The 67% Wedding: What an AI Prodigy's Nuptials Reveal About the Leverage Trade

The number is preposterous. A fund up roughly 80% year-to-date, down 67% in July, and the 24-year-old manager still walks down the aisle in California with Jane Street, a former Tiger Global public-equity chief, and the architect of the "Situational Awareness" advisory network in attendance. The invitation carried roundtable discussions and breakout sessions. This is not a wedding. This is a capital-markets event wearing a tuxedo — scheduled precisely at the moment the fund was conducting "emergency position management." That timing is the entire story. Leopold Aschenbrenner is not a conventional fund manager. Former OpenAI research staff, author of the 165-page manifesto "Situational Awareness," he does not build models. He converts prophecy into a portfolio. His thesis is stark: computation scales, intelligence explodes, and whoever owns the compute owns the future. That worldview translates into a book that is long the highest-beta names in the AI infrastructure complex — GPU designers, liquid-cooling vendors, power and fiber plays, perhaps a fusion pre-IPO or two. In the first half of 2025, that trade printed money. In July, it met the margin clerk. And then there is the marriage itself. Avital Balwit is the chief of staff to Anthropic's CEO, Dario Amodei. She sits at the operational center of the one lab positioned as OpenAI's most credible rival. The guest list extends through Feroz Dewan, the former head of Tiger Global's public-equity desk, and Graham Duncan, the East Rock Capital figure who runs the advisory group orbiting the "Situational Awareness" document. A meaningful portion of the room carries FTX Future Fund lineage — the accelerist capital that reconstituted itself after collapse into smaller, quieter, higher-net-worth structures. Weddings like this are the new boardrooms. Now do the arithmetic. Up 80% by late June. Down 67% in the following four weeks. What portfolio construction produces that trajectory? Not a diversified book. Not a hedged one. A concentrated, leveraged book, almost certainly carrying meaningful options exposure. The mechanics deserve precision. A strategy that returns +80% and then reverses by two-thirds in a month is either running three-times gross leverage or holding long convexity — deep out-of-the-money calls whose value collapses the moment implied volatility spikes. When the AI complex sells off in unison, every down day forces dealer hedging that pushes prices lower, which forces more selling. The margin clerk becomes the co-portfolio manager. The Greeks don't lie; they just don't care about weddings. I recognize this failure architecture. During the 2017 ICO frenzy, I spent nights auditing ERC-20 contracts and found integer-overflow vulnerabilities in projects that had raised millions on white-paper promises. The common variable was never the bug. It was conviction — the founder's absolute commitment to a story that trumped risk controls. A year later came the Great Algorithmic Stablecoin Unwind, and the most interesting thing I learned is that P&L is a lagging indicator. If I see someone drawing down precisely as they double down on the same narrative, I ask which process broke, not which price. In the summer of 2020, I was running delta-neutral yield-farming structures on Compound and Uniswap, borrowing stablecoins against ETH collateral to harvest APY while shorting the tail risk. When the COMP inflation model cracked, I exited within 48 hours. I watched others ride the thesis to zero. The lesson was the same then as it is now: prophecy is not a hedging strategy. Here is what separates this blowup from a routine collapse. The wedding was not a distraction from the drawdown. It was the response to it. When a manager takes a 67% month, LPs start asking hard questions. You do not answer with a memo. You answer with a room — a room full of the most credible names in finance and AI safety, seated around structured discussion tables, with your wife at the center of the web. You signal stability. You signal access. You signal that the volatility is noise on the way to superintelligence. Whether that is true is irrelevant to the immediate capital calculus. The signaling is the strategy. The retail narrative reads this as "AI bubble pops, golden child exposed." That is comfort food. The structural reality is darker. The drawdown is not the trade. The guest list is the trade. When Jane Street attends an accelerist's wedding during a hostile drawdown, the firm is not there for cake. It is there to price the liquidation. Forced selling creates inventory, and sophisticated counterparties are already computing which assets leave the book next week — which GPU names, which compute plays, which over-levered positions — and at what discount they will absorb them. The wedding is an informal clearing event for capital about to change hands. There is also the information question, the actual scandal nobody wants to name. Balwit is the operational center of a frontier lab. Aschenbrenner runs a leveraged fund that trades on frontier-lab trajectory. The dinner-table conversation and the position-taking are not algorithmically separated. The industry pretends a Chinese wall exists. It does not. This is what "AI safety meets AI capital" looks like when formalized: a marriage certificate at the intersection of a hedge fund and its most informed source. Code is law, but bugs are justice. The bug in this system is that the governance structures are decorative, and the people setting policy are sleeping next to the front-running. The uncomfortable comparison is to NFT floor prices. In 2021, I tracked wash-trading patterns in the Bored Ape ecosystem — specific wallets inflating floor prices to trigger liquidations in lending protocols. The NFT floor is a feeling, not a number, and the same logic applies here. The fund's NAV is a narrative. Its "fundamental value" is a story about intelligence explosion that markets will either fund or refuse to fund. In a bull phase, the same narrative that produced +80% is fully capable of producing -90%. The difference between those outcomes has less to do with AI capability progress than with the liquidity cycle that governs all leveraged thematic books. So what do you watch now? The August letter — if the fund still sends one. Whether Balwit leaves Anthropic within twelve months under conflict-of-interest pressure. Whether forced deleveraging spills from his book into the broader infrastructure complex, turning one fund's drawdown into a sector-wide liquidity event. This is not the end of the trade. It is the beginning of the reallocation. The question was never whether the wedding was tone-deaf. The question is who left that room holding the bag. The Greeks don't care about ceremonies. They care about who has to sell into the next gap down.