AI Bet Backfires! Former OpenAI Researcher Suffers Dual Blow in Long and Short Positions, Forcing Fund Liquidation

Situational Awareness, a hedge fund founded by former OpenAI researcher Leopold Aschenbrenner, has suffered significant losses after misjudging investments in artificial intelligence (AI) infrastructure stocks while simultaneously facing losses from short positions in software stocks, leading to severe financial pressure. The fund is now being forced to unwind multiple positions and raise cash to meet margin requirements.

According to sources familiar with the matter, the fund’s prime brokers are accelerating efforts to raise capital, and a major institutional investor is reportedly in agreement to acquire Situational Awareness’s publicly traded assets, though the buyer’s identity remains undisclosed. It is still unclear whether asset sales will fully satisfy the margin call demands.

Situational Awareness reached an asset size of $4.5 billion in early July but quickly fell into distress as its portfolio sharply declined. Sources indicate that major brokers including Bank of America (BAC-US), Goldman Sachs (GS-US), and JPMorgan Chase (JPM-US) are assisting the fund in raising capital or orderly reducing positions, and were marketing a batch of long and short assets to potential buyers ahead of the market open on Thursday, July 30.

The fund’s recent losses stem from two major fronts. On one hand, the decline in AI infrastructure stocks such as SK Hynix (SKHY-US) has severely damaged its long positions. On the other hand, the sharp rise in stock prices of software companies like Adobe (ADBE-US) has inflicted heavy losses on its short positions, creating a scenario of simultaneous losses on both long and short sides.

According to regulatory filings, Situational Awareness’s largest holdings as of the end of Q1 included Nebius Group (NBIS-US), Sandisk (SNDK-US), Micron (MU-US), and CoreWeave (CRWV-US). All four stocks have fallen more than 35% this month, highlighting the rapid reversal in AI infrastructure-related trades.

The fund was reportedly in talks to sell its stake in AI startup Anthropic, though it remains unclear whether the transaction was completed. A spokesperson for Situational Awareness denied the fund is currently selling its Anthropic shares. The exact amount of losses and the scale of funding required remain uncertain, and the situation could evolve rapidly.

This episode underscores the risks of high-leverage trading. When a portfolio’s value drops rapidly, funds may face margin calls. If they cannot raise cash in time, they are forced to sell other assets or reduce positions, which in turn may further depress stock prices and trigger additional margin pressures.

At just 25 years old, Aschenbrenner is one of the most prominent figures in the AI investment world. In 2024, he published a series of articles arguing that rapidly advancing AI systems would require massive expansion in chips, memory, data centers, and power infrastructure. This thesis became the core investment strategy behind Situational Awareness after he left OpenAI.

Aschenbrenner became the valedictorian of Columbia University at age 19 before joining OpenAI’s Superalignment team. He was fired by OpenAI in 2024, with the company alleging he improperly disclosed internal information. Aschenbrenner denies this, stating he only shared a largely non-confidential planning document with external researchers for feedback.

He also warned of security vulnerabilities at OpenAI and claimed the firing was related to tensions over security issues, though OpenAI stated these concerns were unrelated to his departure.

Situational Awareness’s current forced unwinding of positions marks the most severe test of Aschenbrenner’s investment thesis to date. This event not only reflects the leverage risks of a single fund but also illustrates how overly concentrated bets on AI infrastructure can rapidly reverse when valuations of AI概念股 and capital expenditure returns come under scrutiny.