Leopold
America is where dreams are made real, and a margin call can (almost) wipe you out
You have to be a visionary, a dreamer, a storyteller to work in tech. Tech, after all, is making the impossible possible, bringing life to an idea which was dreamt up under the shower, selling it with enthusiasm and conviction even if there is really nothing but your words to back it up. It is about pulling people into that dream, cajoling them, getting them excited. Some people might call this manipulation. But this is venture capitalism, and this is what has build America.
Leopold Aschenbrenner’s story, if you haven’t followed it, begins in Germany. At the age of 15, in late 2016, as a speaker to a conference organized by the Green party at an event organized by the Grüne Jugend group, he sets out his disdain for the backward thinking of the past, to paint a picture of robots displacing jobs, risks to the democracy and a massive change in how people work, including the adoption of universal basic income. His speech emphasized the need for a bigger visions for digitalization/automation. He moves with his parents to the U.S. and gets into Columbia University, where he sets up the university’s Effective Altruism chapter and is a valedictorian graduating at just age 19 in 2021 with a B.A. in Mathematics and Statistics. Before getting a job at OpenAI, he expanded his CV working as a research fellow at Oxford’s Global Priorities Institute and as a grantmaker at the FTX Future Fund. If the words Effective Altruism ring a bell, it is probably because you will remember how Sam Bankman-Fried was a champion of that ideology inspired by the 19th century British philosopher Jeremy Bentham, and articulated by Oxford philosophers William MacAskill and Toby Ord, drawing heavily on Peter Singer.
After Columbia, Leopold joins OpenAI in 2023, joining the Superalignment team under Ilya Sutskever and Jan Leike — and is then let go in April 2024, officially for alleged leaks of confidential information, though Aschenbrenner maintains he was fired after sending a security memo to OpenAI’s board detailing critical cybersecurity vulnerabilities. He then self-publishes a 165-page essay called Situational Awareness in which he sets out the future as he sees it, and how most of us are embarrassingly behind the curve when it comes to technology.
The paper’s key premises are:
- OOMs (Orders of Magnitude) Growth: Rapid trends in compute, algorithmic efficiency, and un-clustering will carry AI capabilities from GPT-4 level to AGI by ~2027.
- The Intelligence Explosion: AI will rapidly automate AI research itself, compressing decades of technical progress into a hyper-accelerated timeframe (from AGI to Superintelligence/ASI in 1–2 years).
- National Security & The “Project”: AGI will become a primary national security priority, inevitably forcing a US government Manhattan Project-style nationalization by 2027–2028 to maintain a lead over China.
- Security Vulnerabilities: Current frontier AI labs suffer from inadequate defense against foreign state-sponsored espionage, putting model weights and key algorithmic secrets at risk.
It is a bold piece for a young man of 22 at the time of publication in June 2024. Leopold then decides to start an AGI-focussed hedge fund which he names “Situational Awareness” after his paper, and he raises ~US$250 million from a number of tech entrepreneurs in Silicon Valley including high-profile seed backers Patrick Collison (Stripe), Daniel Gross, and Nat Friedman. And if you know anything about how hard it is to raise money for a hedge fund when you have no track record and just an essay about your vision of the future, then you have to admire Leopold’s chutzpah.
Anyway, he reportedly grows the fund to $20 billion, then $46 billion and has a portfolio of very concentrated bets in keeping with his uber-bullish AI thesis in his essay. In order to offset his long positions he also has a collection of shorts in ancillary industries and on the Semiconductor ETF as well as a private holding in Anthropic. So far so good. Except he is massively leveraged. The word is he had 3–4x his capital, which means of course, that while positive returns are magnified (the fund was reportedly up 439% on the year in June 2026) the downside is also magnified and that’s when the prime broker — the firms that handle your trades for you and offer you leverage — will force you to liquidate.
Here is where it gets interesting. Under normal conditions, a broker will sell your shares in the market to meet the margin call. But if Leopold was running, say 3.5x leverage on a $46 billion portfolio that means he had approximately $161 billion in open positions. And dumping that much stock into a market which is already a little febrile due to talk of inflation, the risks of a prolonged conflict in the Middle East and concerns about hyperscaler CapEx, would have caused a crash. Instead, the prime brokers quietly called around and in the end, Ken Griffin’s Citadel took on the portfolio at undoubtedly a decent discount. The news of this was then messaged to the market, and the collective sigh of relief caused the Nasdaq to rally ~3% — even while the U.S. Treasury and the Bank of Japan were intervening to stop the yen selling off more against the dollar.
As an important side note, the yen has been the favored vehicle of the carry trade: basically, Japanese interest rates have been at or near (or even below) zero for a long time. This means that traders can borrow yen extremely cheaply and after converting the proceeds into dollars or another currency, can invest at higher returns. The yen carry trade is also synonymous with being “short volatility”: you are betting when you put on this trade, that not only will the yen weakness persist, but the volatility of the currency will remain low or even fall. So the carry trade is essentially a trade where you invest in a risky, higher yielding asset with proceeds borrowed in yen at very low rates.
When the yen strengthens against the dollar, as it did on Friday July 31st 2026, this usually would signal an unwinding of the carry trade. Traders would see the yen strengthening against the dollar and take off the trade in order to minimize any possible losses. This would mean selling risky assets — including tech stocks. Yes, including the same portfolio that Leopold just managed to sell to Ken Griffin in order to meet the margin calls from his prime brokers.
Get Return Seeker’s stories in your inbox
Join Medium for free to get updates from this writer.
The question then to be asking, is: did Citadel get the heads up from Treasury that the coordinated intervention together with the Bank of Japan was about to happen, or did Treasury allow Ken to do the trade, put on his hedges before they intervened? This of course, is conjecture on my part. We will never know, but I would surmise that Treasury would definitely not want to be intervening to prop up the yen for an estimated total of $53.8 billion while the stock market was imploding.
But back to Leopold. While social media is blowing up with memes and enjoying a moment of Schadenfreude for the young hedge fund manager’s fate, a couple of things come to mind.
What Leopold did was pretty unique. He managed to raise a hedge fund and run it with a skeleton staff based on his vision of the future, which may yet not necessarily turn out to be wrong, in fact. His problem was risk management. There is no doubt that Leopold is a very smart individual, with strong convictions and an ability to communicate those convictions in a highly convincing manner. For a tech entrepreneur to invest in his fund, was purely a validation of their own basic thesis. So it was basically a no brainer. But Leopold got over his skis because his bets were all highly correlated and his hedges were ineffective and probably badly sized as well. In a way he was long a single asset with massive leverage and was caught out.
I have a feeling that this will not be the last we will be seeing of Leopold Aschenbrenner. The beauty of American capitalism is that blowing up a hedge fund carries no shame, no stigma. On the contrary, it is kind of a badge of honor, worn by some of the greatest traders in the market. Leopold has just earned his in a very spectacular way.
References
References
Aschenbrenner, L. (2024) Situational Awareness: The Decade Ahead. Self-published essay series. Available at: https://situational-awareness.ai/ (Accessed: 3 August 2026).
Aschenbrenner, L. (2016) Speech to the Green Party
Burns, C., Izmailov, P., Kirchner, J.H., Baker, B., Gao, L., Aschenbrenner, L., Chen, Y., Ecoffet, A., Joglekar, M., Leike, J. and Sutskever, I. (2023) Weak-to-strong generalization: Eliciting strong capabilities with weak supervision, arXiv preprint, arXiv:2312.09390.
Columbia College (2021) Columbia College Announces 2021 Valedictorian and Salutatorian. New York: Columbia University. Available at: https://www.college.columbia.edu/news (Accessed: 3 August 2026).
Goldman, S. (2025) How former OpenAI researcher Leopold Aschenbrenner turned a viral AI prophecy into profit, Fortune, 8 October. Available at: https://fortune.com (Accessed: 3 August 2026).
Inc. Magazine (2026) Former OpenAI Researcher’s Hedge Fund Lost 67 Percent in July, Inc. Magazine, 31 July. Available at: https://www.inc.com (Accessed: 3 August 2026).
MacAskill, W. (2015) Doing Good Better: Effective Altruism and a Radical New Way to Make a Difference. London: Guardian Books.
Rudegeair, P. (2026) The 24-Year-Old AI Wiz Who Counts Jane Street as an Investor, The Wall Street Journal, 8 June.Available at: https://www.wsj.com (Accessed: 3 August 2026).
Originally published at https://returnsseeker.substack.com on August 3, 2026.
