Here are recent rough notes from the site.
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://www.bloomberg.com/news/articles/2026-02-07/natural-gas-traders-fume-after-nymex-glitch-triggers-market-chaos
A new Apollo chartbook (https://www.apolloacademy.com/wp-content/uploads/2026/02/020726_Public-VS-Private_Chart-Book_v3.pdf) 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.
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.
https://epochai.substack.com/p/how-close-is-ai-to-taking-my-job
This went out as my chart of the day (subscribe at 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.
