How a 25-Year-Old AI Trader's $20B Fund Imploded in a Matter of Weeks

The Rise and Fall of a $20 Billion AI Hedge Fund

Leopold Aschenbrenner, a 25-year-old German-born prodigy, turned a $200 million startup into one of Wall Street’s most talked-about AI hedge funds, Situational Awareness LP, which by early summer managed north of $20 billion. His thesis was simple and compelling: the AI boom’s true bottleneck lay in physical infrastructure—chips, memory, data centers, and power generation—not in the software or chip design giants that had already soared. The fund went long on companies like Nebius (cloud), CoreWeave (compute), Bloom Energy (power), Sandisk and Micron (memory), while simultaneously holding massive put options against AI stalwarts such as Nvidia, AMD, Broadcom, Oracle, and ASML, as well as software names like Adobe.

For a time, the bet paid off spectacularly. The fund reportedly returned 439% in the first half of 2026, with Aschenbrenner’s personal take from performance fees likely running into hundreds of millions of dollars. But in July, the trade violently reversed. The infrastructure stocks he owned came under heavy selling pressure—SK Hynix dropped sharply, and Nebius, Sandisk, Micron, and CoreWeave each lost more than 35% in a month. At the same time, his short positions in software companies rallied, hitting both sides of his leveraged, highly concentrated portfolio. Margin calls followed, and by Wednesday night the fund was locked in urgent negotiations with potential buyers.

In the early hours of Thursday, the fund struck a deal to sell the bulk of its approximately $16 billion public equity portfolio to Citadel, the multi-strategy giant run by Ken Griffin. Goldman Sachs and JPMorgan, acting as prime brokers, helped manage the orderly liquidation. The sale removed a huge overhang from the market: when news of the forced disposal broke, shares of IREN (a data center operator) surged nearly 30%, and similar AI infrastructure names saw double-digit gains as short-covering and bargain-hunting kicked in. The episode demonstrated the outsize influence Aschenbrenner’s relatively tiny team of eight employees had come to wield over AI-related stocks.

Behind the scenes, the fund’s structure had always been fragile. Investors had to commit at least $25 million and lock up capital for two years, and the fund employed aggressive leverage to magnify returns. That gearing left no room to ride out a drawdown. The meltdown also revealed a fundamental flaw: even if the macro thesis about AI infrastructure was correct, some of the specific companies chosen—like IREN, which had disclosed controversial executive stock grants—had governance issues that undermined their stock performance. Aschenbrenner plans to continue managing the fund’s remaining private holdings, including a roughly $5 billion stake in AI firm Anthropic, but public market investors who followed his 13F filings after the fact were left nursing heavy losses.

Inside the Unraveling of the Split Bet Strategy

The Split Bet Anatomy: Right Thesis, Wrong Execution

Aschenbrenner’s conviction was that the real value in AI would accrue to the owners of scarce physical resources, while first-wave AI chipmakers and software firms were overvalued. This “split bet” was intellectually appealing but neglected two critical realities. First, the companies he favored were often smaller, less liquid, and more vulnerable to idiosyncratic risks—like IREN’s governance shock, which knocked its shares down nearly 40% independent of AI sentiment. Second, the short side of the trade exposed the fund to momentum-driven rallies in the very tech mega-caps that dominate indices, making simultaneous losses on both sides more likely in a volatile market.

How Leverage Turned a Drawdown into a Fire Sale

The fund’s heavy use of borrowed money amplified every move. When the infrastructure names fell and the shorted stocks rose, margin requirements soared. Prime brokers Goldman Sachs and JPMorgan stepped in to help meet margin calls and liquidate positions in an orderly fashion, but the fund had little choice but to sell. This sequence—a drawdown triggering a liquidity crunch that forced sales at depressed prices—is a classic hedge-fund death spiral. The irony is that Aschenbrenner’s underlying thesis might eventually prove correct; the problem was that his investors’ capital could not survive the interim volatility.

Citadel’s Countertrade and the Smart-Money Advantage

Ken Griffin’s Citadel stands at the opposite end of the hedge-fund spectrum: thousands of employees, sophisticated risk systems, and the balance sheet to absorb large, illiquid blocks. By stepping in as the buyer of last resort for a portfolio that likely had little other natural demand at that moment, Citadel secured positions in AI infrastructure names at distressed valuations. Bloomberg also named Millennium Management and Jane Street as parties that considered acquiring parts of the book. The episode highlights how large, diversified firms with deep pockets can profit from the forced deleveraging of smaller, concentrated rivals.

What the Meltdown Means for the Broader AI Equity Rally

The rapid unwind of a $16 billion public portfolio temporarily rattled a clutch of AI stocks but did not derail the sector’s broader narrative. In fact, the removal of the overhang sparked a relief rally. However, the drama will likely cause institutional investors to reassess the risks of chasing AI themes through highly concentrated, leveraged vehicles. It also underscores that even a correct top-down call on AI spending can be destroyed by poor bottom-up stock selection, company-specific governance failures, and the wrong layer of the value chain—exactly the nuance that can get lost when a compelling macro story meets market reality.

What Market Participants Should Take Away from the Meltdown

• Verify company-level governance before backing a macro thesis: Aschenbrenner’s position in IREN suffered a 40% drop when management granted locked shares worth $800 million, a red flag that undermined an otherwise well-reasoned infrastructure bet.
• Don’t chase 13F filings after the fact: Retail and smaller funds that copied Situational Awareness’s disclosed positions found themselves buying at a premium driven by the fund’s own buying pressure, only to be hit by the subsequent sell-off.
• Leverage magnifies timing risk: The fund’s borrowings meant that a temporary drawdown—even one that could have reversed over time—became a permanent loss when margin calls forced a sale. Investors must size positions so that the time horizon of the thesis matches the patience the portfolio structure allows.
• Counterparty and liquidity planning is critical for concentrated strategies: In a crisis, only a handful of deep-pocketed buyers like Citadel exist for a $16 billion block. The inability to attract multiple competitive bids likely forced the fund to accept a lower price.