The AI trade just ate one of its own prophets. Leopold Aschenbrenner — 24 years old, former OpenAI researcher, author of the essay that convinced a generation that AGI was arriving by 2027 — watched his hedge fund lose most of its value in days. Situational Awareness LP reportedly swelled toward $45 billion at its peak. By the time Ken Griffin's Citadel stepped in to buy its public portfolio, the numbers whispered were closer to $10 billion.
Let that sink in. A man who spent 2024 warning the world about “the most important transition in human history” spent 2026 running one of the most aggressively leveraged momentum bets on Wall Street. The “Nostradamus of AI,” as the press dubbed him, apparently did not see the drawdown coming.
The Golden Child of the AI Trade
Understand the mythology first. Aschenbrenner was the archetypal AI wunderkind: a former OpenAI researcher who wrote “Situational Awareness,” a 165-page manifesto predicting AGI by 2027 and urging Washington to treat AI like a national security matter. The essay made him a celebrity in a community that treats conviction as a virtue.
So when he launched a hedge fund with the same name, the money arrived fast. Really fast. The fund reportedly quadrupled within weeks of its debut.
And that, as it turns out, was the problem. Fast money is not smart money. It is often the same money, arriving in a hurry and leaving faster.
What Actually Happened
The mechanics are boringly familiar: leverage, momentum, a market that refused to cooperate. Reports say the fund was running as much as four times leverage on concentrated AI positions — chipmakers and hyperscalers. For a while, that worked spectacularly. Then the momentum reversed.
July brought a historic rotation out of AI stocks. When the market moves against a four-times-levered book, it does not negotiate. Margin calls followed, reportedly from prime brokers including Bank of America, Goldman Sachs, and JPMorgan. The fund was forced to liquidate its entire public portfolio in what one account called a “margin-fueled fire sale.” Sources put the July loss at roughly 67 percent. Citadel bought what was left.
For the record, here is the timeline, as best as anyone outside the fund can reconstruct it:
- Early 2026: Aschenbrenner launches Situational Awareness LP; assets reportedly surge toward $45 billion within weeks.
- July 2026: AI stocks suffer a sharp momentum reversal; the leveraged fund is hit with margin calls.
- Late July: The fund unwinds its entire public portfolio; a rival buyer picks up the pieces.
- Aftermath: It still holds private stakes — including Anthropic shares — down roughly 67 percent on the month.
Notably, the crash happened in a stock market that was, by most measures, tame. The AI trade did not need a recession to break the fund. It just needed a pullback and a margin call. That should worry everyone, not just his limited partners.
The Uncomfortable Questions
Let's be honest about what this story is really telling us. The AI trade's loudest true believer — the guy who built his brand on predicting the future better than anyone — was, in practice, running a leveraged momentum strategy any quant desk would recognize. The grand superintelligence thesis was just a narrative wrapper around a concentrated bet.
That raises questions the industry would rather not answer:
- Where were the risk limits? A fund managing tens of billions with four-times leverage and no circuit breaker is not a conviction play. It is a gamble dressed in a philosophy.
- Who else is exposed? If the most famous AI investor could not survive a routine rotation, what happens to the leveraged ETFs and margin accounts still riding the same names?
- Is “the bottom is in” just cope? Wall Street is already declaring the worst is over for the AI trade. The people saying it loudest were long the whole way down.
- What was the thesis, really? Was this ever about AGI — or about momentum, flows, and a prophet who turned out to be a very smart day trader?
For the rest of the market, the takeaway is mundane. Momentum strategies that work for six months can be fatal in six days. The AI trade has been the most crowded trade in finance for two years. Crowded trades do not end gently.
The uncomfortable symmetry: the essay that made Aschenbrenner famous warned that AI would concentrate power faster than anyone expected. His fund just demonstrated, in miniature, how quickly value concentrates — then evaporates — when everyone is leveraged to the same narrative.
The timing is also delicious, in a grim way. Governments are stepping in where the market retreated: South Korea is reportedly planning $14 billion for a wealth fund's AI bets, and the EU is funding AI “gigafactories.” When hedge funds get burned and nation-states buy the dip, the cycle has entered a stranger phase.
None of this means AI is a bubble, by the way. The technology can be world-changing and the trade can still be a casino. Both things are true at once. What the Situational Awareness fiasco proves is not that AI is overvalued — it is that the people who claim to see the future most clearly are often just looking at a very leveraged version of the present.
The memes are already flowing, and the “Nostradamus” nickname has acquired a crueler edge. But the real lesson is not about one fund. The AI trade, for all its talk of paradigm shifts, still runs on the oldest financial fuel there is: borrowed money and borrowed confidence. When the margin call comes, prophets are just another position to unwind.
Comments