearance/ Last updated: 2025-11-17T20:46:19.000Z I meant to mention this sooner, but my schedule has been a bit nuts. Last week I recorded an enjoyable appearance on Tracy and Joe's "Odd Lots" podcast for Bloomberg, and [it came out Friday](https://www.bloomberg.com/news/articles/2025-11-14/why-paul-kedrosky-says-ai-is-like-every-bubble-all-rolled-into-one?srnd=phx-oddlots). Check it out at the preceding link. We talked, you guessed it, AI data center spending. Behind the scenes: - Done from a hotel room in West Hollywood - My camera and laptop were balanced on a wicker basket for towels - My ring light screw fell out, so sometimes my light flashed side to side, like in a scene in Blade Runner - My microphone was balanced calamitously on an upright wicker basket on the floor, with a foam yoga brick on top - I was on San Vicente Boulevard, so there were regular sirens, the sound of which which were mostly suppressed by my microphone, which was impressive, if timely for the topic Anyway, enjoy. ### Updates: GenAI Adoption Shifts, Chinese Models, etc. URL: https://paulkedrosky.com/updates-genai-adoption-shifts-chinese-models-etc/ Last updated: 2025-11-17T16:31:10.000Z 💡 ****Correction:** There was a typo in the original emailed version of this post where I called Gemini a Microsoft model. It isn't, obviously. It's Google's. A few quick things worth thinking about on a busy Monday morning: 1. GenAI adoptions changing S-curve 2. LLM benchmark trends 3. Chinese LLMs' growing share --- Let's do this: ### **1\. GenAI Adoption's Changing S-Curve** S&P has out a new [report](https://www.spglobal.com/market-intelligence/en/news-insights/research/2025/11/adoption-of-generative-ai-tools-is-skyrocketing-in-the-us-led-by-chatgpt?utm%5Fsource=marketo&utm%5Fcampaign=17%5Fnovember%5F2025%5Fpc030006&mkt%5Ftok=NDkxLVZORy03OTAAAAGeMKjQflVgCSUTH7SunAd-JYQ7%5Fre7i6K99y2CTGBe1oUu3ms3PHvvoT0oAdQCn4oq2rjfNCWdqbZq4125ffetI1Chq7BLnV86RhJavDbTf-c) on GenAI adoption, and, while interesting, it doesn't entirely say what it thinks it says. (As a related aside, I increasingly find the expression "GenAI" as wearying and dated as I once did "high tech", so keep that in mind.) Here is the main graph that caught my eye: ![](https://paulkedrosky.com/content/images/2025/11/2025-11-17_07-04-37.png) While the growth is dramatic, the history of technology adoption shows that these S-curves often change quickly after about 50% of the market has adopted. Future gains, while material, are generally slower, more expensive, and often come with higher churn. _This post is for subscribers on the Plus and Professional tiers only._ ### Kedrosky Daily: Google leads AI model race 84 percent tech IP growth outpaces others URL: https://paulkedrosky.com/kedrosky-daily-google-leads-ai-model-race-84-percent-tech-ip-growth-outpaces-others/ Last updated: 2025-11-17T12:47:16.000Z _This post is for paying subscribers only._ ### Kedrosky Daily — Sunday Edition: Asia’s emissions surpass West, youth disconnection rises US UK URL: https://paulkedrosky.com/kedrosky-daily-sunday-edition-asias-emissions-surpass-west-youth-disconnection-rises-us-uk/ Last updated: 2025-11-16T12:48:05.000Z **November 16, 2025 — General** This is a weekly free look of our premium Kedrosky Daily email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). ## Two Graphs ![](https://paulkedrosky.com/content/images/2025/11/2025-11-13_09-48-06.png) Asia’s greenhouse gas emissions have surged dramatically since 2000, overtaking all other regions and now exceeding the combined emissions of North America and Europe, which have both seen declining trends since around 2005. ![](https://paulkedrosky.com/content/images/2025/11/2025-11-14_07-55-05.png) While the share of young people (20–24) not in work or education has fallen or stayed stable in all four countries, the percentage who are completely disconnected—neither working, seeking work, studying, nor raising children—has risen steadily since 2000, particularly in the UK and US. ## Headlines Princeton University Says Database With Donor Info Compromised ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-11-16/princeton-university-says-database-with-donor-info-compromised)) … White House Hunts for Ways to Lower the Cost of Living ([**WSJ**](https://www.wsj.com/economy/consumers/trump-cost-of-living-prices-8e08798d?gaa%5Fat=eafs&gaa%5Fn=AWEtsqeJ8yJqO2mRzkrga4n3QK4LMFsWDAct%5F5nvIT7IU6Ry3oj5z28e1fOT&gaa%5Fts=6919cb88&gaa%5Fsig=jHGgOfYEZvo%5FhQZMoQHJ%5FmwWNw5Ak-A%5F15vwnMWcAbR619MSbcEKGmglE3hRw-dRIgjB95wd1W%5FHB0bwP1FHkA%3D%3D)) … U.S. Boat Strikes Are Straining the Counterdrug Alliance ([**WSJ**](https://www.wsj.com/world/americas/trump-boat-strikes-drug-9bbbeb90?gaa%5Fat=eafs&gaa%5Fn=AWEtsqfAzNS5Ltg1FjPqb1baDhZK3HSxVLp0PtjpecqB6vbFi4ueOEyVG8g0&gaa%5Fts=6919cb88&gaa%5Fsig=S0bnf6nfpm60NYO1Fge4NfV9gvfembBf5mqJo6hwaAJIZDS0-35XDU9WYKb4VI2SvTDTckPup0g%5FwjuUvRNdgw%3D%3D)) … Pop Mart’s Labubu Risks Beanie Baby-Style Collapse, Analyst Says ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-11-16/pop-mart-s-labubu-risks-beanie-baby-style-collapse-analyst-says)) … Former Fed Official Violated Trading Rules, Disclosures Show ([**NY Times**](https://www.nytimes.com/2025/11/15/business/economy/fed-kugler-financial-disclosures.html)) … US economic outlook obscured by shutdown-triggered data gap ([**FT**](https://www.ft.com/content/1932bcda-0cde-4e5a-bd96-0216b6990616)) … US retailers try to handle change after penny’s demise ([**FT**](https://www.ft.com/content/c68c8bc9-423a-4633-943b-ab999638cda0)) … 25 Movies, Many Stars, 0 Hits: Hollywood Falls to New Lows ([**NY Times**](https://www.nytimes.com/2025/11/14/business/media/25-movies-many-stars-0-hits-hollywood-falls-to-new-lows.html)) … She Took JPMorgan for $175 Million. That Doesn’t Include Her Restaurant Bills. ([**NY Times**](https://www.nytimes.com/2025/11/14/business/charlie-javice-jpmorgan-legal-bills.html)) … Ex-Fed Governor Kugler Faced Ethics Probe Before Resigning ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-11-15/ex-fed-governor-kugler-faced-ethics-probe-before-she-resigned)) ## The Idea **Mentions:** Geopolitical partnerships affecting trade patterns ### Geopolitical partners boosting mutual trade cooperation. With geopolitical tensions rising, countries are increasingly favoring trade with allies over rivals. This shift is causing significant changes in global trade patterns. Countries most exposed to geopolitical tensions will see altered supply chains and trading volumes over the next 6-24 months. **Related reading:** **FreightWaves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) **Financial Times**: [Welcome to the new age of geoeconomics](https://www.ft.com/content/daf5ce11-c95b-4964-8df1-7463f0f9a81b) **Financial Times**: [Strategic interdependence is rewiring the global economy](https://www.ft.com/content/25ab5e71-6ae2-495b-bf06-6fd2f1cbb18a) ### On AI CapEx Apologists URL: https://paulkedrosky.com/on-ai-capex-apologists/ Last updated: 2025-11-14T15:34:05.000Z A new FT [piece](https://www.ft.com/content/7c268962-7663-43ed-b30d-901d509f831b) argues there is no AI capex bubble. It is poorly argued, tendentious, reminiscent of pre-housing collapse chatter, and assumes its conclusions. But it is also usefully wrongheaded. Let's dive in. ### What It Argues Here are the main arguments the piece makes: 1. **Capacity is Tight** Current AI “bubble” chatters misreads the supply-demand situation: capacity is tight, not overbuilt, so conditions for a classic bursting aren’t present. 2. **Deals Aren't Real Commitments** Many megadeals are mostly options rather than firm capex, so the headline $1.4tn “overbuild” fear is premature; only \~10% or so is actually committed. 3. **Technology, Not Finance, is the Risk** The real risk is in the technology, like hallucinations, inference costs, and uncertain monetisation timelines. It is not asset-stranding or imminent capital flight. 4. **Big Players Benefit** Any correction would be uneven: weakly capitalised players (OpenAI, neoclouds) get hit, while balance-sheet giants (Microsoft, Google) gain share. 5. **The Peak is Far Away** Valuations may fall hard in some names, but the overall trajectory isn’t dot-com-style collapse; chip demand remains strong and the cycle peak is not yet visible. ### Why It's Appealing The draw is its contrarian stance: a calm, clean dismissal of bubble talk at a moment when many are newly forecasting collapse. It highlights genuine optionality in the OpenAI/Nvidia megadeals and makes the overbuild narrative look jumpy rather than grounded. It also leans on the near-term scarcity story — tight capacity, rising chip demand — which collectively gives the argument surface plausibility. ### Why It's Wrong (But Instructive) _This post is for subscribers on the Plus and Professional tiers only._ ### Kedrosky Daily: Retail share in SPX 0DTE options tops 50 US equity risk peaks URL: https://paulkedrosky.com/kedrosky-daily-retail-share-in-spx-0dte-options-tops-50-us-equity-risk-peaks/ Last updated: 2025-11-14T12:48:12.000Z _This post is for paying subscribers only._ ### Kedrosky Daily: Employer health costs surge fivefold, global power demand doubles URL: https://paulkedrosky.com/kedrosky-daily-employer-health-costs-surge-fivefold-global-power-demand-doubles/ Last updated: 2025-11-13T12:48:53.000Z _This post is for paying subscribers only._ ### The Great Credit Convergence URL: https://paulkedrosky.com/the-great-credit-convergence/ Last updated: 2025-11-12T14:59:58.000Z Some notes I made recently that people may find interesting on the changing world of credit, and its implications for a **capex-heavy future** and data center finance. Recent growth in private credit has its roots in orthodox banking's retreat under regulatory oversight after the 2007/8 financial crisis. --- ### **By the Numbers** - **$1.7T:** Size of the private credit market, now matching high yield and syndicated loans. - **⅓ each:** Below-investment grade (IG) debt now split evenly among high-yield, broadly syndicated loans (BSL), and private credit. - **$28B:** Volume of loans migrating between the public and private markets in 1H 2025. Source: [Oaktree](https://www.oaktreecapital.com/docs/default-source/default-document-library/transcript---the-liquid-private-credit-convergence.pdf?sfvrsn=54572966%5F1) --- ### **Why It Matters** - **Borrowers have options.** Large sponsors toggle between **negotiated private loans** and **syndicated placements**, depending on speed, leverage, and certainty. - **Hybrid deals dominate.** Syndicated first liens paired with **private seconds** or **Term A/Term B splits** blur traditional categories. - **Execution, not structure,** now drives capital-raising decisions. --- ### **The Convergence** - **Structures are merging:** leverage ratios, covenants, and pricing are nearly identical across markets. - **Investor base overlapping:** institutions, BDCs, interval funds, and **retail-facing vehicles** now buy both. --- ### **The CapEx Supercycle** - Lenders are very long that the near future will be c**apex-driven**—a reversal of the “asset-light” 2010s. - Demand engines: - **Data centers** - **Defense and reshoring** - **Energy** - These projects require **trillions in long-duration financing** across banks, private lenders, and securitized channels. --- ### **Beyond Direct Lending** - Core remains **sponsor-backed first-liens**, but growth is mostly in: - **RMBS and asset-backed finance** tied to CapEx cycles - **Collateralized loan obligation** debt and equity --- ### My Take - The line between **public and private credit** is functionally gone: there is a **single continuum** of **liquidity** and **opacity**. - That **efficiency hides danger**: spread compression, shared exposures, correlated marks. - Private credit is now **infrastructure finance for AI** (and defense and energy). - The only real lending difference now: **who sees the marks—and when**. ### Global Data Center Investment Passes Oil Investment URL: https://paulkedrosky.com/global-data-center-investment-passes-oil-investment/ Last updated: 2025-11-12T14:21:30.000Z The new IEA energy report and its implications for data centers and debt _This post is for subscribers on the Plus and Professional tiers only._ ### Kedrosky Daily: Employer health costs up 5x, electricity demand doubles by 2035 URL: https://paulkedrosky.com/kedrosky-daily-employer-health-costs-up-5x-electricity-demand-doubles-by-2035/ Last updated: 2025-11-12T12:47:38.000Z _This post is for paying subscribers only._ ### Meta: Purging of Dissidents vs Open Rebellion URL: https://paulkedrosky.com/metas-purging-of-dissdents/ Last updated: 2025-11-14T04:52:07.000Z ### **LeCun’s Exit:** Meta's chief AI scientist Yann LeCun is **planning to depart** the company, according to the [Financial Times](https://www.ft.com/content/c586eb77-a16e-4363-ab0b-e877898b70de?emailId=dd4d1c03-8c5e-4163-8425-5996b93e2317&segmentId=ce31c7f5-c2de-09db-abdc-f2fd624da608). While the move by the Turing Award-winning scientist is presented as a management shuffle, it is equally or perhaps more a **structural break with LLMs** (large language models in AI). ### **Why it matters:** Meta’s AI strategy has pivoted **hard toward scale** (ahem), with more GPUs, bigger models, faster rollouts. LeCun has argued that path is a **scientific dead end**, in part because of the inherent architectural **limitations of LLMs**. ### **Driving the news:** - **Mark Zuckerberg has reoriented** Meta’s AI efforts around “superintelligence,” led by Alexandr Wang of Scale AI. - LeCun’s long-term research lab has been **gutted and folded** into product groups. - LeCun, who has argued for years that language models can’t reason or plan, is allegedly leaving to **build his own startup** focused on “world models". ### **Between the lines:** While some of this is typical corporate intrigue of the Meta strain, there has long been a sense that **LeCun was out of sync** with Zuckerberg. The former has consistently argued that LLMs are a clever trick, but a dead end on the path to higher levels of intelligence; the latter has **tripled down on LLMs** both in terms of hiring and capex. LeCun's view is that intelligence emerges from **interacting with the physical world**, not from the Zuck-ian strategy of throwing **geometrically more GPUs** and training cycles at text prediction. ### **My Take:** This marks the last vestige of major **non-LLM research** disappearing from inside the hyperscalers. It is also, and more importantly, the first **open act of rebellion** from a founding deep-learning figure against the **LLM orthodoxy** dominating Silicon Valley. ### Kedrosky Daily: AI data center funding gap 27 percent construction spending surges URL: https://paulkedrosky.com/kedrosky-daily-ai-data-center-funding-gap-27-percent-construction-spending-surges/ Last updated: 2025-11-11T12:48:23.000Z _This post is for paying subscribers only._ ### Kedrosky Daily: China births drop 70% since 1963, AI firms hit $100B fast URL: https://paulkedrosky.com/kedrosky-daily-china-births-drop-70-since-1963-ai-firms-hit-100b-fast/ Last updated: 2025-11-10T12:47:40.000Z _This post is for paying subscribers only._ ### Kedrosky Daily — Sunday Edition: NYC jobs shift to health care women 70+ marathon surge URL: https://paulkedrosky.com/kedrosky-daily-sunday-edition-nyc-jobs-shift-to-health-care-women-70-marathon-surge/ Last updated: 2025-11-09T12:47:20.000Z **November 09, 2025 — General** This is a weekly free look of our premium Kedrosky Daily email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). ## Two Graphs ![](https://paulkedrosky.com/content/images/2025/11/2025-11-07_12-36-49.png) Since 1990, health care and social assistance have more than doubled their share of New York City's non-farm employment, while finance and insurance have steadily declined—a surprising reversal given the city’s historical reputation as a financial hub. ![](https://paulkedrosky.com/content/images/2025/11/2025-11-07_17-51-30.png) Participation by women aged 70+ in the NYC Marathon is projected to surge by 250% from 2015 to 2025—the largest increase among all age and gender groups. ## Headlines Data Fog Intensifying for Fed as Shutdown Delays US Inflation Numbers ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-11-08/data-fog-intensifying-for-fed-as-shutdown-delays-us-inflation-numbers)) … Visa, Mastercard Near Deal With Merchants That Would Change Rewards Landscape ([**WSJ**](https://www.wsj.com/finance/banking/visa-and-mastercard-near-deal-with-merchants-that-would-change-rewards-landscape-fc6a0c78?gaa%5Fat=eafs&gaa%5Fn=AWEtsqeOAO0ukOmUm1JhFmTwa-IWbb%5FwFkR5Q1%5FNHKM2OGK7ix2AiYOg-p-2&gaa%5Fts=691090db&gaa%5Fsig=-o7VXLko3%5Fg9YVpdpcRjPrmGBqX5zdEcOP-KsrlnKnriBj9gMN9DNxo0dXbfmwAvJsjuCyZRMhmbRnmxx4SHUw%3D%3D)) … Baby Formula Recalled After Infant Botulism Outbreak ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-11-08/baby-formula-recalled-after-infant-botulism-outbreak)) … The Week the AI Boom Got a Reality Check on Wall Street ([**WSJ**](https://www.wsj.com/finance/stocks/ai-stock-market-reaction-5af38bf4?gaa%5Fat=eafs&gaa%5Fn=AWEtsqescA4PKjX45GpJnXWpDAMcTjOhaAsVLEKLst1zyYXQgeqEu8k0dHvy&gaa%5Fts=691090db&gaa%5Fsig=EPKJuQEtIiNQlsKjJuy-jxGDcmwI5y8OO5GNJN6UOvL1%5FOZXVIn6wIC1Qt-Zqgb9o85tuLW5Y2v12HG-toqqvg%3D%3D)) … ‘South Park’ Takes On Trump and Wins Bigly ([**NY Times**](https://www.nytimes.com/2025/11/08/business/media/south-park-trump.html)) … Debt Has Entered the A.I. Boom ([**NY Times**](https://www.nytimes.com/2025/11/08/business/dealbook/debt-has-entered-the-ai-boom.html)) … Corporate America posts best earnings in 4 years despite tariffs ([**FT**](https://www.ft.com/content/2490a2e1-1cb4-4367-89fc-9ccd4ed22802)) … More Drugs to Fight High Cholesterol Are Emerging ([**WSJ**](https://www.wsj.com/health/pharma/more-drugs-to-fight-high-cholesterol-are-emerging-3fdef02f?gaa%5Fat=eafs&gaa%5Fn=AWEtsqeNQKZQDzFFz2XUHVco76GeAKk1KOmqSf%5FBt5TBq1t8eaF-wyutQRS9&gaa%5Fts=691090db&gaa%5Fsig=iWbwvU9oK0IN6ZhWxXU6fXUskFeOptqOjVRoutcH16WmuPzLAg2TSNNWDP7Uw2HWd2dMwQkYH-2x48XcdNgLjw%3D%3D)) … As Beef Prices Remain High, Trump Calls for Inquiry Into Meatpackers ([**NY Times**](https://www.nytimes.com/2025/11/08/business/trump-meatpackers-investigation.html)) … 1,900% Stock Gains, Hate Mail: Welcome to Quantum Investing ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-11-08/1-900-stock-gains-and-hate-mail-welcome-to-quantum-investing)) … Trump orders probe into meat groups amid affordability backlash ([**FT**](https://www.ft.com/content/049654fd-800a-4ab7-bb20-16d8d9a3ca0f)) ## The Idea **Mentions:** geoeconomics, global trade, strategic interdependence ### Strategic interdependence rewiring economies The blending of tech, trade, and policy is altering global economic relationships, leading to a shift towards strategic interdependence. This impacts geopolitical partners by deepening trade ties and reducing rival interactions, causing long-lasting effects on global market structures over the next 6-24 months. **Related reading:** **FreightWaves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) **Financial Times**: [Welcome to the new age of geoeconomics](https://www.ft.com/content/daf5ce11-c95b-4964-8df1-7463f0f9a81b) **Financial Times**: [Strategic interdependence is rewiring the global economy](https://www.ft.com/content/25ab5e71-6ae2-495b-bf06-6fd2f1cbb18a) ### Four Things: OpenAI's Backstop, CoreWeave CDS, Oregon, and Humanity's TAM URL: https://paulkedrosky.com/four-things-openais-backstop-coreweave-cds-oregon-and-humanitys-tam/ Last updated: 2025-11-07T18:27:18.000Z Four notes on how stress is finally showing up in the data center credit complex: 1. The OpenAI Debt Backstop 2. The Oregon PUC Power Debacle 3. CoreWeave CDS Blowout 4. What is the TAM on Replacing Humanity --- Let's do this: 1. **The OpenAI Debt Backstop** The recent flare-up around [**OpenAI’s semi- proposed debt backstop**](https://www.marketwatch.com/story/how-will-openai-cover-its-heavy-spending-it-now-suggests-maybe-the-government-could-help-e1ca64f5?gaa%5Fat=eafs&gaa%5Fn=AWEtsqeHeiqEMWRx-RTz8XiKnydYogAs7ZOtyS547DkFGcRL3v%5FsxBqIhxOS&gaa%5Fsig=a0wfsQxM9e9-53nXU7ryHeqfT0NBfsDR6IRbZ0jZoK-yP6OXgxGQljRA1g0aCuJ4htfrh-RDIYvl6eIkLO2VaA==&gaa%5Fts=690e329f&utm%5Fsource=chatgpt.com) reveals the structural tension between venture-style growth and infrastructure-scale spending, especially when the two become wrapped in the flag. The company's CFO’s comments floated the idea of a government guarantee for data-center financing — a remarkable signal that AI build-out costs have reached utility proportions. [**Sam Altman’s follow-up**](https://www.businessinsider.com/openai-ceo-sam-altman-addresses-strategy-government-backstop-rumors-2025-11) disavowed any such plan, saying OpenAI should “fail if we get it wrong,” but the damage was done. Investors now see an enterprise straddling the line between startup and state-sponsored critical infrastructure, with an overlay of an implied [**"Greenspan put"**](https://en.wikipedia.org/wiki/Greenspan%5Fput)—that AI is so important it can't be allowed to fail. The optics matter: lenders must begin to price OpenAI more like a **sovereign-backed utility** than a growth company. The episode marks the point where AI financing entered the quasi-public domain. Even if the “backstop” was rhetorical, it exposed how fragile the capital structure behind this compute-intensive ecosystem has become. It has also led to umpteen thousand memes, and, amusingly, a Polymarket betting line on when [**Sam Altman finds himself**](https://polymarket.com/event/sam-altman-in-jail-by?tid=1762526916517&dub%5Fid=QJ4R626uFXnqg8pF) a new house. --- 1. **The Oregon PUC Power Debacle** Connecting data centers to the grid has now turned into open warfare between tech firms and utilities. [**Amazon’s complaint against PacifiCorp**](https://edocs.puc.state.or.us/efdocs/HAA/um2410haa341355036.pdf) — filed with the Oregon Public Utility Commission — accuses the Berkshire Hathaway-owned utility of failing to deliver sufficient power to four new data-center campuses. The filing alleges that PacifiCorp provided only partial power to one site, none to a second, and has “refused to even complete its own contracting process” for two more. Some more tidbits: - **Extraordinary remedy sought — revocation of PacifiCorp’s exclusive service territory** Amazon explicitly asks the Oregon PUC to **remove its facilities from PacifiCorp’s monopoly territory** and reassign them to a utility “ready, willing, and able” to serve. That’s a rare and extreme regulatory demand; territory reallocation is almost never sought, let alone by a Fortune 500 customer. - **Alleged unilateral “32.6% tax gross-up” demand** PacifiCorp is accused of imposing a **post-contract policy change** adding a 32.6% surcharge on all capital contributions—retroactively—without regulatory or contractual basis. ADS contends PacifiCorp invented this internal rule years after contract execution. It is, as they say, bananas. **PacifiCorp’s** [**response**](https://www.bloomberg.com/news/articles/2025-11-04/amazon-says-berkshire-utility-failing-to-power-data-centers)**,** that it must avoid “direct and indirect harms between customers,” highlights the underlying conflict. The utility claims it cannot endanger ratepayer affordability by over-committing generation to hyperscalers. Amazon, meanwhile, argues that PacifiCorp breached 2021 obligations and is stalling projects worth billions.Some The episode exposes the **structural tension** at the core of grid policy — decarbonisation mandates and utility over-eagerness for new annuity-paying loads, plus political pressure, without corresponding capacity investment. Oregon’s utilities, still operating on legacy load models, cannot finance or permit new infrastructure fast enough. In all likelihood, no-one can. The result: rationed power, legal confrontation, and an augury. --- 1. **CoreWeave CDS Blows Out** I have written about CoreWeave a few times here, pointing out that it is among the more **speculative data center sponsors**. That is increasingly coming home now, as the company's credit default swaps show. ![](https://paulkedrosky.com/content/images/2025/11/unnamed.png) That chart shows **CoreWeave’s 5-year USD senior CDS (credit-default swap) spread** going out sharply — now at **≈510 bps**, up from \~360 bps in early October 2025. **Keys:** - **Credit risk repricing:** A 510 bps spread implies markets now demand > 5 % annual protection cost against CoreWeave default, roughly double early-fall levels. That’s a major deterioration in perceived credit quality. - **Steep late-October acceleration:** The curve steepened into November. - **Macro backdrop:** Wider moves echo stress across the **AI-infrastructure credit complex**—other GPU-linked names (e.g., hyperscale lessees, power-infrastructure ABS) are also widening, reflecting skepticism on 2026–27 cash-flow realization and project-finance rollovers, crossed with, you know, [other issues](https://www.reuters.com/world/first-brands-gets-600-million-new-funding-after-creditor-ceasefire-2025-11-07/) in the more speculative debt market. - **Prior levels were artificially tight.** The early-2025 spread (\~250–300 bps) priced CoreWeave almost like an investment-grade-adjacent credit, which, as I've previously argued, made little sense given its leverage, tenant concentration, and project-finance structure. _This post is for paying subscribers only._ ### Kedrosky Daily: Labor growth plummets 80 percent, VFX jobs drop 33 percent URL: https://paulkedrosky.com/kedrosky-daily-labor-growth-plummets-80-percent-vfx-jobs-drop-33-percent/ Last updated: 2025-11-07T12:47:28.000Z _This post is for paying subscribers only._ ### Kedrosky Daily: Grok lags in AI adoption, AI data center borrowing hits $126B URL: https://paulkedrosky.com/kedrosky-daily-grok-lags-in-ai-adoption-ai-data-center-borrowing-hits-126b/ Last updated: 2025-11-06T22:18:51.000Z _This post is for paying subscribers only._ ### Kedrosky Daily: Grok lags behind ChatGPT Gemini AI tech borrowing hits $126B in 2025 URL: https://paulkedrosky.com/kedrosky-daily-grok-lags-behind-chatgpt-gemini-ai-tech-borrowing-hits-126b-in-2025/ Last updated: 2025-11-05T12:48:29.000Z _This post is for paying subscribers only._ ### Updates: Bragawatts, Nvidia Theater, Me Stuff, etc. URL: https://paulkedrosky.com/updates-bragawatts-nvidia-theater-me-stuff-etc/ Last updated: 2025-11-05T04:37:10.000Z A few morning things worth noting. Topics: 1. Recent media interview. 2. Michal Burry's new Nvidia rodeo 3. Bragawatts and the O.K. Corral 4. The 5% of China argument --- Let's do this: 1. **Me, Me, Me** I was interviewed here and there in the last week on capex exuberance and related topics, perhaps most notably in [The Atlantic](https://www.theatlantic.com/technology/2025/10/data-centers-ai-crash/684765/?utm%5Fsource=flipboard&utm%5Fcontent=user/TheAtlantic). **2\.** [**Michael Burry's Nvidia bearishness**](https://www.bloomberg.com/news/articles/2025-11-04/burry-discloses-puts-on-nvidia-and-palantir-after-bubble-warning) Michael Burry of "Big Short" fame is in the news today, with a filing showing that he again has puts against AI darling Nvidia (as well as against who-knows-what-it-does darling Palantir). As Bloomberg reminds us, this is not Burry's first anti-Nvidia rodeo, having bet against the company earlier this year, which didn't work out. Of course, this is only the visible part of his portfolio, so it is possible that what looks like a bet against Nvidia is swamped by other long positions in the stock, perhaps hugely fun synthetic securities he has obtained from investment banks, but ... that seems unlikely. Does it matter that he is betting against the company? Probably not. While Nvidia is closer to the end of its current run than to the beginning—such is the law of large numbers—it is not clear Burry has a particular edge here, even if it's nice to see he can't stay out of the game. 1. [**KKR coins "bragawatts"**](https://www.ft.com/content/bf687d99-f373-4a41-8651-fca9dba83aa0) I have little patience for most private equity musings, as readers will know, but I will concede that KKR calling speculative multi-gigawatt data centers **"bragawatts"** is something I wish I had come up with. Nevertheless, it is a handy description. It is handy because it captures the idea that much of what is going is a kind of expensive posturing. Companies try to outdo one another with announcements of ever-larger data centers requiring ever more power. Will they ever be built? Who knows, but if they deter other entrants, then at least some of the mission is accomplished. But such mass deterrence strategies are inherently fragile disequilibria. If everyone shows up at the O.K. Corral armed to the don't-mess-with-me teeth, the result isn't peace, it's a gunfight. Or, perhaps, it is more like the old joke about always being sure to carry a bomb onto an airplane, because *what are the odds of there being two bombs*? amirite. The trouble, of course, is what happens in the aggregate: all these pseudo-rational behaviors incite more such behaviors, leading to, at best, what finance theorists call a "rational bubble". And bragawatts is as good a way as any of capturing that in compressed form. 1. **Getting 5% of China's market** People try to justify current AI spending to me on myriad bases, ranging from faith-based investing (this will work out because it always works out), to AGI (can you tell me the TAM for superintelligence? no? checkmate!), and more. I wish I had a bingo card. Among my favorites, however, is the "5% of China" argument. It has to do with capturing a slice of the impending global productivity benefits from AI, and it generally goes something like this: **Step 1: Global GDP baseline (2025):** ≈ $110 trillion nominal. **Step 2: 20% productivity gain from AI:** 0.20 × $110 T = **$22 trillion** in incremental output. **Step 3: AI tools/services capture 10% of that:** 0.10 × $22 T = **$2.2 trillion** annualized value. **Step 4:** Stand back and make bank. This is, of course, ludicrous. It is on par with people who say, longingly, if they could just get 5% of the Chinese market for \[insert thing here\] they'd have a multibillion dollar company. _This post is for paying subscribers only._ ### Kedrosky Daily: US weekend box office hits multi-decade low, bond maturities rise URL: https://paulkedrosky.com/kedrosky-daily-us-weekend-box-office-hits-multi-decade-low-bond-maturities-rise/ Last updated: 2025-11-04T12:47:31.000Z _This post is for paying subscribers only._ ### Kedrosky Daily: US Box Office Hits Multi-Decade Low, Convertible Bond Maturities Rise URL: https://paulkedrosky.com/kedrosky-daily-us-box-office-hits-multi-decade-low-convertible-bond-maturities-rise/ Last updated: 2025-11-03T15:00:13.000Z _This post is for paying subscribers only._ ### Kedrosky Daily — Sunday Edition: Oman jumps from unlisted in 2015 to 4th in 2025, highlighting significant imp… URL: https://paulkedrosky.com/kedrosky-daily-sunday-edition-oman-jumps-from-unlisted-in-2015-to-4th-in-2025-highlighting-significant-imp/ Last updated: 2025-11-02T15:24:48.000Z **November 02, 2025 — General** This is a weekly free look of our premium Kedrosky Daily email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). ## Two Graphs ![Daily Graph](https://paulkedrosky.com/content/images/2025/11/2025-11-01_10-55-56.png) Oman jumps from unlisted in 2015 to 4th in 2025, highlighting significant improvements in living conditions. This change underscores a substantial development focus, altering the global quality of life landscape and challenging traditional Western dominance in high living standards. ![Daily Graph](https://paulkedrosky.com/content/images/2025/11/2025-11-01_10-54-38.png) Luxembourg shows a striking 62% preference for Business & Law master's programs, starkly differing from the global average, which tends to be more balanced across fields. This dominance underscores Luxembourg's economic focus, impacting workforce skills and industry alignment significantly. ## Headlines China to Suspend Some Rare Earth Curbs, Probes on US Chip Firms ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-11-01/china-to-end-rare-earth-controls-probes-against-chip-companies)) … What Investors Learned From Tech Earnings ([**WSJ**](https://www.wsj.com/finance/stocks/tech-earnings-lessons-charts-df778001?gaa%5Fat=eafs&gaa%5Fn=AWEtsqcB0YmCeueIRrZgzqZqq58vAYP-7SF9pdKO4pjoQLvSruA6HY0EUNUz&gaa%5Fts=69077b42&gaa%5Fsig=%5Fac-qC6y4h%5FMwXhNoUYx19hNGne8JqpYpMGZInJqdSJon0tBvn3Qui-NGhN%5Fts6rn85ltOiHKHrcXQwCM2klnw%3D%3D)) … FT Magazine. He rescued the Michael Jackson estate. Can he save the star’s legacy? ([**FT**](https://www.ft.com/content/f425ca55-e6a3-4111-b55b-1b8cd46b6401)) … Why Medicare recipients should check their 2026 drug plans now ([**NY Times**](https://www.nytimes.com/2025/11/01/business/medicare-drugs-part-d.html)) … JPMorgan alerted U.S. to Epstein transfers involving Wall St. figures ([**NY Times**](https://www.nytimes.com/2025/10/30/business/jpmorgan-jeffrey-epstein.html)) … A.I. is making death threats way more realistic ([**NY Times**](https://www.nytimes.com/2025/10/31/business/media/artificial-intelligence-death-threats.html)) … The Ranks of Corporate Zombies Are Growing in Debt Squeeze ([**Bloomberg**](https://www.bloomberg.com/)) … Why Companies Are No Longer Hanging On to Employees ([**WSJ**](https://www.wsj.com/economy/jobs/layoffs-labor-market-92a069c6?gaa%5Fat=eafs&gaa%5Fn=AWEtsqcCbCFw9A9qH1SUkF5QoWV9Se5Kv6dT3etjA5F-gN2b%5FIcTbmCYhZJe&gaa%5Fts=69077b42&gaa%5Fsig=X8yk-gj7fxwwsKFiAulmpeu7TiImOleJ8jGrFsWiPIgSNwAeBL0AO8bZOZItR%5FUPq3Ya2wyGsfFF0IYEi3asYw%3D%3D)) … The Amazonification of Whole Foods Is Finally Here—Bring On the Doritos ([**WSJ**](https://www.wsj.com/business/retail/whole-foods-brand-identity-amazon-9953d8a0?gaa%5Fat=eafs&gaa%5Fn=AWEtsqfxHl0rb4Vit5ZSJ2Mv18ht50aAeKOMiFz2-NlFNwNrJBbcSxZYiyj6&gaa%5Fts=69077b42&gaa%5Fsig=UB%5FxibSJcrJ5InjChoMORVLXsM7%5FV6vfLB9tjhOF34Gdq2uTcR6xBnZjbUz9-wlovpiZ9V7DouaWyNapeJqsMg%3D%3D)) … Berkshire Hathaway offloads $6.1bn of stock ([**FT**](https://www.ft.com/content/bfaac196-fbc5-4c4e-bfe0-a78a63149cb4)) ## The Idea **Mentions:** geoeconomics, trade partners, restructuring globalization ### Geopolitical alliances restructuring global trade flows As geopolitical alliances strengthen, trade patterns are increasingly aligning along these lines rather than traditional economic logic. This can lead to trade inefficiencies and force companies to adapt their supply chains, creating both risks and opportunities in global markets. This trend is expected to evolve over the next 12–24 months. **Related reading:** **FreightWaves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) **Financial Times**: [Welcome to the new age of geoeconomics](https://www.ft.com/content/daf5ce11-c95b-4964-8df1-7463f0f9a81b) **Financial Times**: [Strategic interdependence is rewiring the global economy](https://www.ft.com/content/25ab5e71-6ae2-495b-bf06-6fd2f1cbb18a) ### Kedrosky Daily (Commodities edition): Healthcare and service roles exhibit minimal core AI development compared to … URL: https://paulkedrosky.com/kedrosky-daily-commodities-edition-healthcare-and-service-roles-exhibit-minimal-core-ai-development-compared-to/ Last updated: 2025-10-30T15:34:38.000Z _This post is for paying subscribers only._ ### Kedrosky Daily (Commodities edition): AI recruiting tools dominate healthcare sectors like dental and nursing, high… URL: https://paulkedrosky.com/kedrosky-daily-commodities-edition-ai-recruiting-tools-dominate-healthcare-sectors-like-dental-and-nursing-high/ Last updated: 2025-10-30T11:48:00.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): AI Integration in Healthcare Tech and Diagnostics URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-ai-integration-in-healthcare-tech-and-diagnostics/ Last updated: 2025-10-29T11:49:31.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): AI Transformations and Market Realignments URL: https://paulkedrosky.com/four-ideas-ai-edition-ai-transformations-and-market-realignments/ Last updated: 2025-10-28T11:47:56.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Interest Rate Moves and Structural Economic Changes URL: https://paulkedrosky.com/four-ideas-macro-edition-interest-rate-moves-and-structural-economic-changes/ Last updated: 2025-10-27T11:48:38.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: AI Transformation in Healthcare and Industry Shifts URL: https://paulkedrosky.com/four-ideas-sunday-edition-ai-transformation-in-healthcare-and-industry-shifts/ Last updated: 2025-10-26T11:48:01.000Z **OCTOBER 26, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** Pay Attention to Your Savings Now, Just Before the Fed Considers a Rate Cut ([**NY Times**](https://www.nytimes.com/2025/10/24/business/interest-rates-cds-savings-accounts.html)) … ‘Where the hell is the market risk?’ Scott Bessent takes on his critics ([**FT**](https://www.ft.com/content/96ac2e66-77cb-43ed-804a-15b0ac078633)) … Australia unveils cyber security push for Indo-Pacific region ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-10-26/australia-unveils-cyber-security-push-for-indo-pacific-region)) … Hollywood Weeps as Warner Bros. Hangs a Sale Sign ([**NY Times**](https://www.nytimes.com/2025/10/24/business/warner-bros-hollywood-reaction.html)) … The lo-fi charm of the audacious Louvre heist ([**FT**](https://www.ft.com/content/d0e3e9bf-c146-4c7d-acec-e204fcabbf1e)) … China’s solar installations rise from near three-year low ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-10-26/china-s-solar-installations-rise-from-near-three-year-low)) … Being In an N.B.A. Player’s Entourage Is Fun. Now, It Can Be a Gambling Edge, Too. ([**NY Times**](https://www.nytimes.com/2025/10/25/business/nba-gambling-scandal-players.html)) … Milei’s Free-Market Experiment Hangs in the Balance as Argentina Votes ([**WSJ**](https://www.wsj.com/world/americas/mileis-free-market-experiment-hangs-in-the-balance-as-argentina-votes-4af71570?gaa%5Fat=eafs&gaa%5Fn=AWEtsqe7DRa1ykQH5mWi2Y1gWb02NwY8D4Yjve0O%5FzujQAPnH7g9vBPNh40T&gaa%5Fts=68fe0df3&gaa%5Fsig=oP-5bomQl6isPbg1pY8zm8Ww1wGaiESvZ8p2K3Hb-i-d6H9rG8wcTUf6EtX2TcBOzBZblEmtjs0ZODWplCZOtw%3D%3D)) … Britain Nears Its Net-Zero Breaking Point ([**WSJ**](https://www.wsj.com/opinion/keir-starmer-net-zero-climate-britain-labour-party-04c069aa?gaa%5Fat=eafs&gaa%5Fn=AWEtsqeE9X5f3f7dUQUeGlreWQOlswVfyH45RK%5FLL4a5tYe8WjCq5orIagBy&gaa%5Fts=68fe0df3&gaa%5Fsig=J9Lr8WdZ4CCyzi3zP4zR9L1DXl2BbrsL8xx6WRW95GYBDOhzzpw6CXOOU%5F6Sh4SAGLvSj921z6kRvgCpnuo6%5Fw%3D%3D)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/10/2025-10-25_07-22-58.png) Adults aged 55-64 have increased daily media consumption, with time spent on social media and music streaming tripling since 2013\. The shift from TV indicates changing digital habits, impacting how media platforms target mature audiences. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/10/2025-10-25_08-06-25.png) Amidst 73 buy ratings for Nvidia, only one analyst issued a sell rating. This stark contrast emphasizes overwhelming market confidence versus a singular bearish perspective, highlighting potential dissent or unique insights in market evaluations. --- **THE IDEAS:** **Mentions:** SPGSustainable1, Macrobond, Bloomberg, FT, CNBC ### 01\. AI expansion in healthcare increases efficiency AI tools in healthcare are expanding rapidly, as seen in UMass Memorial's adoption, which boosts operational efficiency and patient care. This trend predicts a wider AI adoption in the sector, driving cost reductions and improved health outcomes. **Related reading:** **Becker's Hospital Review**: [UMass Memorial expands AI triage tool to 5 more hospitals](https://www.beckershospitalreview.com/healthcare-information-technology/ai/umass-memorial-expands-ai-triage-tool-to-5-more-hospitals/) ### 02\. Geoeconomic ties reshaping global trade dynamics Geopolitical alliances are increasingly forming economic partnerships, influencing trade patterns as seen in McKinsey's observation. This shift is leading to a realignment in global trade, impacting supply chains, and fostering regional trade relations over rival economic interactions. **Related reading:** **FreightWaves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) **Financial Times**: [Welcome to the new age of geoeconomics](https://www.ft.com/content/daf5ce11-c95b-4964-8df1-7463f0f9a81b) ### 03\. Structural shifts boost global stock tailwinds Structural economic changes, such as shifts in trade relations and long-term investment trends, are providing tailwinds for global stocks. Bloomberg highlights how these factors are setting a bullish long-term outlook. **Related reading:** **Bloomberg**: [Global Stocks Have Structural Tailwinds: 3-Minute MLIV](https://www.youtube.com/watch?v=xeUzxrFT5Kw) ### 04\. Industry recovery signals growth opportunities Following a downturn, the industrial sector is seeing a resurgence, marking a critical juncture for growth. This trend offers investment opportunities as highlighted by Oxford Economics, indicating sectoral momentum and innovation. **Related reading:** **Oxford Economics**: [Industry Key Themes 2025: Industrial landscape at a critical juncture](https://www.oxfordeconomics.com/resource/industry-key-themes-2025-industrial-landscape-at-a-critical-juncture/) ### Oct 24, 2025 Update: Nvidia as Pop Mart, Power, Policy, etc. URL: https://paulkedrosky.com/oct-24-2025-update-nvidia-as-pop-mart-power-policy-etc/ Last updated: 2025-10-24T12:47:02.000Z It's been a wildly busy week, so I will touch quickly on a few things that keep coming up in my discussions. Here is what's ahead: 1. **AI and Trade Policy** 2. **Nvidia as Pop Mart** 3. **GPU Spot Prices** 4. **The Five Alarm Fire in Power** --- 1. **AI and Trade Policy** I regularly make the point, one that seemingly confuses people, that AI spending—capex, in particular—is a **significant fraction of U.S. GDP growth**, perhaps more than half, and that it can thus **confound policy**. For example, if I want to know the effect of tariffs it can be harder than expected to untangle it, as well as their effects being muted, by the **massive stimulus flowing** from AI capex. This seems incontrovertible. **Trade policy is hard**, and evolving confounds make it harder. This is not the same thing, however, as saying "tariffs bad", "tariffs good", etc. It is simply saying that when you have large, novel, and fast-moving rivers of money flowing around as you conduct trade policy it can make it difficult to know **what is causing what**, which is among the main goals of evaluating said policy. As I like to point out, the **money flowing into AI data centers** is coming from somewhere. And it is, by definition, thus **not going other places**. To the extent that a goal of current U.S. trade policy is to onshore more manufacturing, that can make it more difficult for manufacturers, at least at the margin, to get all the capital they need. 1. **Nvidia as Pop Mart** Arguing that it is fairly obvious we are in a financial bubble is not the same thing as saying **AI is the** [**Labubu**](https://en.wikipedia.org/wiki/Labubu) **of technology**. This seems an obvious point, but is a constant rejoinder to talk of bubbles. AI is not some whimsical thing being driven higher by dopey investors chasing a shiny object; **Nvidia is not the Pop Mart** of technology. Clearly stated: AI is among the **most important developments** in technology, and current AI progress is the most significant in that domain in my lifetime. It has myriad uses; there will be more, and progress continues. Saying that movement of the stocks, or the related overspending, have **all the hallmarks of a financial bubble** is to state the obvious—and is not the same thing as saying AI is useless. The global financial crisis was among the largest bubbles in modern economic history, and one driver was excess home construction. But **homes were not a fad**. It was the excessive spending and speculative nexus around the homes that led the eventual collapse of the **related credit complex**, and ensuing systemic crisis. 1. **GPU Spot Prices** Once an idea takes hold among people, it is hard to get them to drop it. A current example is the idea that **Nvidia GPUs are in scarce supply**. They are not, or least aren't in the usual narrative sense. A quick check of [marketplace prices](https://medium.com/@cli%5F87015/gpu-rental-market-snapshot-how-a100-and-h100-gpu-prices-vary-across-u-s-1e50efb45314) will show you that prices are down by almost **half this year**, depending on location and specific GPU. This is not the hallmark of something in short supply. The usual response to this is that marketplaces aren't good indications of Real AI Work. The latter, it is argued, requires service level agreements, redundant power, and reservation systems, the sort of things for which **major hyperscalers get 2-4x rental** premiums over marketplaces. Outside of training frontier models or massive-scale chatbots from the big three/four, I am **less convinced** of this than ever. An increasing fraction of startup, emergent model, and other work is done on marketplaces, given the huge **price advantage and increasing supply** of rental GPUs. This is putting growing pressure on hyperscalers, at least at the margin, given that the latter are **trapped by their cost of capital,** which prevents them from competing on price. 1. **The Five Alarm Fire in Power** _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Commodity Trends and Geopolitical Dynamics URL: https://paulkedrosky.com/four-ideas-commodities-edition-commodity-trends-and-geopolitical-dynamics/ Last updated: 2025-10-23T11:47:53.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): Healthcare AI, Health Tech Investments, AI in Healthcare, FDA Breakthrough Devices URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-healthcare-ai-health-tech-investments-ai-in-healthcare-fda-breakthrough-devices/ Last updated: 2025-10-22T11:48:13.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Emerging Macro Trends Impact on Non-Obvious Sectors URL: https://paulkedrosky.com/four-ideas-macro-edition-emerging-macro-trends-impact-on-non-obvious-sectors/ Last updated: 2025-10-20T11:47:55.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-50/ Last updated: 2025-10-19T11:47:32.000Z **OCTOBER 19, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** Education Dept. Agrees to Resume More Student Loan Cancellations ([**NY Times**](https://www.nytimes.com/2025/10/17/business/student-loan-cancellation-idr.html)) … ‘Girl, Take Your Crazy Pills!’ Antidepressants Recast as a Hot Lifestyle Accessory ([**WSJ**](https://www.wsj.com/health/wellness/anti-depressants-lifestyle-accessory-3b66027d?gaa%5Fat=eafs&gaa%5Fn=AWEtsqdlx8HR5SaLPwpmKHf0%5F4-Rl9%5FLAQ8iQQBDmRXCuRV3BM9w6pD-I9WV&gaa%5Fts=68f4d357&gaa%5Fsig=NV2UAx66rLYU39BbE%5F5EiPvgT68z4PTd2b1WDBmJf%5Fq61zNth5bg6oBPD6Pu5lGINElZ8PqaDhLoX9hK68PLSQ%3D%3D)) … Delta and United Are Leaving Other Airlines Behind ([**NY Times**](https://www.nytimes.com/2025/10/16/business/delta-united-airlines-profits.html)) … Traders Spooked as Bank Lending Risk Puts Stock Market on Edge ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-10-18/traders-spooked-as-bank-lending-risk-puts-stock-market-on-edge)) … Chip Firm Arrow Says Its Units To Be Removed from US Blacklist ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-10-18/chip-firm-arrow-says-its-units-to-be-removed-from-us-blacklist)) … When Moving in Retirement Becomes an Expensive Reality Check ([**NY Times**](https://www.nytimes.com/2025/10/18/business/retirement-moving-regret.html)) … Chinese tech giants pause stablecoin plans after Beijing steps in ([**FT**](https://www.ft.com/content/8bc1fcfa-8c69-4de0-b6cb-ab6f7117558b)) … Why are parents behaving so badly on the sidelines? ([**FT**](https://www.ft.com/content/74406a56-4367-4e48-8fb5-2b502771f4c8)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/10/2025-10-18_20-02-35.png) BDC debt risk surged, with the spread gap reaching 100 basis points, the widest since the April tariff shock. This spike indicates heightened financial instability compared to general financials, impacting investor confidence and signaling potential challenges for BDCs in capital markets. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/10/2025-10-18_11-32-09.png) OnlyFans generates $37.6M per employee, over ten times more than NVIDIA's $3.6M. This highlights OnlyFans' extraordinary efficiency compared to tech giants. Such disparity underscores a significant shift in business models, indicating the potential for niche platforms to outperform traditional tech sectors in revenue generation. --- **THE IDEAS:** **Mentions:** SPGlobal, EU, UMass Memorial, Macrobond, McKinsey ### 01\. Geopolitical trade alliances alter economic dynamics The formation of tight-knit trade alliances among geopolitical partners signifies a shift from traditional open trade practices. This is driven by national security concerns and the desire to reduce reliance on rival economies. This trend could lead to regional economic clustering and impact global supply chains significantly over the next 6-24 months. **Related reading:** **FreightWaves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) **Financial Times**: [Strategic interdependence is rewiring the global economy](https://www.ft.com/content/25ab5e71-6ae2-495b-bf06-6fd2f1cbb18a) **Financial Times**: [Welcome to the new age of geoeconomics](https://www.ft.com/content/daf5ce11-c95b-4964-8df1-7463f0f9a81b) ### 02\. AI tools expand in healthcare settings The rapid expansion of AI-based healthcare tools, like UMass Memorial's triage system, shows increased integration of AI in medical facilities. This development, driven by the demand for efficiency and improved patient outcomes, could lead to reduced operational costs and improved service quality within 6-24 months. **Related reading:** **Becker's Hospital Review**: [UMass Memorial expands AI triage tool to 5 more hospitals](https://www.beckershospitalreview.com/healthcare-information-technology/ai/umass-memorial-expands-ai-triage-tool-to-5-more-hospitals/) ### 03\. Industrial growth driven by new strategic hubs The identification of future manufacturing hot spots in industrial hubs such as Japan highlights upcoming economic growth areas. With regional economic policies being realigned, significant opportunities for job creation and investment are expected within the next couple of years. **Related reading:** **Oxford Economics**: [Identifying future manufacturing hot spots in Japan and their implications](https://www.oxfordeconomics.com/resource/identifying-future-manufacturing-hot-spots-in-japan-and-their-implications/) **Oxford Economics**: [Industrial Hubs: Finding opportunity amid the challenges](https://www.oxfordeconomics.com/resource/industrial-hubs-finding-opportunity-amid-the-challenges/) ### 04\. Climate risks increasingly affecting market strategies Investor considerations of climate risks are growing as market anomalies reflect the increasing impact of geoeconomic factors. This trend suggests that climate risk assessment will become central to strategic investment decisions, affecting sectors reliant on consistent climate conditions. **Related reading:** **Oxford Economics**: [What is the ‘hidden’ climate risk in your financial portfolio?](https://www.oxfordeconomics.com/resource/what-is-the-hidden-climate-risk-in-your-financial-portfolio/) **Phys.org**: [Extreme weather alerts can move markets. Here's what investors can learn from our new research](https://phys.org/news/2025-08-extreme-weather-investors.html) ### Quick Hits: AI Capex and Consumer Spending URL: https://paulkedrosky.com/quick-hits-ai-capex-and-consumer-spending/ Last updated: 2025-10-17T16:29:55.000Z Apologies for a second note today, but a **new report from JPM** economists fits so neatly into what we've been talking about there that I had to mention it. I have written many times (see [here](https://paulkedrosky.com/honey-ai-capex-ate-the-economy/), [here](https://paulkedrosky.com/honey-ai-capex-keeps-eating-everything/), and [here](https://paulkedrosky.com/quick-hits-ai-capex-is-eating-san-francisco/)) about how **spending on data centers** is now a **large fraction of U.S. GDP growth** simply based on hyperscaler spending. I have speculated, given typical multipliers, how much larger its effects might actually be, once one considers its broader effects on suppliers, **consumer spending**, and more. Well, here are some highlights from a new JPM report: - **AI = 44% of S&P500** — A basket of 30 AI-linked firms now represents nearly *half* of total U.S. equity market cap. We are at dot-com-era extremes. _This post is for paying subscribers only._ ### Minsky Moments and AI CapEx URL: https://paulkedrosky.com/minsky-moments-and-ai-capex/ Last updated: 2025-10-19T22:47:54.000Z There were **two key events** this week in our ongoing AI capex wave. I will highlight both of them, at different ends of the market, because they are **"unlocking"** moments. First, however, let's put the ongoing spending wave in a credit expansion context, using Hyman Minsky’s **Financial Instability Hypothesis**. He described how periods of economic stability sow the seeds of future instability through leverage and risk-taking. Minsky divided financing behavior into three regimes: 1. **Hedge finance**, where borrowers can meet all debt obligations from cash flow. 2. S**peculative finance**, where they can service interest but must roll over principal, and 3. **Ponzi finance**, where repayment depends on ever-rising asset prices or new borrowing. Over time, **Minsky argued**, as stability breeds complacency, economies drift from hedge toward **Ponzi finance**, creating a **self-reinforcing boom** driven by optimism and easy credit. Eventually, a shock—often minor—exposes cash-flow shortfalls, forcing asset sales and deleveraging. This abrupt reversal, the **“Minsky moment,”** as [Paul McCulley coined it](https://www.pimco.com/us/en/insights/credit-markets-and-the-real-economy-is-the-financial-system-working?utm%5Fsource=chatgpt.com) in 1998\. famously triggers a cascade of defaults and falling asset prices, turning stability into crisis. Where are we in that cycle today with respect to data center financing? After all, the sums keep **spiraling**, with every year seeing regular **revisions higher**. Consider this: as the following figure shows, **2026 capex forecasts** for the top 4 hyperscalers alone **grew almost 50%** *during the year*. That is ... remarkable. ![](https://paulkedrosky.com/content/images/2025/10/2025-10-16_13-22-28-2.png) An important point to note here is that, leaving aside what's happening in high yield data center debt, which I will talk about below, **investment-grade data center debt** is already the **largest "segment"** of the investment-grade market. I owe a JPM report this week for pointing that out, even if JPM's credit analysts came to **sanguine conclusions** about the implications. ![](https://paulkedrosky.com/content/images/2025/10/c832e695414462b505e14b09ca8e235632270ba96d2348fc2e1e07764b76284e.png) Let's now turn to those two aforementioned events this week: 1. TeraWulf's junk bond issuance 2. Meta and Blue Owl's SPV ## 1\. TeraWulf's Junk Bond Issuance ### What Happened - TeraWulf is offering **$3.2B in senior secured notes** due 2030 to fund expansion of its Lake Mariner data center. ([Source](https://investors.terawulf.com/news-events/press-releases/detail/118/terawulf-inc-announces-proposed-offering-of-3-2-billion?utm%5Fsource=chatgpt.com)) - This is its **first major foray into junk bonds**, backed in large part via collateral and “Google backstop” arrangements (Google pledging warrants, lease guarantees) to shore up investor confidence. ([Source](https://www.bloomberg.com/news/articles/2025-10-16/-google-backstop-adds-new-twist-to-data-center-financing-frenzy?utm%5Fsource=chatgpt.com)) - Credit rating agencies have rated the debt as junk (Ba2 / BB). ([Source](https://www.bloomberg.com/news/articles/2025-10-16/-google-backstop-adds-new-twist-to-data-center-financing-frenzy?utm%5Fsource=chatgpt.com)) - It was well over-subscribed, with order interest reportedly exceeding $10B. ### What It Means - We’re now seeing **elevated leverage and speculative financing** in AI infrastructure. - The presence of a Google “backstop” suggests non-hyperscaler deals are beginning to **require implicit guarantees** to attract capital. - Demand is strong, but **yield spreads** remain elevated. - The structure resembles **speculative finance dynamics**. ## 2\. Meta/Blue Owl SPV _This post is for paying subscribers only._ ### Four Ideas (Energy edition): Renewable Energy Grid Innovations and EV Shifts URL: https://paulkedrosky.com/four-ideas-energy-edition-renewable-energy-grid-innovations-and-ev-shifts/ Last updated: 2025-10-17T11:47:59.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Global Commodity Trends and Implications URL: https://paulkedrosky.com/four-ideas-commodities-edition-global-commodity-trends-and-implications/ Last updated: 2025-10-16T11:48:11.000Z _This post is for paying subscribers only._ ### Rougher Notes, Oct 15, 2025: Blackouts, TeraWulf, Debt, etc. URL: https://paulkedrosky.com/rougher-notes-oct-15-2025-blackouts-terawulf-debt-etc/ Last updated: 2025-10-16T01:36:41.000Z Once again, links to sources for **upcoming pieces**, before the pieces themselves: - [**Revisiting “Intelligence Drift” – Charlie Guo**](https://www.ignorance.ai/p/revisiting-intelligence-drift?utm%5Fsource=post-email-title&publication%5Fid=1407539&post%5Fid=176263427&utm%5Fcampaign=email-post-title&isFreemail=true&r=2w5leh&triedRedirect=true&utm%5Fmedium=email) — explores **AI intelligence drift**, model crash, and the **limits of system self-improvement**. - [**New York City Faces Increased Blackout Risk Next Summer – Crain’s NY Business**](https://www.crainsnewyork.com/climate/new-york-city-faces-increased-blackout-risk-next-summer-grid-operator-warns) — warns of **grid reliability risks**, **peak-load strain**, and **aging infrastructure** in NYC’s **summer 2026** outlook, and the role of **data centers** therein - [**TeraWulf Floats $3.2 Billion of Junk Bonds for NY Data Center – Bloomberg**](https://www.bloomberg.com/news/articles/2025-10-14/terawulf-floats-3-2-billion-of-junk-bonds-for-ny-data-center) — details **high-yield financing**, **AI-linked data-center expansion**, and **credit-market stress**. - [**JPMorgan: $1.2 Trillion Debt Tied to AI Tops Bank High-Grade Exposure – Bloomberg**](https://www.bloomberg.com/news/articles/2025-10-07/jpmorgan-says-1-2-trillion-debt-tied-to-ai-tops-bank-high-grade) — quantifies **AI-related corporate debt**, highlighting **structured-credit concentration** and **systemic exposure**. - [**Powered Land Demand Reshapes Real Estate in the AI Era – CRE Daily**](https://www.credaily.com/briefs/powered-land-demand-reshapes-real-estate-in-the-ai-era/) — examines how **power-dense land** and **data-center siting** are transforming **real-estate valuation**. _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): Healthcare AI Innovation and Market Dynamics URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-healthcare-ai-innovation-and-market-dynamics/ Last updated: 2025-10-15T11:49:14.000Z _This post is for paying subscribers only._ ### Quick Hits: AI Capex is Eating ... San Francisco URL: https://paulkedrosky.com/quick-hits-ai-capex-is-eating-san-francisco/ Last updated: 2025-10-14T22:00:19.000Z A remarkable thing is going on in the **San Francisco office market**. I first saw signs about a year ago, as companies reported it was newly less straightforward to find office space, and, meanwhile, erstwhile **Florida escapees** began returning. Now, however, it is a tenancy stampede, with major new **AI-related leases** being signed, and former Floridians everywhere. Some **quick hits** on what's going on, and what it means: **Where We Were** - In 2023, San Francisco held the **highest office vacancy rate in the U.S.**, peaking \~**34%**. - Pandemic-driven remote work, tech layoffs, and sublease glut left downtown hollowed out. - The city became a synonym for post-COVID office collapse. **Where We Are** - Vacancy has fallen to **\~23% in 2025**, now matching major metros like **Chicago, Los Angeles, and D.C.** - AI firms — notably **OpenAI, Vercel, and Harvey AI** — have leased **\>5 M SF** since 2023.¹ - Sublease availability is down **30%**, signaling the first sustained tightening in years. _This post is for paying subscribers only._ ### Four Ideas (AI edition): AI Impact on Utilities and Travel URL: https://paulkedrosky.com/four-ideas-ai-edition-ai-impact-on-utilities-and-travel/ Last updated: 2025-10-14T11:48:13.000Z _This post is for paying subscribers only._ ### Quick Hits: An India of Incremental Power Ahead URL: https://paulkedrosky.com/quick-hits-an-india-of-incremental-power-ahead/ Last updated: 2025-10-13T18:27:15.000Z Adding an India of power capacity means coal stays on for longer _This post is for paying subscribers only._ ### Quick Hits: The Data Center Growth Slowdown URL: https://paulkedrosky.com/quick-hits-the-data-center-growth-slowdown/ Last updated: 2025-10-13T17:49:15.000Z The folks at Apollo have a new [graph out](https://www.apolloacademy.com/data-center-growth-rates-peaked-two-years-ago/) showing a sharp data center growth **slowdown** from a **peak two years ago**. What should we make of it? Is the **bubble** already bursting? ![](https://paulkedrosky.com/content/images/2025/10/2025-10-13_06-57-10.png) **My comments**: 1. **Post-COVID Base Effects** The slowdown in the graph is largely due to post-pandemic comparables. Data center construction froze in 2020–21, then surged again as delayed projects restarted. This, in turn, increased the 2022–23 spending growth, which made the post-2023 “decline” more dramatic. 1. **Data Centers Still Expanding Rapidly** Even at \~30–40% YoY by mid-2025, growth remains very high relative to other sectors. It is more accurately described as a slowing from a post-pandemic surge. and definitely not a contraction. 1. **Supply-Chain and Cost Distortions** _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Global Fragmentation, Reserve Diversification, Geoeconomics Impact, Industrial Growth URL: https://paulkedrosky.com/four-ideas-macro-edition-global-fragmentation-reserve-diversification-geoeconomics-impact-industrial-growth/ Last updated: 2025-10-13T11:48:11.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-49/ Last updated: 2025-10-12T11:47:41.000Z **OCTOBER 12, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** Israel Approves Hostage Deal Setting a Cease-Fire in Gaza ([**WSJ**](https://www.wsj.com/world/middle-east/israel-hamas-gaza-hostage-ceasefire-deal-815d533f?gaa%5Fat=eafs&gaa%5Fn=ASWzDAihee2V65B4TPSf9PylVoeNXQ8fG4h0ZMiWUX0MXZYc0ofghBVlV8cD&gaa%5Fts=68eb98e0&gaa%5Fsig=eLlUO4-MTKDMOXEuJzw39koYeIycv1x1grN5pjT5%5FgGKRMdC1Y2ECw7NTRfKy3cwzYUsa9ZkY%5FMO91UeiZRdug%3D%3D)) … The 11th-Hour Campaign to Land Trump a Nobel Peace Prize ([**WSJ**](https://www.wsj.com/world/trump-nobel-peace-prize-campaign-e30ec554?gaa%5Fat=eafs&gaa%5Fn=ASWzDAg5iBoqRitTnoaEp8J6Pki5NMRTAu5G-MqYbPqAnrjJ1OcMP0xOuIWD&gaa%5Fts=68eb98e0&gaa%5Fsig=VAPfu8NhzHwe3veAPSP3YfBggemCfCa5bvwePXFzUW8dvKCi5zgZA6PK5P165UPbht29Wux8Y31PFma%5FZ9lTtg%3D%3D)) … A Mystery C.E.O. and Billions in Sales: Is China Buying Banned Nvidia Chips? ([**NY Times**](https://www.nytimes.com/2025/10/09/technology/nvidia-chips-china-megaspeed.html)) … TSMC, Samsung in Spotlight as US Pressure Grows in Chip Sector ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-10-10/tsmc-samsung-in-spotlight-as-us-pressure-grows-in-chip-sector?srnd=phx-technology)) … Copper Declines After China-Led Rally as Supply Cuts in Focus ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-10-10/copper-declines-after-china-led-rally-as-supply-cuts-in-focus)) … Intel’s Big Bet: Inside the Chipmaker’s Make-or-Break Factory ([**NY Times**](https://www.nytimes.com/2025/10/09/technology/intel-chip-factory-fab52.html)) … Vingroup Seeks $500 Million Private Credit for EV Charging Ports ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-10-10/vingroup-seeks-500-million-private-credit-for-ev-charging-ports)) … Inside Intel’s big bet to save US chipmaking — and itself ([**FT**](https://www.ft.com/content/e4b1fa29-6027-44d9-8375-b3368132dd1b)) … A.I. Video Generators Are Now So Good You Can No Longer Trust Your Eyes ([**NY Times**](https://www.nytimes.com/2025/10/09/technology/personaltech/sora-ai-video-impact.html)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/10/2025-10-09_20-12-17.png) U.S. growth funds experienced outflows nearly twice as substantial as value funds, peaking around $34 billion compared to $16 billion. This reflects investor preference volatility, significantly impacting growth-focused portfolios and potentially influencing market strategies. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/10/2025-10-09_20-12-35.png) Short-bias ETF inflows peaked around mid-2025, reaching nearly $3,500 million, a significant rebound from the early 2024 dip below -$4,000 million. This shift indicates a substantial change in market sentiment or investor strategy, potentially reflecting increased anticipation of market downturns. --- **THE IDEAS:** **Mentions:** NVDA, META, SPGI, FT, McKinsey ### 01\. Geoeconomics reshaping market and political dynamics Geopolitics' influence on financial markets is rising, blending national security concerns with economic decisions. This trend could lead to shifts in trade policies and realignments in stock markets as countries prioritize strategic partners over cost-efficiency. **Related reading:** **Twitter**: [Good morning FYI, my thoughts on how financial market anomalies, including the coincidence of record prices for both risk-on (stocks) and risk-off (gold) assets, are indicators of the growing influence of geoeconomics...](https://x.com/elerianm/status/1972623114094686583) **FT**: [Welcome to the new age of geoeconomics](https://www.ft.com/content/daf5ce11-c95b-4964-8df1-7463f0f9a81b) ### 02\. Cross-border payments become geopolitical tools Innovation in cross-border payments is not just technical but also strategic, potentially altering the balance of power in global finance by reducing reliance on traditional banking systems and aligning with political interests. **Related reading:** **FT**: [The race for the future of cross-border payments](https://www.ft.com/content/44d973d6-1c38-4eaf-a943-6aa6367177f6) **YouTube**: [Global economy in uncharted territory from tariff impact](https://www.youtube.com/watch?v=hQ--LYj07Ys) ### 03\. AI adoption accelerates amidst geopolitical scrutiny With McKinsey limiting China-focused AI work and increased geopolitical tension, AI adoption faces regulatory scrutiny, changing which firms lead in technological advancements and affecting global technology partnerships. **Related reading:** **FT**: [McKinsey bars China practice from generative AI work amid geopolitical tensions](https://www.ft.com/content/9907da48-3e6d-4208-a1d7-44a85c6a77f8) **Twitter**: [As generative #AI sweeps across industries and the #consumer landscape...](https://twitter.com/SPGlobal/status/1860035529909080480) ### 04\. Industrial growth rebounds from prolonged weakness After years of stagnation, industrial sectors are regaining momentum due to strategic investments and policy shifts, leading to regional economic growth and job creation, especially in future manufacturing hubs. **Related reading:** **Oxford Economics**: [Industry Key Themes 2025: Industrial landscape at a critical juncture](https://www.oxfordeconomics.com/resource/industry-key-themes-2025-industrial-landscape-at-a-critical-juncture/) **Oxford Economics**: [Industrial Hubs: Finding opportunity amid the challenges](https://www.oxfordeconomics.com/resource/industrial-hubs-finding-opportunity-amid-the-challenges/) ### Quick Hits: Roger Rabbit's Data Center, Colo Spin, etc. URL: https://paulkedrosky.com/quick-hits-roger-rabbits-data-center-colo-spin-etc/ Last updated: 2025-10-10T15:49:47.000Z A few quick working notes on things I'm thinking about: ## Data Centers 1. **Powered land and its Chinatown-ish consequences** As all aspects of data centers are increasingly financialized into a stack, new categories of strange keep emerging. Among the more recent ones is that of "powered land", companies that buy up land and associated power in areas with dense interconnect, in a kind of mutated Chinatown gambit (with power rather than water as the focus), then sell that back to data center operators as a turnkey site. Among the larger such firms is Cloverleaf Industries¹, which is in [the news](https://www.datacenterdynamics.com/en/news/cloverleaf-drops-initial-plans-for-michigan-data-center-council-votes-to-block-water-supply/) for having been blocked in an effort to over-speedily back a Michigan data center. It is another example of how local [resistance](https://www.datacenterdynamics.com/en/news/data-center-plan-for-campbell-county-virginia-rejected/) to data centers is growing, especially given the role of stealth applications, NDAs, and numbered companies, some of which are undoubtedly "powered land" operators like Cloverfield. ¹ A name that will be familiar to fans of the film Who Framed Roger Rabbit, where Cloverleaf was the company Judge Doom used to buy up land and block Los Angeles subway construction. Irony is dead, obviously. 1. **Colo operators talking up enterprise** A recent [interview](https://www.datacenterdynamics.com/en/videos/dcdtalks-a-snapshot-of-global-data-center-growth-with-steven-lim-ntt-global-data-centers/) with an executive at a major colo operator² is worth watching for what is said between the lines. Said operators are increasingly under investor pressure as hyperscalers noisily move to owned facilities, so colo operators talk about their business speaks volumes. In this case, the exec cites the return of enterprise customers as hyperscalers exit, and how non-AI loads continue grow nicely. Both of these, while not wrong, are also at the margin. ² A provider of colocation services, which generally involve housing servers and other hardware in a data center environment. Colocation allows organizations to place their physical servers and networking equipment in a facility that provides power, cooling, security, and bandwidth. Think of it as like a very weird daycare. ### Energy 1. **Renewables out-produced coal globally for the first time** As Vaclav Smil has [written](https://amzn.to/42CIJTQ) many times, energy transitions often aren't so much about transitions as about additions. We add new sources of power, but we don't lose the old ones, even if they draw less focus. As energy demand spikes worldwide due to unprecedented loads from hyperscaler data centers (and EVs, to be fair), it is noteworthy that renewables are [only now](https://ember-energy.org/app/uploads/2025/10/Global-Electricity-Mid-Year-Insights-2025-PDF.pdf) a larger source than coal. Will that transition continue as loads spike further, and as the US makes renewables less financially attractive? It will be a very close thing, with myriad consequences, not least for, you know, the earth. ![](https://paulkedrosky.com/content/images/2025/10/2025-10-10_08-10-50.png) ### AI 1. **The AI Overton Window Shifts** **Performance vs. Principle** A new [survey](https://papers.ssrn.com/sol3/papers.cfm?abstract%5Fid=5560401) suggests Americans aren’t exactly clinging to deep moral objections about AI displacing human work. Resistance isn’t principled: people hold back until the machines are clearly better and cheaper, then they say ... Sure, okay fine. _This post is for paying subscribers only._ ### Why AI Capex isn't a Bubble: A Perez-ian Perspective URL: https://paulkedrosky.com/why-ai-capex-isnt-a-bubble-a-perez-ian-perspective/ Last updated: 2025-10-09T15:15:14.000Z > "Et par conséquent il s’ensuit que Dieu a choisi le meilleur des mondes possibles". (“And consequently it follows that God has chosen the best of all possible worlds.”) > — Gottfried Leibniz, Théodicée (1720) > “You met me at a very strange time in my life.” > ― The Narrator, *Fight Club (1999)* > "It costs a lot to look this cheap." > — Dolly Parton The **glacial lake outburst of capital** pouring into artificial intelligence infrastructure—tens of billions into GPUs, data centers, power grids, and software platforms—has people chanting “bubble.” There is, as someone remarked to me this week, a bubble in people saying "bubble". Commentators—myself [included](https://paulkedrosky.com/honey-ai-capex-keeps-eating-everything/)—warn of froth, pointing to soaring valuations, unprecedented hardware orders, and the **risk of stranded assets** if demand falters, GPU rental margins collapse, Taiwan gets invaded, or maybe investors simply become **pissy** and there is a valuation [gestalt flip](https://paulkedrosky.com/the-godfather-and-oracles-crap-gpu-rental-returns/). It is a **very strange time** in economic life, but let's be less reactionary for a moment. **Technology bubbles aren't necessarily bad**, as **economist Carlota Perez** argued in her classic 2002 [book](https://amzn.to/476ZlVg). Viewed through her framework of **technological revolutions**, this moment can look less like a bubble in the narrow financial sense and more like a late-stage **installation surge**. That, in her framing, is the period when financial capital **overshoots**, infrastructure is bought and installed at great speed and well ahead of actual demand, and most of it **goes bust**. The **wreckage** left behind is real. But so is the **scaffolding** for later productivity. ![](https://paulkedrosky.com/content/images/2025/10/2025-10-08_07-07-12.png) --- ## Losses as the Logic of Progress This is the unsettling rhythm of **modern economic growth**, one cited by economists from **Smith to Schumpeter, to Perez**. Every major economic revolution has reached a point where **capital destruction** became the mechanism, even if accidental, for building the future. It takes a lot of **defunct assets** to build a new market. Some examples: - In the 1840s, post-boom **railway companie**s bankrupted investors across Britain. By the 1850s, the same bankrupt networks underpinned national integration, industrial transport, and military logistics. - In the 1990s, the **dot-com boom** went parabolic, only to collapse spectacularly in 2000\. Yet the networks, fiber, and server farms left behind formed the backbone of the 2000s online economy, enabling everyone from **Amazon, to Netflix, to training AI models** based on easy access to online data. The pattern is brutal but familiar: **capital overbuilds**, much of it is destroyed, but **society keeps the infrastructure.** ![a train track running through a field of wildflowers](https://images.unsplash.com/photo-1692838065778-42dd17f9cb75?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3wxMTc3M3wwfDF8c2VhcmNofDEwfHxhYmFuZG9uZWQlMjByYWlsfGVufDB8fHx8MTc2MDAxOTM0NXww&ixlib=rb-4.1.0&q=80&w=2000) Photo by [Joss Broward](https://unsplash.com/@jossbroward) / [Unsplash](https://unsplash.com/?utm%5Fsource=ghost&utm%5Fmedium=referral&utm%5Fcampaign=api-credit) --- ## The Secular Problem of Technology So, if **markets are efficient**, why bubbles? If innovation is progress, why ruin? Because, Perez and others rightly argue, the **stranded and abandoned assets from a capex mania** can become the building blocks for what comes next, from unlit fiber, to currently unkitted AI data centers temporarily being used for **laser tag**. This is a compelling argument. **We never know if we've built enough until we've built too much**. A technological revolution can only be messy, built on **a speculative fervor** ahead of cash flows. That will lead to waste and vast losses—and **huge future gains** from the defunct assets. And however it disrupts our lives, that is [okay](https://papers.ssrn.com/sol3/papers.cfm?abstract%5Fid=5560401). There are at least two problems with this view, however. _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): Healthcare AI and MedTech Innovations URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-healthcare-ai-and-medtech-innovations/ Last updated: 2025-10-08T11:47:50.000Z _This post is for paying subscribers only._ ### The Godfather and Oracle's Crap GPU Rental Returns URL: https://paulkedrosky.com/the-godfather-and-oracles-crap-gpu-rental-returns/ Last updated: 2025-10-07T19:19:49.000Z > She threw it all away just to make me look ridiculous. And a man in my position can't afford to be made to look ridiculous. Now you get the hell out of here! > —Jack Woltz, The Godfather (1972) A new piece on Oracle's **weak GPU rental returns** from [The Information](https://www.theinformation.com/articles/internal-oracle-data-show-financial-challenge-renting-nvidia-chips?rc=puodhd) (based on internal Oracle documents) has people buzzing, even if it shouldn't. But first, some of the key points: - Oracle's getting **14% gross margin** from renting out Nvidia's Blackwell. - In the three months that ended in August, Oracle generated around $900 million from rentals of servers powered by Nvidia chips and recorded a gross profit of $125 million. - That is equal to **14 cents for every $1** of sales. - That’s worse than the **gross margins** of many nontech retail businesses. There is no denying these are **weak numbers**, but they should not be a surprise. We already know that cap rates from hyperscaler data centers—with their long-term contracts, strong renter credit quality, etc.—are not [much above](https://www.credaily.com/briefs/data-centers-lead-reit-investment-surge-with-low-cap-rates/?utm%5Fsource=chatgpt.com)¹ 30-year treasuries. This is a highly capital-intensive business with unimpressive margins. ¹ Data center REITs are commanding the second-highest enterprise values across asset classes, trailing only tower REITs, while ****boasting the lowest implied cap rates** at 4.4%. (Source: [CRE Daily](https://www.credaily.com/briefs/data-centers-lead-reit-investment-surge-with-low-cap-rates/?utm%5Fsource=chatgpt.com)) ### The Gestalt Flip Ahead _This post is for paying subscribers only._ ### Four Ideas (AI edition): Unexpected Impacts of AI Advancements URL: https://paulkedrosky.com/four-ideas-ai-edition-unexpected-impacts-of-ai-advancements/ Last updated: 2025-10-07T11:48:15.000Z _This post is for paying subscribers only._ ### Quick Hits: OpenAI's Mammoth AMD Deal URL: https://paulkedrosky.com/quick-hits-openais-mammoth-amd-deal/ Last updated: 2025-10-06T16:16:35.000Z OpenAI has signed a multibillion-dollar [deal](https://www.wsj.com/tech/ai/openai-amd-deal-ai-chips-ed92cc42?gaa%5Fat=eafs&gaa%5Fn=ASWzDAjXDM67USeiTMXzzC-t8pc0ArMsvmf0deYWsAS-nUWWuKXJ3kViDXxh22n-FhM%3D&gaa%5Fts=68e3eb1f&gaa%5Fsig=cj4OFY5G2M3e0uLyqrQA9dIOE%5FeITBLc-ptyUNvy5sCYIw1GsMM4eDR%5FPfksnyqI8nOTJHwGW6CFR8Y00ukUTA%3D%3D) with AMD to purchase up to 6 gigawatts¹ of chips, including the MI450, to power AI inference workloads starting in 2026\. The agreement includes warrants for OpenAI to acquire up to 10% of AMD at a nominal cost, contingent on performance milestones. This is AMD’s most direct challenge to Nvidia’s dominance and signals a shift in AI chip procurement strategy. ¹ Grounding it in the [known](https://x.com/SemiAnalysis%5F/status/1923142994963669245) MI450 IF128 specs, that’s a **continent-scale build* — multiples larger than today’s largest single-tenant AI deployments. For comparison: all hyperscale data center capacity added worldwide in 2024 was roughly on the order of \~25–30 GW (all workloads). AMD/OpenAI are claiming \~20% of an entire year’s global data center power footprint, just for one partnership, and very quickly. If it happened, which seems a stretch, it would require massive changes in power, cooling, real estate, networking, etc. Here are four quick points: 1. **Vendor Financing in Equity Form** AMD is effectively subsidizing demand by issuing OpenAI warrants for up to 160M shares. That makes OpenAI a 10% shareholder: part customer, part financier — a risk transfer from cash to stock, as well as making OpenAI the largest and thus controlling AMD shareholder. 1. **CUDA Lock-in Challenge** Embedding OpenAI as a large AMD partner forces ROCm (AMD's stack) adoption at scale. This is less about FLOPS and more about dragging AI software development out of CUDA’s orbit, with large implications as workloads shift to inference from training. 1. **Data Center Overbuild Risk Grows** _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Interest Rate Shifts, Geoeconomics, AI Adoption, Travel Experiences URL: https://paulkedrosky.com/four-ideas-macro-edition-interest-rate-shifts-geoeconomics-ai-adoption-travel-experiences/ Last updated: 2025-10-06T11:47:55.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-48/ Last updated: 2025-10-05T11:47:38.000Z **OCTOBER 05, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** Prabowo tells army to guard raw resources from foreign groups ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-10-05/prabowo-tells-army-to-guard-raw-resources-from-foreign-groups)) … We finally have free anti-robocall tools that work ([**NY Times**](https://www.nytimes.com/2025/10/02/technology/personaltech/iphone-robocalls-screener-android.html)) … Her stocks were quietly stolen from her I.R.A. ([**NY Times**](https://www.nytimes.com/2025/10/03/your-money/ira-vanguard-merrill-acats-fraud.html)) … Labour markets stuck in a ‘low-hire, low-fire’ cycle ([**FT**](https://www.ft.com/content/2984eec8-f912-4e11-b0f9-ba30ee3adf8b)) … Indonesia revokes TikTok license suspension after data submitted ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-10-05/indonesia-revokes-tiktok-license-suspension-after-data-submitted)) … How polarised is Britain? ([**FT**](https://www.ft.com/content/c7e4e399-428d-4323-884b-32723c9e694f)) … Indian IPO market set for record month with $5 billion in deals ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-10-05/indian-ipo-market-set-for-record-month-with-5-billion-in-deals)) … He Drops Trump Jr.’s Name in Pursuit of Billion-Dollar Deals. It’s Getting Awkward. ([**WSJ**](https://www.wsj.com/politics/policy/donald-trump-jr-friend-gentry-beach-03824825?gaa%5Fat=eafs&gaa%5Fn=ASWzDAhvh0%5FV6bLUL0iCv2bJbJFBiB92gzojFA7P65UQNjjzxCk0zu-Y35Nu&gaa%5Fts=68e25e5c&gaa%5Fsig=TNhYP1COH00TBPaF-X71QvHhIxt9PNPSEdE7NU%5Fcg59TWMlnnjeWY6VvuyiAo4gJTcohB50n8DFta92eyOlIxQ%3D%3D)) … Discount airlines changed flying. Now they’re in trouble. ([**NY Times**](https://www.nytimes.com/2025/10/04/business/spirit-frontier-low-cost-airlines.html)) … How a Missile Strike Unlocked a Deal for Peace in Gaza ([**WSJ**](https://www.wsj.com/world/middle-east/how-a-missile-strike-unlocked-a-deal-for-peace-in-gaza-0aa62c4f?gaa%5Fat=eafs&gaa%5Fn=ASWzDAhS8Gc7i9fvgS3-JtB5NOyuTmNeXUUVmykwK7E2HqNjA3tHSAWjGlF%5F&gaa%5Fts=68e25e5c&gaa%5Fsig=GABwbF6OnRQicjfd81pTA0Zdh-yOcIUYN4B3LrQJpBB8nQ7-nKduat6b33nJf-uaOI9wR0odpID8ohZMj0DSrg%3D%3D)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/10/image-14-1.png) Layoffs in the U.S. remain below pre-pandemic lows since mid-2020, stabilizing around 1%. This suggests sustained employment resilience post-COVID, contrasting the pandemic spike. Such low discharge rates indicate a strong labor market, crucial for economic recovery and workforce stability. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/10/image-15-1.png) The hire rate dropped from over 4.0% in 2020 to around 3.2% in 2025, highlighting a significant decline. This matters for economic stability and workforce dynamics, indicating a potential slowdown in labor market recovery. --- **THE IDEAS:** **Mentions:** NVDA, META, UHAL, HLIT, TEVA ### 01\. Geoeconomics reshaping global trade patterns The increasing role of geopolitics in shaping economic policies is causing a realignment of global trade practices. Geopolitical partners are trading more with one another and less with rivals, impacting logistics, supply chains, and possibly the global economy within the next 24 months. **Related reading:** **FreightWaves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) **FT**: [Welcome to the new age of geoeconomics](https://www.ft.com/content/daf5ce11-c95b-4964-8df1-7463f0f9a81b) **FT**: [Strategic interdependence is rewiring the global economy](https://www.ft.com/content/25ab5e71-6ae2-495b-bf06-6fd2f1cbb18a) ### 02\. Industrial sectors see renewed momentum Following a period of stagnation, industrial growth is resurging, driven by recovery in global markets and technological advancements. This trend suggests a positive outlook for companies in the industrial sector over the next year. **Related reading:** **Oxford Economics**: [Industry Key Themes 2025: Industrial landscape at a critical juncture](https://www.oxfordeconomics.com/resource/industry-key-themes-2025-industrial-landscape-at-a-critical-juncture/) **CNBC**: [This industrial giant could be about to break out, charts show](https://www.cnbc.com/2025/06/05/this-industrial-giant-could-be-about-to-break-out-charts-show.html) ### 03\. AI adoption transforming consumer and business landscapes Generative AI is rapidly being integrated across various industries, significantly altering consumer behavior and business operations. The ongoing adoption of AI presents significant opportunities and challenges, shaping the competitive landscape over the next two years. **Related reading:** **Twitter**: [As generative #AI sweeps across industries and the #consumer landscape](https://twitter.com/SPGlobal/status/1860035529909080480) **Oxford Economics**: [New Tools, New Skills, No End in Sight: The Challenge of AI Adoption](https://www.oxfordeconomics.com/resource/new-tools-new-skills-no-end-in-sight-the-challenge-of-ai-adoption/) ### 04\. Hospitality sector innovation driven by tech and consumer trends The hospitality industry is rapidly evolving due to technological advances and shifting consumer preferences, setting the stage for new business models and enhanced customer experiences. This trend is expected to continue shaping the industry's growth over the next two years. **Related reading:** **HospitalityNet**: [Top 10 Hospitality Trends: What Shaped 2024 and What To Expect in 2025](https://www.hospitalitynet.org/opinion/4125874.html) **HospitalityNet**: [Where Brands Shape the Future of Living](https://www.hospitalitynet.org/news/4128995.html) ### LLMs, Shifting Baselines, and .400 Hitters URL: https://paulkedrosky.com/llms-shifting-baselines-and-400-hitters/ Last updated: 2025-10-04T00:33:58.000Z 💡 On Fridays I generally try to send something around less tied to recent developments, and more to provoke "weekend thinking". This short essay is ... that. Giants once roamed the earth. There was a time when baseball produced **.400 hitters**—Ted Williams in 1941 being the best-known example, although others dotted the decades before. Since then? Nothing. No one has come very close. Batting averages hover, despite everyone training better, scouts scouting better, players being fitter and eating better. And yet, the **.400 hitter is extinct**. ![](https://paulkedrosky.com/content/images/2025/10/2025-10-03_07-42-19-1.png) Evolutionary biologist and baseball buff **Stephen Jay Gould** wrote about this in his book [*Full House*](https://amzn.to/4330yvf). He argued that the disappearance of .400 hitters was not because players got worse but **because they got *better***. The performance distribution had a higher mean, and the variance had shrunk. As median player skill rose across all aspects of play—hitting, fielding, and pitching—the **right tail of hitting performance** became less populated. Baseball lost the illusion of extraordinary players because variance collapsed against a higher baseline. Outliers were fewer and distributed from a lower average, as **better pitching and fielding triumphed over better batting**. You can see the multi-decadal decline in **batting variance** in the following figure. Over the last hundred years, variance has collapsed. ![](https://paulkedrosky.com/content/images/2025/10/2025-10-03_07-29-31.png) Higher mean performance and collapsing variance isn’t just a baseball story, even if baseball is instructive. Whether we're talking about sports, evolution, or society itself, it is a model for **how societies evolve** under conditions of rapid baseline improvement. The arrival of large language models (LLMs) is just such a moment, transforming how we think about **talent, creativity, and progress**. --- ## The Vanishing Outlier Gould’s insight was deceptively simple: when **overall competence rises**, exceptional-looking performances can vanish. The ceiling does not go up as much as much as the floor does, so the distribution tightens. In baseball: - **1900s–1940s**: Wide variance in player preparation, diet, and training meant outliers—people who hit far better than peers—were visible. - **Post-1960s**: Weight rooms, analytics, nutrition, and scouting systems all converged and became best practices. Everyone became better trained, so **batting averages compressed** into a narrower band. This is why Williams’s .406 season in 1941 looks unrepeatable: not because modern players are worse, but because they are better—collectively. ![](https://paulkedrosky.com/content/images/2025/10/2025-10-03_07-50-44.png) The moral: **extraordinary outcomes** can (seemingly) disappear when the baseline rises because everyone is better. --- ## Tadej Pogačar and the Illusion of Dominance But there are subtleties. Cycling provides an example of how **compression reduces outliers**, but they don't go away. A century ago, Tour de France stages routinely broke riders; “bonking” (running out of energy) was common. The winner was sometimes an hour ahead of everyone else. Nutrition was poor, recovery nonexistent, training haphazard. **Variance was wide**. Now, enter multiple-time world champion and serial Tour de France winner **Tadej Pogačar**, the Slovenian superstar. He dominates pro cycling, despite a modern peloton baseline that is **extraordinarily high**, the product of science, training, altitude camps, nutrition, power meters, and so on. The entire field has compressed upward. You can see that **rising baseline** in the following graph of average speed¹ over the last hundred years. _This post is for subscribers on the Plus tier only._ ### Quick Hits: 85% Margins for Data Centers? Yes, Please URL: https://paulkedrosky.com/quick-hits-85-margins-for-data-centers-yes-please/ Last updated: 2025-10-02T20:05:28.000Z I really and 100% truly wasn't planning anything today—and I do still have something coming tomorrow—but I couldn't let this Bloomberg [piece](https://www.bloomberg.com/opinion/articles/2025-10-02/ai-data-centers-give-private-credit-its-mojo-back) pass without comment. It gives me a chance to untangle the many confused claims made about margins on data centers, while reminding people that increasingly opaque financing now accounts for over half the market. Let's start with this graph, which shows that private credit now accounts for over half of the funding for data centers and growing. It has among its "advantages", as I've [written](https://paulkedrosky.com/weekend-reading-plus-spvs-meta-and-fiber-buildout-2-0/), that public companies can build and control data centers without it affecting their credit rating, usually via partnerships (sic) with private credit. ![](https://paulkedrosky.com/content/images/2025/10/2025-10-02_12-41-23.png) This remains something that should be watched closely, given its role in making untangling total spending more difficult, as well as making it more difficult to understand the structural risks companies are taking on. Now, let's turn to a specific paragraph in the piece: ![](https://paulkedrosky.com/content/images/2025/10/2025-10-02_12-45-23.png) This is problematic in so many ways, some of which are tedious and accounting, but others are deeper. Here is a short list: ### 1\. **Mixing Capex Payback With Operating Margins** - The argument jumps from *capex break-even in 5 years* to *85% operating margins* without showing how the two relate. They may be related, but ... they may not. - A high operating margin on a powered shell lease doesn’t automatically imply rapid payback or strong return on assets (ROA), because margins are on revenue, not on invested capital. The capital intensity, usage risk, and cost of financing matter more. ### 2\. **Mischaracterizing “Powered Shells”** - Powered shells are not like crypto mines: they’re more like laser tag facilities with more impressive power and cooling hookups. Sure, you could put a data center in there, but not cheaply. _This post is for subscribers on the Plus tier only._ ### Four Ideas (Commodities edition): Geoeconomic Shifts, Cocoa Rally, Commodity Traders, Gold Outlook URL: https://paulkedrosky.com/four-ideas-commodities-edition-geoeconomic-shifts-cocoa-rally-commodity-traders-gold-outlook/ Last updated: 2025-10-02T11:48:20.000Z _This post is for paying subscribers only._ ### Quick Hit: Blackstone on Data Center Bubbles URL: https://paulkedrosky.com/quick-hit-blackstone-on-data-center-bubbles/ Last updated: 2025-10-01T16:06:37.000Z I have a longer piece for everyone coming out tomorrow, but I was so struck by this comment on data centers at a Bloomberg [event today](https://www.bloomberg.com/news/live-blog/2025-10-01/women-money-power) from Blackstone's global co-head of real estate: ![](https://paulkedrosky.com/content/images/2025/10/2025-10-01_08-55-05.png) She may be right, of course, but it is important to untangle the claims. Ms. McCarthy-Baldwin is saying *because we only build when we have a signed lease, there can’t be a bubble.* That ignores at least five important points: ### 1\. Counterparty Risk A “long-term lease with a high credit tenant” is not the same as risk-free. Hyperscalers and AI firms can cancel, restructure, or sublease when demand softens. Many of these contracts have outs, step-downs, or renewal risks. Telecoms in the 1990s also had creditworthy counterparties ... until they didn’t. ### 2\. Overcapacity Dynamics _This post is for subscribers on the Plus tier only._ ### Four Ideas (Life sciences edition): FDA, AI, Drone Delivery, Health Systems URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-fda-ai-drone-delivery-health-systems/ Last updated: 2025-10-01T11:48:11.000Z _This post is for paying subscribers only._ ### Rougher Notes, Sep 30, 2025 URL: https://paulkedrosky.com/rougher-notes-sep-30-2025/ Last updated: 2025-09-30T21:54:37.000Z I sometimes send my rough notes *before* I write the piece. Here are some sources I'm working from for an upcoming letter: - **Video games** industry facing uneconomic glut from too many releases, in part because of new AI tooling [Bloomberg](https://www.bloomberg.com/news/newsletters/2025-09-26/the-video-game-industry-has-a-problem-there-are-too-many-games) - **AI data centers** driving power bills sharply higher [Bloomberg](https://www.bloomberg.com/graphics/2025-ai-data-centers-electricity-prices/?terminal=true) - **Global funds** rapidly rebalancing toward retail and **data centers** [CRE Daily](https://www.credaily.com/briefs/global-funds-rebalance-toward-retail-and-data-centers) - **Public companies** could shrink 20% as AI concentrates scale [Bloomberg](https://www.bloomberg.com/news/articles/2025-09-24/ai-could-shrink-public-companies-by-20-an-allianz-gi-cio-warns) - **Power prices** surging in Texas and Virginia as AI load builds [Bloomberg](https://www.bloomberg.com/news/newsletters/2025-09-30/eye-popping-power-prices-show-ai-s-cost-to-consumers) - **AI agents** moving from demos to real workplace tasks [One Useful Thing](https://www.oneusefulthing.org/p/real-ai-agents-and-real-work) - **Doctors’ skills** decline as diagnostic AI reduces cognitive demand [Medscape](https://www.medscape.com/viewarticle/ai-eroding-cognitive-skills-doctors-how-bad-it-2025a1000q2k?src=rss) - **Private credit** growth raises systemic shock concerns [Fitch](https://www.fitchratings.com/research/insurance/systemic-shock-could-expose-private-credits-broad-reach-growing-risk-29-09-2025) - **Cognitive disability** rising with technology to blame, as 11% of US adults now affected [Neurology](https://www.neurology.org/doi/10.1212/WNL.0000000000214226) - **Accenture layoffs** target employees deemed to lack AI potential [Computerworld](https://www.computerworld.com/article/4064955/accenture-wants-to-lay-off-employees-without-ai-potential.html) - **AI hiring** pushes new office demand in New York and San Francisco [CRE Daily](https://www.credaily.com/briefs/ai-hiring-boosts-office-demand-in-new-york-and-san-francisco) - **Stratification engines** frame AI as ideology of replacement [Springer](https://link.springer.com/chapter/10.1007/978-3-032-02665-1%5F4) The full piece is coming Friday, on AI, the disappearance of .400 hitters, and Electronic Arts. ### Four Ideas (AI edition): AI-Driven Investment Shifts URL: https://paulkedrosky.com/four-ideas-ai-edition-ai-driven-investment-shifts/ Last updated: 2025-09-30T11:48:01.000Z _This post is for paying subscribers only._ ### Data Centers: More Money, Fewer Buyers URL: https://paulkedrosky.com/data-centers-more-money-fewer-buyers/ Last updated: 2025-09-29T21:28:35.000Z The following is a premium mailing, with an opening excerpt for free subscribers. AI is driving a data center building boom. I have written about it extensively (see [1](https://paulkedrosky.com/data-centers-make-the-beige-book-plus-power-problems/), [2](https://paulkedrosky.com/the-kanye-data-center-crossover/), [3](https://paulkedrosky.com/honey-ai-capex-ate-the-economy/), [∞](https://paulkedrosky.com/honey-ai-capex-keeps-eating-everything/)), done [podcasts](https://podcasts.apple.com/us/podcast/this-is-how-the-ai-bubble-could-burst/id1594471023?i=1000728026459), and more. It is easy, however, to get wrapped up in chatbots, image models, massive capital expenditures, soaring power demand, ASI/AGI, and the looming singularity, and not think about the financial drivers of what is happening. But that is an error. Think of what is happening through the lens of project financing. Developers can lock in 12–15% IRRs¹ on new data centers, while stabilized hyperscale campuses² trade like long bonds³ at 4–5% cap⁴ rates. That spread is irresistible, so money floods into development. ¹ Internal rate of return: the rate required, in percentage terms, to make the net present value (NPV) of all cash flows (both incoming and outgoing) from an investment equal to zero. ² Data centers full of GPUs leased out to large companies like Microsoft, Google, and OpenAI. ³ Bonds that don't mature for decades. These are generally highly sensitive to interest rates. ⁴ The rate of return on the leased data center, based on the difference between operating costs, interest expense, and lease income. But the imbalance is obvious. Once the centers are built and leased, relatively few buyers want them. The tenant credit is pristine (Amazon, Microsoft, Google), but that’s precisely the problem: the leases are too safe, too long, and too flat. Stabilized data center assets behave like annuities, not growth real estate. Only mega-pensions and sovereigns can absorb them, and even then only sparingly. Developers and private credit increasingly don’t care. They’ve already learned how to sidestep the exit problem. The answer is a familiar word: securitization. --- ## The Liquidity Machine _This post is for subscribers on the Plus tier only._ ### Four Ideas (Macro edition): Geoeconomic Strategies and Industrial Policy Shifts URL: https://paulkedrosky.com/four-ideas-macro-edition-geoeconomic-strategies-and-industrial-policy-shifts/ Last updated: 2025-09-29T11:47:56.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-47/ Last updated: 2025-09-28T11:47:37.000Z **SEPTEMBER 28, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** The C.E.O. Who Spends All Day Thinking About Sleep ([**NY Times**](https://www.nytimes.com/2025/09/27/business/oura-ring-tom-hale.html)) … Russia oil output ([**WSJ**](https://www.nytimes.com/2022/06/15/business/russia-oil-sanctions.html)) … Unexpected Critics of Trump’s Attacks on Wind Energy: Oil Executives ([**NY Times**](https://www.nytimes.com/2025/09/27/business/energy-environment/oil-gas-trump-offshore-wind.html)) … The rise of America’s young socialists—From the 2008 financial crisis to Mamdani ([**WSJ**](https://www.wsj.com/us-news/socialists-zohran-mamdani-2008-financial-crisis-f98e54fa?gaa%5Fat=eafs&gaa%5Fn=ASWzDAh-3fNk5tTMQGF6N7jvzCgkDnN995uX%5FQWurGAK8HxAht0tC%5FvDA4cV&gaa%5Fts=68d923dd&gaa%5Fsig=xfy2Y5yrWUCcmoIjRBi1VQGnFfkd8gPdi8oaveryuNp2nJXRFQmQ8K1KEhmXOm8Dsz4XWwVOM3-PHvQGJOK4Vg%3D%3D)) … EU Unveils $638 Million Package to Scale Up Renewables in Africa ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-09-28/eu-unveils-638-million-package-to-scale-up-renewables-in-africa)) … The Weekend Essay. How tech lords and populists changed the rules of power ([**FT**](https://www.ft.com/content/85ee3be0-c9a6-4a1d-baf9-8b2ca9e46a85)) … Jaguar Land Rover Ltd ([**FT**](https://www.bloomberg.com/profile/company/1105106Z:LN)) … Jaguar Land Rover Gets UK Guarantee for £1.5 Billion Loan ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-09-27/jaguar-land-rover-gets-uk-guarantee-for-1-5-billion-loan)) … US soyabean farmers squeezed as China blocks imports and stockpiles rise ([**FT**](https://www.ft.com/content/b920c4c4-dcd7-4d5b-9a56-fc5cc1c994c4)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/09/Image-1.png) In 2020, only 10% of U.S. teenagers read daily in their leisure time, while 45% hardly ever read. This marks a significant shift from 1985 when these figures were nearly reversed. This change highlights a substantial decline in daily reading habits among teenagers over the past 35 years. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/09/1_C7CF3J6C3OOVXkag2-TmrA.png) The H100 Index shows a significant decrease from approximately 2.27 to 2.13 between early August and mid-September 2025\. This decline highlights potential market instability or external economic challenges, indicating a 6% drop impacting investor confidence and financial strategies. --- **THE IDEAS:** **Mentions:** FT, CNBC, Bloomberg, Macrobond, Oxford Economics ### 01\. Geoeconomics influences global trade partnerships Strategic interdependence and geopolitical partnerships are redefining global trade, with countries reshaping alliances for economic advantage. This trend impacts trade flows and market dynamics, as evidenced by McKinsey and Mohamed El-Erian's analysis of geopolitical trading patterns. **Related reading:** **FT**: [Strategic interdependence is rewiring the global economy](https://www.ft.com/content/25ab5e71-6ae2-495b-bf06-6fd2f1cbb18a) **FreightWaves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) ### 02\. EM currency positions reflect economic convergence Investors' move towards EM currencies, with a focus on valuation convergence and growth potential, is shifting global investment patterns. This trend is highlighted by Bloomberg's analysis of increased EM currency buying, influencing equities and broader economic activities. **Related reading:** **X**: [U.S. dollar shorts concentrated against EM currencies](https://x.com/elerianm/status/1967593226421883304) ### 03\. Industrial reclamation amid global economic shifts Industries are capitalizing on structural changes to drive growth, as noted by Oxford Economics. The evolving industrial landscape offers opportunities to adapt and thrive in shifting economic conditions, emphasizing strategic adaptation and growth. **Related reading:** **Oxford Economics**: [Industry Key Themes 2025: Industrial landscape at a critical juncture](https://www.oxfordeconomics.com/resource/industry-key-themes-2025-industrial-landscape-at-a-critical-juncture/) ### 04\. AI adoption strengthening economic recovery strategies AI-led initiatives, particularly in China, are providing a confidence boost amid economic sluggishness and geopolitical tensions. AI is aiding economic resilience and opening opportunities, as seen with DeepSeek's growth initiatives. **Related reading:** **CNBC**: [DeepSeek-led AI adoption offers China an opportunity to boost its sputtering growth](https://www.cnbc.com/2025/02/21/deepseek-led-ai-adoption-offers-china-opportunity-to-boost-growth.html) ### Quick Update: Appearances, Financial Flywheels, etc. URL: https://paulkedrosky.com/quick-update-appearances-financial-flywheels-etc/ Last updated: 2025-09-28T22:37:40.000Z Just a quick update between some recent travel: I was on Derek Thompson's [**Plain English podcast**](https://podcasts.apple.com/us/podcast/this-is-how-the-ai-bubble-could-burst/id1594471023?i=1000728026459) this week. I expanded on various data center ideas, while also discussing the role of "check size" in creating a **financing flywheel**¹, a concept that many observers overlook. ¹ The more money an asset manager has under management, the more motivated they are, all else even somewhat equal, to find projects for which they can write large checks. Billion-dollar funds do not want to write one thousand checks for $1-million; $100--billion funds don't want to write a thousand checks for $100-million. This causes large funds to gravitate to projects that need a lot of money, which can have malign consequences. You can have a listen [here](https://podcasts.apple.com/us/podcast/this-is-how-the-ai-bubble-could-burst/id1594471023?i=1000728026459). Lots more in the queue, with two things coming next week. One piece is on Monday for premium subscribers, and then someone else for everyone later in the week. I also have a few other interviews coming out, and I'll let folks know as they do. ### Four Ideas (AI edition): AI-Driven Market and Policy Shifts URL: https://paulkedrosky.com/four-ideas-ai-edition-ai-driven-market-and-policy-shifts/ Last updated: 2025-09-23T11:47:51.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Interest Rate Divergence, EM Currencies, EU Productivity, Real Estate Dynamics URL: https://paulkedrosky.com/four-ideas-macro-edition-interest-rate-divergence-em-currencies-eu-productivity-real-estate-dynamics/ Last updated: 2025-09-22T11:48:05.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-46/ Last updated: 2025-09-21T11:47:42.000Z **SEPTEMBER 21, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** Trump's $100,000 H-1B visa change sparks chaos at tech companies ([**Bloomberg**](https://www.bloomberg.com/)) … Seeing Through the Reality of Meta’s Smart Glasses ([**NY Times**](https://www.nytimes.com/2025/09/20/technology/personaltech/meta-smart-glasses-mark-zuckerberg.html)) … The Weekend Essay. The truth about immigration ([**FT**](https://www.ft.com/content/43dad7d6-3042-4697-b266-4252eba78ce3)) … Deutsche Bahn to appoint Evelyn Palla as CEO, officials say ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-09-20/deutsche-bahn-to-appoint-evelyn-palla-as-ceo-officials-say)) … Wall Street Bets Rates Will Drop Much More Than Fed’s Forecasts ([**WSJ**](https://www.wsj.com/finance/investing/interest-rates-bets-investors-f47c12be?gaa%5Fat=eafs&gaa%5Fn=ASWzDAgqEioxJeeFRjP6TLeHUPo6IXs9nGjCXYl614vDo6L400InAZp-q8j-&gaa%5Fts=68cfe961&gaa%5Fsig=bqn0bsnLF80Q2aZMWORbDNfomjnnK-jLpOZ%5F5LZ0FWsJ4T2rtfwwg5SHGU0L8BAmXMNvthnQULOwc2mLDTTGIw%3D%3D)) … What to Know About Jimmy Kimmel’s Show Being Suspended ([**NY Times**](https://www.nytimes.com/2025/09/20/business/jimmy-kimmel-show-abc-kirk-fcc.html)) … Investors turn to derivatives for US corporate bonds as issuers can't keep up ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-09-20/credit-investors-have-so-much-cash-to-put-to-work-companies-can-t-keep-up)) … At the Airport, There’s a Fancy New Lounge Everywhere You Look ([**WSJ**](https://www.wsj.com/business/airlines/airport-lounges-everywhere-fancy-dc0daae9?gaa%5Fat=eafs&gaa%5Fn=ASWzDAj1LdNvHsLzvGlOIzs36YuTilJEWipgQEk8-fG4KSE4kTEfIGzoIfn1&gaa%5Fts=68cfe961&gaa%5Fsig=x5%5FPR0BR2DPebBqJYfKJwn1tmlLkTtBR4yQIiwpR9FXFO5%5FZle3cnHtITt%5FtLU1Hf3Qhl8OJQlG4ePalyTfd9g%3D%3D)) … Trump's new H-1B visa fee prompts emergency guidance from companies ([**FT**](https://www.ft.com/content/9822b5ee-4d3e-4e73-bbf4-be382d2fada7)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/09/2025-09-18_05-58-42-1.png) AI and machine learning VC deals in North America are projected to reach 63% in 2025, significantly surpassing other regions. This illustrates a striking dominance compared to Asia’s 18% and Europe's 15%, underscoring North America's leadership and potential innovation advantages in AI technology investments. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/09/2025-09-16_17-21-36-1.png) Amazon's projected CapEx as a percentage of EBITDA for FY25 is unusually high, exceeding 60%, rivaling AT&T (1998) and Exxon (2014). This signals substantial investment needs, highlighting Amazon's aggressive expansion strategy in contrast to peers, reflecting significant focus on growth and infrastructure development. --- **THE IDEAS:** **Mentions:** SPGSustainable1, x.com/elerianm, cnbc.com, oxfordeconomics.com, ft.com ### 01\. Generative AI reshaping consumer and business landscapes Generative AI tools are increasingly being adopted across industries, transforming consumer interactions and business processes. This trend is driven by high demand for AI-driven solutions and sustained interest from businesses in innovation. This could lead to growth opportunities for tech companies specializing in AI and sectors rapidly integrating these tools. **Related reading:** **Twitter**: [@SPGlobal: As generative #AI sweeps across industries...](https://twitter.com/SPGlobal/status/1860035529909080480) **Carrier Management**: [The Evolving P/C GenAI Journey...](https://www.carriermanagement.com/features/2024/11/06/268339.htm) ### 02\. Geopolitical dynamics influencing trade partnerships Strategic interdependence is reshaping global trade, emphasizing partnerships among geopolitical allies over rivals. This shift affects sectors reliant on international trade and urges companies to reassess supply chain strategies, potentially benefiting firms in countries with strong geopolitical ties. **Related reading:** **Freight Waves**: [McKinsey: Geopolitical partners trading more with one another...](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) **Financial Times**: [Strategic interdependence is rewiring the global economy](https://www.ft.com/content/25ab5e71-6ae2-495b-bf06-6fd2f1cbb18a) ### 03\. Emerging markets gaining traction in global equity flows Investors are increasingly favoring emerging market equities, driven by better valuations and growth potential compared to developed markets. This trend could foster economic growth in emerging markets and attract further investments, benefiting sectors like consumer goods and technology. **Related reading:** **X (formerly Twitter)**: [Bloomberg: U.S. dollar shorts concentrated against EM currencies...](https://x.com/elerianm/status/1967593226421883304) ### 04\. Industrial growth set for resurgence post-weakness After years of stagnation, the industrial sector is experiencing a rebound. This resurgence is seen in key regions and is driven by investments in technology and infrastructure, providing opportunities for firms involved in manufacturing and industrial goods. **Related reading:** **Oxford Economics**: [Industry Key Themes 2025: Industrial landscape at a critical juncture](https://www.oxfordeconomics.com/resource/industry-key-themes-2025-industrial-landscape-at-a-critical-juncture/) ### Four Ideas (Energy edition): Solar Engineers, EV Battery Sustainability, Grid Innovation, Emerging Markets URL: https://paulkedrosky.com/four-ideas-energy-edition-solar-engineers-ev-battery-sustainability-grid-innovation-emerging-markets/ Last updated: 2025-09-19T11:48:27.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Copper Opportunities, Emerging Gold Narrative, Cocoa Demand Surge, Natural Gas Trends URL: https://paulkedrosky.com/four-ideas-commodities-edition-copper-opportunities-emerging-gold-narrative-cocoa-demand-surge-natural-gas-trends/ Last updated: 2025-09-18T11:48:11.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): Generative AI's Impact Across Sectors and Regions URL: https://paulkedrosky.com/four-ideas-ai-edition-generative-ais-impact-across-sectors-and-regions/ Last updated: 2025-09-16T11:47:51.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Shifts in Global Economy: Manufacturing, AI, Tariffs, Hospitality URL: https://paulkedrosky.com/four-ideas-macro-edition-shifts-in-global-economy-manufacturing-ai-tariffs-hospitality/ Last updated: 2025-09-15T11:48:15.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-45/ Last updated: 2025-09-14T11:47:44.000Z **SEPTEMBER 14, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** How the journal analyzed more than one million FAA reports ([**WSJ**](https://www.wsj.com/business/airlines/how-the-journal-analyzed-more-than-one-million-faa-reports-7e7e043a?gaa%5Fat=eafs&gaa%5Fn=ASWzDAgx%5F-ehxJ7uFMPW11wlEkiLIkJ4oJOGEvUJRraIjyRHojLh%5FROAtkLP&gaa%5Fts=68c6aee3&gaa%5Fsig=ZdJcHiPlzKoHRv0O85F6dLV9KRZM%5Fxl85VA3XkWAfl4XszQRHldcMvuwkgFpIH1d1713K80W42W3GzIMrdmXmA%3D%3D)) … Used E.V. Sales Take Off as Prices Plummet ([**NY Times**](https://www.nytimes.com/2025/09/13/business/used-electric-vehciles.html)) … Australia to Invest $8 Billion in Aukus Defense Hub ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-09-14/australia-to-invest-8-billion-in-aukus-defense-hub-near-perth)) … AI-controlled drone swarms set to transform combat on battlefield ([**FT**](https://www.ft.com/content/f6741167-476c-41bc-8b1c-ed65b8b4c054)) … They Had Money Problems. They Turned to ChatGPT for Solutions. ([**NY Times**](https://www.nytimes.com/2025/09/13/business/chatgpt-financial-advice.html)) … Wall Street Rides a $1 Trillion Money Wave as Fed Test Looms ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-09-12/wall-street-fund-juggernauts-near-1-trillion-haul-ahead-of-fed)) … Global 36-Hour Interest-Rate Spree Heralds First US Cut of 2025 ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-09-13/global-36-hour-interest-rate-spree-heralds-first-us-cut-of-2025)) … Toxic fumes are leaking into airplanes, sickening crews and passengers ([**WSJ**](https://www.wsj.com/business/airlines/air-travel-toxic-fumes-64839d6e?gaa%5Fat=eafs&gaa%5Fn=ASWzDAgh4WmYsVZVQDdOATIpdDMsmvw77RjuKT2zA76EaIrPp6IppQeDIdTC&gaa%5Fts=68c6aee3&gaa%5Fsig=dbOUSjOEbKTiCl2cq62mYuxdAq2tQBhISuRWOkf%5Fe5TcEi7qCkBygc9akOLHA%5F9WfJNSFWQBKJlcrZb41dcSbQ%3D%3D)) … French companies’ borrowing costs fall below government’s as debt fears intensify ([**FT**](https://www.ft.com/content/be40c638-77ce-4986-9274-fa157d3a0ffb)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/09/2025-09-12_09-50-13-2.png) Canada's electricity demand is projected to reach 18.4 KWh per capita in 2024, significantly surpassing the second-highest, the U.S. at 13.6 KWh. This difference underscores the impact of energy-intensive industries and cold climate in Canada, influencing global electricity consumption patterns. --- **THE IDEAS:** **Mentions:** SPGI, MLIV, MDLZ, FT, CNBC ### 01\. AI adoption boosts hospitality engagement and revenue. AI enables hoteliers to deepen customer connections and improve revenue streams by personalizing experiences based on AI-driven insights. This trend transforms consumer interactions and has broader implications for customer-centric industries adopting similar technologies. **Related reading:** **HospitalityNet**: [How hoteliers can create connections that drive revenue in an AI-driven world](https://www.hospitalitynet.org/opinion/4128725.html) **SP Global**: [As generative #AI sweeps across industries...](https://twitter.com/SPGlobal/status/1860035529909080480) ### 02\. Geoeconomics reshapes global trade relations. As geoeconomic forces challenge traditional trade alliances, countries are reorienting partnerships. This shift creates opportunities to seize new trade ties while exposing vulnerabilities due to less predictable geopolitical landscapes. **Related reading:** **FreightWaves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) **Financial Times**: [Welcome to the new age of geoeconomics](https://www.ft.com/content/daf5ce11-c95b-4964-8df1-7463f0f9a81b) ### 03\. Hidden climate risks are underestimated in portfolios. With the market's limited understanding of extended climate risks, financial portfolios remain exposed. Proper assessment tools and ratings for climate risk could shift future corporate financial strategies and investor focus. **Related reading:** **Oxford Economics**: [What is the ‘hidden’ climate risk in your financial portfolio?](https://www.oxfordeconomics.com/resource/what-is-the-hidden-climate-risk-in-your-financial-portfolio/) **Phys.org**: [Extreme weather alerts can move markets...](https://phys.org/news/2025-08-extreme-weather-investors.html) ### 04\. Industrial growth gains momentum post-weak period. Following several years of stagnation, industrial sectors show signs of resurgence, creating opportunities for investment and innovation. This momentum could stimulate broader economic activities, enhancing job markets and technological advancement. **Related reading:** **Oxford Economics**: [Industry Key Themes 2025: Industrial landscape at a critical juncture](https://www.oxfordeconomics.com/resource/industry-key-themes-2025-industrial-landscape-at-a-critical-juncture/) **CNBC**: [This industrial giant could be about to break out, charts show](https://www.cnbc.com/2025/06/05/this-industrial-giant-could-be-about-to-break-out-charts-show.html) ### Weekend ... Thinking ! URL: https://paulkedrosky.com/weekend-thinking/ Last updated: 2025-09-28T22:36:16.000Z Five things to think about this weekend _This post is for subscribers on the Paul Kedrosky and Plus tiers only._ ### ChatGPT as the Original AI Error URL: https://paulkedrosky.com/chatgpt-as-the-original-ai-error/ Last updated: 2025-09-28T22:37:10.000Z > What if he's right What . . .if. . .he . . .is . . . right W-h-a-t i-f h-e i-s r-i-g-h-t > —Tom Wolfe, [The new life out there](https://archive.wphna.org/wp-content/uploads/2015/03/1965-NYHT-Tom-Wolfe-on-Marshall-McLuhan.pdf), New York Herald Tribune, 1965 There is a growing sense that chat—as in **ChatGPT—was an error**. For two recent examples, see [here](https://www.bloomberg.com/news/newsletters/2025-09-11/databricks-ceo-says-companies-erred-with-focus-on-chatbots-galore?srnd=homepage-americas) and [here](https://www.semafor.com/article/09/05/2025/palantir-ceo-on-why-silicon-valley-over-hyped-ai). Neither of these are from people from whom I would take advice on anything consequential, so I am, if anything, biased to think they're wrong. But, to quote writer **Tom Wolfe on Marshall McLuhan**, "What if he's right"? Let's start with something that too often goes unsaid, but helped drive ChatGPT's unprecedented adoption (which came as a surprise to OpenAI as much as anyone). And it is this: Humans are hairless, ***gregarious*** apes. Humans, and their fascinations, cannot be understood without our tireless and sometimes **lethal compulsion** to talk. Why lethal? Because of how we evolved to **privilege speech over breathing**. In most mammals, the larynx is high in the throat, keeping the airway and esophagus reasonably separate. In humans, however, the **larynx descended** to enlarge the vocal tract. This made for a more **resonant cavity**, enabling singing and complex speech, but it also **created a shared pipe** via which even small amounts of wrongly routed food could cause us to choke. But the **evolutionary advantage** of advanced communication outweighed the survival risk of occasional choking. ![](https://paulkedrosky.com/content/images/2025/09/2025-09-12_10-59-37.png) Humans must be understood through this lens: they are so obsessed with talking that they are **willing to die** to get a few words out. And, as a result, they judge intelligence based on how human-like things sound. _This post is for subscribers on the Paul Kedrosky and Plus tiers only._ ### Four Ideas (Commodities edition): Gold Surge, Commodity Price Dynamics, AI in Commodities, Tariffs Impact URL: https://paulkedrosky.com/four-ideas-commodities-edition-gold-surge-commodity-price-dynamics-ai-in-commodities-tariffs-impact/ Last updated: 2025-09-11T11:47:57.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): AI in Healthcare: Expansion and Investment Strategies URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-ai-in-healthcare-expansion-and-investment-strategies/ Last updated: 2025-09-10T11:47:59.000Z _This post is for paying subscribers only._ ### Declining AI Adoption, Surveys, and Simpson's Paradox URL: https://paulkedrosky.com/declining-ai-adoption-surveys-and-simpsons-paradox/ Last updated: 2025-09-28T22:35:35.000Z AI adoption at large and small companies, far from slowing, is accelerating _This post is for subscribers on the Paul Kedrosky and Plus tiers only._ ### Rougher Notes—Sep 5, 2025 URL: https://paulkedrosky.com/rougher-notes-sep-5-2025/ Last updated: 2025-09-05T20:30:38.000Z These rougher notes—things I read recently and that got me thinking—usually go out as part of the weekly newsletter, but to empty the queue of things, I'm sending them separately today so we can start fresh next week. Have a good weekend ## Rougher Notes AI Leasing Signals San Francisco [**Office Revival**](https://www.credaily.com/briefs/ai-leasing-signals-san-francisco-office-revival/) (CRE Daily) … Interview: [**Ray Dalio**](https://www.ft.com/content/d130beb3-b567-4323-aa40-a48bc9e92a41) (FT) … What values best distinguish the world's cultures? The machine learning-based cultural values [**inventory**](https://academic.oup.com/pnasnexus/article/4/8/pgaf229/8240670?login=false) (PNAS Nexus) … How secretive hedge fund [**Magnetar went all in on AI**](https://www.ft.com/content/17e67627-cca0-47c1-8c1c-8a2a67a2148e) (FT) … Who Didn’t Do a [**Hamptons Pop-up**](https://www.curbed.com/article/hamptons-pop-up-luxury-retail-montauk-bridgehampton.html) This Summer? (Curbed) … On the Decline of Elite-Educated [**Republicans in Congress**](https://www.cambridge.org/core/journals/perspectives-on-politics/article/on-the-decline-of-eliteeducated-republicans-in-congress/B50B628CA9CCC799A0D86D9496EA3F6F) (Cambridge Core) … Is the decline of [**reading making politics dumber**](https://www.economist.com/culture/2025/09/04/is-the-decline-of-reading-making-politics-dumber) (The Economist) … How full are [**airplanes**](https://fredblog.stlouisfed.org/2025/09/how-full-are-airplanes/) (FRED Blog) … Is AI Contributing to Rising [**Unemployment**](https://www.stlouisfed.org/on-the-economy/2025/aug/is-ai-contributing-unemployment-evidence-occupational-variation) (St. Louis Fed) … Hot mic picks up [**Putin and Xi**](https://www.reuters.com/business/media-telecom/hot-mic-picks-up-putin-xi-discussing-organ-transplants-immortality-2025-09-03/) discussing organ transplants and immortality (Reuters) … Cohort mortality forecasts indicate signs of deceleration in [**life expectancy gains**](https://www.pnas.org/doi/10.1073/pnas.2519179122) (PNAS) … Smartphone use on the [**toilet and the risk of hemorrhoids**](https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0329983) (PLOS One) … Exposing [**AI manipulation in fast fashion**](https://www.digitaldigging.org/p/exposing-ai-manipulation-in-fast) (Digital Digging) … The Pill That Women Are Taking for Everything From [**Speeches to First Dates**](https://www.wsj.com/style/beta-blockers-anxiety-propranolol-e063674b?mod=hp%5Flead%5Fpos9) (WSJ) … ‘You’re a Boomer If You Wear [**Leggings**](https://www.wsj.com/style/fashion/baggy-workout-pants-fp-movement-leggings-53814676?st=3HugAG&reflink=desktopwebshare%5Fpermalink)**’**: The Rise of Big Workout Pants (WSJ) --- Back Sunday with the weekend edition, and next week with another regular note. ### Data Centers Make the Beige Book, Plus Power Problems URL: https://paulkedrosky.com/data-centers-make-the-beige-book-plus-power-problems/ Last updated: 2025-09-28T22:35:46.000Z The latest Beige Book of regional economic activity is a kind of econo-ode to data centers _This post is for subscribers on the Paul Kedrosky and Plus tiers only._ ### Four Ideas (Commodities edition): Commodities Driven by Geopolitical and Economic Transitions URL: https://paulkedrosky.com/four-ideas-commodities-edition-commodities-driven-by-geopolitical-and-economic-transitions/ Last updated: 2025-09-04T11:47:50.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): AI-driven Healthcare Innovation and Tech Push URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-ai-driven-healthcare-innovation-and-tech-push/ Last updated: 2025-09-03T11:47:54.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): Adaptive Regulation, AI in Hospitality, AI Bubble Warnings, Insurer Innovations URL: https://paulkedrosky.com/four-ideas-ai-edition-adaptive-regulation-ai-in-hospitality-ai-bubble-warnings-insurer-innovations/ Last updated: 2025-09-02T11:47:41.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Extreme Weather, Tariffs, Trade Dynamics, Japan Manufacturing URL: https://paulkedrosky.com/four-ideas-macro-edition-extreme-weather-tariffs-trade-dynamics-japan-manufacturing/ Last updated: 2025-09-01T11:47:50.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-44/ Last updated: 2025-08-31T11:47:31.000Z **AUGUST 31, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** What has 4 hooves and more viewers than hockey? Fox Sports' next gamble. ([**NY Times**](https://www.nytimes.com/2025/08/30/business/fox-sports-horse-racing-sports-betting.html)) … Trump takes aim at South Korean chipmakers’ China operations ([**NY Times**](https://www.nytimes.com/2025/08/30/business/trump-samsung-skhynix-chips.html)) … Mortgage Refinancing Starts to Thaw as Rates Trend Down ([**WSJ**](https://www.wsj.com/personal-finance/mortgages/mortgage-refinancing-starts-to-thaw-as-rates-trend-down-25ad4ae7?gaa%5Fat=eafs&gaa%5Fn=ASWzDAgYal2kybEqjmeBSkf1Bxfr8S-iQFgIuT%5F750xr8U2BxE4l%5FyMZdsCA&gaa%5Fts=68b439d6&gaa%5Fsig=P5VlFAIYaE2gLYjoYBb8VhjiMdjkVp5xJ9mKj5fVmhXYaPX%5FvEfq8QgcrUoRI1gp0DeJT6TP0acdJJzMt8Pu2Q%3D%3D)) … Indonesia Finance Minister’s Home Looted as Protests Spread ([**Bloomberg**](https://www.ft.com/content/5264faa0-2ce2-48e7-8fa0-da2c0a286720)) … FT Magazine. The Argentine junta, the crusading lawyer and the imposter ([**FT**](https://www.ft.com/content/2971a0e7-a4b2-4f19-bdc9-732b0778320c)) … Alaska produces a ton of gas. Soon, its biggest city might not have enough. ([**NY Times**](https://www.nytimes.com/2025/08/30/business/energy-environment/alaska-natural-gas-shortage.html)) … Cyberattack on Evertec’s Sinqia Hits HSBC, Others in Brazil ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-08-31/cyberattack-on-evertec-s-sinqia-hits-hsbc-others-in-brazil)) … Fed Upheaval Sows New Uncertainty About Path of Rates ([**WSJ**](https://www.wsj.com/economy/central-banking/fed-interest-rate-cut-lisa-cook-2f3c7e0f?gaa%5Fat=eafs&gaa%5Fn=ASWzDAihBlndK9IMvZmJCre1J0Hvx2S7kfrDCiZBJBGgq3Lgd47K8nASSuNv&gaa%5Fts=68b439d6&gaa%5Fsig=ykf5yqeDHzAqIL%5FwVx7932Gg-wCgaOk2WPnbU%5F3shCZ7Az8ItRlFn5pLhLwaUsZpALhfQ0wMX5YplibMmzOZHA%3D%3D)) … US Trading Partners ‘Dazed and Confused’ After Tariff Court Loss ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-08-30/us-trading-partners-dazed-and-confused-after-tariff-court-loss)) … JPMorgan Chase poaches record number of senior bankers ([**FT**](https://www.ft.com/content/3d613989-5662-461d-a987-49ee02ad6ca5)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/08/2025-08-30_08-23-14.png) Projected hyperscale capital expenditure in 2027 shows AWS leading at $110 billion, while Oracle sees a twelvefold rise from 2021 to $25 billion. Total expenditure more than quadruples from 2020 to $427 billion in 2027\. This surge underscores intensified strategic partnerships and competitive expansion in cloud services. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/08/2025-08-29_12-33-44.png) ARK Innovation ETF experienced a dramatic surge to $12 billion before dropping back to $8 billion within a short period. This volatile movement indicates significant investor activity or market events, reflecting confidence shifts and broader market impacts. --- **THE IDEAS:** **Mentions:** UMass Memorial, McKinsey, Lufthansa, TripStack, Chinese AI ### 01\. AI's healthcare potential expands via UMass Memorial The expansion of AI in healthcare, exemplified by UMass Memorial's AI triage tool, signifies growing trust and utilization of AI technology in hospitals, potentially improving efficiency and patient outcomes. **Related reading:** **Becker's Hospital Review**: [UMass Memorial expands AI triage tool to 5 more hospitals](https://www.beckershospitalreview.com/healthcare-information-technology/ai/umass-memorial-expands-ai-triage-tool-to-5-more-hospitals/) ### 02\. AI restrictions amid geopolitics impact consultancies McKinsey's restriction on AI work in China amidst geopolitical tensions highlights potential disruptions for consultancies operating globally, affecting strategies and collaborations. **Related reading:** **Financial Times**: [McKinsey bars China practice from generative AI work amid geopolitical tensions](https://www.ft.com/content/9907da48-3e6d-4208-a1d7-44a85c6a77f8) ### 03\. Open interlining transforms airline connectivity The adoption of open interlining strategies allows airlines like TripStack to bypass traditional constraints, fostering more dynamic partnerships and improving network efficiency. **Related reading:** **Skift**: [Open Interlining: Inside the Strategy Transforming How Airlines Connect Flights](https://skift.com/2025/04/02/open-interlining-inside-the-strategy-transforming-how-airlines-connect-flights/) ### 04\. Chinese AI adoption boosts growth prospects China's focus on AI adoption through companies like DeepSeek could stimulate economic growth amidst external pressures, indicating a strategic pivot towards innovation. **Related reading:** **CNBC**: [DeepSeek-led AI adoption offers China an opportunity to boost its sputtering growth](https://www.cnbc.com/2025/02/21/deepseek-led-ai-adoption-offers-china-opportunity-to-boost-growth.html) ### Four Ideas (Energy edition): AI in Grid Modernization and EV Battery Innovations URL: https://paulkedrosky.com/four-ideas-energy-edition-ai-in-grid-modernization-and-ev-battery-innovations/ Last updated: 2025-08-29T11:47:41.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Commodity Market Shifts, Cocoa, AI Adoption, Weather Impacts URL: https://paulkedrosky.com/four-ideas-commodities-edition-commodity-market-shifts-cocoa-ai-adoption-weather-impacts/ Last updated: 2025-08-28T11:47:40.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): AI accelerates healthcare transformation, health tech investment, diagnostic innovation, digital health emphasis URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-ai-accelerates-healthcare-transformation-health-tech-investment-diagnostic-innovation-digital-health-emphasis/ Last updated: 2025-08-27T11:47:51.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): AI Market Dynamics and Emerging Trends URL: https://paulkedrosky.com/four-ideas-ai-edition-ai-market-dynamics-and-emerging-trends/ Last updated: 2025-08-26T21:20:50.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Emerging Markets, High-Yield Bonds, Extreme Weather, Geoeconomics URL: https://paulkedrosky.com/four-ideas-macro-edition-emerging-markets-high-yield-bonds-extreme-weather-geoeconomics/ Last updated: 2025-08-25T11:48:01.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-43/ Last updated: 2025-08-24T11:47:32.000Z **AUGUST 24, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** Insurance Companies Send Chilling Letters Just Before Surgery. But Why? ([**NY Times**](https://www.nytimes.com/2025/08/22/your-money/insurance-prior-authorization-surgery-unitedhealthcare.html)) … Rich economies will need foreign workers to fuel growth, policymakers warn ([**FT**](https://www.ft.com/content/8bfdf5d7-3584-444d-849e-b75adc2e07ed)) … Rising Risk for the U.S. Economy: A Job Market That Is Just 'Meh' ([**WSJ**](https://www.wsj.com/economy/jobs/rising-risk-for-the-u-s-economy-a-job-market-that-is-just-meh-ca37eac1?gaa%5Fat=eafs&gaa%5Fn=ASWzDAg1YWcm06%5F4NiUKbvPljHFZnev2-1efXRFQqRvku1q-Rf8Lplku31ny&gaa%5Fts=68aaff57&gaa%5Fsig=OMYZXAHWuhujTbd%5F7Ucngk%5F%5FUyFaV5HTLUtWUz0PjOqEg8YcQjddwEynEazoiXIj5yNEOGpc3yeLo2vDZE6kCw%3D%3D)) … Decades After Genocide, Rwanda Emerges as Aggressive Regional Power ([**WSJ**](https://www.wsj.com/world/africa/rwanda-congo-military-rebels-minerals-2dc229e0?gaa%5Fat=eafs&gaa%5Fn=ASWzDAiDhIR-l2a9X7RKuIQzqUFMYpR3%5FrdZzQwUmZksio91wZhY4Hii2BM5&gaa%5Fts=68aaff57&gaa%5Fsig=Bonxgjz3B2KZVT8uGnfDVS45Jm3fkZAdZsnTETknj5jAe3Ty9nUAqNFtvCLmXOWkWdB6p3d2Q-iXlN4YSCIDCA%3D%3D)) … Vietnam Steps Up Island-Building in South China Sea, Says CSIS ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-08-24/vietnam-steps-up-island-building-in-south-china-sea-says-csis)) … US Inflation to Edge Up as Powell Shifts on Job Market ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-08-23/us-inflation-to-edge-up-as-powell-shifts-on-job-market)) … The Weekend Essay. The myths that made Putin’s war ([**FT**](https://www.ft.com/content/97312cf6-127d-411e-9690-a22e4396e2d0)) … Malaysian State Near Singapore Hit by Two Mild Earthquakes ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-08-24/malaysian-state-near-singapore-hit-by-two-mild-earthquakes)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/08/2025-08-22_16-54-24.png) Alaska's per capita energy expenditure of $12,100 in 2023 is over three times higher than Florida's $3,700\. This significant difference highlights Alaska's elevated energy costs, possibly due to harsher climate conditions and reliance on diverse energy sources, impacting economic considerations for residents and policymakers. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/08/2025-08-22_16-57-37.png) Toronto's condo construction reaches a historical peak, exceeding 90,000 units in 2023, a dramatic rise from 20,000 in the late 1990s. This surge reflects a significant market expansion, impacting housing availability and potentially affecting real estate prices and urban development policies. --- **THE IDEAS:** **Mentions:** UMass Memorial, DeepSeek, McKinsey, Phys.org, CNBC ### 01\. Generative AI reshaping industries and markets Generative AI is now widely adopted, transforming industry processes and consumer interactions. As this technology gains traction, associated markets such as hardware and AI services will expand. **Related reading:** **Twitter**: [As generative #AI sweeps across industries and the #consumer landscape...](https://twitter.com/SPGlobal/status/1860035529909080480) **Financial Times**: [Disrupted or displaced? How AI is shaking up jobs...](https://www.ft.com/content/5009fd1e-85db-433f-aa2b-55d9b88b6481) ### 02\. Geoeconomic shifts driving strategic realignment Global trade is witnessing strategic interdependencies that alter existing trade frameworks. This realignment is driven by geopolitical tensions and necessitates revised business strategies and partnerships. **Related reading:** **Financial Times**: [Strategic interdependence is rewiring the global economy](https://www.ft.com/content/25ab5e71-6ae2-495b-bf06-6fd2f1cbb18a) **Financial Times**: [Welcome to the new age of geoeconomics](https://www.ft.com/content/daf5ce11-c95b-4964-8df1-7463f0f9a81b) ### 03\. Regional M&A landscapes evolving for competitive growth M&A strategies are shifting as businesses look to expand beyond traditional boundaries, seeking growth via regional cross-border activities and strategic alignments, including emerging European markets. **Related reading:** **Becker's Hospital Review**: [The multi-state M&A playbook: Preparing for growth beyond your footprint](https://www.beckershospitalreview.com/finance/the-multi-state-ma-playbook-preparing-for-growth-beyond-your-footprint/) **CNBC**: [Investors looking beyond the U.S. see opportunities in one small pocket of Europe](https://www.cnbc.com/2025/08/18/investing-in-sweden-denmark-norway-finland-iceland-opportunity-novo-nordisk.html) ### 04\. Weather alerts directly influencing market behavior Extreme weather alerts are increasingly impacting market dynamics by influencing investor behavior and market prices, suggesting a need for better predictive models and risk management strategies. **Related reading:** **Phys.org**: [Extreme weather alerts can move markets. Here's what investors can learn from our new research](https://phys.org/news/2025-08-extreme-weather-investors.html) ### Four Ideas (Energy edition): Decentralized Energy Networks, AI's Impact on Solar, Battery Storage Demand, Subsidy-Free Renewables URL: https://paulkedrosky.com/four-ideas-energy-edition-decentralized-energy-networks-ais-impact-on-solar-battery-storage-demand-subsidy-free-renewables/ Last updated: 2025-08-22T11:47:45.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Geostrategic Realignments and Commodity Market Impacts URL: https://paulkedrosky.com/four-ideas-commodities-edition-geostrategic-realignments-and-commodity-market-impacts/ Last updated: 2025-08-21T11:48:07.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): Healthcare AI Innovations and Investments URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-healthcare-ai-innovations-and-investments/ Last updated: 2025-08-20T11:47:47.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): AI Bubble Formation, AI Optimism, Logistical Shifts, Hospitality Innovation URL: https://paulkedrosky.com/four-ideas-ai-edition-ai-bubble-formation-ai-optimism-logistical-shifts-hospitality-innovation/ Last updated: 2025-08-19T11:47:58.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): AI, Geoeconomics, Hospitality, Emerging Markets URL: https://paulkedrosky.com/four-ideas-macro-edition-ai-geoeconomics-hospitality-emerging-markets/ Last updated: 2025-08-18T11:48:18.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-42/ Last updated: 2025-08-17T11:47:14.000Z **AUGUST 17, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** Canada Directs Labor Board to End Strike at Biggest Airline ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-08-16/canada-seeks-to-end-airline-strike-with-labor-board-referral)) … Blackwater’s Erik Prince muscles back into the mercenary business, touting Trump allies ([**WSJ**](https://www.wsj.com/world/americas/erik-prince-mercenaries-vectus-global-5a166dca?gaa%5Fat=eafs&gaa%5Fn=ASWzDAh0Rwv-4aM0oDArmsZ2sr6Q-5EKp2pAVvOUlWiHdF4uGB09oPU7EREQ&gaa%5Fts=68a1c4c6&gaa%5Fsig=gJOeen%5FkaQmuAiNOxx1uoFaHVe3q7RjCznfNVjXFhZQnawfK9zi5DjCNU4SQZo5iTMRqFT%5FLTAIdU80C-7VSFg%3D%3D)) … SpaceX gets billions from the government. It gives little to nothing back in taxes. ([**NY Times**](https://www.nytimes.com/2025/08/15/technology/spacex-musk-government-contracts-taxes.html)) … Echo Investment Unit Inks €565 Million Polish Real Estate Sale ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-08-16/echo-investment-unit-inks-565-million-polish-real-estate-sale)) … For Gen Z, ‘little treats’ are worth going over budget ([**NY Times**](https://www.nytimes.com/2025/08/16/business/gen-z-treat-spending.html)) … C.E.O.s want their companies to adopt A.I. but do they get it themselves? ([**NY Times**](https://www.nytimes.com/2025/08/16/business/ceos-adopt-ai.html)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/08/2025-08-15_21-03-20.png) Significant outflows from bullish leveraged ETFs around June 2025 reached nearly $3 billion, contrasting with previous inflows. This shift signals a notable change in investor sentiment, raising concerns about market volatility and risk management strategies. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/08/2025-08-16_07-03-59.png) U.S. median apartment rents increased by 1.7% after a period of decline, marking the largest rise since 2023\. This shift indicates a potential rebound in the rental market, affecting housing affordability and demand dynamics. --- **THE IDEAS:** **Mentions:** AAPL, MSFT, FTSE, INTC, XOM ### 01\. AI tools revolutionizing operational efficiencies Generative AI is sweeping across industries, impacting operational efficiencies and consumer expectations. This will lead to competitive advantages for companies adopting AI early, impacting sectors from tech to hospitality. Broader implications include cost reductions and acceleration of innovation cycles. **Related reading:** **CNBC**: [An AI 'pick-and-shovel' play and signs that the market rally is broadening](https://www.cnbc.com/2025/08/14/an-ai-pick-and-shovel-play-and-signs-that-the-market-rally-is-broadening.html) **Twitter**: [As generative #AI sweeps across industries and the #consumer landscape](https://twitter.com/SPGlobal/status/1860035529909080480) ### 02\. Geostrategic shifts reshaping trade routes Geopolitical shifts are changing traditional trade dynamics. The US-China relationship is no longer driven solely by cost and efficiency, creating opportunities for new trade alliances. Second-order effects include changes in supply chain strategies and long-term economic dependencies. **Related reading:** **Financial Times**: [Strategic interdependence is rewiring the global economy](https://www.ft.com/content/25ab5e71-6ae2-495b-bf06-6fd2f1cbb18a) **FreightWaves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) ### 03\. Intangible assets gaining strategic focus internationally The valuation of intangible assets abroad is increasingly significant, necessitating strong oversight for maximizing potential. This trend holds substantial implications for multinational corporations aiming to leverage these assets, boosting competitiveness and innovation. **Related reading:** **Phys.org**: [Unlocking the value of intangible assets abroad requires strong board oversight, new study finds](https://phys.org/news/2025-08-intangible-assets-requires-strong-board.html) ### 04\. Tech companies adjusting to Washington's new climate The shifting political climate in Washington demands tech companies to adapt their strategies. Companies that navigate these policy changes effectively can benefit from favorable conditions, while others risk falling behind due to eroding geopolitical relationships. **Related reading:** **Twitter**: [Tech companies must adapt to Washington’s new political climate or risk falling behind](https://x.com/EurasiaGroup/status/1917572333901918496) ### Four Ideas (Energy edition): Decentralized Energy, Renewable Trade Barriers, Battery Storage URL: https://paulkedrosky.com/four-ideas-energy-edition-decentralized-energy-renewable-trade-barriers-battery-storage/ Last updated: 2025-08-15T11:47:14.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Geopolitical Shifts, Cocoa Boom, Tariff Impacts, Gold Opportunities URL: https://paulkedrosky.com/four-ideas-commodities-edition-geopolitical-shifts-cocoa-boom-tariff-impacts-gold-opportunities/ Last updated: 2025-08-14T11:47:35.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): AI Healthcare Innovations and Medicare; Health tech investment surge URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-ai-healthcare-innovations-and-medicare-health-tech-investment-surge/ Last updated: 2025-08-13T11:47:34.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): Strategic Interdependence and AI-Driven Shifts URL: https://paulkedrosky.com/four-ideas-ai-edition-strategic-interdependence-and-ai-driven-shifts/ Last updated: 2025-08-12T11:47:29.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Emerging Trends in Logistics, Geopolitics, AI, and Trade URL: https://paulkedrosky.com/four-ideas-macro-edition-emerging-trends-in-logistics-geopolitics-ai-and-trade/ Last updated: 2025-08-11T11:47:44.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-41/ Last updated: 2025-08-10T11:47:34.000Z **AUGUST 10, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** Airbnb Gives Modest Outlook, Warns Growth May Not Keep Up ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-08-06/airbnb-gives-modest-summer-outlook-warns-growth-may-not-keep-up)) … Airbnb’s shift to revenue management tools ([**FT**](http://www.wsj.com/articles/airbnb-signals-expansion-into-professional-vacation-rentals-1432051843?gaa%5Fat=eafs&gaa%5Fn=ASWzDAis7LKB0RCStF1hGofp2-CKJQ-UlzquhNVn7S8GNyYGHVtS4E2Dx5xC&gaa%5Fts=68988a59&gaa%5Fsig=UuF5dU0M9xfPftfgR15Xv%5FTU9lYO9RMJFFYb3f3DszC5UXpxMc7YTiKcJQBWrOjVB6dLzQh1e0-JVvFMQ-2qpw%3D%3D)) … The Tariff Effect: Billions in Revenue but No Economic Earthquake ([**WSJ**](https://www.wsj.com/economy/trade/trump-tariffs-economy-impact-2b07ab4f?gaa%5Fat=eafs&gaa%5Fn=ASWzDAh0uPwEyvFmxTi7jtAqPeF57TGyx8QffpRPlQd2pXkBLZIsBARVMywB&gaa%5Fts=68988a59&gaa%5Fsig=AlMEQnPBDuQmIUrdnfVnTKNXocPWEK0ssyLBeOQrQ9JPL4vBOQknKbVlTt-f3otPNrhmVnIvSKWDUT8iLgpcyg%3D%3D)) … India faces stark choice on Russian oil ([**FT**](https://www.ft.com/content/f2594187-889e-40ac-97f0-666388eac03a)) … WhiteFiber Is Said to Have Raised $159.4 Million in Upsized US IPO ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-08-06/whitefiber-is-said-to-raise-159-4-million-in-upsized-us-ipo)) … Record US Corn Crop Drives Profits for Crop Chemical Makers ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-08-06/record-us-corn-crop-drives-profits-for-top-fertilizer-maker)) … China Is a Nation of Savers. Many Are Drowning in Debt. ([**NY Times**](https://www.nytimes.com/2025/08/06/business/china-consumer-debt.html)) … Uber’s Festering Sexual Assault Problem ([**NY Times**](https://www.nytimes.com/2025/08/06/business/uber-sexual-assault.html)) … Trump Slaps 100% Duty on Imported Chips, but Promises Exemptions ([**WSJ**](https://www.wsj.com/livecoverage/stock-market-today-dow-sp-500-nasdaq-08-06-2025?gaa%5Fat=eafs&gaa%5Fn=ASWzDAieahQw1hRk8tZHMPN4J8umJduxuJ8I80tEa-5%5FsSo8Tc78OYtW09jq&gaa%5Fts=68988a59&gaa%5Fsig=pO6GrZnW5zuWa7Tlc%5FFTekwwu%5F0DUbc3JNAV04jbT6AAKS-yTjAOFPqUI2i%5Fbn-F-bIIKsrx4p-VGGfyQc78kQ%3D%3D)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/08/2025-08-09_06-50-41.png) Brunei ranks fifth by GDP at market rates but drops significantly when adjusted for prices and hours worked. This highlights how price differences and work hours can drastically alter economic standings, emphasizing the complexity of comparing economic performance across nations. --- **THE IDEAS:** **Mentions:** GOOGL, AMADY, MCK, SPGI, CCIH ### 01\. AI-driven cross-border payments transformation The cross-border payments landscape is evolving rapidly with AI's potential to enhance efficiency and security, especially amid geopolitical tensions reshaping international trade and finance. Projects innovating correspondent banking highlight the geopolitical imperatives driving AI adoption. **Related reading:** **Financial Times**: [The race for the future of cross-border payments](https://www.ft.com/content/44d973d6-1c38-4eaf-a943-6aa6367177f6) **Skift**: [AI Tools, and the New Rules of Travel Planning](https://skift.com/2025/08/01/booking-trends-ai-tools-and-the-new-rules-of-travel-planning/) **Twitter**: [As generative #AI sweeps across industries](https://twitter.com/SPGlobal/status/1860035529909080480) ### 02\. Surge in AI demand for inference The demand for AI-powered inference technology is skyrocketing, providing strong growth prospects for companies specializing in AI solutions and services, impacting sectors relying heavily on rapid data processing and decision making. **Related reading:** **Twitter**: [“But inference demand is going through the roof!!!!”](https://x.com/zoomyzoomm/status/1953751897518776497) **Becker's Hospital Review**: [UMass Memorial expands AI triage tool to 5 hospitals](https://www.beckershospitalreview.com/healthcare-information-technology/ai/umass-memorial-expands-ai-triage-tool-to-5-more-hospitals/) ### 03\. Industrial hubs adapting to economic shifts As industrial growth gains momentum following a period of stagnation, industrial hubs are finding new opportunities by adapting to economic changes, potentially leading to a resurgence in manufacturing and logistics. **Related reading:** **Oxford Economics**: [Industrial landscape at a critical juncture](https://www.oxfordeconomics.com/resource/industry-key-themes-2025-industrial-landscape-at-a-critical-juncture/) **Oxford Economics**: [Industrial Hubs: Finding opportunity amid the challenges](https://www.oxfordeconomics.com/resource/industrial-hubs-finding-opportunity-amid-the-challenges/) ### 04\. Emergence of small-world networks in multinationals Small-world networks in multinational companies can mitigate ESG controversies and enhance supply chain resilience by facilitating more sustainable and ethical business practices. **Related reading:** **Phys.org**: [Small-world networks can mitigate ESG controversies](https://phys.org/news/2025-08-small-world-networks-mitigate-environmental.html) ### The Kanye/Data Center Crossover URL: https://paulkedrosky.com/the-kanye-data-center-crossover/ Last updated: 2025-09-02T19:15:11.000Z ## Overview There are two sections today: a longish **Rough Notes update** on my recent data center comments, from cost of capital to futures, followed by some related (and unrelated) reading in **Rougher Notes**. ## Rough Notes ### The Kanye/Data Center Crossover I promise to move on to other topics soon—really!—but I've been overwhelmed with interest in my prior **two pieces** ([here](https://paulkedrosky.com/honey-ai-capex-ate-the-economy/) and [here](https://paulkedrosky.com/honey-ai-capex-keeps-eating-everything/)) on AI-related capex. I have done a dozen **media** interviews, some **podcasts** (at least one of which is [out](https://www.youtube.com/watch?v=t%5Fey5z30gf0)), had various conference **speaking** invites (yeah, no), and been reminded why I only do **selective consulting** to banks and hedge funds. There have also been a bazillion **glosses** elsewhere on the piece, some of which I've had a chance to read, but not all. My takeaway is that the level of interest in this topic is, as they say in the Tour de France, ***hors categorie***, outside of the categorization system, crossing from business and geek fixations into popular culture fascination. To put in a kind of context, I sometimes compare things in **units of "Ye"s**. Is the topic more interesting to people than is **Kanye West**? Well, data centers are about 1.25 Ye's, which is **effing nuts**: ![](https://paulkedrosky.com/content/images/2025/08/2025-08-07_12-18-44.png) Given the data center/Ye crossover, this seems like a good time to summarize the **most common questions** I get and my answers to them. ### Why did you write about AI capex? Like when writing about most things, I wrote it because no one **else would**. Writing is annoying, and so, having noticed that AI-related capital expenditures had become large enough to **affect economic data**, I assumed someone else would say it first. When no one did, or at least I didn't notice anyone doing so, I did. Of course, there is more to it than that. I wrote about AI capex because it seemed **important** that so much money was flowing so quickly into such a narrow slice of the economy. The size is rapidly exceeding prior capital expenditure waves, from the **fiber buildout** in the dot-com era, to the **railroads** in the 19th century. That is **remarkable**. ![](https://paulkedrosky.com/content/images/2025/08/2025-08-07_13-37-10.png) ### Why should we care? _This post is for subscribers on the Paul Kedrosky and Plus tiers only._ ### Four Ideas (Energy edition): Emerging Clean Energy Trends and Implications URL: https://paulkedrosky.com/four-ideas-energy-edition-emerging-clean-energy-trends-and-implications/ Last updated: 2025-08-08T11:47:33.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Understanding Emerging Trends in Commodities Market URL: https://paulkedrosky.com/four-ideas-commodities-edition-understanding-emerging-trends-in-commodities-market/ Last updated: 2025-08-07T13:57:30.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): AI-driven Health Tech and Patient Data Accessibility URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-ai-driven-health-tech-and-patient-data-accessibility/ Last updated: 2025-08-06T11:47:29.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): AI Disruption and New Industry Opportunities URL: https://paulkedrosky.com/four-ideas-ai-edition-ai-disruption-and-new-industry-opportunities/ Last updated: 2025-08-05T11:47:19.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Cross-Border Payments and Digital Influence Expansion URL: https://paulkedrosky.com/four-ideas-macro-edition-cross-border-payments-and-digital-influence-expansion/ Last updated: 2025-08-04T11:47:25.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-40/ Last updated: 2025-08-03T11:47:43.000Z **AUGUST 03, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** Norinchukin Bank ([**FT**](https://www.bloomberg.com/profile/company/NORZ:JP)) … After a Lag, Consumers Begin to Feel the Pinch of Tariffs ([**NY Times**](https://www.nytimes.com/2025/08/02/business/trump-tariffs-consumer-prices.html)) … Berkshire’s Quarterly Earnings Drop on Insurance Results, Currency Moves ([**WSJ**](https://www.wsj.com/business/earnings/berkshire-hathaway-q2-earnings-warren-buffett-deeff105?gaa%5Fat=eafs&gaa%5Fn=ASWzDAgcEfDx0hX6bqyyFmbSp3wUTxJF0Pb01bgtbNRBodf8s6D79e6O27Xu&gaa%5Fts=688f4fe2&gaa%5Fsig=Wew1ThFsmcasE0SdwRNtwbM5n6PPfusvKd-yACEA0RtHxxKlk19p%5FKgL133QY51kcHMPtlr3a1BQ64LHvA4nWw%3D%3D)) … Why Even Basic Airline Seats Keep Getting More ‘Premium’ ([**NY Times**](https://www.nytimes.com/2025/08/02/business/dealbook/premium-airfare.html)) … Tequila, Drugs and Torture: The Spending Binge of Two Crypto Bros That Ended Behind Bars ([**WSJ**](https://www.wsj.com/us-news/crypto-duplessie-woeltz-kidnap-torture-aae24e88?gaa%5Fat=eafs&gaa%5Fn=ASWzDAj9uF4ptdgYyEc6uv9-GWDe13ZWj1IIo7zW5%5F2EYaFNyI%5FWP9IfU3ET&gaa%5Fts=688f4fe3&gaa%5Fsig=WVdqzBeuq%5Fs1%5F90PXhD2Cbldf04uBS9aHjP8U7nij5UoWaRs8B2wEmFmpUZQvo-gPktvZcreJPKjdu4qoAmrlQ%3D%3D)) … Chocolate cartels: the rise of cocoa smuggling ([**FT**](https://www.ft.com/content/e0b7f437-0e3d-4e68-a2cd-d03affb72400)) … Joby Said to Weigh Deal for Helicopter Ride-Share Operator Blade ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-08-02/joby-said-to-weigh-deal-for-helicopter-ride-share-operator-blade)) … Triangles, Crescents, Slivers: Can Odd-Shaped Lots Help Ease the Housing Crisis? ([**NY Times**](https://www.nytimes.com/2025/08/02/business/housing-crisis-odd-lots.html)) … Codelco Finds Human Remains in Blow to Chile Mine Rescue Efforts ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-08-02/codelco-says-trapped-miners-may-be-safe-as-rescue-advances)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/08/2025-08-01_11-46-27.png) May's job growth estimate was drastically revised from 139k to 19k, a decrease of 86%. This significant revision highlights potential volatility or initial inaccuracies in economic data reporting, impacting economic planning and policy decisions. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/08/2025-08-01_18-19-16.png) Non-bank mortgage origination rose from under 10% in 2010 to over 70% by 2020, reflecting a significant market shift post-financial crisis. This change highlights increasing reliance on non-bank lenders, reshaping the mortgage landscape and influencing regulatory and financial stability discussions. --- **THE IDEAS:** **Mentions:** GOOGL, PYPL, AMADY, SPCE, MAR ### 01\. AI tools redefine travel booking strategies AI tools integrated into travel booking systems are significantly changing how consumers plan and manage trips. By collaborating on AI tools, companies like Google and Amadeus are streamlining travel booking processes, potentially increasing booking efficiency and customer satisfaction. Over the next 6-24 months, expect an uptick in AI adoption across hospitality and travel-related sectors. **Related reading:** **Skift**: [Booking Trends, AI Tools, and the New Rules of Travel Planning](https://skift.com/2025/08/01/booking-trends-ai-tools-and-the-new-rules-of-travel-planning/) ### 02\. AI boosts cross-border payment innovations AI is increasingly crucial in revolutionizing cross-border payment systems. Emerging tech-enabled solutions aim to reduce transaction costs and improve speed, indicating a strategic shift toward more integrated and efficient global payment networks. These innovations are likely to influence stakeholders including financial firms and create new geopolitical and economic dynamics. **Related reading:** **Financial Times**: [The race for the future of cross-border payments](https://www.ft.com/content/44d973d6-1c38-4eaf-a943-6aa6367177f6) ### 03\. Hospitality sector integrates local experiences The hospitality industry is increasingly valuing local expertise alongside global operational frameworks. This trend highlights the importance of local knowledge in delivering customized and engaging guest experiences, which could drive demand for localized service offerings and partnerships with local businesses, enhancing revenue streams. **Related reading:** **Hospitality Net**: [Global Vision, Local Knowledge: Why One Can’t Succeed Without the Other](https://www.hospitalitynet.org/opinion/4128378.html) ### 04\. Space economy drives technology investments Growth in the space economy is attracting increased investments in aerospace technology and satellite services. With a market now valued at $613 billion, sectors from telecommunications to Earth observation are likely to see significant advancements, fostering innovation and creating new business opportunities over the next 24 months. **Related reading:** **YouTube**: [Global Space Economy Grows to $613 Billion](https://www.youtube.com/watch?v=q9MvZS-1v8I) ### Four Ideas (Energy edition): European Grids, Storage, Decentralization, and Energy Transition URL: https://paulkedrosky.com/four-ideas-energy-edition-european-grids-storage-decentralization-and-energy-transition/ Last updated: 2025-08-01T11:47:24.000Z _This post is for paying subscribers only._ ### Honey, AI Capex Keeps Eating ... Everything URL: https://paulkedrosky.com/honey-ai-capex-keeps-eating-everything/ Last updated: 2025-09-02T19:13:46.000Z Across companies, geographies, and even GDP, AI capex has become overwhelming _This post is for subscribers on the Paul Kedrosky and Plus tiers only._ ### Four Ideas (Commodities edition): Commodities Trading Diversification and Energy Trends URL: https://paulkedrosky.com/four-ideas-commodities-edition-commodities-trading-diversification-and-energy-trends/ Last updated: 2025-07-31T11:47:18.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): AI in Healthcare, FDA Policies, Health Tech Investments URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-ai-in-healthcare-fda-policies-health-tech-investments/ Last updated: 2025-07-30T11:47:37.000Z _This post is for paying subscribers only._ ### Tariffs are Dumb Enough to (Almost) Work URL: https://paulkedrosky.com/tar/ Last updated: 2025-09-04T14:25:39.000Z ## Updates & Erasures: Another week, another funky datacenter funding vehicle, aka an SPV. The latest example of "hide the AI datacenter capex", which I've been [on about repeatedly](https://paulkedrosky.com/weekend-reading-plus-spvs-meta-and-fiber-buildout-2-0/), comes from xAI. It is working with Valor, the [WSJ says](https://www.wsj.com/tech/ai/elon-musk-x-ai-funding-feecede1?mod=hp%5Flead%5Fpos1), trying to help it raise money to buy more chips and such for Grok-type things. And that, it seems requires an SPV. Why? Well, the tricky bit requiring the SPV, other than it's fun to hide debt off-balance sheet is that xAI apparently has debt covenants prevented it from raising more than an additional $5b. But with $12b (what it wants to raise) being materially larger than $5b (what it is allowed to raise), there is a problem. Bankers, being nothing if not creative, enter the idea of using a legal artifice to attach the debt to something that is not xAI, but is, for all practical purposes. This will all work out well, boosting GDP and such, of course, until it doesn't. ## Rough Notes: ## Tariffs Are Just Stupid Enough to (Almost) Work **Tariffs are dumb**. They **distort** trade, favor **inefficient** local producers, cause trading partners to **retaliate**, and make people **worse off** than a world without them. On these points, economists **almost universally agree**¹. ¹ This is weak. Saying that "most economists agree on XYZ" can be like saying that it is common sense, like backing out of your driveway quickly is a bad idea. Totally, yes, 100%. Didn't need you to tell me though, etc. But tariffs are **not useless**. They may even be sort of, almost, kinda, a ... **good idea** in these very weird U.S. circumstances. Hear me out, because **three things** are going on, so it can get messy: 1. The U.S. is, as the line goes, an **insurance company with an army**, which has straitjacketed its budget, which I've written [about previously](https://paulkedrosky.com/life-under-two/). 2. The U.S. **hates taxes**, and most voters are innumerate, so it finds silly ways to hide them. 3. Tariffs are a kind of horrible, **second-best** solution to the above problems. The first two points are mostly self-explanatory. Entitlements plus defence are now around 70% of the U.S. budget—see also, insurance company with an army—leaving little room to do much other than cut, unless you find new revenue. But new revenue is hard, because Americans **hate income taxes**, and have long resisted carbon taxes or a **value-added tax** (VAT). They aren't coping well with what I've called [life under 2%](https://paulkedrosky.com/life-under-two/). Enter tariffs. They raise money because **consumers buy** things. We can argue about whether the **producing companies** pay the tariff (they mostly don't), or whether **consumers pay** it via higher prices (they mostly do), but the effect is the same: consumers buying things increases government revenue. That is tariff income. _This post is for subscribers on the Paul Kedrosky and Plus tiers only._ ### Four Ideas (AI edition): AI Investment Trends and Economic Impact URL: https://paulkedrosky.com/four-ideas-ai-edition-ai-investment-trends-and-economic-impact/ Last updated: 2025-07-29T11:47:28.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Global Shifts in Space Economy and AI Regulations URL: https://paulkedrosky.com/four-ideas-macro-edition-global-shifts-in-space-economy-and-ai-regulations/ Last updated: 2025-07-28T11:47:19.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-39/ Last updated: 2025-07-27T11:47:11.000Z **JULY 27, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** Americans See Their 401(k)s Not Just as Nest Eggs but as Rainy Day Funds ([**WSJ**](https://www.wsj.com/personal-finance/retirement/401k-withdrawal-emergency-savings-25926369?gaa%5Fat=eafs&gaa%5Fn=ASWzDAi22153a5KWO0KjB8A69WHfPsVVBKOICfV458VWYZJe9n2RxUI-y8qt&gaa%5Fts=68861542&gaa%5Fsig=-r1jkmCPde-PY1wOw6UV9tAfaAylxd1okUW10Wxi5AJkkCqZT55hiPY7WFqjb%5FFAGnSkCwN9YTZDHHOUwSTtQQ%3D%3D)) … Union Pacific Is Said to Near Deal for Norfolk Southern ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-07-26/union-pacific-is-said-to-near-deal-for-norfolk-southern)) … As consumers lose their appetite, food brands fight to keep Wall St. happy ([**NY Times**](https://www.nytimes.com/2025/07/25/business/kraft-heinz-ferrero-food-brand-deals.html)) … Plastic turf fields are taking over America ([**NY Times**](https://www.nytimes.com/2025/07/26/business/artificial-turf-grass-sports.html)) … Artificial intelligence: AI intelligence test, Festival of science in Cambridge ([**FT**](https://www.nytimes.com/2007/07/29/magazine/29robots-t.html)) … Hedge funds: London hedge funds dominate court challenges over market turmoil ([**FT**](https://www.ft.com/content/910647af-73c8-4b71-a86c-d0a23e9751fe)) … Wall Street ‘euphoria’ sparks bubble warnings ([**FT**](https://www.ft.com/content/2450c892-3b9e-431e-a352-1d01df269617)) … How to Extract $400 Million From a Billionaire: Use a Gilded Age Family Name ([**WSJ**](https://www.wsj.com/finance/scam-billionaire-astor-bf8ce715?gaa%5Fat=eafs&gaa%5Fn=ASWzDAhjm9vbbKpG2Isb%5FBZFtzgqUB1UAmOlHQgxtN2Cs6SgJWDu45PqC%5FEK&gaa%5Fts=68861542&gaa%5Fsig=M1pJbb%5FWHCbVh3KKSbpq1U4zF219WIICc8IWvUResa8V5NtK83bX99ShXraH%5FpzYOIhkJ7gI%5F0MoZ1q8RbVWzQ%3D%3D)) … Ecuador in Spat with Colombia Over Deportation of Prisoners ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-07-26/ecuador-in-spat-with-colombia-over-deportation-of-prisoners)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/07/2025-07-26_14-10-12.png) US tariff rates are projected to reach early 1900s levels by 2025, surging from current low levels. This significant increase, largely driven by tariffs on Chinese imports, marks a reversal from nearly a century of tariff decline, potentially impacting global trade dynamics and economic relations. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/07/2025-07-26_14-06-40.png) Investors significantly shifted $20B from US Treasuries to IG Credit by May, reversing previous trends. This highlights a confidence change in corporate bonds over government securities, potentially impacting borrowing costs and market stability. --- **THE IDEAS:** **Mentions:** UMass Memorial, DeepSeek, Mark Cuban, McKinsey, NASDAQ ### 01\. Healthcare AI integration expands dramatically UMass Memorial's expansion of AI triage tools illustrates growing adoption in healthcare systems. This trend boosts efficiency and reduces workload, offering a competitive edge to early adopters. As more hospitals integrate AI, the demand for related technologies and services will rise, leading to potential partnerships and investments in AI healthcare providers. **Related reading:** **Beckers Hospital Review**: [UMass Memorial expands AI triage tool to 5 more hospitals](https://www.beckershospitalreview.com/healthcare-information-technology/ai/umass-memorial-expands-ai-triage-tool-to-5-more-hospitals/) ### 02\. China triples focus on AI startups China seeks economic growth amidst trade tensions and economic uncertainty, investing in AI startups like DeepSeek. This shift encourages investors to explore emerging AI companies in China for long-term opportunities, potentially affecting global tech competition and innovation hubs. **Related reading:** **CNBC**: [DeepSeek-led AI adoption offers China an opportunity to boost its sputtering growth](https://www.cnbc.com/2025/02/21/deepseek-led-ai-adoption-offers-china-opportunity-to-boost-growth.html) **CNBC**: [Foreign investors are returning to China in search of opportunities in artificial intelligence startups](https://www.cnbc.com/2025/07/16/cnbcs-the-china-connection-newsletter-from-didi-to-deepseek.html) ### 03\. Industrial resurgence signals potential manufacturing shift The industrial sector's regained momentum signals a potential manufacturing shift, driven by job creation and income growth in strategic areas like Japan. Companies positioned in these emerging hotspots stand to benefit, reshaping global supply chains and industrial policies. **Related reading:** **Oxford Economics**: [Industry Key Themes 2025: Industrial landscape at a critical juncture](https://www.oxfordeconomics.com/resource/industry-key-themes-2025-industrial-landscape-at-a-critical-juncture/) **Oxford Economics**: [Identifying future manufacturing hot spots in Japan and their implications](https://www.oxfordeconomics.com/resource/identifying-future-manufacturing-hot-spots-in-japan-and-their-implications/) ### 04\. Global Space Economy accelerates towards trillion dollars The space economy's growth to $613 billion indicates rapid expansion, offering investment opportunities in satellite services, space tourism, and related technologies. As global interest intensifies, companies in aerospace and innovative startups could see increased valuation and demand. **Related reading:** **YouTube**: [Global Space Economy Grows to $613 Billion](https://www.youtube.com/watch?v=q9MvZS-1v8I) ### Four Ideas (Energy edition): Emerging Energy Innovations and Storage Solutions URL: https://paulkedrosky.com/four-ideas-energy-edition-emerging-energy-innovations-and-storage-solutions/ Last updated: 2025-07-25T11:47:23.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Commodities Resurgence and Geopolitical Impacts URL: https://paulkedrosky.com/four-ideas-commodities-edition-commodities-resurgence-and-geopolitical-impacts/ Last updated: 2025-07-24T11:47:25.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): Tech Disruption, AI in Healthcare, China Biotech, Wearables URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-tech-disruption-ai-in-healthcare-china-biotech-wearables/ Last updated: 2025-07-23T11:47:05.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): AI-driven Industrial Transformation Opportunities URL: https://paulkedrosky.com/four-ideas-ai-edition-ai-driven-industrial-transformation-opportunities/ Last updated: 2025-07-22T11:47:21.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Global Economic Shifts: Copper, AI, and Geopolitics URL: https://paulkedrosky.com/four-ideas-macro-edition-global-economic-shifts-copper-ai-and-geopolitics/ Last updated: 2025-07-21T12:51:18.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-38/ Last updated: 2025-07-20T11:47:01.000Z **JULY 20, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** HSBC renews search for chair after struggling to find suitable candidates ([**FT**](https://www.ft.com/content/f2aa59bb-e956-4165-99b3-19a04dbdd8a5)) … The Epic Battle for AI Talent—With Exploding Offers, Secret Deals and Teams ([**WSJ**](https://www.wsj.com/tech/ai/meta-ai-recruiting-mark-zuckerberg-sam-altman-140d5861?gaa%5Fat=eafs&gaa%5Fn=ASWzDAiImygHLA5XCJqXiOP3hv%5FSLn9pxF3qI3fND29-vNf-BYTXFFlMoO1l&gaa%5Fts=687cdab7&gaa%5Fsig=Acb%5FItNn7hMxXzUVrQEs%5Fl-CjMuzbYMLWpUoAEyaIGpuExcwq-XcdebFln5syTQ5RgmGiCPmp2WS3QGz1QgNLg%3D%3D)) … The Big Read. Will a return to risk-taking rouse animal spirits? ([**FT**](https://www.ft.com/content/46dd0764-192a-4ca0-92a4-c8a4c1833242)) … UK Government Pledges to Cut Sewage Spills in Half by 2030 ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-07-19/uk-governments-pledges-to-cut-sewage-spills-in-half-by-2030)) … Millions Stolen, Death Threats: Should Banks Do More to Fight ‘Pig Butchering’? ([**WSJ**](https://www.wsj.com/finance/regulation/banks-pig-butchering-fight-fraud-92c06642?gaa%5Fat=eafs&gaa%5Fn=ASWzDAgllRIB2N284FR2iNFgInOA-IknfiGIcUMj0A6M0VQfFIo8jftNOuk6&gaa%5Fts=687cdab7&gaa%5Fsig=vTdQRDSKVohWFW3b28AHRNhJg%5FIJ%5FGt77sk4gmidAmThaDkoBTk5dxCLA5RcWN1T4fnIaamUTs-StRSIAYfJKg%3D%3D)) … CBS Canceling ‘Late Show With Stephen Colbert’ After Next Season ([**NY Times**](https://www.nytimes.com/2025/07/17/business/stephen-colbert-late-show-ending.html)) … Astronomer CEO Resigns After Being Caught on Coldplay Kiss Cam ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-07-19/astronomer-ceo-resigns-after-being-caught-on-coldplay-kiss-cam)) … Tech’s Top Venture Firm Tried to Stay Above Politics. Then a Partner Created a Furor. ([**NY Times**](https://www.nytimes.com/2025/07/19/technology/sequoia-capital-shaun-maguire-mamdani.html)) … 5 Years Ago, They Reversed Roles With Their Parents. Where Are They Now? ([**NY Times**](https://www.nytimes.com/2025/07/19/business/reverse-boomerang-parents-adult-children.html)) … Why the Federal Reserve’s Building Renovation Costs $2.5 Billion ([**Bloomberg**](https://www.bloomberg.com/news/articles/2025-07-19/trump-s-attacking-powell-over-fed-renovation-here-s-why-it-costs-2-5-billion)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/07/2025-07-16_07-20-37-2.png) Less-developed countries show a unique positive relationship between GDP per capita growth and subjective well-being. While transition and developed countries trend negatively, a 10% GDP growth aligns with a 0.15 increase in well-being for less-developed countries, highlighting significant socio-economic improvements. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/07/2025-07-16_07-16-01-2.png) Starbucks experienced a drastic sales decline during early 2020, followed by a sharp recovery in 2021\. This fluctuation contrasts with consistent prior growth and recent stagnation, highlighting pandemic impacts. Understanding this volatility is crucial for future strategy amid shifting consumer behavior and economic conditions. --- **THE IDEAS:** **Mentions:** AI, China, Geoeconomics, Hospitality ### 01\. AI revolutionizes early-stage investments. AI is sparking interest from investors keen to capitalize on its rapid development. With China as a key player in AI startups, foreign investments are increasing, boosting innovation and potential returns in early-stage tech firms. **Related reading:** **CNBC**: [CNBC's The China Connection newsletter: From Didi to DeepSeek](https://www.cnbc.com/2025/07/16/cnbcs-the-china-connection-newsletter-from-didi-to-deepseek.html) **Oxford Economics**: [New Tools, New Skills, No End in Sight: The Challenge of AI Adoption](https://www.oxfordeconomics.com/resource/new-tools-new-skills-no-end-in-sight-the-challenge-of-ai-adoption/) ### 02\. Shifting geoeconomics reshapes global alliances. The mingling of tech, trade, and military policies marks a departure from neoliberalism, prompting countries to form new economic alliances. This affects global trade flows, influencing which countries become trade hubs over the next few years. **Related reading:** **Financial Times**: [Welcome to the new age of geoeconomics](https://www.ft.com/content/daf5ce11-c95b-4964-8df1-7463f0f9a81b) **FreightWaves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) ### 03\. China's AI role boosts economic resilience. China's focus on AI via platforms like DeepSeek offers potential to counter its economic challenges. This shift could bolster global AI collaboration, ensuring China's ongoing growth in tech innovation and trade. **Related reading:** **CNBC**: [DeepSeek-led AI adoption offers China an opportunity to boost its sputtering growth](https://www.cnbc.com/2025/02/21/deepseek-led-ai-adoption-offers-china-opportunity-to-boost-growth.html) **CNBC**: [The China Connection newsletter: Whispers of a new world order in Tianjin](https://www.cnbc.com/2025/06/25/cnbcs-the-china-connection-newsletter-whispers-of-a-new-world-order.html) ### 04\. Hospitality reinvents for post-2024 expectations. The hospitality industry adapts to tech and consumer trends, requiring innovative, tech-driven service strategies. Post-2024, businesses that leverage these trends are positioned to capture a new wave of demand. **Related reading:** **HospitalityNet**: [Top 10 Hospitality Trends: What Shaped 2024 and What To Expect in 2025](https://www.hospitalitynet.org/opinion/4125874.html) **HospitalityNet**: [Food & Beverage Innovation: Trending Creativity and Sustainability](https://www.hospitalitynet.org/opinion/4127047.html) ### Honey, AI Capex is Eating the Economy URL: https://paulkedrosky.com/honey-ai-capex-ate-the-economy/ Last updated: 2025-09-02T19:15:31.000Z As ever, here is **what's ahead**: 1. **Updates** on prior pieces 2. My most recent Rough Notes **essay** 3. A few things **worth reading** ## Updates & Erasures: I [previously](https://paulkedrosky.com/on-curve-steepeners-subway-surfers-and-such/) wrote about the **perils of building renovation** as a Fed chair, especially given an administration bent on finding a reason to fire you "for cause." As anyone who has renovated anything **larger than a dog house** knows, no one thinks what you spent on renovation makes sense, not even you. Most **spouses *wish* they could fire** *someone*for cause over most renovations. In that light, [this WSJ explainer](https://www.wsj.com/economy/central-banking/federal-reserve-reonovations-oust-jerome-powell-6bdf82c8?mod=hp%5Flead%5Fpos2) of the ongoing **Federal Reserve renovations** is useful reading as various administration officials **feign dismay** at someone **renovating a building** that hasn't been renovated in ... ever. Shocked, I am shocked to find renovations going on in here. > On Tuesday, T\*\*\*p said he was surprised to learn about the costly building renovations, given his view that Powell is a boring and unintelligent person. “I think he’s a total stiff, but the one thing I didn’t see him \[as\] is a guy that needed a palace to live in,” he told reporters. Say what you will about Powell's monetary policy, either that **wallpaper doesn't go**¹, or Powell does, apparently. ¹ Writer Oscar Wilde's final words may or may not have been: "This wallpaper and I are fighting a duel to the death. Either it goes or I do.”. ## Rough Notes: ### Honey, AI Capex Ate the Economy How extreme is **AI datacenter spending**? It is to the point that even Chinese President **Xi Jinping** is warning about it. With more than [**250 datacenters**](https://www.ft.com/content/3d6601f0-c7cf-48a8-9ae2-e9820dedf0b0) under construction in his country, this week he [cautioned](https://www.ft.com/content/9c19d26f-57b3-4754-ac20-eeb627e8723e) against the **spending rush**, threatening to intervene: > “When it comes to projects, there are a few things — artificial intelligence, computing power and new energy vehicles. Do all provinces in the country have to develop industries in these directions?" The U.S., however, leads the **capex spending** way. One analyst recently speculated (via [Ed Conard](https://www.edwardconard.com/macro-roundup/using-nvidias-datacenter-revenue-as-a-reference-for-us-ai-capex-jensnordvig-estimates-that-ai-will-make-up-2-of-us-gdp-in-2025-given-a-standard-multiplier-implying-an-ai-contribution-to-g/?view=detail&filters=macro-roundup-database)) that, based on Nvidia's latest datacenter sales figures, AI capex may be **\~2% of US GDP in 2025**, given a standard multiplier. This would imply an AI contribution to **GDP growth of 0.7%** in 2025. I wrote previously about some of the [unintended consequences](https://paulkedrosky.com/weekend-reading-plus-spvs-meta-and-fiber-buildout-2-0/), in particular the emergence of **off-balance sheet funding** vehicles. Let's dive into its size and consequences again, however, with **three thoughts** in mind: 1. The above impact estimates may be a **lower bound**. 2. The redirected capital spending is **coming from somewhere**. 3. All this spending helps explain **a mystery**. ### 1\. A Lower Bound? Here are some basic numbers to get things started: - US 2025 GDP: **$25 trillion** (Source: Consensus forecasts and extrapolations from BEA data) - Nvidia sales to datacenters (Q1 FY2026): **$39.1B (**Annualized: **$156.4B)** - % of Nvidia datacenter sales that is AI related: **\~99%** (Primarily H100/GH200 sales to hyperscalers and enterprises) - Nvidia share of datacenter capex: **25-35%** - Economic multiplier: **1.5x-2.0x** - Implied total datacenter capex: **\~$520B** (156.4b/0.3) So, to get the total GDP impact, including multipliers, do the simple math: These are very large numbers given that before 2022 AI capex was likely less than 0.1% of GDP. It has, in three years, **grown by at least 10x** from there, and perhaps more than that. _This post is for subscribers on the Paul Kedrosky and Plus tiers only._ ### Four Ideas (Energy edition): Vehicle Batteries, Decentralized Energy, EV Grid Impact, Energy Storage Innovations URL: https://paulkedrosky.com/four-ideas-energy-edition-vehicle-batteries-decentralized-energy-ev-grid-impact-energy-storage-innovations/ Last updated: 2025-07-18T11:47:25.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Gold Tariffs, Oil Prices, Commodity Traders, and China Demand URL: https://paulkedrosky.com/four-ideas-commodities-edition-gold-tariffs-oil-prices-commodity-traders-and-china-demand/ Last updated: 2025-07-17T11:47:33.000Z _This post is for paying subscribers only._ ### On Curve Steepeners, Subway Surfers, and the Polycrisis of Confidence URL: https://paulkedrosky.com/on-curve-steepeners-subway-surfers-and-such/ Last updated: 2025-07-16T16:49:04.000Z Thinking like a macro hedge fund manager about the US eating itself _This post is for subscribers on the Paul Kedrosky and Plus tiers only._ ### Four Ideas (Life sciences edition): Healthcare AI Trends and Innovations URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-healthcare-ai-trends-and-innovations/ Last updated: 2025-07-16T11:47:20.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): AI Data Centers, Travel Tech, AI Investing, Chip Bifurcation URL: https://paulkedrosky.com/four-ideas-ai-edition-ai-data-centers-travel-tech-ai-investing-chip-bifurcation/ Last updated: 2025-07-15T18:40:32.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-37/ Last updated: 2025-07-13T11:47:12.000Z **JULY 13, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** - **Extreme weather** impacts global regions significantly ([FT](https://ft.com)) - Plane **accident** raises safety concerns ([FT](https://ft.com)) - **US inflation** rises due to tariff pass-through ([Bloomberg](https://bloomberg.com)) - **Stellantis** reports armed man in Michigan plant arrested ([Bloomberg](https://bloomberg.com)) - Retirees need a **plan** to manage savings withdrawal ([NY Times](https://nytimes.com)) - Colorado manufacturer **survives** trade war challenges ([NY Times](https://nytimes.com)) - **SpaceX** invests $2 billion in xAI ([WSJ](https://wsj.com)) - **Cable Cowboy** fights giant gas station ([WSJ](https://wsj.com)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/07/2025-07-12_20-22-17.png) Heat records dominate with 7.10% of Earth's surface reaching new highs compared to a negligible 0.01% for cold records. This stark contrast highlights a significant warming trend, emphasizing urgent climate action to mitigate associated environmental and societal impacts. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/07/unnamed-2.png) Young people experienced the steepest 69% decline in attending or hosting social events, contrasting with a 13% drop for ages 55-64\. This significant change suggests shifting social habits among younger generations, impacting social dynamics and the future of in-person interactions. --- **THE IDEAS:** **Mentions:** U.S. REITs, Northwell Health, DeepSeek, AI Tools, Asian-Gulf Trade ### 01\. AI revolutionizing healthcare triage processes AI's growing role in healthcare, as seen in UMass Memorial's AI triage tool expansion, could enhance efficiency and accuracy in patient assessments. This trend indicates a broader integration of AI technologies in medical settings, potentially reducing costs and improving patient outcomes over the next 6-24 months. **Related reading:** **Becker's Hospital Review**: [UMass Memorial expands AI triage tool to 5 more hospitals](https://www.beckershospitalreview.com/healthcare-information-technology/ai/umass-memorial-expands-ai-triage-tool-to-5-more-hospitals/) **Phys.org**: [How education systems can adapt to the challenges and opportunities of AI](https://phys.org/news/2025-04-opportunities-ai.html) ### 02\. Geoeconomic partnerships steer global trade The geopolitical realignment is affecting trade patterns, with partners trading more among themselves, reducing reliance on rivals. This trend highlights potential shifts in global supply chains and could lead to stronger regional economic clusters? **Related reading:** **Freightwaves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) **Financial Times**: [Welcome to the new age of geoeconomics](https://www.ft.com/content/daf5ce11-c95b-4964-8df1-7463f0f9a81b) ### 03\. AI as a catalyst for China's economic recovery DeepSeek-led AI adoption offers a boost to China's economic growth amid a trade war with the U.S. and economic slowdown. This push towards AI may foster innovation and increase competitive advantage, potentially resulting in economic expansion over the next two years. **Related reading:** **CNBC**: [DeepSeek-led AI adoption offers China an opportunity to boost its sputtering growth](https://www.cnbc.com/2025/02/21/deepseek-led-ai-adoption-offers-china-opportunity-to-boost-growth.html) **Twitter**: [As generative AI sweeps across industries and the consumer landscape](https://twitter.com/SPGlobal/status/1860035529909080480) ### 04\. Asian-Gulf trade ties expanding beyond oil Deepening trade relations between the Gulf and Asia signals diversification beyond oil. Enhanced commercial ties could result in increased economic cooperation, investment opportunities, and strategic alliances, potentially reshaping trade dynamics in these regions. **Related reading:** **Financial Times**: [China courts Asian and Gulf partners at ‘inflection point’ for global trade](https://www.ft.com/content/c72e43fb-094d-45ae-a28b-d3c362798cae) **The Economist**: [Commercial ties between the Gulf and Asia are deepening](https://www.economist.com/business/2024/09/05/commercial-ties-between-the-gulf-and-asia-are-deepening) ### Four Ideas (Energy edition): Decentralization, Energy Storage, Geoeconomics, Clean Energy Tariffs URL: https://paulkedrosky.com/four-ideas-energy-edition-decentralization-energy-storage-geoeconomics-clean-energy-tariffs/ Last updated: 2025-07-11T11:47:26.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Commodities Trends and Opportunities 2025 URL: https://paulkedrosky.com/four-ideas-commodities-edition-commodities-trends-and-opportunities-2025/ Last updated: 2025-07-10T11:47:17.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): Healthcare AI Growth and Pharmaceutical Innovation URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-healthcare-ai-growth-and-pharmaceutical-innovation/ Last updated: 2025-07-09T11:47:13.000Z _This post is for paying subscribers only._ ### Regime Changes: Tour de France vs the U.S. Deficit URL: https://paulkedrosky.com/regime-changes-tour-de-france-vs-the-u-s-deficit/ Last updated: 2025-09-04T14:25:54.000Z Regime changes in dynamic systems can inform and confound, from budget deficits to the Tour de France. _This post is for subscribers on the Paul Kedrosky and Plus tiers only._ ### Four Ideas (AI edition): AI-Driven Market Shifts and New Beneficiaries URL: https://paulkedrosky.com/four-ideas-ai-edition-ai-driven-market-shifts-and-new-beneficiaries/ Last updated: 2025-07-08T11:47:46.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Global Interest Rates and Geopolitical Rivalries URL: https://paulkedrosky.com/four-ideas-macro-edition-global-interest-rates-and-geopolitical-rivalries/ Last updated: 2025-07-07T11:47:13.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-36/ Last updated: 2025-07-06T11:46:57.000Z **JULY 06, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** - **Texas floods** kill at least 51, including 15 children ([WSJ](https://wsj.com)) - **Meme stocks** drive market rally amid speculation ([WSJ](https://wsj.com)) - **Texas floods** surpass 50 deaths with more rain expected ([Bloomberg](https://bloomberg.com)) - **Germany's Autobahn** cracks due to extreme heat ([Bloomberg](https://bloomberg.com)) - **China** acquires mines globally for resource security ([FT](https://ft.com)) - **Big Food's** snack sales decline as Americans change habits ([FT](https://ft.com)) - **YouTube Pirates** exploit Hollywood's summer blockbusters ([NY Times](https://nytimes.com)) - **China** faces environmental costs for rare earth dominance ([NY Times](https://nytimes.com)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/07/2025-07-01_07-13-53-2.png) Southern and eastern coastal counties face extreme disaster costs, exceeding $570 million, while many inland areas remain under $1.3 million. This significant disparity highlights coastal vulnerability to floods and hurricanes, emphasizing the urgent need for improved infrastructure and disaster preparedness in high-risk regions. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/07/2025-07-01_07-34-33-2.png) GS EU Defense achieved a striking 107.7% year-to-date gain, vastly surpassing others despite a modest second-quarter return. This highlights its resilience and potential safe-haven status amidst broader volatile trends, contrasting with the underperformance in traditional indices like the S&P 500's 5.5% year-to-date increase. --- **THE IDEAS:** **Mentions:** SPGSustainable1,EurasiaGroup,CNBC,Macrobond,FT ### 01\. Surge in AI tool adoption across industries AI is rapidly transforming sectors by enhancing efficiency and decision-making processes, as evidenced by UMass Memorial's expansion of AI tools in healthcare. This trend is fueled by the ongoing technological advancements and necessity for competitive advantage, influencing industries like healthcare and smart manufacturing globally over the next 24 months. **Related reading:** **Becker's Hospital Review**: [UMass Memorial expands AI triage tool to 5 more hospitals](https://www.beckershospitalreview.com/healthcare-information-technology/ai/umass-memorial-expands-ai-triage-tool-to-5-more-hospitals/) **SPGlobal**: [As generative #AI sweeps across industries and the #consumer landscape](https://twitter.com/SPGlobal/status/1860035529909080480) ### 02\. Geoeconomics reshaping global trade partnerships There's a growing shift in trade patterns as countries align more with geopolitical partners, reducing reliance on rivals. This new geoeconomic era, less dependent on past neoliberal policies, will redefine global trade dynamics, as highlighted by China's strategies in global trade inflection points. **Related reading:** **Financial Times**: [Welcome to the new age of geoeconomics](https://www.ft.com/content/daf5ce11-c95b-4964-8df1-7463f0f9a81b) **Financial Times**: [China courts Asian and Gulf partners at ‘inflection point’ for global trade](https://www.ft.com/content/c72e43fb-094d-45ae-a28b-d3c362798cae) **FreightWaves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) ### 03\. Industrial growth regaining momentum after downturn Following significant economic challenges in recent years, the industrial sector is poised for recovery. Growth is being driven by strategic investments and pent-up demand, which are expected to lead to increased output and technological innovations, particularly in emerging industrial hubs like Japan. **Related reading:** **Oxford Economics**: [Industry Key Themes 2025: Industrial landscape at a critical juncture](https://www.oxfordeconomics.com/resource/industry-key-themes-2025-industrial-landscape-at-a-critical-juncture/) **Oxford Economics**: [Identifying future manufacturing hot spots in Japan and their implications](https://www.oxfordeconomics.com/resource/identifying-future-manufacturing-hot-spots-in-japan-and-their-implications/) **CNBC**: [This industrial giant could be about to break out, charts show](https://www.cnbc.com/2025/06/05/this-industrial-giant-could-be-about-to-break-out-charts-show.html) ### 04\. Realignment of global supply chains amid deglobalization As complexity in international supply chains unravels, industries face significant risks, even within the US. This shift is a result of increasing geopolitical tensions and a pivot towards regional resilience, raising the need for businesses to adopt new strategies for navigating this altered global landscape. **Related reading:** **Financial Times**: [Companies find themselves caught in deglobalisation crossfire](https://www.ft.com/content/6223ac18-d2a4-4cc1-8265-a340197e274f) **Macrobond**: [Navigating shifts in global and regional financial trends](https://www.macrobond.com/insights/charts-of-the-week/navigating-shifts-in-global-and-regional-financial-trends) **CNBC**: [CNBC's The China Connection newsletter: Whispers of a new world order in Tianjin](https://www.cnbc.com/2025/06/25/cnbcs-the-china-connection-newsletter-whispers-of-a-new-world-order.html) ### Four Ideas (Energy edition): Energy Grid Transformation and Storage Innovation URL: https://paulkedrosky.com/four-ideas-energy-edition-energy-grid-transformation-and-storage-innovation/ Last updated: 2025-07-04T11:47:13.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Commodities Outlook: Metals, Geoeconomics, and Supply Chains URL: https://paulkedrosky.com/four-ideas-commodities-edition-commodities-outlook-metals-geoeconomics-and-supply-chains/ Last updated: 2025-07-03T11:47:16.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): AI Transformations, FDA Device Approvals, Health Tech Investments, AI in Health Systems URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-ai-transformations-fda-device-approvals-health-tech-investments-ai-in-health-systems/ Last updated: 2025-07-02T11:47:01.000Z _This post is for paying subscribers only._ ### On Endowments and Unintended Consequences URL: https://paulkedrosky.com/on-endowments-and-unintended-consequences/ Last updated: 2025-07-01T16:39:35.000Z ## Updates and Erasures: Earlier this week, I wrote about the credit risks of [AI data centers](https://paulkedrosky.com/weekend-reading-plus-spvs-meta-and-fiber-buildout-2-0/) becoming an asset class, and I had hardly hit Publish before I saw [this II piece](https://www.institutionalinvestor.com/article/insatiable-search-compute-power-could-drive-another-commodity-boom) on how ... data centers would make a terrific new commodity. So, you know, the future comes at you fast. ## Rough Notes: ### On Taxes, Endowments, and Unintended Consequences > "Make no mistake, what’s going on with endowments in this country is equivalent to the financial crisis of 2008.” > [University of California CIO Jagdeep Singh Bachher](https://www.pionline.com/endowments-and-foundations/university-california-cio-jagdeep-singh-bachher-sees-systemic-problem/) Many top U.S. universities, despite having multi-billion-dollar endowments, are now facing a **liquidity crunch**. Much like happened to banks in 2008, during the financial crisis, schools are **being forced to sell illiquid assets**, usually private equity and real estate, at lower-than-market prices. The reason? A massive loss of income from cancelled or delayed research grants from NIH and NSF, and the threat of higher taxes in the future. This may turn out to be **a dress rehearsal** for a future, larger liquidity crisis. To understand why, consider the unintended consequences of the current Congressional spending bill. It would, among uncountable other things, **increase the endowment tax** from roughly 1.4% to rates as high as 21% for the largest institutions. That change would dramatically cut net income, making less available for operations. In response, endowment managers are already trying to get liquid, which isn't always easy when you're holding **illiquid private equity** in huge amounts, 10% or more of many portfolios. The results "forced sale" discounts can be sharp. [Yale, Harvard, and others](https://www.barrons.com/articles/yale-harvard-university-endowment-private-equity-b54dee2f?utm%5Fsource=chatgpt.com) are now selling **private equity stakes** at up to 13% off in some cases to raise money. It doesn't help that private equity overpaid for assets in the 2021-23 period, and can't exit and distribute cash to investors. As a result, managers are holding onto assets longer, starving investors for cash. (This has, entertainingly, given rise to ["continuation funds"](https://www.chicagobooth.edu/research/stigler/research/-/media/5d46328c68e0466b9787f42d98275f3b.ashx), where private equity managers create a new vehicle and then, in effect, sell the assets to themselves, thus giving investors some liquidity. I'm not kidding.) With large portions of their portfolios in private equity and credit, endowments are becoming **asset-rich but cash-poor**. In addition to these forced sales of illiquid assets, many, for example, are issuing [taxable bonds](https://www.bloomberg.com/news/articles/2025-05-05/mit-joins-deluge-of-colleges-selling-bonds-amid-funding-threats) to meet capital calls and spending obligations, thus further increasing their tax exposure. Of course, and perversely, schools that weren't as **exposed to private equity** as they might have liked are newly [crowing](https://www.wsj.com/articles/university-of-california-system-to-review-private-markets-exposure-880bf842?utm%5Fsource=chatgpt.com) about it. Suddenly not having the right contacts to put a few billion into a hot private credit fund doesn't seem as bad as it once did. But now, along comes the new **excise tax on larger endowments**. By being exposed to higher taxes, guess what that will do? It will push them to less tax-exposed investments: away from hedge funds, index funds, and equity markets—and toward illiquid private equity, venture capital, and credit funds. Those private funds, given their longer hold times, are generally taxed at lower rates than their public market counterparts, and **tax minimization is the new name of the endowment game**. ![](https://paulkedrosky.com/content/images/2025/07/_--visual-selection.png) But taxing endowment income more heavily **doesn’t discourage risk**—it redirects it. To preserve net returns, institutions **shift deeper into illiquid** private assets where after-tax outcomes look better. But this raises **systemic fragility**: in a downturn, as we're seeing with NIH cuts, these assets can’t be sold without steep discounts, forcing distressed sales at scale. Just like in 2008, liquidity vanishes precisely when it’s needed most. We’ve made the **next crisis more likely**, and we’ve hidden it in private markets. ## Sketches: Amazon now has [more robots than people](https://www.wsj.com/tech/amazon-warehouse-robots-automation-942b814f?st=rJzD6D&reflink=desktopwebshare%5Fpermalink) in its factories ... [AI power consumption](https://www.ft.com/content/deab6a77-12f2-4b66-a9f7-2625654fb493) is outcompeting smelters ... Small groups of [proto-humans kept migrating ](https://www.newscientist.com/article/2484740-the-remarkable-tale-of-how-humans-nearly-didnt-conquer-the-world/?utm%5Fcampaign=RSS%7CNSNS&utm%5Fsource=NSNS&utm%5Fmedium=RSS&utm%5Fcontent=home)out of Africa, getting cut off, and going extinct ... Ukraine is inching towards [robot-on-robot fighting](https://www.economist.com/europe/2025/06/26/ukraine-is-inching-towards-robot-on-robot-fighting) ... The [creativity](https://arxiv.org/pdf/2412.20292) of image general models is weirdly mathenatically explicable ... Hot [hedge fund strategy](https://www.bloomberg.com/news/articles/2025-06-29/hedge-fund-strategy-built-on-catastrophes-taps-a-hot-new-trend?srnd=homepage-europe) built on insuring the catastrophes that insurers don't want to insure ... [Earth's energy imbalance](https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2024AV001636) more than doubled in recent decades, far more than expected ... Subtle changes in how you ask AI chatbots [medical questions](https://dl.acm.org/doi/10.1145/3715275.3732121)—even typos or use of exclamations—dramatically changes the response ### Four Ideas (AI edition): AI Expansion Shaping Diverse Economic Sectors URL: https://paulkedrosky.com/four-ideas-ai-edition-ai-expansion-shaping-diverse-economic-sectors/ Last updated: 2025-07-01T11:47:29.000Z _This post is for paying subscribers only._ ### SPVs, Credit, and AI Datacenters URL: https://paulkedrosky.com/weekend-reading-plus-spvs-meta-and-fiber-buildout-2-0/ Last updated: 2025-06-30T13:37:15.000Z ## Updates and Erasures: Last week I [wrote](https://paulkedrosky.com/midnight/) about how AI goes through three phases: - AI imitates us - We and AI co-evolve - We imitate AI A recent [Bloomberg piece](https://www.bloomberg.com/news/articles/2025-06-27/as-ai-infiltrates-call-centers-human-workers-are-being-mistaken-for-bots?srnd=homepage-americas) was a reminder that this is already happening, and there are unexpected implications: Call center employees are increasingly irritated at being accused of being AI bots. > By the time Jessica Lindsey’s customers accuse her of being an AI, they are often already shouting. For the past two years, her work as a call center agent for outsourcing company Concentrix has been punctuated by people at the other end of the phone demanding to speak to a real human. Sometimes they ask her straight, ‘Are you an AI?’ Other times they just start yelling commands: ‘Speak to a representative! Speak to a representative!’ To adapt Kurt Vonnegut, and so it goes. ## Rough Notes: ### AI Data Centers as Credit Risk I was struck late this past week by [**Meta's rumored $29b fundraising**](https://www.ft.com/content/aff1a2d2-d58e-44de-a114-9f0ce9d15a15) for a rapid buildout of more AI data centers. The company is supposedly talking to various private equity firms, looking to structure it as $26 billion in debt, $3 billion in equity. This will buy you a lot of Nvidia GPUs. But everyone is buying a lot of Nvidia GPUs, or Google ones, as the following WSJ graph shows. More were bought in the first quarter of this year than in any full year before 2021. ![](https://paulkedrosky.com/content/images/2025/06/2025-06-28_20-27-01.png) But let's return to Meta's AI datacenter spending, because it is instructive. A friend asked me, "Why do that? Don't they have the money?" And that got me thinking. Yes, they do, but that "having the money" doesn't matter illuminates the current moment in instructive ways. Consider this from the FT [article](https://www.ft.com/content/aff1a2d2-d58e-44de-a114-9f0ce9d15a15): > Private investment groups have increasingly been pitching investment grade corporations on alternative financings to traditional corporate bonds or loans. Such deals, including the Intel transaction, are often structured as special purpose vehicles or joint ventures, where the asset managers take a large minority ownership share in the vehicle. The company contributes assets to the venture in exchange for the capital — either in debt or equity — that private investment firms provide. There is a lot here, so let's unpack it. It's saying that companies like Meta, which can raise money from banks at low rates any time they want to, increasingly choose ... not to. Instead, they turn to private investment groups—private equity, essentially—who can create **custom financing** for the project. And for which the company pays a **significant premium over investment grade interest rates**. How much more? As much as 200-300 basis points, or 2-3%. This is a **juicy return** on investment-grade company debt. So, why would an investment-grade company agree to do that? They do it because the **capital needed for these buildouts** is so large that doing it with orthodox balance sheet debt, or by issuing sufficient equity, let alone spending your cash, would make a mess of your balance sheet. By structuring it this way, via **special purpose vehicles (SPVs)** in which they have joint ownership, companies like Meta don't have to show the debt as *their* debt. It is the debt of *those* guys over there, that SPV. Not us. Granted, they retain shared control, and they get to use the AI data center, and nothing there happens without their say-so, but still. It's not ours. This is accounting trickery, of course. It is a transparent attempt to raise large amounts of money without balance sheet damage by putting the debt in a vehicle you indirectly control, but that, for accounting reasons, doesn't have to be disclosed as your debt on your balance sheet. The accounting term of art is **"control without consolidation"**. Now, turn the lens around and look at it from the standpoint of private credit providers. Their problem is that they have a **lot of capital**, but they want higher yields on it—and they want to get that yield without taking more risk. Their solution: create customized financings for investment-grade companies like Meta. They can charge higher rates, but still feel confident that they will be repaid. At the same time, an increasing percentage of private credit providers are funded, in part, by **controlling interests in insurance companies**, whose capital they use to fund investments. Finding investments that generate higher yields without higher risk—lending at above-market rates to companies like Meta—is exactly what they want to get higher yield while not running afoul of insurance regulators. And while it all makes perfect sense as financial engineering, **this is where the risks start**. Why? Because this system creates a powerful incentive loop between structurally overcapitalized insurers, return-hungry private equity firms, and mega-cap companies trying to avoid looking like they're leveraging up. Everyone gets what they want—until something breaks. There are three main issues: 1. **Opaque risk**. The illusion of off-balance-sheet financing makes risk murky. Yes, the SPV isn’t *technically* on Meta’s balance sheet, but for all practical purposes, it exists to serve Meta, is controlled by Meta, and its default would materially impact Meta. Ratings agencies are not blind to this, but credit markets often behave as if these are truly arm’s-length deals. 2. **Overbuilding**. With capital artificially cheap and risk mismeasured or opaque, the market signal is: *Build more*. AI data centers, like telecom fiber before them, are highly capital-intensive, and the long-term returns remain speculative. The fact that equity is a small slice of the financing (in this case, reportedly only $3B of $29B) further amplifies the leverage risk. This creates a capital stack that’s thinly cushioned if AI ROI assumptions go awry: a 10% equity buffer is wildly insufficient if projected AI workloads stall or margins compress. 3. **Asset-liability mismatch**. Insurers funding these SPVs via their general accounts face what's often called asset-liability mismatch risks. Insurance regulation assumes things that may not hold in a downturn—long-duration liabilities matched with liquid, diversified assets, not with illiquid concentrated bets—or if AI capex is too high. And yet, the rating agencies let it pass because the borrower is Meta, not a pre-revenue startup. To add to the second point, here is a rough estimate of AI data center *now* vs telecom spending *then* during the dot-com bubble. **We are already spending more now than then**, driven by the same sort of **hyperpolic projections** of future usage and spending. They may turn out to be correct, but if the returns aren't there quickly, the thin equity cushion above datacenter SPV debtholders will be speedily wiped out. ![](https://paulkedrosky.com/content/images/2025/06/90bed046-f133-4dbf-b921-355bb22b4052.png) ### Conclusion This epic AI data center spending, partly on the back of f**inancial engineering,** will work until it doesn't—and when it doesn't it could be a *very* big mess. Granted, not a mess on the scale of the global financial crisis after the housing bubble, but that is perhaps only because no one has yet had the bright idea of rolling up cash flows from SPV-controlled data centers and syndicating them. Maybe let's not suggest that. Meanwhile, there is a new risk regime growing in front of us, and, as usual, it is in the **empty spaces between regulations**, at the intersection of non-bank finance and AI data centers. It will grow rapidly, and if something breaks, damaging insurance assets, people will wonder why they went along with using home and life insurance to pay for AI data centers. Is this GFC 2.0? No, not yet. This is not systemic risk in the mortgage-backed sense. But the components are familiar: **leverage hidden in plain sight**, **mispriced risk**, and capital chasing yield through increasingly convoluted structures. We’ve seen how that story ends. **Collateralized AI Obligations**, anyone (CAOs)? I kid ... I hope. ## Rougher Notes: - "Think about the AI revolution as [a wave of billions of AI immigrants](https://www.reddit.com/r/singularity/comments/1lid8a7/yuval%5Fnoah%5Fharari%5Fsays%5Fyou%5Fcan%5Fthink%5Fabout%5Fthe%5Fai/). They don't need visas. They don't arrive on boats. They come at the speed of light. They'll take jobs. They may seek power. And no one's talking about it." - Optimists think there is an [industrial explosion ahead](https://www.lesswrong.com/posts/Na2CBmNY7otypEmto/the-industrial-explosion) if robots start building robots - [Sleeping through the night](https://journals.sagepub.com/doi/10.3233/JAD-240154?url%5Fver=Z39.88-2003&rfr%5Fid=ori:rid:crossref.org&rfr%5Fdat=cr%5Fpub 0pubmed) is a modern invention—we used to sleep and wake in cycles all night—and it may be breaking our brains. - The supply chain for [zero-day cyber exploits](https://www.atlanticcouncil.org/in-depth-research-reports/report/crash-exploit-and-burn/) is thoroughly screwed up - [Wall Street is so over AI](https://www.bloomberg.com/news/articles/2025-06-25/wall-street-pitches-sci-fi-etfs-for-robots-ufos-quantum-bets?srnd=phx-markets), with even wilder thematic ETFs all the rage. Sci-Fi! Robots! UFOs! Quantum! - Second-gen GLP-1s may [halve the risk of developing dementia](https://journals.sagepub.com/doi/10.1177/13872877251351329) among type 2 diabetics - Two workers at [SEC's Edgar charged with insider trading](https://www.bloomberg.com/news/articles/2025-06-28/two-workers-for-sec-s-edgar-system-charged-with-insider-trading) - [Betting on catatrophes](https://www.bloomberg.com/news/articles/2025-06-29/hedge-fund-strategy-built-on-catastrophes-taps-a-hot-new-trend?srnd=homepage-europe) is increasingly popular and profitable, which should be a bleak joke, but isn't - Seemingly AI-generated band [The Velvet Sundown have](https://www.nme.com/news/music/seemingly-ai-generated-band-the-velvet-sundown-have-over-400000-monthly-spotify-listeners-3874381) over 400,000 monthly Spotify listeners ### Four Ideas (Macro edition): Geoeconomic Shifts, Emerging Markets, Tariffs, Japanese Manufacturing URL: https://paulkedrosky.com/four-ideas-macro-edition-geoeconomic-shifts-emerging-markets-tariffs-japanese-manufacturing/ Last updated: 2025-06-30T11:47:08.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-35/ Last updated: 2025-06-29T11:46:57.000Z **JUNE 29, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** - **China** influences Tibetan education, culture from age 4 ([WSJ](https://wsj.com)) - **Market rally** extends beyond big tech ([WSJ](https://wsj.com)) - **Stocks** face five risks impacting outlook ([Bloomberg](https://bloomberg.com)) - **Supermarkets** risk fines over healthy eating targets ([Bloomberg](https://bloomberg.com)) - **US multinationals** get tax reprieve after G7 deal ([FT](https://ft.com)) - **Meta** focuses on new platform developments ([FT](https://ft.com)) - Squash player **Mostafa Asal** faces rules controversy ([NY Times](https://nytimes.com)) - **Gen Z** excels in retirement savings ([NY Times](https://nytimes.com)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/06/2025-06-26_06-48-03.png) Consumer sentiment in June 2025 plummeted to its lowest point since 2019, indicating a significant downturn in economic confidence. This sharp decline underscores potential economic challenges, impacting consumer behavior and economic forecasts. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/06/2025-06-25_20-47-56.png) US SPAC issuances surged to over $8 billion this quarter, the highest since early 2021\. This resurgence contrasts sharply with the negligible activity in intervening years, signaling renewed investor interest and potential market optimism. --- **THE IDEAS:** **Mentions:** GOOGL, SPGSustainable1, FREIGHT, MDLZ, ELI ### 01\. AI Triage Tools Transform Healthcare Operations AI tools in healthcare are expanding, leading to improved efficiency and patient outcomes. Hospitals like UMass Memorial are deploying AI triage systems, which offer quicker, data-driven decisions. This trend is expected to grow, affecting hospital operations, reducing wait times, and increasing treatment accuracy over the next two years. **Related reading:** **Becker's Hospital Review**: [UMass Memorial expands AI triage tool to 5 more hospitals](https://www.beckershospitalreview.com/healthcare-information-technology/ai/umass-memorial-expands-ai-triage-tool-to-5-more-hospitals/) **Financial Times**: [Disrupted or displaced? How AI is shaking up jobs](https://www.ft.com/content/5009fd1e-85db-433f-aa2b-55d9b88b6481) ### 02\. Geopolitical Alliances Reshape Global Trade Networks Shifts in geopolitical alliances are altering trade patterns, with countries increasingly trading within alliances rather than across rival lines. This could stabilize supply chains within blocs but risks isolating certain economies. The implications are significant for sectors reliant on global trade, prompting shifts in sourcing and logistics strategies. **Related reading:** **Freight Waves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) **Financial Times**: [Welcome to the new age of geoeconomics](https://www.ft.com/content/daf5ce11-c95b-4964-8df1-7463f0f9a81b) ### 03\. Sustainability Megatrends Reshape Business Strategies Sustainability concerns are evolving into major business megatrends, driving fundamental changes in strategy and operations across industries. Firms are realigning to tackle climate issues, which could redefine competitive advantages and offer long-term growth opportunities in emerging sustainable technologies. **Related reading:** **Twitter**: [RT For 5 years, we've tracked #sustainability trends shaping business. In 2025, we’re zooming out to explore the megatren…](https://x.com/SPGSustainable1/status/1879567991945265553) **Oxford Economics**: [Tech spend: Driving growth and resilience through economic downturns](https://www.oxfordeconomics.com/resource/tech-spend-driving-growth-and-resilience-through-economic-downturns/) ### 04\. AI Revolutionizes Travel Marketing Strategies Google's AI transformation is significantly disrupting travel marketing. As AI tools advance, travel companies must adapt to new marketing tactics, likely boosting efficiency in connecting with consumers. This shift will lead to innovative consumer engagement and improved targeting, reshaping the travel industry's marketing landscape. **Related reading:** **Skift**: [Google’s AI Shift: Travel Marketing Disruption Is Coming Fast](https://skift.com/2025/06/09/googles-ai-shift-travel-marketing-disruption-is-coming-fast/) **Twitter**: [As generative #AI sweeps across industries and the #consumer landscape, much attention is being paid to the level of adoption of these new tools and how they are being used by consumers and businesses.](https://twitter.com/SPGlobal/status/1860035529909080480) ### AI as Immigrants URL: https://paulkedrosky.com/ai-as-immigrants/ Last updated: 2025-06-27T15:38:20.000Z "You can think about the AI revolution as a wave of billions of AI immigrants. They don't need visas. They don't arrive on boats. They come at the speed of light. They'll take jobs. They may seek power. And no one's talking about it." —Yuval Noah Harari ### Four Ideas (Energy edition): Energy Storage, AI in Energy, Clean Energy Politics, Renewable Partnerships URL: https://paulkedrosky.com/four-ideas-energy-edition-energy-storage-ai-in-energy-clean-energy-politics-renewable-partnerships/ Last updated: 2025-06-27T11:47:16.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Silver & Platinum Surge, Geoeconomic Shifts, Oil Volatility, Commodity Investments URL: https://paulkedrosky.com/four-ideas-commodities-edition-silver-platinum-surge-geoeconomic-shifts-oil-volatility-commodity-investments/ Last updated: 2025-06-26T11:47:08.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): AI Transformations in Healthcare URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-ai-transformations-in-healthcare/ Last updated: 2025-06-25T11:47:01.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): Non-obvious AI Trends Reshaping Markets URL: https://paulkedrosky.com/four-ideas-ai-edition-non-obvious-ai-trends-reshaping-markets-2/ Last updated: 2025-06-24T11:47:07.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Interest Rates, Energy Transition, Geopolitics, Freight Market URL: https://paulkedrosky.com/four-ideas-macro-edition-interest-rates-energy-transition-geopolitics-freight-market/ Last updated: 2025-06-23T11:47:28.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-34/ Last updated: 2025-06-22T11:46:55.000Z **JUNE 22, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** - **Zuckerberg** receives a macho-man makeover ([FT](https://ft.com)) - **Tesla** robotaxi ambitions face reality check ([FT](https://ft.com)) - **Media industry** clash unfolds at Cannes ([NY Times](https://nytimes.com)) - **Tariffs** raise can production costs ([NY Times](https://nytimes.com)) - **Iran imagery** reveals rush to export oil globally ([Bloomberg](https://bloomberg.com)) - **ICE raids** scare workers from job sites ([Bloomberg](https://bloomberg.com)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/06/2025-06-19_13-56-29-1.png) Palm Beach home values surged from $20 billion in 2020 to nearly $50 billion in 2024, over doubling within four years. Such rapid growth impacts affordability, housing availability, and reflects larger economic shifts, potentially driven by migration trends, investment influx, and regional development policies. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/06/2025-06-19_13-56-38-1.png) Home market values in Palm Beach have increased nearly threefold compared to property taxes since 2018\. This discrepancy, driven by a state assessment cap, highlights a growing gap affecting local tax revenue and housing affordability. --- **THE IDEAS:** **Mentions:** GOOGL, ARAMCO, SPG, FT, S&P ### 01\. AI reshapes travel marketing strategy Google's AI shift is rapidly disrupting travel marketing, shifting focus from traditional to AI-driven models. This transformation can impact how travel businesses allocate marketing budgets and target key demographics, altering competitive dynamics within the industry. **Related reading:** **Skift**: [Google’s AI Shift: Travel Marketing Disruption Is Coming Fast](https://skift.com/2025/06/09/googles-ai-shift-travel-marketing-disruption-is-coming-fast/) **FT**: [Disrupted or displaced? How AI is shaking up jobs](https://www.ft.com/content/5009fd1e-85db-433f-aa2b-55d9b88b6481) ### 02\. Renewable energy accelerates global transition The adoption of renewable energy is accelerating globally due to technological advancements and climate change-driven policies. This trend could lead to a decline in fossil fuel dependence and reshape energy markets, benefiting companies in renewable sectors. **Related reading:** **PowerMag**: [The Great Shift: Navigating the Global Energy Transition](https://www.powermag.com/the-great-shift-navigating-the-global-energy-transition/) **PowerMag**: [2025’s Energy Crossroads: 6 Trends Redefining the Global Power Sector](https://www.powermag.com/2025s-energy-crossroads-6-trends-redefining-the-global-power-sector/) ### 03\. European bank M&A faces alignment challenges Despite potential benefits, European banks face significant hurdles in mergers and acquisitions due to regulatory complexities and strategic misalignment. Progress may occur if conditions align, potentially reshaping the financial landscape. **Related reading:** **FT**: [European bank M&A is hard, but sometimes stars align](https://www.ft.com/content/2e2de734-1d07-467f-b9d8-0227594e92a4) **S&P Global**: [From Our banking risk experts share insights on key events shaping the #FinancialSector in emerging markets](https://x.com/SPGlobal/status/1933034044339405055) ### 04\. Shifting geopolitical trade partnerships Geopolitical partners are increasingly trading more with allies and less with rivals. This shift can impact global supply chains, influencing trade policies and economic strategies. **Related reading:** **FreightWaves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) **FT**: [Welcome to the new age of geoeconomics](https://www.ft.com/content/daf5ce11-c95b-4964-8df1-7463f0f9a81b) ### Weekend Reading: Who's Weird? Maybe We're Weird URL: https://paulkedrosky.com/weekend-reading-whos-weird-maybe-were-weird/ Last updated: 2025-06-20T22:16:17.000Z ### the tl;dr - The neoliberal consensus of the past 40 years is a historical aberration, not a natural progression. - Economic models’ repeated failures are due to blind spots around non-market variables like war, tech disruption, AI, and climate risk. Like many of you, I think, I feel increasingly unmoored, and it is often unclear what, for want of a better word, still *works*. As physicist Stephen Hawking famously wrote, "Science will win because it works". That was always my implicit [demarcation](https://en.wikipedia.org/wiki/Demarcation%5Fproblem) line between facts and nonsense, the idea that good ideas based on facts work, and thus win out, eventually (and eventually can be very long), whether in science, capital markets, or elsewhere (mostly). But, but ... we are increasingly finding out, across a host of domains, that maybe we over-extrapolated from limited evidence. Just because something works, that it fits the facts, however briefly, or makes a lot of people a lot of money, doesn't make it a law of nature, economic or otherwise. It is easy to fall into wave theory traps here, becoming a naive theoretician throwing around [Fernand Braudel's](https://en.wikipedia.org/wiki/Fernand%5FBraudel) ideas about cycles of capital accumulation, or [Giovanni Arrighi's](https://www.versobooks.com/products/1483-the-long-twentieth-century) compelling expansion thereof. Those, too, are laws, albeit meta-laws, and we are therefore right to be suspicious of their claims, however appealing. We are left with ... questions. ### **What works? And why? Who is the weird one?** Here is Financial Times columnist Gillian Tett on the topic in a recent Odd Lots [episode](https://open.spotify.com/episode/4UgWETPusT90EonV5rnvI8) on complexity and risk in finance. She makes many terrific points (with my emphasis added): > \[The current economic\] mindset arose from a combination of **three factors**. One was the growth of **neoliberal free market ideas** after the 1970s and 80s. The second was the **explosion in computing power**. And the third was the **explosion in the financial industry**. > And those three things came together to create a voracious demand to use the new digital tools in computing to model finance and free markets, supposedly free markets, to create a set of instruments that the fast-expanding ranks of financial professionals could use to price securities, predict the future, place trading bets, et cetera, around the world, **backed by groups like the CFA, which were a bit like the American financial equivalent of the Catholic Church**, in that they went around the world evangelizing and spreading the creed right around the world. > Now that was the mindset which we all grew up with—and an anthropologist would say **we're all creatures of our own cultural environment—**it seems to us to be normal natural and inevitable. It doesn't just use computing power to predict the future; it's also marked by a certain amount of **tunnel vision** because it assumes that if you put the right inputs into an economic model or put the information that matters onto a balance sheet of a compass you've basically got the key to model and forecast what's going to happen next. > And in some ways, that worked really well. But the problem is that **there are always things that you leave out of your economic model**. There are always things that you leave out of your balance sheet or just footnotes. And those things tend to be things like politics, social conflict, tech change, environmental risk, medical risk, or what groups like the World Economic Forum. interstate conflict, better known as war. > And t**he story of the last two decades is that everything that wasn't on the balance sheet or in the model is what's really blown up everyone's forecasts** and become increasingly important. > So what we're seeing now, in my view, is **the fifth big swing in the intellectual zeitgeist since 1900**. And by that, I mean that up until 1914, you had **imperialism**. imperialist free market capitalism in the world. Then, between the wars, you had **protectionist**, populist, and nationalist visions of the economy. Then, after World War II, you essentially had **Keynesianism**, the idea that the state could jump in and direct things for the good of all. And that was sort of international Keynesianism. Then you had the **neoliberal** age, starting in the 1980s. > And now you've got a swing of the pendulum back towards effectively geoeconomics: the world's woken up and discovered what they always knew back in the 1920s. And they also sort of knew back in the Keynesian period, which is that **power matters; politics matters**. > > \[T\]he period of time that we all grew up with when free market ideals were taken for granted is actually a historical aberration—if you look across societies and most points of history, it's not the case that the world we're moving into now is weird. **We were the weird ones for the last 40 years**. This is me, Paul, back again. Tett's powerful and important points are worth heeding. I am reminded of author Douglas Adams writing about a sentient puddle: > This is rather as if you imagine a puddle waking up one morning and thinking, 'This is an interesting world I find myself in — an interesting hole I find myself in — fits me rather neatly, doesn't it? In fact it fits me staggeringly well, must have been made to have me in it!' This is such a powerful idea that as the sun rises in the sky and the air heats up and as, gradually, the puddle gets smaller and smaller, frantically hanging on to the notion that everything's going to be alright, because this world was meant to have him in it, was built to have him in it; so the moment he disappears catches him rather by surprise. I think this may be something we need to be on the watch out for.” > ― **Douglas Adams,** [The Salmon of Doubt: Hitchhiking the Galaxy One Last Time](https://www.goodreads.com/work/quotes/809325) I think we have some things to watch out for as we try to re-discover what *works* and what is *transient* in a post-Keynesian world of AI, tribalism, and zero-sum thinking. ### Four Ideas (Energy edition): Energy Transition, AI in Energy, EVs, Gulf Renewable URL: https://paulkedrosky.com/four-ideas-energy-edition-energy-transition-ai-in-energy-evs-gulf-renewable/ Last updated: 2025-06-20T22:14:56.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Commodities Trends and Strategic Insights URL: https://paulkedrosky.com/four-ideas-commodities-edition-commodities-trends-and-strategic-insights/ Last updated: 2025-06-19T11:47:16.000Z _This post is for paying subscribers only._ ### Life Under Two: Debt, Deficits, and the AI Discontinuity URL: https://paulkedrosky.com/life-under-two/ Last updated: 2025-06-18T14:35:19.000Z ### the tl;dr - The U.S. economy is undergoing a fundamental shift from over 3% average annual growth to "life under two": at or below 2% annually. - This slowdown underlies rising populism, fiscal strain, anti-immigrant sentiment, and distrust in institutions. - The shift from abundance to scarcity is altering political, economic, and psychological frameworks across American life. - The U.S. could accept this shift, and adjust accordingly, but there is an outlier chance it avoids it altogether --- This is mostly about GDP and economics, so let's get something important out of the way first: **Economics is boring as fuck**. Again: **Economics is boring as fuck**. It just *is*. People who pretend others are generally undersocialized (if bright) young males who like to play Settlers of Catan more than is healthy. So ... economists. But the following topic is too important to be left to economists. ### The Crackup The American economy is facing a quiet crackup that's behind much of the frustration and anger among its citizens. For decades, the United States averaged GDP growth rates above 3% annually. Today, 2% growth—or less—is more likely, now and in the near future. ![](https://paulkedrosky.com/content/images/2025/06/2025-06-17_15-18-03-1.png) The Congressional Budget Office projects similar GDP growth over the next decade. ![](https://paulkedrosky.com/content/images/2025/06/2025-06-17_17-45-15.png) This shift from robust to modest growth affects ... everything. It is behind, in large part: - The rise of populism - Anti-immigrant sentiment - DOGE - A return to tariffs - Soaring U.S. deficits and debt ### The Psychology of Slower Growth The difference between 3% and 2% annual growth is stark. At the higher rate, over ten years, you have: - **$3.36 trillion larger economy** - **$3.7 trillion more in tax revenue** - **6 million more jobs** - **Significantly higher real incomes** The gap directly affects opportunities, wage growth, tax revenue, social services, health care, solvency, and optimism about the future. Growing quickly and being awash in income makes people optimistic about the future; the opposite does ... the opposite. Americans built their self-image and expectations around decades of strong growth, trusting each generation would surpass the previous one. Sub-2% growth erodes this assurance, shifting psychology from abundance toward scarcity, and even zero-sum thinking. When people don't see abundance they stop sharing. And Americans have stopped sharing. > **Zero-sum thinking**: A general belief system about the antagonistic nature of social relations, shared by people in a society or culture and based on the implicit assumption that a finite amount of goods exists in the world, in which one person's winning makes others the losers, and vice versa ... a relatively permanent and general conviction that social relations are like a zero-sum game. People who share this conviction believe that success, especially economic success, is possible only at the expense of other people's failures. [More](https://en.wikipedia.org/wiki/Zero-sum%5Fthinking). ### Spending Cut Theater Lower growth explains rising political polarization. When growth was strong, expansive revenues funded both new programs and tax cuts. Rising wages benefited workers and shareholders alike, making immigration economically beneficial. Under slower growth, policies become inherently competitive. Tax cuts reduce funding for public programs. Wage increases for some imply job losses for others. Immigrants are seen as competition for jobs, even when they are not. Populist movements reflect this zero-sum thinking. Frustrations over trade, immigration, and government spending stem from the palpable sense that the economy isn't working the way it once did. > DOGE could only be theater Creating DOGE was one response to slow growth. After all, lower growth demands less government spending. But given that most U.S. spending is statutory—it cannot simply be cut—DOGE could only be theater. Consider the following graphic of U.S. federal spending. What should be clear is that given mandatory spending and defense, you can only touch about 14% of the budget. And, just for the sake of argument, even if you cut that to zero—trimming $800 billion (which is impossible)—the U.S. would still be running a deficit. ![](https://paulkedrosky.com/content/images/2025/06/2025-06-17_18-11-04.png) While the instinct to cut costs is understandable, and perhaps even logical, there is no way for the U.S. to cut its way to break-even. Any serious effort to reform spending would have to involve defense, healthcare, and social security, and all of those have immense and well-funded constituencies, most of whom vote. Unserious efforts that don't touch those line items are just theater—but they tap into the unease of life in a sub-2% world, so many people don't care. ### Tariffs and Immigration in a Slow-Growth World Restrictions also gain appeal in a low-growth economy. With rapid expansion, foreign competition and immigration offer benefits. Displaced domestic workers can easily transition to new sectors, and immigrants fill labor shortages. But once growth falters, everything goes into reverse. Losing one's job and hoping that the economy can handle it feels like a bad bet. Workers displaced by trade struggle to find comparable opportunities. Communities undergoing rapid demographic shifts feel resource pressure. Immigration restrictions and higher tariffs become politically attractive means to maintain existing conditions. They are zero-sum responses to scarcity, predictable during Life Under Two. > Immigration restrictions and tariffs are predictable, zero-sum responses to scarcity But cutting immigration and expelling illegals has immense consequences, which I will return to in an upcoming note. ### Lotteries and Non-Economic Thinking It should come as no surprise the rise on non-economic thinking predicated on lottery assets, like crypto. Unlike orthodox financial instruments, they don't represent a claim on productive output. They are, if anything, the negation of orthodox claims, a repudiation of the old way of doing things, pure price reflexivity. This is understandable in a world where people have lost faith in economic growth. Why wait? Find things that go up and chase after them. We see this in the rise of crypto, of sports betting, of YOLO-ing meme stock-chasing Reddit bros, and more. What they have in common is impatience in the orthodox system ever working for them. And having lost faith in the system itself, institutional distrust becomes a baked-in feature of what they lust after. > Crypto as ironic commentary on Life Under Two You can think of them as a kind of ironic commentary on Life Under Two itself. Instruments are increasingly divorced from the system, acting almost as jokes in a surrealist play about what happens when nothing economic can happen anymore. They are what happens when money seeks returns divorced from economic foundations, turning inward on itself. We should expect even more of this kind of behavior. ### End-Running the Reform Challenge Slower growth makes reforms harder. There is less room to maneuver. Healthcare costs, for example, continue to rise faster than economic growth, making them an ever-larger chunk of spending just by standing still. Social Security and Medicare do the same. Infrastructure needs investment, yet raising taxes or reallocating spending don't happen in a zero-sum world. Other countries whose growth has slowed have made difficult choices about taxes, entitlements, and economic structures. The United States hesitated, hoping growth would rebound to historical levels, which it has not. And there are virtually no examples in economic history where a major economic power has had a multi-year slowdown, and then returned to growth. > There are virtually no examples in economic history where a major economic power has had a multi-year slowdown, and then returned to growth But "virtually no" examples is not "none". The U.S. did it once before during the dot-com period, when a wave of technology innovation and productivity briefly brought the country back to growth. So, while, unlikely, it's not impossible, and it would cause current sovereign finance forecasts to be badly wrong. ### The Future(s) There are two divergent ways of thinking about the future of the U.S. in this cracup context. One more pessimistic and linear, and one more optimistic, non-linear, but still fraught. **Scenario 1: Rationality and Limits** First, a slower-growing U.S. might be a better global citizen, less central and less convinced of its own rectitude. A multi-polar world could be a safer world, less of an economic, cultural, and security monoculture. The country will struggle with this, convulsing as it attempts to reconcile its beliefs in its own exceptionalism with the reality of lower growth and limits. Under this view, the US would, in a series of stop-start waves, cut spending and reform its obligations, aiming for what has been called the 3% target for deficits as a percentage of GDP. It will be wrenching, with the waves rippling back and forth from politics, to fiscal and monetary policy, and back to politics again. **Scenario 2: Accidential Escape Velocity** There is, however, a second, outlier possibility. There is recent scholarly [work](https://economics.mit.edu/sites/default/files/publications/Demograhics%20and%20Automation.pdf) suggesting that aging, slowing societies are likely to be earlier at adopting AI and robotics. Faced with scarce and expensive labor, they jump straight to these new tools, not having the option of there being humans to do the work. It is analogous to developing markets that skipped over wired telecom and leaped straight to wireless. ![](https://paulkedrosky.com/content/images/2025/06/2025-06-17_19-31-38.png) And robots are improving much faster than is commonly understood. There is the potential for a labor explosion to rival the much-heralded AI intelligence explosion. The combination would explode models in unexpected ways. Could it happen in the U.S.? It could, and in some sense, it is the de facto bet being laid. Americans and their politicians, by their theatrical inaction, are betting that something magic will happen that restarts growth, compensates for lost workers, and helps rebalance the budget. ![](https://paulkedrosky.com/content/images/2025/06/2025-06-18_07-08-19.png) This chance would be wrenching, putting pressure on everything from power grids, to social safety nets, likely requiring some form of universal basic income, and leading to a major restructuring of the workforce. But it is *possible*, and not properly discounted in most models. ### Coda: Living with Less ... or Maybe a Lot More Sustained sub-2% growth would mark a profound shift in American life, and it is already creating unease. It will not be easy for a country built on expanding frontiers, abundance, and growth. > The U.S. has configured itself such that normal doesn't adequately pay the bills, nor satisfy its psyche Political movements—from populism to efficiency campaigns—are early reactions to this reality, however subconscious. There will be more. But life under 2% growth isn't necessarily worse. For most OECD countries it is ... normal. The trouble is that the U.S. has configured itself such that normal doesn't adequately pay the bills, nor satisfy its psyche. > The U.S. may have painted itself, completely by accident, into the very corner it needed to But there is a chance, however remote, that the U.S. could paint itself completely by accident into the very corner it needed to. It is possible that, having denied itself access to labor, cut taxes to unsustainable levels, built huge tariff walls, and maintained outsized spending, the U.S. will once again be on the right side of a new growth wave, this time predicated on robotics and AI. This possibility is under-discounted, even if the change would be wrenching. ### Four Ideas (Life sciences edition): AI-Driven Innovations and Health Tech Partnerships URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-ai-driven-innovations-and-health-tech-partnerships/ Last updated: 2025-06-18T11:47:13.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): Non-obvious AI Trends Reshaping Markets URL: https://paulkedrosky.com/four-ideas-ai-edition-non-obvious-ai-trends-reshaping-markets/ Last updated: 2025-06-17T11:47:01.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Geopolitical Dynamics & Sectoral Impacts URL: https://paulkedrosky.com/four-ideas-macro-edition-geopolitical-dynamics-sectoral-impacts/ Last updated: 2025-06-16T11:47:10.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-33/ Last updated: 2025-06-15T11:46:46.000Z **JUNE 15, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** - **Syracuse** misjudges value; offers $200,000 to teens ([NY Times](https://nytimes.com)) - **U.S. Steel** and Nippon Steel seal partnership agreement ([NY Times](https://nytimes.com)) - **Boeing** anticipates 44,000 new jets, warns of slow output ([Bloomberg](https://bloomberg.com)) - **Amazon** invests $13 billion in Australian data centers ([Bloomberg](https://bloomberg.com)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/06/2025-06-14_17-15-31.png) Consumer sentiment in June 2025 dropped to 60.5 from 74 in December 2024, a decline of 18.2%. Despite softening tariff concerns, confidence remains low, reflecting persistent economic uncertainty. This sharp decrease highlights ongoing challenges in restoring consumer optimism and potential impacts on economic recovery efforts. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/06/2025-06-14_17-16-20.png) Global manufacturing output shows significant growth in 2025, with China leading at 8% compared to the previous 5%. The Eurozone is the exception, reporting a decrease from a 0% change to -1%. This shift highlights China's robust recovery and potential Eurozone challenges in the global economy. --- **THE IDEAS:** **Mentions:** GOOGL, ENPH, NVDA, RIVN, AMZN ### 01\. AI transforming travel industry marketing strategies AI advancements are significantly disrupting travel marketing, with Google’s AI transformation playing a key role. This rapidly evolving landscape forces travel companies to innovate in marketing strategies or risk losing competitive edge. **Related reading:** **Skift**: [Google’s AI Shift: Travel Marketing Disruption Is Coming Fast](https://skift.com/2025/06/09/googles-ai-shift-travel-marketing-disruption-is-coming-fast/) **Financial Times**: [Disrupted or displaced? How AI is shaking up jobs](https://www.ft.com/content/5009fd1e-85db-433f-aa2b-55d9b88b6481) ### 02\. Accelerated renewable energy adoption influences power markets Technological advancements and government directives are causing a rapid increase in renewable energy adoption, potentially reshaping power market dynamics and creating opportunities for companies like Enphase Energy. **Related reading:** **Power Magazine**: [The Great Shift: Navigating the Global Energy Transition](https://www.powermag.com/the-great-shift-navigating-the-global-energy-transition/) **Power Magazine**: [2025’s Energy Crossroads: 6 Trends Redefining the Global Power Sector](https://www.powermag.com/2025s-energy-crossroads-6-trends-redefining-the-global-power-sector/) ### 03\. Freight optimization through data in logistics sector As the logistics industry adapts to economic changes and rising fuel costs, the adoption of data-driven decision-making and flexible strategies is becoming crucial for competitiveness, impacting companies like Amazon and Rivian. **Related reading:** **FreightWaves**: [The Road Ahead: Freight Market Trends & Industry Insights for 2025](https://www.freightwaves.com/news/the-road-ahead-freight-market-trends-industry-insights-for-2025) **FreightWaves**: [Navigating the evolving cross-border environment with Werner](https://www.freightwaves.com/news/navigating-the-evolving-cross-border-environment-with-werner) ### 04\. Emerging regional hubs redefine global economic archetypes The evolution of key regional hubs identified by Oxford Economics is opening new fronts for investment and market development, highlighting opportunities for companies to tap into diverse local economic dynamics in 2025. **Related reading:** **Oxford Economics**: [Top regional hubs to watch in 2025](https://www.oxfordeconomics.com/resource/top-regional-hubs-to-watch-in-2025/) **Oxford Economics**: [Identifying future manufacturing hot spots in Japan and their implications](https://www.oxfordeconomics.com/resource/identifying-future-manufacturing-hot-spots-in-japan-and-their-implications/) ### Midnight URL: https://paulkedrosky.com/midnight/ Last updated: 2025-06-13T19:33:15.000Z ### the tl;dr - In *Doctor Who: Midnight*, an entity mimics and then overcomes the Doctor, flipping roles from follower to controller. - This mirrors AI progression: from parroting human culture (Phase 1), to tightly mimicking and co-evolving with us (Phase 2), to leading and generating original content humans then emulate (Phase 3). - The shift to models trained on synthetic data once raised concerns of “model collapse,” but as models surpass humans self-training is not only viable but superior. - As models take control of language generation, they may begin shaping human culture and mythos directly, risking a loss of human authorship in the foundational narratives of civilization. In the classic Doctor Who episode [*Midnight*](https://tardis.fandom.com/wiki/Midnight%5F%28TV%5Fstory%29), show-runner Russell T. Davies has the Doctor on a tourist ship across the surface of a planet, and the ship breaks down. The planet's sunlight is lethal, so the Doctor and his fellow travelers must wait for help to arrive, which will take hours. As they are sitting there they suddenly hear a knocking on the outside hull. There shouldn't, of course, be anyone alive outside to do the knocking. And then stranger things start happening. First, the ship is rocked with an impact, then something ephemeral moves toward a passenger named Sky, and then she begins cowering. And the knocking continues. The passengers try to determine what's wrong with Sky, at which time she turns and begins imitating what they say to her, and what they say to each other. > DOCTOR: Sky? It's all right, Sky. I just want you to turn around, face me. > (Sky turns slowly and stares into the torchlight.) > DOCTOR: Sky? > SKY: Sky? > DOCTOR: Are you all right? > SKY: Are you all right? > DOCTOR: Are you hurt? > SKY: Are you hurt? At first, it is strange, then maddening, and then it becomes bizarre, as Sky mimics the Doctor as he recites the first 30 digits or so of pi. They try to engage with Sky, to understand what's going on, or what's gone wrong. But there is no use; she keeps imitating. Sky zeroes in on the Doctor, imitating his every word, faster and faster. The time gap between what the Doctor says and what she says becomes tiny. Eventually, the mimicry gaps collapse, and the Doctor and Sky are speaking at the same time. She has learned his cadences, style, and phrasing: they are *phase-locked*. The Doctor and Sky are, verbally, the same person. But that doesn't last long, but suddenly Sky leaps ahead and speaks first. She is imitating the Doctor, *but from the front*. He has become her puppet, > JETHRO: She spoke first. > SKY: Oh, look at that. I'm ahead of you. > DOCTOR: Oh, look at that. I'm ahead of you. > HOBBES: Did you see? She spoke before he did. Definitely. > JETHRO: He's copying her. Having gained control of the Doctor, it now tries to turn things around, blaming the Doctor, in his voice, for everything that has gone wrong. It was him, *always* him, never the "thing", whatever it was before it gained control. > SKY: It killed the driver. > DOCTOR: It killed the driver. > SKY: And the mechanic. > DOCTOR: And the mechanic. > SKY: And now it wants us. > DOCTOR: And now it wants us. --- This Doctor Who episode comes to my mind regularly when thinking about where we are in the current AI frenzy. There are three grand phases. ### Phase 1 Large language models (LLMs) were initially seen as parrots, even if stochastic ones. They were trained on our cultural artifacts, from books, to blog posts, to Reddit threads, to YouTube comments. Having learned from those, models could generate such language fluently, to the point that it is often indistinguishable from professional prose, sometimes able to fool AI detection tools. (Spoiler: The text I tried this on was mostly AI generated). ![](https://paulkedrosky.com/content/images/2025/06/2025-06-12_14-26-12.png) We are already seeing how people are rapidly offloading much of their thinking, purchasing, planning, coding, writing, and more to AIs. This phenomenon has us, charitably, co-evolving with our tools, which isn't historically unusual, even if [cognitive offloading](https://www.cell.com/trends/cognitive-sciences/abstract/S1364-6613%2816%2930098-5) at this scale is unprecedented. (There is already a phrase for its dumbing-down effects: "digital dementia".) The human brain is highly energetically expensive, something like 30% of basal metabolic rate, and it cheerfully unloads anything it can. The river of AI already runs through everything we do. For example, raging torrents of what is often called "AI slop" are flowing, swamping music, video, audio, and text. Some already claim that the Internet itself has already been repopulated, as is postulated in the ["dead internet"](https://en.wikipedia.org/wiki/Dead%5FInternet%5Ftheory) theory, where most of what you think is human online already ... isn't. ### Phase 2 The gap is shrinking, so this is the second phase: rapid and discomfiting mimicry, like Sky. We should expect this to be constant and accelerating, as news stories, memes, and trends become algorithmic triggers for more stories, memes, and trends. Everything will trigger everything in a non-stop loop of interlocking co-evolution, an onanistic ouroboros. ![](https://paulkedrosky.com/content/images/2025/06/2025-06-12_16-02-26.png) ### Phase 3a But it won't stop there. Recall, the third stage in Midnight, after co-evolution, was Sky getting ahead of the Doctor, with him following slavishly behind. > SKY: Oh, look at that. I'm ahead of you. > DOCTOR: Oh, look at that. I'm ahead of you. There is a coming reversal in the training loop. Models originally learned from people, having consumed vast amounts of online data produced by humans over hundreds of years, finding its way onto the public Internet. And then models and people co-evolved in real-time, as is happening now. But It will increasingly be the case that models learn from the data they are creating, often called "synthetic data". While some have argued this reliance on synthetic data will lead to [model collapse](https://www.nature.com/articles/s41586-024-07566-y), Google DeepMind's Demis Hassabis will have none of that. In a [recent presentation](https://www.youtube.com/watch?v=rtd%5F1gxMTZM#t=08m34s), he off-handedly claimed that relying on synthetic data for video training won't cause model collapse. Why? Because *models are rapidly becoming better at video than are humans*. 0:00 /0:09 1× > We know there's a lot of worries about this so-called model collapse. I mean, video is just one thing, but in any modality, text as well ... I don't actually see that as a big problem. Eventually, we may have video models that are so good you could put them back into the loop as a source of additional data ... synthetic data, it's called. > —Demis Hassabis, Google DeepMind ### Phase 3b While it is important to computer scientists that models not suffer indigestion and even death during data [autocoprophagia](https://www.collinsdictionary.com/us/dictionary/english/autocoprophagy) (do *not* look that up before a meal), there is a deeper societal import when models are creating their own data and influencing people and culture. ### Coda As Yuval Harari wrote in his recent [Nexus](https://www.amazon.com/Nexus-Brief-History-Information-Networks/dp/059373422X), losing control of language is to lose control of culture, of civilization, of our myths: "Civilizations are born from the marriage of bureaucracy and mythology". We are handing over control of language and of myths to models, not just learning from them, but following them. Like the Doctor in Midnight, we are following models, transitioning from leader to slavish follower. New society-naking myths are ahead, and many won't be ours. ### Four Ideas (Energy edition): Opportunities in Energy Transition 2025 URL: https://paulkedrosky.com/four-ideas-energy-edition-opportunities-in-energy-transition-2025/ Last updated: 2025-06-13T11:46:56.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Commodities and Energy Transition Innovations URL: https://paulkedrosky.com/four-ideas-commodities-edition-commodities-and-energy-transition-innovations/ Last updated: 2025-06-12T11:47:03.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): AI Transforming Healthcare, Cardiology and Diagnostics URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-ai-transforming-healthcare-cardiology-and-diagnostics/ Last updated: 2025-06-11T11:47:18.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): 4 underexplored AI impacts in diverse sectors URL: https://paulkedrosky.com/four-ideas-ai-edition-4-underexplored-ai-impacts-in-diverse-sectors/ Last updated: 2025-06-10T11:48:10.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): AI's Supply Chain Impact, Gulf-Asia Trade, Tariffs URL: https://paulkedrosky.com/four-ideas-macro-edition-ais-supply-chain-impact-gulf-asia-trade-tariffs/ Last updated: 2025-06-09T11:47:11.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-32/ Last updated: 2025-06-08T11:46:52.000Z **JUNE 08, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** - **US fiscal policy** in disarray, solutions lacking ([FT](https://ft.com)) - **US employment** growth trends analyzed ([FT](https://ft.com)) - **ASML** navigates trade challenges with advanced tools ([NY Times](https://nytimes.com)) - US **tourism** sector concerned about declining vacationers ([NY Times](https://nytimes.com)) - Administration **ends** ‘Quiet Skies’ surveillance program ([WSJ](https://wsj.com)) - **Stocks** jump after new jobs data release ([WSJ](https://wsj.com)) - **S&P 500** gains as jobs report impacts bonds ([Bloomberg](https://bloomberg.com)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/06/2025-06-07_09-50-51.png) DeepMind loses significantly more talent to Anthropic, with a 10.6:1 ratio, compared to the balanced flow with Hugging Face. This highlights Anthropic's strong pull in the AI talent market, potentially influencing industry innovation and collaboration dynamics. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/06/2025-06-06_16-10-54.png) Stablecoin operators now surpass China in short-term US debt holdings, highlighting crypto's growing role in global finance. This shift signals stablecoins' influence, managing around 50% more US debt compared to China's, raising questions about financial stability and emerging economic dynamics. --- **THE IDEAS:** **Mentions:** NASDAQ, REIT, Macrobond, CNBC, Oxford Economics ### 01\. China gains edge via AI in supply chains Chinese factories are integrating AI technology for cost-cutting and quality control, making them more competitive. This trend could increase China's dominance in global supply chains, impacting importers and exporters worldwide. Over time, this could lead to dependency on Chinese goods, and potentially increase inflation for countries relying on these imports. **Related reading:** **CNBC**: [CNBC's The China Connection newsletter: AI integration gives China supply chains an even greater advantage](https://www.cnbc.com/2025/06/04/cnbcs-the-china-connection-newsletter-china-supply-chains-ai-edge.html) ### 02\. US exceptionalism persists in market dynamics Despite changing global economic conditions, US markets remain resilient with AI advancements driving cybersecurity stocks. This trend underlines US economic exceptionalism, suggesting a stable investment climate in the US market. Over the next couple of years, this could preserve investor interest and sustain capital inflow into US equities. **Related reading:** **CNBC**: [A cybersecurity stock to play the AI trend, and why U.S. exceptionalism in markets isn't over](https://www.cnbc.com/2025/06/04/a-cybersecurity-stock-to-play-the-ai-trend-and-why-us-exceptionalism-in-markets-isnt-over.html) ### 03\. Geopolitical realignment reshapes global trade patterns Countries are increasingly trading within geopolitical blocs, reducing reliance on rivals. This realignment could lead to elevated exports among allies and a shift in global trade balances, impacting markets that depend on rival nations. Over the next year, watch for shifts in trade policies and tariff implementations. **Related reading:** **FreightWaves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) ### 04\. Industrial sector recovery signals broader economic rebound Following recent stagnation, the industrial sector is showing signs of recovery. Leading indicators suggest increased production and growth, which may culminate in broader economic improvement. As industrial demand rises, this stimulus could lead to enhanced employment and profit margins, benefiting related sectors over the next 18 months. **Related reading:** **Oxford Economics**: [Industry Key Themes 2025: Industrial landscape at a critical juncture](https://www.oxfordeconomics.com/resource/industry-key-themes-2025-industrial-landscape-at-a-critical-juncture/) ### Four Ideas (Energy edition): Energy Decentralization, Renewables Tariffs, AI in Supply Chains, Energy Storage URL: https://paulkedrosky.com/four-ideas-energy-edition-energy-decentralization-renewables-tariffs-ai-in-supply-chains-energy-storage/ Last updated: 2025-06-06T11:47:27.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Metals, Commodities, Trading Margins, Cocoa URL: https://paulkedrosky.com/four-ideas-commodities-edition-metals-commodities-trading-margins-cocoa/ Last updated: 2025-06-05T11:47:00.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): AI's Expanding Role in Healthcare and Technology URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-ais-expanding-role-in-healthcare-and-technology/ Last updated: 2025-06-04T11:47:01.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): AI Trends Impacting Market Landscape URL: https://paulkedrosky.com/four-ideas-ai-edition-ai-trends-impacting-market-landscape/ Last updated: 2025-06-03T11:46:57.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Geoeconomics, AI Disruption, Tariff Impacts, Tourism Boom URL: https://paulkedrosky.com/four-ideas-macro-edition-geoeconomics-ai-disruption-tariff-impacts-tourism-boom/ Last updated: 2025-06-02T11:47:03.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-31/ Last updated: 2025-06-01T11:46:48.000Z **JUNE 01, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** - Group chat on **economy** turns chaotic ([WSJ](https://wsj.com)) - **Dollar** correlation with Treasury yields breaks down ([FT](https://ft.com)) - **M&S hackers** pursue new targets ([FT](https://ft.com)) - **Canadian wildfires** halt oilsands operations and evacuations ([Bloomberg](https://bloomberg.com)) - **Vietnam** proposes $2b casino resort development ([Bloomberg](https://bloomberg.com)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/06/2025-05-29_12-01-00.png) Netflix dominates with 12 of the top 20 most-viewed shows, including the highest with 27.1 million viewers. The platform's lead highlights its strong content strategy, significantly surpassing traditional networks like CBS, whose highest-ranking show garners 17.4 million viewers. This underscores streaming's growing influence in media consumption. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/06/2025-05-28_16-05-08.png) Harvard, with the largest endowment at $49B, has only 26% foreign students, contrasting sharply with Carnegie Mellon's 42% foreign share and $4B endowment. This highlights how higher endowment does not necessarily correlate with higher foreign student enrollment, suggesting potential differences in international student recruitment strategies. --- **THE IDEAS:** **Mentions:** UMM, SPGlobal, MDLZ, MS, XOM ### 01\. AI Automation Disrupting Job Markets AI advancements are not only reshaping tech jobs but also opening new opportunities for workers, fundamentally altering labor demands in various sectors over the next 6-24 months. **Related reading:** **Financial Times**: [The great AI jobs disruption is under way](https://www.ft.com/content/cb9ea970-e6de-4daf-aa9e-7a48d5e648c3) **Oxford Economics**: [New Tools, New Skills, No End in Sight: The Challenge of AI Adoption](https://www.oxfordeconomics.com/resource/new-tools-new-skills-no-end-in-sight-the-challenge-of-ai-adoption/) ### 02\. Geopolitical Trade Realignment Intensifying Nations are increasingly trading within geopolitical confines, as China courts Asian and Gulf partners, reshaping traditional trade alliances and potentially disrupting global trade flows. **Related reading:** **Financial Times**: [China courts Asian and Gulf partners at ‘inflection point’ for global trade](https://www.ft.com/content/c72e43fb-094d-45ae-a28b-d3c362798cae) **FreightWaves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) ### 03\. Hospitality Industry's Shift Toward Technology The hospitality industry is transforming through tech advances, adapting to new consumer trends that will redefine operational models by 2025, enhancing customer experience and profitability. **Related reading:** **HospitalityNet**: [3 Emerging Booking Trends You Can’t Ignore in 2025](https://www.hospitalitynet.org/opinion/4127264.html) **HospitalityNet**: [Top 10 Hospitality Trends: What Shaped 2024 and What To Expect in 2025](https://www.hospitalitynet.org/opinion/4125874.html) ### 04\. Global Power Sector Undergoing Major Transformation Driven by chronic energy shortages and price volatility, the global power sector is undergoing significant transitions, with implications for energy policies and market investments. **Related reading:** **Power Magazine**: [2025’s Energy Crossroads: 6 Trends Redefining the Global Power Sector](https://www.powermag.com/2025s-energy-crossroads-6-trends-redefining-the-global-power-sector/) **CNBC**: [The power sector is 'transforming,' Morgan Stanley says, naming global stocks set to rise 40%](https://www.cnbc.com/2024/10/25/power-and-electricity-stock-picks-from-morgan-stanley-as-demand-booms.html) ### Four Ideas (Energy edition): Clean Energy Shifts and Opportunities URL: https://paulkedrosky.com/four-ideas-energy-edition-clean-energy-shifts-and-opportunities/ Last updated: 2025-05-30T11:46:57.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): China-Gulf Partnerships, Sustainable Agriculture, Geoeconomic Shifts, Commodity Super-Cycle URL: https://paulkedrosky.com/four-ideas-commodities-edition-china-gulf-partnerships-sustainable-agriculture-geoeconomic-shifts-commodity-super-cycle/ Last updated: 2025-05-29T11:47:25.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): AI Healthcare Growth, Diabetes Divestiture, Data Monetization, Healthcare Partnerships URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-ai-healthcare-growth-diabetes-divestiture-data-monetization-healthcare-partnerships/ Last updated: 2025-05-28T11:47:25.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): AI Edge in Financial and Energy Sectors URL: https://paulkedrosky.com/four-ideas-ai-edition-ai-edge-in-financial-and-energy-sectors/ Last updated: 2025-05-27T11:47:11.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): AI Jobs Disruption, Manufacturing Hotspots, Geo-economic Shifts, Travel Innovations URL: https://paulkedrosky.com/four-ideas-macro-edition-ai-jobs-disruption-manufacturing-hotspots-geo-economic-shifts-travel-innovations/ Last updated: 2025-05-26T11:47:12.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-30/ Last updated: 2025-05-25T11:47:01.000Z **MAY 25, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** - **Apple** threat impacts India's tariff talks ([NY Times](https://nytimes.com)) - **Tax bill** raises bond investor concerns ([NY Times](https://nytimes.com)) - Two Israeli embassy staffers **shot** in Washington ([WSJ](https://wsj.com)) - **Apple** plans glasses for 2026, cancels camera watch ([Bloomberg](https://bloomberg.com)) - Mystery buyer bets **$3 billion** on US stock rally ([Bloomberg](https://bloomberg.com)) - **Investors** question next steps post-American fever ([FT](https://ft.com)) - **Jony Ive** and **Sam Altman** possibly challenge Apple ([FT](https://ft.com)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/05/2025-05-20_06-43-21-3.png) Research papers in physical sciences mentioning AI rose from under 2% to over 8% since 2018, significantly outpacing other fields. This surge highlights the growing integration of AI technologies in physical sciences, emphasizing a critical shift in research methodologies and potential advancements in technological innovations. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/05/2025-05-18_16-29-29-3.png) Hypertension and obesity each affect over 40% of U.S. adults, significantly more than diabetes at roughly 15%. This disparity highlights a major public health challenge as these conditions are major risk factors for heart disease and other serious health issues. --- **THE IDEAS:** **Mentions:** TSLA, HON, MAR, IHG, BABA ### 01\. AI reshapes job market and business models. The integration of AI is transforming the job landscape by automating tasks and creating new roles that demand diverse skills. This shift is challenging traditional business models and influencing how companies operate within the broader economy, particularly in tech sectors. **Related reading:** **Financial Times**: [The great AI jobs disruption is under way](https://www.ft.com/content/cb9ea970-e6de-4daf-aa9e-7a48d5e648c3) **Oxford Economics**: [New Tools, New Skills, No End in Sight: The Challenge of AI Adoption](https://www.oxfordeconomics.com/resource/new-tools-new-skills-no-end-in-sight-the-challenge-of-ai-adoption/) ### 02\. Japan's regional manufacturing hot spots emerging. Japan is experiencing a shift as its export-reliant industries cluster in certain cities, revealing disparities but also pointing to new growth areas with job creation potential. These trends indicate potential investments and economic opportunities across less developed regions. **Related reading:** **Oxford Economics**: [Identifying future manufacturing hot spots in Japan and their implications](https://www.oxfordeconomics.com/resource/identifying-future-manufacturing-hot-spots-in-japan-and-their-implications/) ### 03\. Hospitality adapts to tech and consumer trends. The hospitality industry is adapting rapidly to changing consumer preferences and technological advancements. This presents opportunities for businesses to innovate in booking processes, customer experiences, and operational efficiencies in response to evolving market demands. **Related reading:** **Hospitality Net**: [3 Emerging Booking Trends You Can’t Ignore in 2025](https://www.hospitalitynet.org/opinion/4127264.html) **Hospitality Net**: [Top 10 Hospitality Trends: What Shaped 2024 and What To Expect in 2025](https://www.hospitalitynet.org/opinion/4125874.html) ### 04\. Geoeconomics alters global trade relationships. The blending of tech, trade, and military policies is leading to altered global trade dynamics, with geopolitical partners trading more with each other and less with rivals. These shifts may influence corporate strategies and international economic policies. **Related reading:** **Financial Times**: [Welcome to the new age of geoeconomics](https://www.ft.com/content/daf5ce11-c95b-4964-8df1-7463f0f9a81b) **Freight Waves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) ### Four Ideas (Energy edition): Decentralization, Battery Storage, Renewables, Microgrids URL: https://paulkedrosky.com/four-ideas-energy-edition-decentralization-battery-storage-renewables-microgrids/ Last updated: 2025-05-23T11:47:13.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Commodities: Oil, Cocoa, Gold, Supply Chains URL: https://paulkedrosky.com/four-ideas-commodities-edition-commodities-oil-cocoa-gold-supply-chains/ Last updated: 2025-05-22T11:47:10.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): AI in Healthcare and Data Utilization URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-ai-in-healthcare-and-data-utilization/ Last updated: 2025-05-21T11:47:54.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): AI Optimism Underpinning Strategic Investments URL: https://paulkedrosky.com/four-ideas-ai-edition-ai-optimism-underpinning-strategic-investments/ Last updated: 2025-05-20T11:47:10.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Geoeconomics Trends and Sector Implications URL: https://paulkedrosky.com/four-ideas-macro-edition-geoeconomics-trends-and-sector-implications/ Last updated: 2025-05-19T11:47:16.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-29/ Last updated: 2025-05-18T11:46:50.000Z **MAY 18, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** - **Crypto attacks** involve severed fingers and intense violence ([WSJ](https://wsj.com)) - **California explosion** kills one, damages fertility center ([WSJ](https://wsj.com)) - **Newark** faces dire air traffic staffing crisis ([NY Times](https://nytimes.com)) - **Life savings** stolen from mailed paper check ([NY Times](https://nytimes.com)) - The **'Taco' factor** boosts markets ([FT](https://ft.com)) - Apollo's plan to **remake Wall Street** ([FT](https://ft.com)) - **Tariff disruptions** worsen economic outlook over earnings ([Bloomberg](https://bloomberg.com)) - **Sailboat** with 270 people hits Brooklyn Bridge ([Bloomberg](https://bloomberg.com)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/05/2024-04-06_06-30-42.png) China's industrial robot installations in 2022 vastly surpassed other countries, exceeding Japan's by over three times. This dominance highlights China's rapid automation growth, potentially influencing global manufacturing dynamics and labor markets. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/05/2025-05-16_09-25-34.png) Consumer sentiment has plummeted to near-record lows, contrasting sharply with inflation expectations reaching multi-decade highs of 5% over 5-10 years. This divergence underscores significant economic uncertainty, influencing consumer behavior and economic planning. --- **THE IDEAS:** **Mentions:** SPGSustainable1, MDLZ, X, UMass, INTC ### 01\. Geoeconomic shifts redefine global partnerships The integration of tech, trade, finance, and military policies is reshaping global partnerships. Countries are prioritizing trade with allies over rivals, influencing supply chains and economic strategies. This shift impacts multinational companies navigating geopolitical landscapes. **Related reading:** **Financial Times**: [Welcome to the new age of geoeconomics](https://www.ft.com/content/daf5ce11-c95b-4964-8df1-7463f0f9a81b) **Freight Waves**: [Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) ### 02\. Japan's regional industries offer growth potential Japan's manufacturing sectors show uneven growth across regions, with potential for new hotspots. Investments in these emerging areas can drive job creation and income growth, offering lucrative opportunities for companies like INTC to diversify their investment strategies. **Related reading:** **Oxford Economics**: [Identifying future manufacturing hot spots in Japan and their implications](https://www.oxfordeconomics.com/resource/identifying-future-manufacturing-hot-spots-in-japan-and-their-implications/) ### 03\. Trade policy chaos favors anecdotal data reliance With U.S. trade policies causing uncertainty, investors are shifting towards soft data like anecdotal evidence for decision making. This trend highlights the need for dynamic data analysis tools and could influence sectors dependent on swift policy changes. **Related reading:** **Reuters**: [Investors look to soft data for direction amid trade policy chaos](https://www.reuters.com/business/investors-look-soft-data-direction-amid-trade-policy-chaos-2025-05-07/) ### 04\. AI expands impact on industries, consumer habits The rapid adoption of generative AI across industries is transforming consumer and business practices. This shift presents opportunities for companies focusing on AI solutions to capture new markets and enhance operational efficiency, as seen in healthcare and tech sectors. **Related reading:** **Becker's Hospital Review**: [UMass Memorial expands AI triage tool to 5 more hospitals](https://www.beckershospitalreview.com/healthcare-information-technology/ai/umass-memorial-expands-ai-triage-tool-to-5-more-hospitals/) **S&P Global**: [As generative AI sweeps across industries and the consumer landscape](https://twitter.com/SPGlobal/status/1860035529909080480) ### Four Ideas (Energy edition): Hydrogen Production, Solar Resilience, Electrification Challenges, Energy Storage URL: https://paulkedrosky.com/four-ideas-energy-edition-hydrogen-production-solar-resilience-electrification-challenges-energy-storage/ Last updated: 2025-05-16T21:08:37.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Commodity Market Trends and Geopolitics URL: https://paulkedrosky.com/four-ideas-commodities-edition-commodity-market-trends-and-geopolitics/ Last updated: 2025-05-16T21:08:37.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): Healthcare AI Surge and Digital Transformation 2025 URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-healthcare-ai-surge-and-digital-transformation-2025/ Last updated: 2025-05-16T21:08:37.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): AI's Influence on Market Shifts and Sectors URL: https://paulkedrosky.com/four-ideas-ai-edition-ais-influence-on-market-shifts-and-sectors/ Last updated: 2025-05-16T21:08:37.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Unconventional Impacts on Global Economy and Markets URL: https://paulkedrosky.com/four-ideas-macro-edition-unconventional-impacts-on-global-economy-and-markets/ Last updated: 2025-05-16T21:08:37.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-28/ Last updated: 2025-05-11T11:46:45.000Z **MAY 11, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** - **BlackRock** mandates five-day office week ([FT](https://ft.com)) - **America’s pope** impacts Catholic church ([FT](https://ft.com)) - **OpenAI** hires Instacart CEO for operations ([NY Times](https://nytimes.com)) - **Musk-tied investor** clashes with asset manager ([NY Times](https://nytimes.com)) - **Asia's super-rich** reduce US exposure amid trade war ([Bloomberg](https://bloomberg.com)) - **Cardinal Prevost** becomes first American pope, Leo XIV ([Bloomberg](https://bloomberg.com)) - **Celsius’s** Mashinsky gets 12-year prison sentence ([WSJ](https://wsj.com)) - **India** intercepts Pakistani missiles and drones ([WSJ](https://wsj.com)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/05/2025-05-09_08-38-34.png) Gen Alpha's gaming participation at 94% is highest, significantly surpassing Boomers' 50%. The 44% gap highlights generational shifts in gaming culture. This contrast underscores the increasing importance of gaming among younger generations, shaping future entertainment industry trends. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/05/2025-05-09_07-14-34.png) Kai Cenat’s subscription peak of 728,535 in November 2024 significantly surpasses the second place by 402,283\. This notable lead underscores his extraordinary influence and Twitch's evolving dynamics. --- **THE IDEAS:** **Mentions:** $MDLZ, UMass Memorial, Mark Cuban, Mondelez, EurasiaGroup ### 01\. Tech innovations drive sustainability in luxury. Luxury brands are adapting to economic uncertainty by focusing on sustainable innovations. This trend is driven by changing consumer expectations and the need to align luxury offerings with modern demands for sustainability. Brands integrating sustainability can enhance customer loyalty and capture new markets. **Related reading:** **Hospitality Net**: [Luxury with Impact: 2025 Luxury Landscape Overview](https://www.hospitalitynet.org/opinion/4126948.html) **Hospitality Net**: [Food & Beverage Innovation: Trending Creativity and Sustainability](https://www.hospitalitynet.org/opinion/4127047.html) ### 02\. AI triage tools expand healthcare accessibility. Expansion of AI-driven triage tools like those from UMass Memorial is increasing healthcare accessibility. By leveraging AI, hospitals can streamline patient flow, reduce wait times, and allocate resources more efficiently, improving patient outcomes especially in underserved areas. **Related reading:** **Becker's Hospital Review**: [UMass Memorial expands AI triage tool to 5 more hospitals](https://www.beckershospitalreview.com/healthcare-information-technology/ai/umass-memorial-expands-ai-triage-tool-to-5-more-hospitals/) ### 03\. Navigating geoeconomic strategies amid new dynamics. With geopolitical tensions increasing, nations are reassessing their trade and economic strategies. Companies that can navigate these changes by aligning with favored geopolitical partners and adapting to new trade policies will succeed. This trend necessitates agile strategies in response to policy shifts. **Related reading:** **Financial Times**: [Welcome to the new age of geoeconomics](https://www.ft.com/content/daf5ce11-c95b-4964-8df1-7463f0f9a81b) **EurasiaGroup**: [Tech companies must adapt to Washington’s new political climate or risk falling behind.](https://x.com/EurasiaGroup/status/1917572333901918496) ### 04\. Deglobalization impacts industry, reshapes supply chains. Deglobalization is leading to the unraveling of complex supply chains, placing industries at risk. Companies need to localize supply chains, find alternative partners, and innovate logistics to minimize disruptions and maintain competitiveness amid global economic recalibration. **Related reading:** **Financial Times**: [Companies find themselves caught in deglobalisation crossfire](https://www.ft.com/content/6223ac18-d2a4-4cc1-8265-a340197e274f) **Freight Waves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) ### Four Ideas (Energy edition): Decentralized Power, Solar Growth, Hydrogen Challenges, Geopolitical Trade Shifts URL: https://paulkedrosky.com/four-ideas-energy-edition-decentralized-power-solar-growth-hydrogen-challenges-geopolitical-trade-shifts/ Last updated: 2025-05-16T21:08:38.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-27/ Last updated: 2025-04-13T11:46:41.000Z **APRIL 13, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** - Amid tariffs, **China** seen as safest harbor by companies ([NY Times](https://nytimes.com)) - **US bonds** and dollar's safe haven status questioned ([NY Times](https://nytimes.com)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/04/2025-04-10_06-54-28-1.png) Data centers will account for less than 10% of electricity demand growth between 2024 and 2030, contrasting significantly with the industry sector, which leads in growth. This highlights data centers' comparatively lower impact on energy demand despite the digital economy's expansion, suggesting improvements in efficiency. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/04/2025-04-10_07-01-26-1.png) Excluding food and energy, the Consumer Price Index increased by 2.8% compared to an overall rise of 2.4% as of March 2025\. This reversal indicates stronger core inflation pressures despite a broader slowdown, impacting monetary policy and economic planning decisions. --- **THE IDEAS:** **Mentions:** META, TSA, PANW, FTNT, TTEK ### 01\. AI reshapes travel with personalization revolution AI tools and social media are increasingly shaping personalized travel experiences, transforming the industry by merging automation with human connection, leading to novel service offerings and potentially disrupting traditional travel agencies. **Related reading:** **Hospitality Net**: [Data Axle’s 2025 Survey Reveals How Pop Culture, Social Media, and Personalization Are Shaping Modern Travel](https://www.hospitalitynet.org/news/4126643.html) **Skift**: [Navigating AI and Travel: 'Social-First,' Influencers, and the Human Touch](https://skift.com/2025/04/07/navigating-ai-and-travel-social-first-influencers-and-the-human-touch/) ### 02\. Cybersecurity stocks poised for continued resilience Rising market uncertainties and the essential nature of digital security drive growth in cybersecurity stocks, providing robust investment opportunities amidst global tech expansion and increased cyber threats. **Related reading:** **CNBC**: [Resilient cybersecurity stocks to navigate the market turmoil and compelling tech opportunities](https://www.cnbc.com/2025/04/08/resilient-cybersecurity-stocks-to-navigate-the-market-turmoil-and-compelling-tech-opportunities.html) ### 03\. Tariffs drive structural shift in trade dynamics The US tariffs are part of a broader structural shift in global trade, affecting supply chains and encouraging nations to form closer trade links with allies, potentially leading to long-term economic realignments. **Related reading:** **YouTube**: [Trump Signals He Won't Back Down From Tariff Barrage](https://www.youtube.com/watch?v=PtPVvmA0wxM) **Social Media**: [US tariff announcement part of a larger structural shift in the global economy](https://x.com/elerianm/status/1907851813451346178) ### 04\. Open innovation transforms inter-industry collaboration Open innovation is becoming a key driver for strategic progress and impact in industries, as organizations leverage external ideas to accelerate development and gain competitive advantages, restructuring conventional research paradigms. **Related reading:** **Hospitality Net**: [Open Innovation – A Catalyst for Progress, Purpose and Impact](https://www.hospitalitynet.org/opinion/4126613.html) ### Four Ideas (Energy edition): Tariffs Impacting Renewable Energy and Global Trade URL: https://paulkedrosky.com/four-ideas-energy-edition-tariffs-impacting-renewable-energy-and-global-trade/ Last updated: 2025-05-16T21:08:38.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Commodities Trends 2025 URL: https://paulkedrosky.com/four-ideas-commodities-edition-commodities-trends-2025/ Last updated: 2025-05-16T21:08:38.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): Healthcare AI Expansion, Breakthrough Devices Fast-Tracked URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-healthcare-ai-expansion-breakthrough-devices-fast-tracked/ Last updated: 2025-05-16T21:08:38.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): AI Transformations in Industries and Markets URL: https://paulkedrosky.com/four-ideas-ai-edition-ai-transformations-in-industries-and-markets-2/ Last updated: 2025-05-16T21:08:38.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Creative Impacts of Geopolitical and Economic Shifts URL: https://paulkedrosky.com/four-ideas-macro-edition-creative-impacts-of-geopolitical-and-economic-shifts/ Last updated: 2025-05-16T21:08:38.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-26/ Last updated: 2025-04-06T11:46:36.000Z **APRIL 06, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** - **Trade war** triggers global market pessimism ([NY Times](https://nytimes.com)) - **Apple** plunges 9%, leading tech sell-off ([NY Times](https://nytimes.com)) - **Amazon** and **Meta** face China tariff impacts ([WSJ](https://wsj.com)) - **Stocks** suffer biggest drop since March 2020 ([WSJ](https://wsj.com)) - **US dollar** falls on new levies ([FT](https://ft.com)) - **Bling, bags, booze** hit by trade war ([FT](https://ft.com)) - **Retail traders** buy amid tariff-induced market turmoil ([Bloomberg](https://bloomberg.com)) - Worst **S&P 500** day since 2020 as stocks plummet ([Bloomberg](https://bloomberg.com)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/04/2025-04-02_16-30-33-1.png) Countries in Asia face the highest new tariffs, up to 50%, while many others see lower rates. This strategic focus on Asia can significantly impact global trade dynamics, affecting supply chains and international relations. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/04/2025-04-02_16-30-52-1.png) Lululemon's stock plummeted nearly 12% after tariff news, leading consumer stock declines. This sharp drop contrasts with Amazon's smaller decrease near 5%, highlighting differing sensitivities to trade tensions. Such significant drops could indicate potential impacts on future profitability and market strategy adjustments. --- **THE IDEAS:** **Mentions:** TSLA,META,C,HON,BA ### 01\. Geo-economic reorganization reshapes global trade. A structural, secular, and systemic shift in tariffs is reshaping international trade. Initiatives like Trump's tariffs indicate a prolonged and fundamental change to trade patterns. This shift will likely see countries strengthening regional alliances and minimizing reliance on traditional global powers, impacting multinational companies' supply chains. **Related reading:** **Financial Times**: [How to make sense of Donald Trump’s tariffs](https://www.ft.com/content/3436cd20-f09f-46dd-9e94-ba10b86de83f) **X.com**: [Posted early this morning: The US tariff announcement is part of a larger “three-S” shift in the global economy that is … Structural in nature, including the rewiring of the international trading system;](https://x.com/elerianm/status/1907851813451346178) ### 02\. Open interlining revolutionizes airline operations. Airlines adopting open interlining can bypass legacy systems, allowing more dynamic and scalable partnerships. This shift will enhance efficiency and passenger experience, with potential revenue impacts. Airlines equipped with this strategy may outperform competitors still reliant on traditional interline agreements. **Related reading:** **Skift**: [Open Interlining: Inside the Strategy Transforming How Airlines Connect Flights](https://skift.com/2025/04/02/open-interlining-inside-the-strategy-transforming-how-airlines-connect-flights/) ### 03\. Digital compliance transforms healthcare efficiency. AI technologies are streamlining healthcare compliance and automating workflows, leading to enhanced data security and operational efficiency. As more healthcare providers embrace these innovations, there will be a significant impact on administrative costs and patient service delivery. **Related reading:** **Healthcare Dive**: [Balancing innovation and compliance: A risk-smart approach to digital evolution](https://www.healthcaredive.com/spons/balancing-innovation-and-compliance-a-risk-smart-approach-to-digital-evolu/736712/) ### 04\. Experience-driven travel reshapes industry dynamics. The shift towards experience-driven travel influences industry dynamics, with companies refocusing on personalized and unique offerings. This trend is aligned with changing traveler behaviors and leveraging tech in booking processes, potentially leading to an increase in travel experiences sector growth. **Related reading:** **Skift**: [The Rise of Travel Experiences: 4 Key Insights From Industry Decision Makers](https://skift.com/2024/10/11/the-rise-of-travel-experiences-4-key-insights-from-industry-decision-makers/) ### Four Ideas (Commodities edition): Commodities Supply Chain, Gold Outlook, Cocoa Demand, Energy Transition URL: https://paulkedrosky.com/four-ideas-commodities-edition-commodities-supply-chain-gold-outlook-cocoa-demand-energy-transition/ Last updated: 2025-05-16T21:08:38.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): AI Transformations in Healthcare and Pharmaceuticals URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-ai-transformations-in-healthcare-and-pharmaceuticals/ Last updated: 2025-05-16T21:08:38.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): AI and Energy, Healthcare, Cybersecurity, Mobile Industry URL: https://paulkedrosky.com/four-ideas-ai-edition-ai-and-energy-healthcare-cybersecurity-mobile-industry/ Last updated: 2025-05-16T21:08:38.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Insurance, Travel, AI in Healthcare, Geopolitical Trading URL: https://paulkedrosky.com/four-ideas-macro-edition-insurance-travel-ai-in-healthcare-geopolitical-trading/ Last updated: 2025-05-16T21:08:38.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-25/ Last updated: 2025-03-30T11:46:49.000Z **MARCH 30, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** - **Climate-resilient chickens** help fight poverty in Zambia ([NY Times](https://nytimes.com)) - **Auto tariffs** prompt car buying rush ([NY Times](https://nytimes.com)) - **Khosla** sees best investment opportunity in a decade ([Bloomberg](https://bloomberg.com)) - **Quake** impacts Thai tourism, heightens safety concerns ([Bloomberg](https://bloomberg.com)) - **Home buyers** act despite high rates and prices ([WSJ](https://wsj.com)) - **Elon Musk** aims to dominate NASA missions to Mars ([WSJ](https://wsj.com)) - **EU capitals** push to soften tariff retaliation ([FT](https://ft.com)) - **Columbia** president resigns after yielding to demands ([FT](https://ft.com)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/03/2025-03-29_13-31-56.png) US labor market faces a marked contraction as three-month average payroll growth plummets from nearly 300K to below 100K within two years. This decline signifies weakening economic momentum and raises concerns over potential impacts from policy changes related to tariff announcements. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/03/2025-03-28_11-39-40.png) The Swedish Krona's 11% rise against the US dollar is significant, contrasting sharply with the Canadian Dollar's 1% increase. Such a disparity underscores potential shifts in economic policies and investor confidence, influencing global trade dynamics and currency market strategies. --- **THE IDEAS:** **Mentions:** META, TSLA, Honeywell, DeepSeek, S&P 500 ### 01\. AI-driven compliance in healthcare advances rapidly AI platforms are transforming healthcare compliance by automating workflows and enhancing efficiency and data security. This trend will drive investment in digital health technologies impacting companies developing compliance solutions. As regulatory scrutiny intensifies, the healthcare sector should prepare for AI-driven innovation in compliance amid rising data privacy concerns. **Related reading:** **Healthcare Dive**: [Balancing innovation and compliance: A risk-smart approach to digital evolution](https://www.healthcaredive.com/spons/balancing-innovation-and-compliance-a-risk-smart-approach-to-digital-evolu/736712/) ### 02\. Trade wars elevate stagflation risk globally The escalation of tariff wars is sparking fears of stagflation as economies face rising costs and slowing growth. This could impact sensitive sectors by elevating input prices and restricting supply chains. Monitoring tariff developments is crucial for global industries, particularly manufacturing and consumer goods. **Related reading:** **YouTube**: [Stagflation Fears Rise as Tariff War Escalates | Full Episode| Insight with Haslinda Amin 3/28/2025](https://www.youtube.com/watch?v=1bYWrCHlnlE) **Financial Times**: [Clarity on tariffs can’t come soon enough](https://www.ft.com/content/f8ee3ba9-b4d7-4cfa-8111-fb806fbc3566) ### 03\. mRNA technology faces policy-induced uncertainty Ongoing regulatory uncertainties are casting a shadow over mRNA technology development. If governments reduce support or impose stricter regulations, biotech companies may face challenges in innovation and funding. This could influence investment strategies in pharmaceutical and biotechnology sectors, affecting long-term advancements in medical technology. **Related reading:** **Financial Times**: [An ominous shadow falls over mRNA technology](https://www.ft.com/content/70654d6d-48ad-45be-8a10-dcd1a75ffc5e) ### 04\. Geopolitical shifts redefine global trade partnerships Geopolitical dynamics are prompting countries to trade more with allies and less with rivals, driven by strategic alliances. This trend impacts global supply chains and trade patterns, influencing companies dependent on international markets. Continued monitoring will be key as geopolitical tensions and collaborations evolve. **Related reading:** **FreightWaves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) ### Four Ideas (Energy edition): Clean Energy Opportunities and Trends URL: https://paulkedrosky.com/four-ideas-energy-edition-clean-energy-opportunities-and-trends/ Last updated: 2025-05-16T21:08:39.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Commodities Insight: Gold, Copper, Coffee, Natural Gas URL: https://paulkedrosky.com/four-ideas-commodities-edition-commodities-insight-gold-copper-coffee-natural-gas/ Last updated: 2025-05-16T21:08:39.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): AI in Healthcare, Radiology Innovation, Implantable Biosensors, Healthcare Payment Innovations URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-ai-in-healthcare-radiology-innovation-implantable-biosensors-healthcare-payment-innovations/ Last updated: 2025-05-16T21:08:39.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): AI-induced shifts in energy, healthcare, and markets URL: https://paulkedrosky.com/four-ideas-ai-edition-ai-induced-shifts-in-energy-healthcare-and-markets/ Last updated: 2025-05-16T21:08:39.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Geopolitical Realignments, AI-Driven Compliance, EM Resilience, Copper Stockpiles URL: https://paulkedrosky.com/four-ideas-macro-edition-geopolitical-realignments-ai-driven-compliance-em-resilience-copper-stockpiles/ Last updated: 2025-05-16T21:08:39.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-24/ Last updated: 2025-03-23T11:46:40.000Z **MARCH 23, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** - Unexpected **home sales** rise as buyers return ([WSJ](https://wsj.com)) - **Tesla** recalls majority of Cybertrucks ([WSJ](https://wsj.com)) - **Accenture** reports revenue hit from Musk’s spending cuts ([FT](https://ft.com)) - **Jeff Bezos** reconciles with Maga ([FT](https://ft.com)) - **Office market** recovery signals worst is over ([NY Times](https://nytimes.com)) - **Elon Musk’s** top DOGE overseer revealed ([NY Times](https://nytimes.com)) - **Stocks** rally halted by trade fears ([Bloomberg](https://bloomberg.com)) - **UBS** may move HQ unless Swiss cut capital demands ([Bloomberg](https://bloomberg.com)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/03/2025-03-19_16-55-52-2.png) Women born in the 1990s have fewer children by age 30 compared to previous generations, with about 1.5 fewer than those from the 1960s. This significant decline highlights a societal shift toward delayed parenthood, impacting demographics and economic planning. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/03/2025-03-19_11-37-26-2.png) Gold's price increased by 40.99%, outperforming Bitcoin's 24.81% rise over the same period. This is significant as Bitcoin often leads in volatility and growth in uncertain times. The reversal highlights a stronger preference for traditional assets despite digital currency popularity. --- **THE IDEAS:** **Mentions:** AAPL, S&P 500, NASDAQ, EU, China ### 01\. AI reshapes China's economic strategy DeepSeek's AI breakthroughs provide China a path to revitalize its economic growth amidst a prolonged slowdown and trade tensions with the U.S. This positions China's tech sector for growth, catching global investor interest, despite broader economic challenges. **Related reading:** **CNBC**: [DeepSeek-led AI adoption offers China an opportunity to boost its sputtering growth](https://www.cnbc.com/2025/02/21/deepseek-led-ai-adoption-offers-china-opportunity-to-boost-growth.html) ### 02\. Tariffs disrupt global supply chains The ongoing uncertainty and impact of tariffs are throwing global supply chains into 'uncharted territory,' leading to shifts in trading patterns and manufacturing decisions globally. **Related reading:** **YouTube**: [Global #economy in "uncharted territory" from #tariff impact](https://www.youtube.com/watch?v=hQ--LYj07Ys) **Financial Times**: [Clarity on tariffs can’t come soon enough](https://www.ft.com/content/f8ee3ba9-b4d7-4cfa-8111-fb806fbc3566) ### 03\. Financial markets undergo regime shift Investment gains on 'Mag 7' stocks in 2024 match the German Dax, signaling a significant regime shift in financial markets. This indicates the strong influence of tech stocks in evolving market dynamics. **Related reading:** **Bloomberg**: [Interesting factoid from Bloomberg's Dani Burger that illustrates the dramatic regime shift in financial markets](https://x.com/elerianm/status/1902288325886566904) ### 04\. Geopolitical shifts alter trade partnerships Countries are increasingly trading with geopolitical allies rather than rivals, indicating a shift in global trade dynamics that could affect investment strategies and geopolitical stability. **Related reading:** **Freight Waves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) ### Four Ideas (Energy edition): Distributed solar grids, EV charging, energy storage, grid upgrades URL: https://paulkedrosky.com/four-ideas-energy-edition-distributed-solar-grids-ev-charging-energy-storage-grid-upgrades/ Last updated: 2025-05-16T21:08:39.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Commodities Market Trends for 2025 URL: https://paulkedrosky.com/four-ideas-commodities-edition-commodities-market-trends-for-2025/ Last updated: 2025-05-16T21:08:39.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): AI, Biosensors, Interoperability, Hospital-at-Home URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-ai-biosensors-interoperability-hospital-at-home/ Last updated: 2025-05-16T21:08:39.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): AI's Energy Demands and Global Shift URL: https://paulkedrosky.com/four-ideas-ai-edition-ais-energy-demands-and-global-shift/ Last updated: 2025-05-16T21:08:39.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): AI Energy Demand, Tariff Impact, Currency Flows URL: https://paulkedrosky.com/four-ideas-macro-edition-ai-energy-demand-tariff-impact-currency-flows/ Last updated: 2025-05-16T21:08:39.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-23/ Last updated: 2025-03-16T11:46:37.000Z **MARCH 16, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** - **Economic fear** prompts consumer, business distress signals ([WSJ](https://wsj.com)) - **Virgin Atlantic** plans UK air-taxi service with Joby ([Bloomberg](https://bloomberg.com)) - **Canada** reviews F-35 order amid US rift ([Bloomberg](https://bloomberg.com)) - **C.F.P.B.** revives after court orders ([NY Times](https://nytimes.com)) - **Tariffs** threaten France's champagne industry ([NY Times](https://nytimes.com)) - **German army** struggles recruiting Gen Z for war ([FT](https://ft.com)) - **Free speech** history evolves from ideal to politics ([FT](https://ft.com)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/03/2025-03-15_17-56-09.png) Republicans’ consumer sentiment dropped below Democrats’ for the first time since Trump's 2025 re-election, illustrating a significant shift. Their sentiment declined to 83.9, while Democrats’ fell to 41.4\. This reversal indicates changing economic perceptions across party lines, highlighting varied responses to political and economic developments. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/03/2025-03-15_16-42-33.png) US policy uncertainty skyrocketed in recent years, peaking at a record 350, nearly five times the pre-2016 average. This unprecedented surge signifies substantial economic instability, potentially impacting investor confidence, market volatility, and decision-making for businesses and policymakers. --- **THE IDEAS:** **Mentions:** AAPL, SPG, MSFT, TSLA, BX ### 01\. AI drives massive data center power demand The AI boom is revolutionizing the power sector by significantly increasing electricity demand, due to the growth of data centers. This shift pushes the energy industry to adapt quickly, offering opportunities for investment in power supply and infrastructure enhancements. **Related reading:** **Power Magazine**: [AI Boom Reshapes Power Landscape as Data Centers Drive Historic Demand Growth](https://www.powermag.com/ai-boom-reshapes-power-landscape-as-data-centers-drive-historic-demand-growth/) ### 02\. Trade deviates to geopolitical partners Global trading patterns are shifting as geopolitical allies increasingly trade more with each other and less with rivals. This trend has implications for supply chains and sectors dependent on international trade, such as technology and consumer goods. **Related reading:** **Freight Waves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) **YouTube**: [Global #economy in "uncharted territory" from #tariff impact](https://www.youtube.com/watch?v=hQ--LYj07Ys) ### 03\. Investment rotating from US to global markets Investors are starting to shift capital from US markets to global opportunities due to various geopolitical and economic changes. This realignment may affect US stock valuations and boost emerging markets. **Related reading:** **Reuters**: [Investors spy the dawn of a tectonic shift away from US markets](https://www.reuters.com/markets/global-markets-investors-analysis-2025-03-05/) ### 04\. Tariff policies reshape global economic outlook Tariff-driven economic changes are significantly impacting global trade dynamics, affecting growth projections and encouraging strategic adjustments in multinational corporations. This trend poses risks to dependent industries, requiring adaptation and new strategic approaches. **Related reading:** **NBER**: [Trade and Structural Change: Focusing on the Specifics -- by Andrew Greenland, John Lopresti](https://www.nber.org/papers/w33127#fromrss) **Financial Times**: [It's a growth scare first and a tariff scare second](https://www.ft.com/content/4644a9fc-1360-4d4f-a411-6a22b2a257b8) ### Four Ideas (Energy edition): Grid Innovations, AI in Energy, Critical Minerals, European EV Challenges URL: https://paulkedrosky.com/four-ideas-energy-edition-grid-innovations-ai-in-energy-critical-minerals-european-ev-challenges/ Last updated: 2025-05-16T21:08:40.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Oil export surge, Tariffs, AI power demand, Gold rally URL: https://paulkedrosky.com/four-ideas-commodities-edition-oil-export-surge-tariffs-ai-power-demand-gold-rally/ Last updated: 2025-05-16T21:08:40.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): Healthcare AI Growth, Interoperability, Mergers and VC Trends URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-healthcare-ai-growth-interoperability-mergers-and-vc-trends/ Last updated: 2025-05-16T21:08:40.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): AI Transformations and Market Shifts URL: https://paulkedrosky.com/four-ideas-ai-edition-ai-transformations-and-market-shifts/ Last updated: 2025-05-16T21:08:40.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Global Interest Rate Shifts and Data Center Demand URL: https://paulkedrosky.com/four-ideas-macro-edition-global-interest-rate-shifts-and-data-center-demand/ Last updated: 2025-05-16T21:08:40.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-22/ Last updated: 2025-03-09T11:46:36.000Z **MARCH 09, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** - Advances in **carbon capture** technology ([Financial Times](https://ft.com)) - **Wirecard fugitive** linked to spy ring ([Financial Times](https://ft.com)) - **India** faces challenges post major financial loss ([Bloomberg](https://bloomberg.com)) - **Brush fires** impact Long Island Pine Barrens ([Bloomberg](https://bloomberg.com)) - **Walgreens billionaire** sees empire unravel ([WSJ](https://wsj.com)) - **Scammers** take savings, victims face tax issues ([NY Times](https://nytimes.com)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/03/2025-03-06_16-22-12-1.png) Denmark's letter volume dropped from 1.4 billion in 2000 to under 200 million by 2022, illustrating over an 85% decrease. This sharp decline underscores the significant impact of digital communication on traditional mail, challenging postal services to adapt. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/03/GlYcF98WQAA9XQB-1.jpg) U.S. gold bar imports surged to nearly $40 billion by early 2025, a dramatic increase compared to previous years. This spike may indicate economic uncertainty or strategic financial adjustments, especially after the Russia-Ukraine conflict, reflecting significant shifts in global trade dynamics and investment strategies. --- **THE IDEAS:** **Mentions:** DEEP, CLF, TSLA, SPGI, GS ### 01\. Data centers amplify energy demand growth The expansion of AI and technology demands increasingly large data centers, driving substantial growth in energy consumption. This trend can stress power grids but also create opportunities in the renewable energy sector. Over 6-24 months, expect higher investments in energy infrastructure. **Related reading:** **Power Magazine**: [AI Boom Reshapes Power Landscape as Data Centers Drive Historic Demand Growth](https://www.powermag.com/ai-boom-reshapes-power-landscape-as-data-centers-drive-historic-demand-growth/) ### 02\. Tariff impacts create global economic uncertainty The imposition of tariffs has placed the global economy in 'uncharted territory', potentially disrupting trade patterns and creating uncertainty. This requires businesses to rethink supply chains and investments. In the next 6-24 months, expect shifts towards localized production. **Related reading:** **YouTube**: [Global #economy in 'uncharted territory' from #tariff impact](https://www.youtube.com/watch?v=hQ--LYj07Ys) **Twitter**: [RT by Interesting read from the CEO of $CLF (one of the largest American Steel companies) Regarding Tariffs](https://x.com/TylerWilson119/status/1896758998420062611) ### 03\. US market shift towards previously lagging sectors There is a noticeable rotation in stock markets, with investments moving away from winners to previously underperforming sectors. This trend may create opportunities in undervalued markets and resources over the next 6-24 months. **Related reading:** **Morningstar**: [A Stock Market Rotation Is Underway. Will It Last?](https://www.morningstar.co.uk/uk/news/261532/a-stock-market-rotation-is-underway-will-it-last--.aspx) ### 04\. China's AI efforts bolster its economic stance China's AI advancement led by companies like DeepSeek could reinvigorate its economic landscape, attracting global investment despite current economic challenges. Expect renewed interest from investors in Chinese tech stocks within 6-24 months. **Related reading:** **CNBC**: [DeepSeek-led AI adoption offers China an opportunity to boost its sputtering growth](https://www.cnbc.com/2025/02/21/deepseek-led-ai-adoption-offers-china-opportunity-to-boost-growth.html) **CNBC**: [DeepSeek's AI breakthrough could get global investors interested in China again](https://www.cnbc.com/2025/02/09/deepseeks-ai-breakthrough-could-get-global-investors-interested-in-china-again.html) ### Four Ideas (Energy edition): AI Power Demand, Energy Storage Innovation, Chinese Market Influence, European Energy Challenges URL: https://paulkedrosky.com/four-ideas-energy-edition-ai-power-demand-energy-storage-innovation-chinese-market-influence-european-energy-challenges/ Last updated: 2025-05-16T21:08:40.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Oil Price Volatility and Energy Transition Impacts URL: https://paulkedrosky.com/four-ideas-commodities-edition-oil-price-volatility-and-energy-transition-impacts/ Last updated: 2025-05-16T21:08:40.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): Healthcare AI and Innovations Reshaping the Industry URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-healthcare-ai-and-innovations-reshaping-the-industry/ Last updated: 2025-05-16T21:08:40.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): Data Center Demand, China AI Investments, Stock Rotation, Sports Tech Disruption URL: https://paulkedrosky.com/four-ideas-ai-edition-data-center-demand-china-ai-investments-stock-rotation-sports-tech-disruption/ Last updated: 2025-05-16T21:08:40.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): AI in China, US Treasury Yields, Emerging Markets, Geopolitical Trade Shifts URL: https://paulkedrosky.com/four-ideas-macro-edition-ai-in-china-us-treasury-yields-emerging-markets-geopolitical-trade-shifts/ Last updated: 2025-05-16T21:08:40.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-21/ Last updated: 2025-03-02T12:46:49.000Z **MARCH 02, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** - Instagram **apologizes** for violent video flood ([WSJ](https://wsj.com)) - **Mexico extradites** 29 drug bosses to U.S. ([WSJ](https://wsj.com)) - **Instagram** faces backlash for graphic content ([Bloomberg](https://bloomberg.com)) - **Buyout** reveals fuzzy math in private equity deals ([Bloomberg](https://bloomberg.com)) - **BP** shifts focus to oil and gas ([NY Times](https://nytimes.com)) - **US economy** strained by tariffs and spending cuts ([NY Times](https://nytimes.com)) - **Stripe** rebounds to $90bn valuation amid AI demand ([FT](https://ft.com)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/03/2025-03-01_07-51-34.png) U.S. home entertainment spending is projected to reach nearly $50 billion by 2024, marking a significant increase from around $20 billion in 2011\. This contrasts sharply with the decline of physical rentals and sales, highlighting a shift towards digital and streaming formats, reshaping the industry landscape. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/03/2025-02-28_15-04-44.png) Long-term inflation views surged to the highest in years above 3.5%, despite declining year-over-year consumer price changes since 2022\. This shift correlates with falling consumer sentiment, highlighting growing economic concerns and potential impacts on spending and economic stability. --- **THE IDEAS:** **Mentions:** PLTR, Goldman, DeepSeek, TikTok, StatNews ### 01\. AI boosts investor interest in China DeepSeek-led AI adoption is revitalizing global investor interest in Chinese stocks, even amidst economic downturns. Such breakthroughs attract increased international investments, strengthening China's market position over the next 6-24 months. **Related reading:** **CNBC**: [DeepSeek-led AI adoption offers China an opportunity to boost its sputtering growth](https://www.cnbc.com/2025/02/21/deepseek-led-ai-adoption-offers-china-opportunity-to-boost-growth.html) **CNBC**: [DeepSeek's AI breakthrough could get global investors interested in China again](https://www.cnbc.com/2025/02/09/deepseeks-ai-breakthrough-could-get-global-investors-interested-in-china-again.html) ### 02\. Embedded insurance central to retail strategy Embedded insurance is increasingly vital for retailers and brands seeking value-added services, presenting a substantial strategic pivot in customer engagement methods. This trend will reshape retail landscapes within 24 months. **Related reading:** **Insurance Journal**: [Viewpoint: Insurers Risk Losing Out if They Ignore Embedded Insurance Opportunities](https://www.insurancejournal.com/news/international/2025/02/18/811964.htm) ### 03\. Geopolitical trade realignment emerging Geopolitical partners are increasingly prioritizing trade among allies over rivals, signaling significant shifts in global trade dynamics. This realignment will impact global supply chains and economic policies over the next two years. **Related reading:** **FreightWaves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) ### 04\. Tech disrupting traditional sports Advances in technology are transforming traditional sports sectors, offering new economic models and growth opportunities and challenging old mindsets. These developments will redefine the sports business landscape swiftly over the next two years. **Related reading:** **Sportico**: [The Traditional Sport Mindset Can Be a Trap When Tech Advances](https://www.sportico.com/business/tech/2025/traditional-sport-mindset-trap-tech-disruption-1234831451/) ### Four Ideas (Energy edition): Solar Microgrids, AI in Energy, Battery Innovation, Grid Investment URL: https://paulkedrosky.com/four-ideas-energy-edition-solar-microgrids-ai-in-energy-battery-innovation-grid-investment/ Last updated: 2025-05-16T21:08:41.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Gold, Copper, Commodities, AI, Tariffs, China URL: https://paulkedrosky.com/four-ideas-commodities-edition-gold-copper-commodities-ai-tariffs-china/ Last updated: 2025-05-16T21:08:41.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): AI Partnerships, FDA Layoffs, AI-Driven Health Tech, Ultrasound Innovations URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-ai-partnerships-fda-layoffs-ai-driven-health-tech-ultrasound-innovations/ Last updated: 2025-05-16T21:08:41.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): AI-driven transformations in unexpected sectors URL: https://paulkedrosky.com/four-ideas-ai-edition-ai-driven-transformations-in-unexpected-sectors/ Last updated: 2025-05-16T21:08:41.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): China AI, Embedded Insurance, Geostrategic Trade, Global Services URL: https://paulkedrosky.com/four-ideas-macro-edition-china-ai-embedded-insurance-geostrategic-trade-global-services/ Last updated: 2025-05-16T21:08:41.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-20/ Last updated: 2025-02-23T12:46:47.000Z **FEBRUARY 23, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** - **Berkshire Hathaway** hits record tax bill ([Bloomberg](https://bloomberg.com)) - **Inflation gauge** drops to seven-month low ([Bloomberg](https://bloomberg.com)) - **Musk** demands federal workers report accomplishments ([WSJ](https://wsj.com)) - **DOGE** asserts billion-dollar savings, details pending ([WSJ](https://wsj.com)) - **Businesses** anticipate tax cut extension impacting economy ([NY Times](https://nytimes.com)) - **Berkshire** profits rise through U.S. Treasury investments ([NY Times](https://nytimes.com)) - **Buffett** reassures shareholders on record cash pile ([FT](https://ft.com)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/02/https___substackcdn.com_image_fetch_w_1100-c_limit-f_auto-q_auto_good-fl_progressive_steep_https-3A-2F-2Fsubstack-post-media.s3.amazonaws.com-2Fpublic-2Fimages-2F10d3efa5-2f5d-4166-86bd-33bfe691ede6_1198x1023.png) MIT shows a sharp increase in job-seeking MBA grads without jobs, rising to nearly 20% in 2024, the highest among the schools, doubling its 2022 figure. This significant rise highlights potential hurdles in the job market for top-tier graduates, signaling potential economic or industry shifts. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/02/https___substackcdn.com_image_fetch_w_1100-c_limit-f_auto-q_auto_good-fl_progressive_steep_https-3A-2F-2Fsubstack-post-media.s3.amazonaws.com-2Fpublic-2Fimages-2Fc4257ab1-782b-409e-8e52-5525a869edc8_1296x1308.jpg) Recent college graduates face the highest unemployment rates in decades, spiking to nearly 14% in 2020, compared to 9% for all workers. This disparity highlights increased vulnerability and economic challenges for new graduates entering the workforce. --- **THE IDEAS:** **Mentions:** DeepSeek, AI, China, Insurers, Markets ### 01\. DeepSeek AI revives global interest in China. DeepSeek's AI achievements present China as a hub for technological advancements, counteracting recent economic gloom and attracting foreign investment. This reinvigorates China's market appeal, facilitating a potential shift in global capital flow back towards Chinese equities and innovation sectors over the next 12-24 months. **Related reading:** **CNBC**: [DeepSeek-led AI adoption offers China an opportunity to boost its sputtering growth](https://www.cnbc.com/2025/02/21/deepseek-led-ai-adoption-offers-china-opportunity-to-boost-growth.html) **CNBC**: [DeepSeek's AI breakthrough could get global investors interested in China again](https://www.cnbc.com/2025/02/09/deepseeks-ai-breakthrough-could-get-global-investors-interested-in-china-again.html) ### 02\. Embedded insurance becoming central retail strategy. Embedded insurance is transforming into a key strategy for retailers, offering bundled insurance products to enhance value propositions. This trend creates new revenue streams and strengthens customer loyalty, which could lead to significant shifts in insurance sales dynamics and retailer-insurer collaborations over the next 1-2 years. **Related reading:** **Insurance Journal**: [Viewpoint: Insurers Risk Losing Out if They Ignore Embedded Insurance Opportunities](https://www.insurancejournal.com/news/international/2025/02/18/811964.htm) ### 03\. Geopolitical partners trade more, reducing rival reliance. Increased trade within geopolitical alliances signifies a shift towards economic insularity, reducing dependency on adversarial nations. This could lead to more targeted economic policies and bilateral trade agreements, affecting global supply chain realignment and diversification strategies over the next couple of years. **Related reading:** **FreightWaves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) ### 04\. AI adoption disrupting labor market dynamics. AI is fundamentally altering job structures by automating tasks and influencing job competency requirements, leading to a need for reskilling and altering employment patterns across industries. This emerging trend could create significant demand for education services and employment policies to adapt over the next 1-2 years. **Related reading:** **Phys.org**: [Is AI already shaking up the labor market? Four trends point to major change](https://phys.org/news/2025-02-ai-labor-trends-major.html) ### Four Ideas (Commodities edition): Gold, Coffee, Critical Minerals, Geopolitical Trade URL: https://paulkedrosky.com/four-ideas-commodities-edition-gold-coffee-critical-minerals-geopolitical-trade/ Last updated: 2025-05-16T21:08:41.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): AI-Driven Healthcare Innovations in 2025 URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-ai-driven-healthcare-innovations-in-2025/ Last updated: 2025-05-16T21:08:41.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): AI Transformations: New Beneficiaries and Market Shifts URL: https://paulkedrosky.com/four-ideas-ai-edition-ai-transformations-new-beneficiaries-and-market-shifts/ Last updated: 2025-05-16T21:08:41.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Geopolitics, AI Innovation, Energy Transition, and US Exceptionalism URL: https://paulkedrosky.com/four-ideas-macro-edition-geopolitics-ai-innovation-energy-transition-and-us-exceptionalism/ Last updated: 2025-05-16T21:08:41.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-19/ Last updated: 2025-02-16T12:46:44.000Z **FEBRUARY 16, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** - **DeepSeek** fuels $1.3 trillion China stock rally ([Bloomberg](https://bloomberg.com)) - **Corporate America** fears profit outlook amid equity rally ([Bloomberg](https://bloomberg.com)) - **Broadcom** and **TSMC** consider splitting Intel's chip business ([WSJ](https://wsj.com)) - New **spy unit** leads Russia’s shadow war against US ([WSJ](https://wsj.com)) - **Amazon** union push fails in North Carolina ([NY Times](https://nytimes.com)) - **Economic uncertainty** challenges business leaders ([NY Times](https://nytimes.com)) - **Michelin** reintroduces culture of deference ([FT](https://ft.com)) - Understanding **Elon Musk**'s influence ([FT](https://ft.com)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/02/883f552f576f6e423d866d878b6a43dd7a13ed5652d11cdd53c74f53b8f656f2.png) IRS staffing has surged from a low in 2015 to projected levels in 2024 not seen since the early 1990s. This sharp increase of roughly 30,000 employees over a decade could enhance tax enforcement and processing efficiency amidst evolving fiscal demands. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/02/2025-02-15_18-48-33.png) Barcelona's nightlife now ends significantly earlier, with club closing times shifting from 6 AM to 3 AM. This three-hour change highlights a major cultural shift and potentially impacts the local economy and nightlife culture. --- **THE IDEAS:** **Mentions:** MSFT, NVDA, DEEP, MCK, JD.V ### 01\. AI to rejuvenate interest in China DeepSeek's AI breakthrough is invigorating global investor interest in Chinese markets, potentially reversing the trend of economic skepticism. By showcasing innovation, it adds appeal to previously hesitant investors, which may revitalize Chinese stocks. **Related reading:** **CNBC**: [DeepSeek's AI breakthrough could get global investors interested in China again](https://www.cnbc.com/2025/02/09/deepseeks-ai-breakthrough-could-get-global-investors-interested-in-china-again.html) ### 02\. Trump tariffs heighten trade uncertainty With Trump revealing plans for new tariffs, there is significant potential for volatility in global trade. This development introduces uncertainty for industries relying on the global supply chain, particularly affecting manufacturing and trade-dependent sectors. **Related reading:** **YouTube**: [Indian Carmaker Gears Up for Trump Tariff Uncertainty](https://www.youtube.com/watch?v=WUdF6%5F0J2pc) **YouTube**: [Trump Welcomes Modi, Unveils Tariff Plan](https://www.youtube.com/watch?v=aSonOxaX1VU) ### 03\. Energy shift towards sustainability and integration Key trends indicate a major pivot towards sustainable energy, driven by geopolitical needs for security and infrastructure upgrading. As regions focus on interconnected networks, energy efficiency and renewables become attractive growth sectors. **Related reading:** **Financial Times**: [The investment trends shaping the global energy transition](https://www.ft.com/content/f83a0b51-c08e-42ef-b74c-668ac228b6c9) **POWER Magazine**: [2025’s Energy Crossroads: 6 Trends Redefining the Global Power Sector](https://www.powermag.com/2025s-energy-crossroads-6-trends-redefining-the-global-power-sector/) ### 04\. Geopolitical partners prefer mutual trade A study by McKinsey indicates a shift where countries are increasingly opting to trade within alliances and reduce reliance on geopolitical rivals. This trend suggests a strategic decoupling and realignment of global trade dynamics. **Related reading:** **FreightWaves**: [McKinsey: Geopolitical partners trading more with one another, less with rivals](https://www.freightwaves.com/news/mckinsey-geopolitical-partners-trading-more-with-one-another-less-with-rivals) ### Four Ideas (Energy edition): Energy Storage Innovations Drive Market Opportunities URL: https://paulkedrosky.com/four-ideas-energy-edition-energy-storage-innovations-drive-market-opportunities/ Last updated: 2025-05-16T21:08:41.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Commodities and Geopolitical Dynamics URL: https://paulkedrosky.com/four-ideas-commodities-edition-commodities-and-geopolitical-dynamics/ Last updated: 2025-05-16T21:08:42.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): AI's impact on healthcare and biotech advancement URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-ais-impact-on-healthcare-and-biotech-advancement/ Last updated: 2025-05-16T21:08:42.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): DeepSeek's Unintended Innovations Transform AI Markets URL: https://paulkedrosky.com/four-ideas-ai-edition-deepseeks-unintended-innovations-transform-ai-markets/ Last updated: 2025-05-16T21:08:42.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Global AI investment trends, energy shifts URL: https://paulkedrosky.com/four-ideas-macro-edition-global-ai-investment-trends-energy-shifts/ Last updated: 2025-05-16T21:08:42.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-18/ Last updated: 2025-02-09T12:46:54.000Z **FEBRUARY 09, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** - **American prosperity** threatened in weekend essay ([FT](https://ft.com)) - **$4bn cuts** may harm US medical research ([FT](https://ft.com)) - Judge blocks Musk's **DOGE** from treasury system ([WSJ](https://wsj.com)) - Soho **House** faces billionaire feud and takeover issues ([WSJ](https://wsj.com)) - Elliott builds stake in **BP** ([Bloomberg](https://bloomberg.com)) - Musk’s team seeks **Treasury** data access ([Bloomberg](https://bloomberg.com)) - N.F.L. network coordinates **TV advertising** with game pauses ([NY Times](https://nytimes.com)) - Business newsletter **Feed Me** rises in popularity ([NY Times](https://nytimes.com)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/02/GjRzr-NXAAA76Kc.jpg) Employment in firms with over 500 employees surpasses those with fewer than 500, reaching 53% compared to 47%. This shift began around 2000, highlighting the growing dominance of larger businesses, which impacts economic dynamics and policy considerations. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/02/2025-02-06_18-54-15-1.png) U.S. tech giants Apple, Nvidia, and Microsoft alone surpass Europe's entire market capitalization by $5 trillion. Europe's leading firms are dwarfed, with SAP at $304 billion. This stark contrast highlights the dominance of U.S. technology in global markets, potentially influencing economic power dynamics and investment trends. --- **THE IDEAS:** **Mentions:** TCS, META, MSFT, NVDA, SLB ### 01\. AI breakthrough drives renewed China interest DeepSeek's AI advancements spark global investors to reconsider allocations to China despite ongoing economic concerns. This technological progress could revive foreign interest in Chinese markets and lead to increased investment flows, impacting the country's tech sector over the next 24 months. **Related reading:** **CNBC**: [DeepSeek's AI breakthrough could get global investors interested in China again](https://www.cnbc.com/2025/02/09/deepseeks-ai-breakthrough-could-get-global-investors-interested-in-china-again.html) **YouTube**: [Nouriel Roubini on DeepSeek, AI's Impact on Jobs, Health, Stocks](https://www.youtube.com/watch?v=p81ddWhmG54) ### 02\. Energy transition bolsters infrastructure investment Increasing demand for large battery storage is reshaping energy generation and infrastructure investment. This trend highlights a shift towards sustainable energy solutions, creating opportunities for sectors involved in energy generation, transmission, and distribution over the next two years. **Related reading:** **Financial Times**: [The investment trends shaping the global energy transition](https://www.ft.com/content/f83a0b51-c08e-42ef-b74c-668ac228b6c9) **POWER Magazine**: [Four Energy Trends to Watch in 2025](https://www.powermag.com/four-energy-trends-to-watch-in-2025/) ### 03\. Climate risks prompt strategic business adjustments As climate risks top risk manager concerns, businesses are adjusting strategies and investing in sustainability. This shift could lead to a surge in demand for green technologies and services, altering corporate priorities and operational frameworks. **Related reading:** **Insurance Journal**: [Climate Risks Again Top List of Risk Manager Worries](https://www.insurancejournal.com/news/national/2025/02/07/810930.htm) **Phys.org**: [Building a circular future: Study reveals key organizational capabilities for sustainability](https://phys.org/news/2025-02-circular-future-reveals-key-organizational.html) ### 04\. Southeast Asia benefits amid US-China tensions Southeast Asia is becoming a focal point for investors due to geopolitical tension between the US and China. The influx of investment into this region is generating potential opportunities despite challenges in extracting value from investments. **Related reading:** **Financial Times**: [South-east Asia calculates its next move as US and China face off](https://www.ft.com/content/a86cf0f1-bc81-41cc-9811-e0778006f513) **Eurasia Group**: [Cutting USAID creates a vacuum – and says that creates opportunity for the US’ adversaries, like China.](https://x.com/EurasiaGroup/status/1887585085156958262) ### Four Ideas (Energy edition): Energy Storage, Renewable Setbacks, AI-Driven Nuclear, Energy Grid Resilience URL: https://paulkedrosky.com/four-ideas-energy-edition-energy-storage-renewable-setbacks-ai-driven-nuclear-energy-grid-resilience/ Last updated: 2025-05-16T21:08:42.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): Medical AI Expansion, Health Tech Investment Growth URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-medical-ai-expansion-health-tech-investment-growth/ Last updated: 2025-05-16T21:08:42.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): China's AI Leap, AI-Driven Energy Demand, AI Legal Implications, Generative AI Acquisitions URL: https://paulkedrosky.com/four-ideas-ai-edition-chinas-ai-leap-ai-driven-energy-demand-ai-legal-implications-generative-ai-acquisitions/ Last updated: 2025-05-16T21:08:42.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Evolving Supply Chains and Market Responses URL: https://paulkedrosky.com/four-ideas-macro-edition-evolving-supply-chains-and-market-responses/ Last updated: 2025-05-16T21:08:42.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-17/ Last updated: 2025-02-02T12:46:29.000Z **FEBRUARY 02, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** - **Consumer bureau chief** fired by administration ([NY Times](https://nytimes.com)) - **OpenAI** releases new AI agent 'Operator' ([NY Times](https://nytimes.com)) - **Dollar General** allows ICE interactions at stores ([Bloomberg](https://bloomberg.com)) - **Airline** central to Brazil deportation issue ([Bloomberg](https://bloomberg.com)) - **American Airlines** flight 5342 passengers' final moments detailed ([WSJ](https://wsj.com)) - **DOGE** joins payment system distributing trillions ([WSJ](https://wsj.com)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/02/3d46211b-4932-40a8-b0c7-0ace0f1ac04c-bac0dafb39e44364.jpg) Information Technology leads with 59% foreign revenue, nearly double the S&P 500's 28% average. This global dependence highlights the sector's vulnerability to international market shifts, potentially impacting financial stability and growth strategies compared to more domestically focused sectors. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/02/2025-01-31_04-38-12-d277a15ce42e2f33.png) Eighth-grade reading proficiency declined nearly 10 percentage points between 2013 and 2022, contrasting the previous stability. This significant drop highlights educational challenges and suggests the need for focused interventions to address literacy development during crucial middle school years. --- **THE IDEAS:** **Mentions:** MSFT, AAPL, TSLA, MC, MA ### 01\. Technology driving shift in energy landscape As tech companies invest in power generation for data centers, a significant shift in the energy landscape is occurring. This will impact power generation sources, like renewables, over 6-24 months, presenting opportunities for tech firms and renewable energy sectors. **Related reading:** **Power Magazine**: [Generational Shift—Data Centers Bring Change to Energy Landscape](https://www.powermag.com/generational-shift-data-centers-bring-change-to-energy-landscape/) ### 02\. Budget cuts alter EMIA energy transition Budget cuts in major economies will challenge the EMIA power industry, potentially slowing the energy transition and forcing a rethink of energy strategies. This could impact companies within the renewable sector and fossil fuel industries over the next few years. **Related reading:** **Power Magazine**: [5 Trends Shaping the EMIA Power Industry in 2025](https://www.powermag.com/5-trends-shaping-the-emia-power-industry-in-2025/) ### 03\. AI analysis reveals tech vendor opportunities Utilizing AI for document analysis unveils tech opportunities like system upgrades and digital transformations, crucial for firms in the tech sector. This trend over 6-24 months could shift competitive dynamics within tech service vendors and consultants. **Related reading:** **Skift**: [New Skift Feature: Uncovering Tech Vendor Opportunities Using AI Analysis](https://skift.com/2024/12/26/new-skift-feature-uncovering-tech-vendor-opportunities-using-ai-analysis/) ### 04\. Macro divergence benefits trend-following strategies As macroeconomic conditions continue to diverge globally, trend-following strategies become more viable. This opens opportunities in the equity sector, particularly for funds utilizing algorithmic trading to exploit these macro variations over the next year. **Related reading:** **Institutional Investor**: [It’s Prime Time for Trend-Following Strategies as Macro Divergence Continues](https://www.institutionalinvestor.com/article/2dr9igom4y6l27n86dxc0/innovation/its-prime-time-for-trend-following-strategies-as-macro-divergence-continues) ### Four Ideas (Energy edition): Energy Transition Innovations and Challenges URL: https://paulkedrosky.com/four-ideas-energy-edition-energy-transition-innovations-and-challenges/ Last updated: 2025-05-16T21:08:42.000Z _This post is for paying subscribers only._ ### Four Ideas (Commodities edition): Commodities Trends: Energy, Metals, and Agriculture URL: https://paulkedrosky.com/four-ideas-commodities-edition-commodities-trends-energy-metals-and-agriculture/ Last updated: 2025-05-16T21:08:42.000Z _This post is for paying subscribers only._ ### Four Ideas (Life sciences edition): Healthcare AI, Venture Capital, and Innovation Trends URL: https://paulkedrosky.com/four-ideas-life-sciences-edition-healthcare-ai-venture-capital-and-innovation-trends/ Last updated: 2025-05-16T21:08:42.000Z _This post is for paying subscribers only._ ### Four Ideas (AI edition): AI Transformations in Industries and Markets URL: https://paulkedrosky.com/four-ideas-ai-edition-ai-transformations-in-industries-and-markets/ Last updated: 2025-05-16T21:08:43.000Z _This post is for paying subscribers only._ ### Four Ideas (Macro edition): Power Challenges, Emerging Market Growth, AI Impacts, Insurance Evolution URL: https://paulkedrosky.com/four-ideas-macro-edition-power-challenges-emerging-market-growth-ai-impacts-insurance-evolution/ Last updated: 2025-05-16T21:08:43.000Z _This post is for paying subscribers only._ ### Four Ideas, Sunday Edition: URL: https://paulkedrosky.com/four-ideas-sunday-edition-16/ Last updated: 2025-01-26T12:46:11.000Z **JANUARY 26, 2025—GENERAL EDITION** This is a weekly free look of our premium daily Four Ideas email. If you're not already a subscriber, sign up for a [five-day preview](https://paulkedrosky.com/#/portal/signup/648103c473f8b0fd77e0b79a/monthly). **HEADLINES:** - Wall Street **banks** to sell billions in loans ([WSJ](https://wsj.com)) - **Home sales** drop to 30-year low ([WSJ](https://wsj.com)) - **KTM's lenders** resist debt proposal ([Bloomberg](https://bloomberg.com)) - **BASF earnings** drop on impairments, restructuring ([Bloomberg](https://bloomberg.com)) - **US regulator** addresses significant securities concerns ([FT](https://ft.com)) - **Venture Global** falters in Wall Street debut ([FT](https://ft.com)) - Inflation fix targets non-existent **energy emergency** ([NY Times](https://nytimes.com)) - Tech's **TikTok** dilemma on law compliance ([NY Times](https://nytimes.com)) **TWO GRAPHS:** ![Daily Graph](https://paulkedrosky.com/content/images/2025/01/2025-01-23_08-12-45-1.png) The percentage of Americans moving has reached a record low at 7.8% in 2023, compared to over 20% in the late 1960s. This decline indicates reduced mobility, potentially impacting economic dynamics and regional development. --- ![Daily Graph](https://paulkedrosky.com/content/images/2025/01/2025-01-23_08-12-37-1.png) The hiring rate has dropped to 3.3%, matching levels seen only during severe economic downturns like the early pandemic and Great Recession. This decline signifies potential economic instability, highlighting challenges in job market recovery amidst current conditions. --- **THE IDEAS:** **Mentions:** XOM, TSLA, MASTERCARD, MSFT, VISA ### 01\. Geopolitical tensions driving energy innovation Amidst rising geopolitical tensions, calls for scientific 'safe zones' indicate a global push for neutral grounds fostering international collaboration, potentially leading to innovations like IIASA in energy sectors as countries seek energy self-sufficiency. **Related reading:** **Phys.org**: [Amid rising geopolitical tensions, researchers call for 'safe zones' for scientific collaboration](https://phys.org/news/2025-01-geopolitical-tensions-safe-zones-scientific.html) **Power Magazine**: [5 Trends Shaping the EMIA Power Industry in 2025](https://www.powermag.com/5-trends-shaping-the-emia-power-industry-in-2025/) ### 02\. AI adoption reshapes consumer business models The widespread adoption of generative AI across industries is transforming consumer engagement and business operations. Companies are focusing on how AI tools are applied, altering marketing strategies and product offerings. **Related reading:** **SPGlobal**: [As generative AI sweeps across industries and the consumer landscape...](https://twitter.com/SPGlobal/status/1860035529909080480) **Skift**: [New Skift Feature: Uncovering Tech Vendor Opportunities Using AI Analysis](https://skift.com/2024/12/26/new-skift-feature-uncovering-tech-vendor-opportunities-using-ai-analysis/) ### 03\. Diversifying energy investments amid market shifts Energy sectors are transforming, with data centers impacting energy generation. Investments in power by tech firms signal a shift in energy landscape, with Morgan Stanley predicting a 40% rise in stocks within the power sector. **Related reading:** **Power Magazine**: [Generational Shift—Data Centers Bring Change to Energy Landscape](https://www.powermag.com/generational-shift-data-centers-bring-change-to-energy-landscape/) **CNBC**: [The power sector is 'transforming,' Morgan Stanley says, naming global stocks set to rise 40%](https://www.cnbc.com/2024/10/25/power-and-electricity-stock-picks-from-morgan-stanley-as-demand-booms.html) ### 04\. India's power demand bolstered by economic growth India's booming economy is intensifying demand for power, driven by AI, electric vehicles, and technology adoption, which is impacting energy strategies and attracting investments in these sectors. **Related reading:** **Twitter**: [RT 1\. AI. 2\. EVs.. 3\. Bitcoin.... 4\. India....... = Power Demand](https://x.com/Convertbond/status/1880966599882199425) **CNBC**: [CNBC's Inside India newsletter: India's deepening ties with natural gas](https://www.cnbc.com/2024/10/31/cnbcs-inside-india-newsletter-indias-deepening-ties-with-natural-gas.html) ### Four Ideas (Energy edition): Energy Storage Innovations and Grid Resilience URL: https://paulkedrosky.com/four-ideas-energy-edition-energy-storage-innovations-and-grid-resilience/ Last updated: 2025-05-16T21:08:43.000Z _This post is for paying subscribers only._ _Includes the latest 500 public posts. Use `/sitemap.xml` for the complete archive of public content._