Situational Awareness Sells Most Public Holdings After Leveraged AI Losses

Situational Awareness, the hedge fund created by former OpenAI researcher Leopold Aschenbrenner, has reportedly sold most of its public-stock portfolio to Citadel after a sharp reversal in leveraged AI infrastructure bets. The transaction follows a period in which assets had grown rapidly before losses forced the fund to reduce risk.

Aschenbrenner built the strategy around the belief that artificial intelligence would require enormous spending on semiconductors, memory, energy and compute. The fund reportedly returned 439% through June and at one point managed as much as $45 billion. Positions in companies including SK Hynix, Sandisk, Bloom Energy and Nebius then fell more than 30% during a broader selloff.

Borrowed money amplified the decline. Citadel purchased much of the public portfolio, and reported assets at Situational Awareness fell to roughly $10 billion. Citadel already held some related infrastructure investments and has a history of buying assets from leveraged investors that need to unwind quickly.

The fund retained private-company positions, including an Anthropic stake reportedly valued at $5 billion, along with investments in MatX and Fluidstack. Anthropic's expected public offering could generate a significant gain, but private valuations are less liquid and may not provide immediate protection during a public-market drawdown.

Early investors reportedly included Jane Street, Stripe's founders and senior Meta figures, giving the fund credibility before it established a long public record. The rapid rise attracted additional capital that increased the size of each decision. The unwind is a reminder that prestigious backers and technical intelligence do not substitute for risk controls developed through multiple market cycles.

The episode highlights the difference between a public stock portfolio and private holdings that cannot be sold easily. Anthropic shares may preserve substantial value even after a hedge fund changes its liquid positions, but that value depends on the company's next financing or exit. Investors should examine concentration, lockups and how the fund marks private assets. A strategy associated with AI expertise can still face ordinary market and liquidity risk. Transparent reporting will be essential if the fund raises again or offers investors a new vehicle with similar exposure.

The episode does not disprove the long-term demand for AI infrastructure. It demonstrates the difference between a durable thesis and a survivable portfolio. Even a correct view can fail when leverage, timing and concentration force a sale before the underlying market recovers.