Aschenbrenner's AI thesis could be correct, his timing and leverage were not
Quick Answer
Leopold Aschenbrenner's hedge fund, Situational Awareness, faced a rapid decline from a 439% return to a fire sale due to leverage and falling AI stock prices, leading to a forced sell-off of nearly its entire portfolio.
Quick Take
Despite his thesis on AI-driven infrastructure growth, rising costs and margin calls undermined investor confidence.
Key Points
- Aschenbrenner's fund peaked at $45 billion with only eight employees.
- He reported a 439% return for the first half of 2024 before the downturn.
- The fund's leverage amplified losses, triggering an emergency sell-off.
- Falling AI stocks like SK Hynix contributed to the fund's rapid decline.
- Aschenbrenner retains private holdings, including a stake in Anthropic.
DeepSignal Analysis
What happened
Leopold Aschenbrenner's hedge fund, Situational Awareness, experienced a dramatic decline from a 439% return to a forced sell-off of nearly its entire portfolio due to leverage and falling AI stock prices. His thesis on AI infrastructure growth was overshadowed by rising costs and margin calls, leading to diminished investor confidence.
Key evidence
- Aschenbrenner's hedge fund peaked at $45 billion in assets but employed only eight people, four of whom were involved in investments.
- The fund reported a 439% return for the first half of the year before falling AI stocks and margin calls triggered an emergency sell-off.
- Aschenbrenner's investments included a stake in Anthropic, while his bet against software stocks like Adobe did not yield positive results.
Why it matters
This case illustrates the risks associated with high leverage in investment strategies, particularly in volatile markets like AI. Aschenbrenner's experience highlights how quickly investor sentiment can shift, especially when faced with rising costs and market corrections. The situation raises questions about the sustainability of AI-driven growth and the financial strategies employed by hedge funds.
📖 Reader Mode
~2 min readLeopold Aschenbrenner's AI hedge fund Situational Awareness had to sell nearly its entire publicly traded stock portfolio to Ken Griffin's firm Citadel after racking up steep losses.
Aschenbrenner, now in his mid-20s and with no prior trading experience, started working for the FTX Future Fund in February 2022. The fund was the philanthropic arm of Sam Bankman-Fried's crypto exchange, which later collapsed. Aschenbrenner left before the implosion. In 2023, he joined OpenAI's "Superalignment" team but was fired in April 2024 over an alleged information leak. Aschenbrenner disputed that account, saying he'd only shared a largely harmless document with three outside researchers and had previously flagged security gaps at OpenAI.
That same year, Aschenbrenner gained a large following with his essay "Situational Awareness," which predicted rapid AI growth. In 2024, he launched the hedge fund by the same name. According to CNBC, it peaked at $45 billion in assets but had just eight employees, four of them on the investment side.
From 439 percent returns to a fire sale in days
As recently as last Friday, Aschenbrenner reported a 439 percent return for the first half of the year, according to the FT, and called on investors to put in fresh capital by August 1. Days later, falling AI stocks like SK Hynix and margin calls from banks forced the heavily leveraged fund into an emergency sell-off. His bet against software stocks like Adobe also backfired. Aschenbrenner is keeping his private holdings, including a stake in Anthropic.
Aschenbrenner's thesis was that increasingly powerful AI would drive massive buildouts of chips, memory, data centers, and power infrastructure. That buildout is happening. But the enormous spending and rising financing costs made investors question returns, triggering a sell-off. What did him in was the leverage. Borrowed money amplified the losses and forced sales right before some of the stocks bounced back.
— Originally published at the-decoder.com
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from The Decoder
See more →
An AI model programmed nonstop for 19 days on a single MirrorCode task that cost $2,600 to run
Epoch AI's MirrorCode benchmark reveals Claude Opus 4.7 as the leader with a 56% solve rate, reconstructing a 16,000-line toolkit in 14 hours. Despite this, all models tested struggle with the most complex tasks, highlighting limitations in current AI capabilities. The single task consumed $2,600 over 19 days, raising questions about cost-effectiveness in AI development.
