An AI Hedge Fund Put $400 Million Into a Secret Chip Startup Right After Nearly Collapsing

Situational Awareness, an AI-focused hedge fund founded by former OpenAI researcher Leopold Aschenbrenner, invested $400 million in a secretive chip manufacturing startup called Source Foundry. The investment came within days of the fund nearly collapsing. The Wall Street Journal first reported the investment, which TechCrunch confirmed brought the fund's total commitment to Source Foundry to $500 million, following an earlier $100 million investment. Bloomberg reported on August 8 that the previously undisclosed $400 million had gone to Source Foundry, citing people familiar with the matter.

The timing is what makes this story unusual. At the end of July 2026, Situational Awareness sold most of its publicly traded investments to Citadel, a large investment firm run by Ken Griffin. The fund kept only its shares in Anthropic, a private AI company. TechCrunch reported that the fund's total assets fell from $20 billion to $10 billion after that sale. The Business Times reported that the $400 million investment in Source Foundry came within days of that near-collapse. So the fund, having just lost half its value and sold off most of the investments it could easily cash out, immediately put money into a private company that has not publicly shared its technology or its approach to making chips.

Source Foundry is a private startup operating in what the industry calls "stealth mode," meaning it is keeping its work secret while it develops its product. According to The Wall Street Journal, the company aims to reinvent how computer chips are manufactured. It was founded by Stanford researchers with the goal of making chip production faster and cheaper, per TechCrunch. Beyond that, the startup has not disclosed what specific technology it uses, when its products will be ready, or how it plans to reach customers.

Situational Awareness itself is a young fund. Aschenbrenner, in his mid-twenties, founded it in 2024 after leaving OpenAI, with no prior experience in investing or trading. The fund reached $20 billion in assets in about two years, making it one of the fastest-growing hedge funds in recent memory. The sale to Citadel and the resulting drop in assets happened in a short window, and the Source Foundry investment came right after.

The logic behind these moves is unusual but has a certain consistency. A fund built around AI investments sold its publicly traded AI holdings but kept its shares in Anthropic, a private company it likely could not easily sell. Then it doubled down on a private chip startup at a moment when most funds would be focused on protecting what they have left. Both decisions point in the same direction: Aschenbrenner is putting the fund's money into private, AI-related companies where he believes the potential payoff is much larger than anything available in public stock markets.

Whether that belief is correct is a separate question from whether it is a sensible risk. A $500 million total investment in a single startup that has not yet shared its technology carries a level of risk that most professional investors would find hard to justify under standard rules for spreading risk across different investments. Making computer chips is one of the most expensive and technically difficult businesses in technology. Startups in this field often take years before they produce anything usable, and many fail. Source Foundry's Stanford roots and its goal of lowering chip production costs are the only public signs of what it is trying to do, and those signs are thin.

One detail worth noting: the drop from $20 billion to $10 billion does not necessarily mean investors pulled out $10 billion in cash. The sale of the fund's public investments to Citadel, along with any drop in the value of its remaining holdings, could explain much of that decline without any investors actually withdrawing their money. The exact breakdown between investor withdrawals, sold investments, and value losses is not clear from what has been reported.

The broader context is that money flowing into AI is increasingly going in two different directions. Large, established companies and investors are pouring tens of billions of dollars into proven chip manufacturing projects and leading AI labs. Meanwhile, smaller funds making big bets on unproven startups have far less room for error. Situational Awareness, with half the assets it once had and its largest private investments still locked up, is now clearly in that second category.

The optimistic view is simple. If Source Foundry can make chip manufacturing cheaper and faster, the fund's $500 million bet could look brilliant. Lower chip production costs would make it easier for new companies to design custom chips, grow the chip design community, and push down prices charged by existing chip factories. That would benefit the entire AI hardware industry. The pessimistic view is equally simple: a fund that just lost half its value has made a large bet on a company that has shared almost nothing about itself, led by someone with no track record in investing. The next few quarters will show whether Aschenbrenner's confidence reflects real insight or a risky gamble made from a weak position.