Situational Awareness Bets US$ 400 Million on Chips After Losing Half of Its Fund

A fund that has lost half of its assets under management in just a few months typically retreats, reduces exposure, and tries to stem the bleeding. Leopold Aschenbrenner, founder of Situational Awareness, chose the opposite path. This week, his hedge fund focused on artificial intelligence invested US$ 400 million in Source Foundry, a startup founded by Stanford researchers that promises to make chip manufacturing faster and cheaper.

With this new investment, the total investment of the fund in Source Foundry reaches US$ 500 million. The move comes weeks after Situational Awareness sold most of its public equity portfolio to Citadel, Ken Griffin's firm, and saw its assets under management shrink from US$ 20 billion to US$ 10 billion.

The decision reveals a clear thesis: even in the face of significant losses in publicly traded AI infrastructure stocks, Aschenbrenner is betting on the most fundamental layer of the chain, the physical manufacturing of semiconductors. It is a bet that deserves attention.

Aschenbrenner is a former OpenAI researcher who, in his early twenties, founded Situational Awareness in 2024. With no prior experience in the financial market, he built a hedge fund entirely based on the premise that artificial intelligence would radically transform the global economy and that assets linked to this transformation were undervalued.

The initial returns were significant. The AI narrative drove up stocks of companies like Nvidia, Broadcom, and data center infrastructure providers. The fund rode this wave and grew rapidly, accumulating US$ 20 billion in assets under management. For those following the technology sector and its reflections in the markets, Aschenbrenner's initial trajectory seemed to validate the thesis that the race for AI infrastructure would be the trade of the decade.

The problem arose when the cycle turned. AI infrastructure stocks suffered sharp declines in recent months, pressured by stretched valuations and doubts about the actual speed of technology monetization. Situational Awareness, highly concentrated in these stocks, faced steep losses.

At the end of July, the fund sold most of its public equity portfolio to Citadel, Ken Griffin's conglomerate. It was a forced deleveraging move: when losses accumulate, investor redemptions compress assets under management and force liquidation of positions.

Two details stand out in this restructuring. First, the fund retained its shares in Anthropic, OpenAI's rival. This suggests that Aschenbrenner sees more value in cutting-edge language models than in the hardware companies that made up the bulk of his public portfolio. Second, even after losing half of its assets, the fund still manages US$ 10 billion, a volume that keeps Situational Awareness among the relevant funds in the sector.

The question that arises is: with less capital and a tarnished reputation, why allocate US$ 400 million to a single private chip startup? The answer lies in the long-term thesis about where value will be captured in the AI chain.

Source Foundry was founded by Stanford researchers with a straightforward proposal: to reduce the cost and time of chip manufacturing. In a market dominated by a few players like TSMC, Samsung, and Intel Foundry, any alternative that can democratize semiconductor production has enormous disruptive potential.

The geographical and corporate concentration of chip manufacturing is one of the world's biggest strategic bottlenecks. More than 90% of the most advanced chips are manufactured in Taiwan. Governments in the United States, Europe, and Japan are pouring hundreds of billions of dollars in subsidies to diversify this chain, as we have shown in analyses of Nvidia's dominance and the race for semiconductors.

If Source Foundry can deliver on its promises of more agile and accessible manufacturing, it could benefit from both the growing demand for AI chips and government incentives for reshoring production. It is a bet on the physical substrate of the technological revolution, not on surface applications.

Aschenbrenner's decision reflects a shift in posture that other institutional investors are also adopting. After two years betting on liquid AI infrastructure stocks, some smart capital is migrating to private investments in earlier stages of the chain.

The logic is that public AI stocks have already priced in much of the future growth, while startups that solve real manufacturing, energy, and efficiency bottlenecks still offer asymmetry. The move also signals that, despite recent corrections, long-term conviction in the AI thesis remains intact among the largest capital allocators.

For investors following the technology sector, the case of Situational Awareness serves as a practical study of the risks of excessive concentration. A fund that grew rapidly by betting everything on a single narrative saw its assets cut in half when market sentiment changed. But the reaction, to concentrate even more on a high-conviction bet, shows that Aschenbrenner is playing a different game than the average investor.

It remains to be seen whether Source Foundry will deliver results before the patience of the fund's investors runs out. With US$ 500 million committed to a single private startup and a drastically reduced public portfolio, Situational Awareness has staked its future on the most challenging and essential layer of the artificial intelligence chain: silicon.

This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.