Editor’s note: the story has been updated with the latest reporting about Situational Awareness’s portfolio.
Leopold Aschenbrenner turned a viral essay about artificial intelligence into Situational Awareness, a $20 billion hedge fund that returned 439% net through June.
This week, after losses in the AI selloff, he was asking his investors and lenders for more cash.
He also offered some of them the chance to buy assets straight out of the portfolio, according to a Financial Times report. One person briefed on the discussions called them ad hoc rather than a coordinated raise.
The fund did not respond to the FT.
CNBC’s David Faber reported Thursday that Situational Awareness has exited all of its public long and short positions in essentially one transaction with another major hedge fund.
The positions represented roughly two-thirds of the firm’s holdings.
The fund will continue operating with private investments, although it has also been marketing stakes in private companies, including Anthropic, to raise liquidity.
CNBC reported interest from prospective buyers but did not say an Anthropic sale had been completed.
Leverage Cuts Both Ways
The fund used borrowed money to magnify its returns, according to the FT, and in a letter to investors Aschenbrenner conceded it had not been immune to the swings, particularly in Asia.
South Korea may explain much of that Asian pain. The Kospi has fallen nearly 40% from its June record, tripping circuit breakers on consecutive days for the first time in its history and forcing the finance minister into an emergency meeting.
Disappointing results from SK Hynix (NASDAQ:SKHY) deepened the selloff, while forced selling in single-stock leveraged products accelerated it. Regulators in Seoul have announced new restrictions on single-stock leveraged products.
The FT reports the fund is staffed by eight people, four of them investment professionals, and run by a manager who had never traded professionally before launching it in 2024.
Traders Are Repricing the Bubble
Traders on Polymarket now put the odds of an AI bubble burst before year-end at 24%, up from 17% earlier this week, on a contract that has drawn nearly $2.3 million in volume.
The contract only pays out if three of six severe events happen inside a 90-day window, including Nvidia Corp. (NASDAQ:NVDA) closing 50% below its record high, the main semiconductor ETF falling 40%, or a bankruptcy at OpenAI or Anthropic.
What the Last Filing Showed
Aschenbrenner’s March 31 filing listed roughly $8.46 billion in underlying-share value tied to puts across AI linked stocks, including Nvidia, Oracle Corp. (NYSE:ORCL) and Advanced Micro Devices Inc. (NASDAQ:AMD).
That is not the amount invested, and because the filing omits premiums, strikes, offsetting positions and everything private, it says little about the fund’s net exposure.
His pitch to investors for the second half rests on an Anthropic IPO, a stake The Wall Street Journal has reported at roughly one-fifth of the fund’s assets.
Polymarket gives a 71% chance of Anthropic completing its IPO this year.
Not everyone reads this as a blowup.
Market commentator Citrini argued these LPs bought an AI maximalist thesis and remain up several times their original investment, making them unlikely sellers.
Under that theory, Aschenbrenner raises the cash, removes his hedges and marks the bottom rather than the top.
Steve Eisman, who shorted subprime before 2008, once told a Harvard audience that bank executives kept earning more every year as their firms quietly got more levered. They “mistook leverage for genius,” he said.
Image: Shutterstock
Login to comment