Leopold Aschenbrenner’s hedge fund just made a $400 million bet on a chip startup, days after being forced to dump most of its public stock portfolio. According to TechCrunch AI, Situational Awareness poured the money into Source Foundry this week, a company founded by Stanford researchers working to make chip manufacturing faster and cheaper. That brings the fund’s total stake in Source Foundry to $500 million.
The timing is what makes this worth watching.
What happened
Situational Awareness is an AI-focused hedge fund started in 2024 by Aschenbrenner, a former OpenAI researcher in his mid-twenties who had no trading experience when he launched it. TechCrunch AI reports that early returns were strong, but the fund got hammered in recent months as AI infrastructure stocks slid.
The damage was steep. At the end of July, the fund sold off the majority of its public portfolio to Ken Griffin’s Citadel. Its assets under management reportedly fell from $20 billion to $10 billion, citing The Wall Street Journal. One position it kept: its Anthropic shares.
So the picture is a fund cutting exposure to public AI stocks while doubling down on a private semiconductor bet. That’s not a retreat. That’s a repositioning.
Why this matters
What stands out here is the direction of the money. Aschenbrenner didn’t shrink his conviction after the losses. He moved it.
A few things worth pulling apart:
- Public pain, private conviction. Selling liquid AI stocks to Citadel while writing a $400 million private check tells you where he thinks the durable value sits. Not in volatile infrastructure equities, but in the companies making the physical layer AI runs on.
- Chips are the chokepoint. Source Foundry is chasing cheaper, faster chip manufacturing. Every AI lab, cloud provider, and model builder is constrained by silicon supply and cost. A startup that meaningfully cuts either one sits at the center of the whole industry.
- The Anthropic hold is a tell. Keeping Anthropic shares while offloading nearly everything else signals a bet on frontier labs plus the hardware underneath them. That’s a barbell strategy: back the model makers and the metal, skip the middle.
The context you need
Aschenbrenner isn’t a random money manager. He wrote Situational Awareness: The Decade Ahead, the 2024 essay series arguing that AI progress will keep compounding and that compute, power, and chips become the scarcest resources on the board. This investment is that thesis with a dollar figure attached.
The status quo before this: most AI capital flowed into models, apps, and the big public infrastructure names like the GPU makers and cloud providers. Betting $500 million on a chip manufacturing startup is a step further down the stack, into the process of making the hardware itself. That’s a harder, slower, more capital-intensive place to play. It’s also where a real edge, if it works, is tough to copy.
What to watch next
A few signals to track from here:
- Whether the losses keep coming. A fund that halved its AUM is under pressure. If AI infrastructure stocks keep sliding, expect more portfolio moves and more scrutiny of Aschenbrenner’s calls.
- Source Foundry’s roadmap. $500 million buys a lot of runway. The question is whether the Stanford team can actually deliver cheaper, faster manufacturing at scale, or whether this joins the long list of hard-tech bets that burned capital.
- More private, less public. If this pattern holds, other AI-focused funds may follow the same path: trim liquid AI exposure, concentrate on private hardware and frontier labs.
My read: this is a conviction trade dressed up as a comeback. Aschenbrenner is betting that the next phase of the AI buildout is won at the chip level, not the app level. If he’s right, $500 million into Source Foundry looks early. If he’s wrong, it’s another line in a rough year.
For the full details, check the original report at TechCrunch AI.