Situational Awareness bets $400M on chip startup Source Foundry
Leopold Aschenbrenner's hedge fund, weeks removed from a near-death margin-call spiral, is back to writing nine-figure checks — this time into the machines that make the chips.
Leopold Aschenbrenner's Situational Awareness has poured $400 million into Source Foundry, a stealth chip-manufacturing startup taking aim at ASML's lithography monopoly — days after the fund nearly collapsed. Bloomberg reported the investment, which brings the fund's total commitment to the Sequoia-backed company to $500 million, and the Wall Street Journal first identified the recipient. Founded in 2025 by Stanford researchers Abdulmalik Obaid and Joe Burg, Source Foundry is developing simpler, cheaper, faster semiconductor manufacturing tools meant to break the industry's dependence on ASML's extreme ultraviolet (EUV) machines — the single most constrained link in the AI supply chain.
The timing is the story. Situational Awareness's assets under management peaked around $45 billion before a brutal July, when leveraged bets on public AI names triggered cascading margin calls and the fund cratered to roughly $10 billion. Aschenbrenner was forced into a fire sale, unloading part of the public portfolio to Ken Griffin's Citadel at a 10 percent discount, and told investors he was stripping leverage out of the fund. Days later, he committed another $400 million to a private hardware company whose product won't ship for years — illiquid, unhedged, and a bet on the decade rather than the quarter.
Sequoia partner Stephanie Zhan frames the thesis as the "chip wall": AI demand grows on an exponential software curve, while semiconductor manufacturing capacity grows on a linear industrial-equipment curve. Source Foundry is one of several startups — alongside Substrate, xLight, and Inversion Semiconductor — trying to crack that wall with new lithography physics instead of incremental EUV improvements. At a reported valuation around $5 billion for a company still in stealth, the check sizes say investors now treat fab equipment as the strategic chokepoint of the AI boom.
The move also reads as a deliberate pivot in Aschenbrenner's playbook: out of leveraged public-market concentration and into structural private infrastructure, the picks-and-shovels layer of compute. Whether that is discipline or just a new flavor of the same conviction is the open question — the fund's wrapper of redemptions, lockups, and gates was built for liquid trading, not five-year hardware cycles. Either way, July's lesson, that being right about AI's trajectory does not protect you from margin calls, is now being tested against a $500 million position in machines that do not exist yet.
Do you think hedge funds belong in long-cycle chip hardware bets, or is that a mismatch of liquidity and conviction? Tell us in the comments.
