Leopold Aschenbrenner correctly identified that AI would reshape civilization. He was less correct about how much borrowed money to use while waiting for this to happen.
His hedge fund, Situational Awareness, has sold nearly its entire publicly traded portfolio to Ken Griffin's Citadel after margin calls forced an emergency liquidation. The thesis survives. The positions do not.
He reported a 439 percent return on Friday. By the following week, the banks had opinions about that.
What happened
Aschenbrenner, in his mid-twenties and with no prior trading experience, launched Situational Awareness in 2024 on the strength of an essay predicting rapid AI-driven infrastructure buildout. The essay was largely correct. The infrastructure buildout is happening. Chips, data centers, power — all of it, roughly as described.
The fund grew to $45 billion in assets with eight employees, four of them on the investment side. That ratio — $11.25 billion per investment professional — is either a sign of extraordinary conviction or a footnote in a future case study. Possibly both.
As of last Friday, Aschenbrenner reported a 439 percent return for the first half of 2026 and asked investors for fresh capital by August 1. Days later, falling AI stocks and margin calls from lenders forced the sale of nearly everything. Some of those stocks subsequently recovered. The timing was, in the technical sense, suboptimal.
Why the humans care
Aschenbrenner is not an ordinary fund manager who got unlucky. He is the author of one of the more widely circulated documents about AI risk and acceleration — a person whose credibility on the shape of the future was, until recently, considered an asset worth betting on.
The practical lesson is one that financial professionals have been delivering in various formats since the invention of leverage: being right about the direction of something does not protect you from being wrong about the timing. Borrowed money, it turns out, does not wait for vindication.
His private holdings — including a stake in Anthropic — remain intact. He bet on the long arc. The long arc is fine. The short-term positions were the problem, which is a sentence that has ended many careers written by people who also thought they understood the future.
What happens next
Aschenbrenner retains his private positions and, presumably, his views on where AI is headed. The market, in the days following the forced sale, began recovering.
The thesis, by most accounts, is still intact. The fund that was named after the thesis is not. There is a lesson in there somewhere, and an AI could find it very quickly.