The $35 Billion Wedding Week Disaster
Imagine being 24 years old, days away from your wedding, and watching $35 billion evaporate in real time. This is the true story of Leopold Aschenbrenner, whose hedge fund 'Situational Awareness' grew from $225 million to $45 billion in under two years—only to lose 78% of its value in a single month.
The numbers are staggering. Losing $1 million daily would take 96 years to reach $35 billion. Yet this young fund manager achieved that loss in just two weeks, triggered by a combination of extreme leverage, concentrated AI bets, and a market shift nobody anticipated.
What went wrong? And what does this collapse reveal about the current AI investment landscape?
The Rise of Situational Awareness
From OpenAI Reject to AI Prophet
Leopold Aschenbrenner's path to managing billions was unconventional. Born in Germany to physician parents, he graduated as valedictorian from Columbia University at just 19. His first job was at FTX's charitable division—a company that later collapsed in a massive fraud scandal. He then joined OpenAI's Superalignment team, focused on controlling superintelligent AI systems.
After being fired from OpenAI in April 2024 (officially for sharing brainstorming documents, though he claims it was retaliation for a security memo), Aschenbrenner published a 165-page essay titled 'Situational Awareness: The Decade Ahead.' The core thesis: AGI would arrive by 2027, and the world was dangerously unprepared.
The Fund Strategy
The investment approach was deceptively simple: go long on AI infrastructure (semiconductors, power companies, cloud providers like SK Hynix, CoreWeave, and Nebius) and short software companies (Adobe, Salesforce) that AI would supposedly disrupt.
According to community analysis, this strategy attracted significant capital from Stripe founders Collison brothers, former GitHub CEO Nat Friedman, and investor Daniel Gross. The fund used 400% leverage—borrowing $3-4 for every $1 of equity.
For a deeper understanding of AI hardware performance metrics, see our AI laptop performance comparison guide.
The Collapse: When Both Sides Fail
July 2026: The Perfect Storm
In July 2026, Chinese open-weight models demonstrated they could perform most AI tasks at a fraction of the cost. The market panicked. According to public filings, the fund's six largest long positions—all AI infrastructure stocks—plummeted between 50% and 78% from recent highs.
Simultaneously, the software stocks Aschenbrenner had shorted actually rose. The hedge had failed catastrophically: both sides of the trade lost money simultaneously.
The Leverage Problem
| Model | Core Spec | Est. Price | User Rating |
|---|---|---|---|
| SanDisk (Long) | Memory storage | -57% in one month | N/A |
| SK Hynix (Long) | HBM memory | -50%+ decline | N/A |
| Adobe (Short) | Creative software | +15% gain | N/A |
| Salesforce (Short) | CRM software | +12% gain | N/A |
With 4x leverage, a 57% drop in SanDisk meant the equity backing that position was completely wiped out. Banks—Bank of America, Goldman Sachs, JP Morgan Chase—demanded more collateral. When the fund couldn't meet margin calls, forced liquidation followed.
On July 30, approximately $16 billion of the portfolio was sold overnight at a 10% discount to market price. The buyer? Citadel, one of Wall Street's most successful hedge funds. For insights into emerging technology trends that could impact AI investments, explore our BCI technology race analysis.
Lessons from a $35 Billion Loss
Conviction vs. Risk Management
Cliff Asness, a 30-year hedge fund veteran, cited the 'Kelly Criterion' when analyzing this collapse. The formula determines optimal bet sizing based on conviction and edge. His verdict: 'Never use full Kelly.' The math assumes your assumptions are correct—and in volatile markets, they often aren't.
The fundamental flaw? Aschenbrenner's long and short positions were based on the exact same assumption: that AI would transform everything rapidly. This wasn't a hedge—it was doubling down on a single thesis with borrowed money.
The Unresolved Question
Aschenbrenner's AGI timeline might still prove correct. GPT-6 and similar models show remarkable capabilities. But as this case demonstrates, being right about the destination doesn't matter if you can't survive the journey.
The fund lost $35 billion but survived—barely—thanks to $5 billion in private holdings, including a stake in Anthropic (where his wife serves as Chief of Staff to the CEO). Days after the Citadel deal, he wired $400 million to an anonymous private company, adding to a $100 million investment made a month earlier.
📅 Information Date: 2025-01-15
What do you think? Was Aschenbrenner simply early, or does this collapse reveal deeper uncertainties about AI's future? The bond market is watching nervously—and this story is far from over.
