Situational Awareness hedge fund collapse and Citadel's rescue deal

Situational Awareness ran a long-AI-infrastructure, short-software thesis, betting that compute buildout would outpace enterprise software monetization, funded with reported leverage of roughly 3-4x, with some accounts citing leverage as high as 400% ^[1][2]. When AI infrastructure and semiconductor names fell more than 30% in a month and software shorts such as Adobe moved against the fund's position, the fund's prime broker - which had already placed it on a watch list - issued repeated margin calls in its final week, forcing it to unwind positions it could not otherwise afford to hold ^[3]. Assets under management collapsed from roughly $45 billion at the peak to about $10 billion, a 67% loss for July alone ^[1][4].

Ken Griffin, together with Citadel's co-Chief Investment Officer Pablo Salame, Chief Operating Officer Gerald Beeson, Perry Vais, and Chief Legal Officer Shawn Fagan, pulled an all-nighter to evaluate and then acquire roughly $16 billion of Situational Awareness's public stock book, beating out rival bidders Millennium and Jane Street ^[1][3]. Citadel bought the portfolio at more than 10% below market value just before markets opened ^[3]- and importantly, it bought the stock book, not the fund or Aschenbrenner himself. Some market commentary framed the off-market, block-purchase structure itself as a form of liquidity arbitrage: with the fund already overleveraged, dumping the $16 billion book directly into the open market risked triggering further panic and price damage, so an off-market sale let Citadel capture the discount while containing the fallout. The move continues a recognizable pattern for Griffin, who has stepped in on distressed rivals before: recruiting Enron's energy traders after its 2001 bankruptcy, absorbing part of Sowood Capital's book in 2007, and injecting capital into Melvin Capital alongside Point72 after its 2021 GameStop short squeeze ^[1].

Hours before the Citadel deal, Situational Awareness had a different escape route lined up: on July 29 it agreed to sell $3.5 billion of its private Anthropic shares to a consortium led by Greenoaks and Sequoia Capital, only to withdraw from that sale the next morning ^[5]. The fund instead kept its private Anthropic stake, reportedly worth about $5 billion, along with other private holdings in chipmaker MatX and data-center startup Fluidstack ^[6]. That decision leaves Aschenbrenner's investors with continued exposure to Anthropic's trajectory even as the public equity book that generated the headlines is gone.

Commentary split sharply on how to read the collapse. Former hedge fund manager Martin Shkreli argued the fund's leverage made a blowup inevitable, saying 'he sort of had to blow up. There was no other ending' ^[3], while CNN framed the episode as evidence of systemic risk from leveraged, crowded AI trades across the hedge fund industry ^[7]. Aschenbrenner himself compared the margin-call spiral to a bank run - 'vulnerability begetting more vulnerability' - and told investors directly, 'we let you down' ^[1][8]. Yet the fund reportedly remained up roughly 80% year-to-date even after the 67% July loss, since the underlying long-infrastructure, short-software thesis was largely directionally correct; the failure was leverage and risk sizing, not the trade itself ^[8]. That tension surfaced in online discussion too, where a widely-upvoted but unverified community theory speculated that Citadel may have manufactured or benefited from a rate-hike scare that pressured the fund before it bought the distressed book at a discount, while other commenters pushed back that the thesis remained directionally correct - the fund was still up roughly 80% year-to-date - and that the real failure was leverage and risk-sizing rather than market manipulation.