Aschenbrenner's AI hedge fund unwinds after a chip drawdown

The man who wrote the AI boom’s most-quoted manifesto just watched his own fund fall apart on an AI trade.

Leopold Aschenbrenner, a former OpenAI researcher, became famous in 2024 for “Situational Awareness”, the essay that convinced a lot of Silicon Valley that the AI buildout was underpriced. He then launched a hedge fund on that thesis.

The drawdown

The fund, estimated at $20 billion to $24 billion by mid-2026, lost roughly 67% in July on tumbling semiconductor stocks and mounting margin calls, with gross exposure reported at about 4x. On July 30 it sold its entire public equities book, longs and shorts together and including SK Hynix and CoreWeave, to Ken Griffin’s Citadel in a single block trade before the open.

Not a zero

The obituaries overstate it, and this part matters. The fund was not fully liquidated after the unwind: it kept its private holdings, including its stake in Anthropic, and was reportedly seeking fresh capital. Through June 30 it had returned 439% net, and its disclosed public holdings fell more than 35% during July.

So the accurate description is a brutal drawdown and a destroyed public book, not a zero. Those are different outcomes, even if they photograph the same.

The lesson in the leverage

Being right about a technology and being right about the trade are two different jobs. The thesis behind the fund was not obviously wrong; the AI buildout it bet on is still happening. The leverage was the problem. At about 4x gross exposure, a sharp move in semiconductor stocks does not test your conviction, it removes your ability to hold the position at all.

Is this a crack in the AI trade itself, or just one over-levered book meeting a volatile week? The distinction will matter to everyone else still running the same thesis with less margin.