🔥 The Situational Awareness Saga: A Masterclass in Crisis Management
The hedge fund world got its most candid crisis letter in years this week when Leopold Aschenbrenner, founder of Situational Awareness LP, sent an update to investors detailing the fund's dramatic July drawdown and subsequent portfolio restructuring. After days of speculation about liquidation, blow-up, and career implications, the letter cut through the noise with remarkable clarity.
The facts: Situational Awareness LP experienced a 67% drawdown for the month of July, triggered by extreme moves in AI-related positions and what Aschenbrenner described as "increasingly adverse trading in names publicly associated with us" — a dynamic he compared to a bank run. The fund had grown from an initial raise of $250 million less than two years ago to $4.5 billion in AUM, delivering extraordinary returns before this month's volatility.
"We came closer to permanent capital impairment than is acceptable to us. While we ultimately found a solution that protected the fund and you as investors... our fund must always be structured such that we can take a loss and fight another day."
The resolution came via a block transaction with Citadel, where a portion of the public equity portfolio was sold to remove all leverage from the fund. All short positions were closed, and the fund now operates with a fully paid-for public book — long stock and fully paid-for options with no margin or liquidity risk. The private book, including positions in Anthropic, Maddox, and Fluid Stack, remains intact.
Despite the monthly carnage, the fund remains up 80% year-to-date — a performance that would still rank among the best in the industry.
💼 The Letter That Restored Confidence
Aschenbrenner's communication strategy deserves attention. Rather than dodge, deflect, or disappear, he published a direct, unflinching letter that immediately became a case study in crisis leadership. Key elements:
- Full ownership: "I take full responsibility for these events" — a standalone paragraph with zero equivocation
- Clarity on structure: The fund is not shutting down, not liquidating, and not transforming into a private-only vehicle
- Specific risk acknowledgment: The letter explicitly used the term "bank run" to describe the dynamic, acknowledging the vulnerability without minimizing it
- Aligned interests: "Virtually all of my capital is in the fund, and I intend to work relentlessly to demonstrate that the events of this month have made me a wiser and stronger investor"
Market reaction was mixed but largely supportive. Delian Asparouhov predicted the fund would be "bigger than Citadel by the end of the decade," citing Aschenbrenner's ability to combine predictive insight with "very expensive lessons in risk." John Arnold, founder of Centaurus Energy, offered perspective: "My philosophy when I used to hire traders was that the optimal number of past blowups was one."
🎯 The Trade That Almost Wasn't
One of the more fascinating details emerged from Wall Street Journal reporting: Situational Awareness attempted to sell a $3.5 billion stake in Anthropic to a group led by Green Oaks and Sequoia. The parties reached a deal late Wednesday, but Aschenbrenner pulled out Thursday morning, choosing instead to preserve the private book while unwinding the public positions through Citadel.
This decision — to keep the crown jewel private positions rather than liquidate everything — signals conviction in the underlying thesis even as the execution strategy gets rebuilt.
📈 Big Tech Delivers, Markets React Selectively
While one AI-focused fund was navigating turbulence, the companies actually building AI infrastructure posted blockbuster results. The disconnect between fundamentals and market reaction, however, underscored the volatility Aschenbrenner referenced.
Microsoft led the pack with a historic performance:
- Q4 revenue: $64 billion, up 15% year-over-year, beating the $62.5 billion consensus
- Azure revenue accelerated 29% year-over-year
- The stock surged, adding $184 billion in market cap in a single day — the largest one-day gain for any US company in history
- Shares are up 7% over the past month
Apple also exceeded expectations:
- Quarterly revenue: $124.3 billion
- EPS: $2.40
- Market cap briefly crossed $3.9 trillion before retreating
- Despite the strong print, shares fell sharply in subsequent trading
Amazon impressed with AWS momentum:
- Total revenue climbed 10% to $170 billion
- AWS grew 19% to $27.5 billion
- Shares rallied in after-hours trading and are up over the period
Meta faced a different reaction despite solid top-line performance:
- Revenue: $48.4 billion, up 19% year-over-year
- However, the stock sold off as investors focused on elevated capex guidance and one-time costs
Alphabet delivered across the board:
- Revenue: $88.3 billion
- Google Cloud revenue surged, though specific figures varied by reporting period
The pattern is clear: even with strong fundamentals, markets are hyper-focused on AI infrastructure spending trajectories and near-term return visibility. Trillion-dollar companies are experiencing 10-15% single-day swings based on guidance nuances — unprecedented volatility at this scale.
🤖 The Model Wars: Benchmarks, Harnesses, and the Frontier
While earnings dominated headlines, the AI development race continued at breakneck pace. OpenAI made two significant moves:
1. Massive price cuts on GPT-4o mini: The company reduced costs dramatically on its efficiency-focused model, undercutting many open-source alternatives on a cost-per-task basis when accounting for token efficiency.
2. ARC-AGI v3 investigation: After surprisingly low scores on the ARC-AGI v3 benchmark, OpenAI discovered the evaluation harness was preventing the model from retaining learned patterns. Enabling two API settings tripled scores while using 6x fewer output tokens.
"We found that the harness was not letting it remember what it had learned. This is fascinating — we've seen over the past year and a half that the harness really matters a lot."
The finding highlights an under-discussed reality: evaluation infrastructure is becoming as important as model capability. As models approach the ceiling on certain benchmarks, how they're tested, what they're allowed to remember, and how tasks are structured can produce dramatically different results.
The broader implication: the "god model" that simply predicts the next token perfectly may not be the end state. Integration, scaffolding, and task-specific optimization are becoming critical layers on top of raw model capability.
🎬 The Narrative Arc
The Situational Awareness story offers several lessons beyond risk management:
- Communication matters: Transparency and directness can rebuild confidence faster than evasion
- Conviction has value: Choosing to keep Anthropic while unwinding public positions signals belief in the thesis
- Volatility is structural: AI-driven market dynamics are creating swings unprecedented for assets at this scale
- Comebacks are real: The consensus view shifted quickly from "career over" to "expensive tuition" once the facts clarified
As one observer noted, comparing this to Bill Hwang's Archegos collapse misses the point entirely. Hwang turned $200 million into $36 billion with pure leverage, ran prayer circles before trading, and caused $10 billion in bank losses when it unwound in two days. Situational Awareness executed an orderly transaction with a willing counterparty, preserved capital, and maintained operations.
The fund is down, but definitively not out. And if Aschenbrenner's track record on AI predictions holds, the second act may be more interesting than the first.
📊 Market Snapshot
Situational Awareness LP:
- July performance: -67%
- Year-to-date performance: +80%
- AUM peak: $4.5 billion
- Current structure: Fully paid-for public book, no leverage
Big Tech Moves:
- Microsoft: $184 billion single-day market cap gain
- Apple: Briefly crossed $3.9 trillion market cap
- Amazon: Strong AWS performance driving momentum
- Meta: Elevated capex concerns despite revenue beat
AI Development:
- OpenAI: Major price cuts on efficiency models
- Benchmark scores: Harness configuration can triple performance
- Evaluation infrastructure emerging as critical bottleneck
The through-line connecting all these stories: AI remains the dominant force driving both opportunity and volatility across markets, and the players who can navigate both will define the next era of finance and technology.
