An AI hedge fund that soared 439 percent in six months just lost most of its 45 billion dollars in a matter of days. Leopold Aschenbrenner built the most aggressive bet on the artificial intelligence boom, then leverage turned the trade inside out. The unwind rattled chip stocks from Seoul to Silicon Valley.

By Blumefield | August 1, 2026

The Fastest Fortune in Modern Finance

Six months ago Leopold Aschenbrenner ran the hottest AI hedge fund on the planet. Situational Awareness LP had climbed 439 percent in the first half of 2026, swelling to roughly 45 billion dollars and turning its 24-year-old founder into the most talked-about money manager since the meme-stock era. This week it became something else entirely. After a brutal run on semiconductor stocks and a cascade of margin calls, the fund shed most of its assets in a matter of days, falling to around 10 billion dollars and forcing a fire sale of its entire public portfolio. The collapse of this AI hedge fund is the clearest sign yet that the market's favorite trade has grown dangerously crowded.

Aschenbrenner was never a conventional investor. He built his reputation in 2024 with a sprawling essay series arguing that artificial intelligence would demand an almost unimaginable expansion of computing power, advanced chips, memory and energy. He then did what few essayists dare. He put the entire thesis into a single, highly leveraged book of stocks, betting that anything touching AI infrastructure would keep rising. For a while, spectacularly, it did. His investors were up more than fourfold before the summer even ended.

How Four Times Leverage Became a Trap

The mechanics of the blowup are almost textbook. According to accounts from people close to the fund, Situational Awareness ran gross exposure of roughly four times its capital, with its top five positions making up more than three-quarters of its disclosed long book. That is extreme concentration layered on extreme borrowing. When a portfolio is built that way, the math is unforgiving. A 30 percent drop across the core holdings is enough to wipe out the equity cushion entirely, and that is precisely what happened in July.

The names Aschenbrenner loved most turned on him at once. SK Hynix, CoreWeave, Nebius, Micron and Bloom Energy, the physical backbone of the AI buildout, fell between 35 and 47 percent over the month. Compounding the pain, the fund had shorted software companies including Adobe as a hedge, and those positions moved sharply the wrong way at the same time. Every leg of the trade broke simultaneously. This is the hidden fragility of a concentrated AI hedge fund. The same conviction that produces a 439 percent gain on the way up produces a total wipeout on the way down.

The timing had a cruel edge. The sharp July decline coincided with the US listing of SK Hynix, one of the fund's largest holdings, while a broader unwind of leveraged positions dragged down South Korean equities. What began as an ordinary pullback in richly valued chip stocks quickly became a forced deleveraging event, as prime brokers moved to protect themselves.

The Block Trade That Ended a Thousand Percent Run

Goldman Sachs, JPMorgan Chase and Bank of America, the fund's three prime brokers, issued the margin calls that sealed its fate. They demanded additional collateral to support the borrowing, and Situational Awareness did not have it. Rather than sell piecemeal into a falling market and signal distress, the fund unloaded its entire public equity portfolio in a single block trade. The buyer was Ken Griffin's Citadel, which picked up the hard-hit AI names at a discount and now inherits the positions that a rival AI hedge fund could no longer hold.

It is difficult to overstate how fast this reversal was. A fund that had returned roughly 1,000 percent since inception effectively liquidated its public book in one transaction. For the young manager who had become a symbol of the AI age, the sale was a public humbling. The essays that made him famous warned that the race to build artificial intelligence would strain every input in the economy. They did not anticipate that the strain would break his own portfolio first.

For the wider market, the fire sale had an unexpected side effect. Once Citadel absorbed the block and the forced selling cleared, several of the battered semiconductor stocks actually bounced. The overhang that traders feared, an unstable fund holding enormous concentrated positions, was gone. A dangerous seller had been replaced by one of the steadiest hands on Wall Street.

What Survives, and Why It Matters

Situational Awareness is not dead. What remains is a leaner private vehicle built around a single crown jewel, a stake in the AI safety company Anthropic valued at roughly 5 billion dollars. Aschenbrenner acquired the position when Anthropic was valued near 60 billion dollars in an early 2025 round, and with the company now worth vastly more on paper, that holding has become the anchor of the fund's future. The public trades are gone. The private bet on one of the leaders of the field survives.

That split tells you something important about how the smart money now views artificial intelligence. The liquid, leveraged way to play the theme, buying chipmakers and data-center suppliers with borrowed money, has just proven treacherous. The illiquid, patient way, owning equity in the labs building frontier models, is what Aschenbrenner chose to keep. Whether that judgment looks wise or merely lucky will depend on how the next year of the AI cycle unfolds. Readers following the sector can track the fallout through ongoing coverage at Blumefield.

The Warning Inside the AI Trade

The lesson of this AI hedge fund is not that the artificial intelligence boom is over. Demand for computing, memory and power remains real, and the companies at the center of it are still growing fast. The lesson is about structure. When a single narrative becomes so consensus that investors are willing to lever it four times over and pile three-quarters of a book into five names, the trade stops being an investment and becomes a machine for amplifying volatility. Aschenbrenner did not misjudge the technology. He misjudged the fragility of betting on it with borrowed money at the top.

Regulators and prime brokers will study this episode closely, because a single fund unwinding 20 billion dollars in exposure can move entire markets. The speed of the collapse, from record highs to fire sale in weeks, shows how quickly concentration and leverage can turn a winning thesis into a systemic risk. Aschenbrenner's own writing predicted that AI would test the limits of the financial system. His fund just delivered the first stress test, and it failed in public.

For everyone still riding the AI trade, the message from this AI hedge fund is sobering and simple. Being right about the technology is not the same as surviving the volatility it creates. The future Aschenbrenner described may well arrive. Getting there in one piece is the harder problem, and leverage is rarely the way to do it. More market analysis is available at Blumefield.