Situational Awareness just put another $400 million into Source Foundry, a chip manufacturing startup trying to attack the part of AI that money alone can't quickly fix.
Situational Awareness, the fund run by 24-year-old Leopold Aschenbrenner, has invested $500 million in Source Foundry, including a fresh $400 million infusion this week. The Wall Street Journal reported the deal on August 7, days after the AI-focused hedge fund sold most of its public stock portfolio to Citadel following heavy losses. That timing is the story. A fund that just got punished for crowded AI trades is now putting real money into the machines behind the machines.
Source Foundry isn't designing another accelerator. It's trying to change how advanced chips are made. According to the Journal, the California startup was founded in 2025 by Stanford researchers Abdulmalik Obaid and Joe Burg and is working on lithography tools, the equipment used to print circuit patterns onto silicon. That puts it in the path of ASML, the Dutch company whose extreme ultraviolet systems sit at the center of leading-edge chip production.
ASML is not an easy target. Its EUV machines took decades to build, depend on a deep supplier network, and sell into a customer base that includes the world's largest chipmakers. So a $5 billion valuation for a young, pre-revenue lithography startup is not a small compliment. It's a bet that AI demand has made even an outside chance at loosening ASML's grip worth paying for now, not later.
Look at the money trail. Source Foundry has already been backed by Sequoia Capital, according to the Journal. California business registry data also lists Source Foundry Inc. as filed on July 18, 2025, with Abdulmalik Obaid as chief executive and chief financial officer and Joe Burg as secretary at a San Francisco address on Harrison Street. That dry detail matters. This isn't just a fund name attached to a pitch deck.
The strange part is who wrote the latest check.
Situational Awareness had been one of the loudest financial expressions of the AI infrastructure trade. Bloomberg reported on July 30 that its assets had fallen to about $10 billion after Citadel stepped in to buy most of its public equity investments. That's a steep fall. The same Bloomberg report said the fund had more than $20 billion in assets at the end of May, based on earlier Wall Street Journal reporting, before AI-related stocks rolled over and brokers pulled back the credit that had propped up its bets.
Reuters also reported that Situational Awareness sold the bulk of its stock portfolio to Ken Griffin's Citadel after losses in tech holdings, with people familiar with the matter naming the pressure from heavy AI exposure. Other reports put the July drawdown around 67%. However you phrase it, the fund didn't trim a position. It had to get smaller fast.
That makes the Source Foundry deal sharper. This isn't a comfortable victory lap. It's a wounded fund choosing private semiconductor tooling over the public AI names that just hurt it.
The bet is on fabrication - not another model #
Here's the thing: this is more interesting than the headline number. Hedge funds normally like positions they can sell when the facts change. Lithography tools don't work like that. They take time, manufacturing skill, customer trust and a tolerance for expensive failure. You can't trade your way out of a half-built machine.
But you can see why the bet exists. AI companies can talk about model progress all they want, but the hard limit keeps showing up in physical form: chips, memory, power, servers and the tools needed to make them. If Source Foundry can simplify or speed up any part of advanced lithography, it touches a real bottleneck. If it can't, the valuation will look like another peak-AI check written when fear of missing out outran proof.
Aschenbrenner is not new to that argument. His 2024 essay series, Situational Awareness: The Decade Ahead, made him a well-known voice in AI circles before his fund became a Wall Street spectacle. Axios described the work at the time as a 165-page paper arguing that powerful AI systems could arrive faster than most institutions were ready for, and Aschenbrenner's own site says he previously worked on OpenAI's Superalignment team.
The fund's public blow-up does not make the chip thesis wrong. It does make the next swing less forgiving. Source Foundry is barely a year old, it has not shipped a commercial lithography tool, and ASML did not become central to semiconductor manufacturing by accident. That is the risk sitting under the $400 million check.
Neither Situational Awareness nor Source Foundry has publicly laid out detailed terms beyond the reported investment. The open question is simple enough: whether Source Foundry is a real manufacturing breakthrough in its earliest form, or just the next place AI money went after public stocks got too painful to hold. For now, Aschenbrenner's fund has answered with capital, not commentary.
Also read: Unitree's Shanghai IPO Prices China's First Humanoid Robot Maker at $9 Billion; Xbox Series X Price Jumps to $799.99 as AI Chip Demand Drains Memory Supply; AI Model Distillation Becomes the New Battleground Between the US and China