AI Has Doubled Computing's Share of U.S. GDP; Nvidia Under Inference Pressure
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
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
Hyperscaler and data-center investment, negligible through 2022, is now tracking toward $1.5 trillion annually by 2030. At that level, it will surpass, in real terms, the residential construction
A new paper 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
A new Anthropic paper 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
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 new NBER working paper 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
There is a new paper 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
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
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
The following data is from a NYT piece about OpenRouter's latest funding round. The company has apparently seen a 5x increase in token volume in six months, which
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
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—