Aschenbrenner Pours $400M Into ASML Challenger Source Foundry

Leopold Aschenbrenner just put another $400 million into a year-old stealth startup that says it will take on ASML, and he did it days after his hedge fund lost roughly two-thirds of its assets. That is the shape of the story the Wall Street Journal reported this week, with the fresh check taking his fund Situational Awareness's total position in Source Foundry to about $500 million at a $5 billion valuation.

Source Foundry was incorporated in California in July 2025 by Stanford alumni Abdulmalik Obaid and Joe Burg, and it is backed by Sequoia Capital. It is not a chip maker in the TSMC sense. The pitch, as described in the reporting, is to redesign the tooling and manufacturing processes upstream of the fab so they suit AI workloads, and specifically to challenge ASML's extreme ultraviolet lithography machines, which are today's chokepoint for cutting-edge AI silicon.

The context makes the size of the bet startling. Situational Awareness peaked near $45 billion in AUM this summer and, according to Bloomberg, fell to about $10 billion after a 67% July loss forced a fire sale of its public stock book to Ken Griffin's Citadel. Prime brokers Bank of America, Goldman Sachs and JPMorgan had come knocking with margin calls after up to 400% leverage on concentrated AI hardware names, including SK Hynix, Nebius and Micron, turned against the fund. What is left is a roughly $10 billion book weighted toward private positions such as Anthropic, and now a much larger relative stake in Source Foundry.

The honest caveat is that Source Foundry is stealth and, on the reporting, pre-product. Displacing ASML has been attempted before and the physics is unforgiving; the $5 billion valuation is a bet on a team and a thesis, not a shipped tool. What the reporting doesn't give you is which manufacturing step Source Foundry is actually attacking, whether any foundry customer is engaged, or whether this week's $400 million was fresh primary issuance or a secondary from an existing holder.

The interesting thing to watch is not the mark. It is whether Aschenbrenner is signaling a broader move into private AI infrastructure now that his leveraged public thesis has been forcibly unwound, and whether any credible fab customer ever validates Source Foundry's approach. If either happens, the shape of the AI-hardware capex debate changes.