Key Takeaways
- • Leopold Aschenbrenner's Situational Awareness AI hedge fund collapsed 67% in July 2026, wiping out approximately $35 billion in value as assets plummeted from roughly $45 billion to $10 billion in a single month.
- • SEC filings dated August 14 reveal that nearly 50% of the fund's US equity portfolio was concentrated in SanDisk and Micron as of June 30, exposing the fund to severe concentration risk in memory semiconductor stocks before the liquidation event.
- • The fund's March 31 quarterly filing showed $13.68 billion in 13F securities, with Bloom Energy as the top long position at $879 million, alongside significant put option positions totaling $8.7 billion across chip companies, indicating a complex hedging strategy that failed to protect against the drawdown.
- • Market shifts from AI training to inference workloads drove enterprise SSD adoption to 48% of total NAND shipments in Q2 2026, initially supporting memory chip demand but ultimately insufficient to prevent the fund's collapse and forced liquidation of its entire public stock portfolio.
- • The fund reportedly executed a block trade selling its entire remaining stock portfolio to Citadel as part of the unwinding process following the catastrophic July losses.
Summary
Leopold Aschenbrenner's Situational Awareness LP experienced a dramatic collapse in July 2026, losing 67% of its value in a single month—a reversal that transformed the AI-focused hedge fund from approximately $45 billion in assets to roughly $10 billion. SEC filings submitted on August 14 provide critical detail on the positioning that preceded this liquidation event, revealing a fund heavily concentrated in semiconductor exposure at precisely the wrong moment.
According to the fund's 13F-HR filing dated June 30, nearly half of Situational Awareness' US equity portfolio was invested in just two memory chip manufacturers: SanDisk and Micron. This extreme concentration in a single subsector of the AI supply chain left the fund acutely vulnerable to adverse price movements in memory semiconductors. The fund's March 31 quarterly filing, which disclosed $13.68 billion in 13F securities, showed Bloom Energy as the largest long position at $879 million, with SanDisk and CoreWeave also comprising significant holdings. Additionally, the fund maintained $8.7 billion in put options related to chip companies, suggesting management attempted to hedge downside risk through derivatives.
The fund's heavy weighting toward memory stocks reflected a thesis about AI infrastructure demand, specifically betting on the transition from AI model training to inference workloads. Counterpoint Research data supports this logic: enterprise SSD adoption reached 48% of total NAND shipments in Q2 2026 as the market shifted inference-focused architectures requiring substantial memory storage capacity. However, this structural thesis proved insufficient to insulate the fund from broader market dislocation or execution challenges in July.
The liquidation forced Aschenbrenner to unwind the entire public stock portfolio. According to reports, the fund executed a block trade of its remaining equity holdings with Citadel, representing a fire-sale exit from public markets. The gap between the fund's apparent positioning as of June 30 and its July collapse suggests either rapid market deterioration in semiconductor equities, broader redemption pressures, or margin calls tied to the fund's options positions that forced accelerated deleveraging.
According to the fund's 13F-HR filing dated June 30, nearly half of Situational Awareness' US equity portfolio was invested in just two memory chip manufacturers: SanDisk and Micron. This extreme concentration in a single subsector of the AI supply chain left the fund acutely vulnerable to adverse price movements in memory semiconductors. The fund's March 31 quarterly filing, which disclosed $13.68 billion in 13F securities, showed Bloom Energy as the largest long position at $879 million, with SanDisk and CoreWeave also comprising significant holdings. Additionally, the fund maintained $8.7 billion in put options related to chip companies, suggesting management attempted to hedge downside risk through derivatives.
The fund's heavy weighting toward memory stocks reflected a thesis about AI infrastructure demand, specifically betting on the transition from AI model training to inference workloads. Counterpoint Research data supports this logic: enterprise SSD adoption reached 48% of total NAND shipments in Q2 2026 as the market shifted inference-focused architectures requiring substantial memory storage capacity. However, this structural thesis proved insufficient to insulate the fund from broader market dislocation or execution challenges in July.
The liquidation forced Aschenbrenner to unwind the entire public stock portfolio. According to reports, the fund executed a block trade of its remaining equity holdings with Citadel, representing a fire-sale exit from public markets. The gap between the fund's apparent positioning as of June 30 and its July collapse suggests either rapid market deterioration in semiconductor equities, broader redemption pressures, or margin calls tied to the fund's options positions that forced accelerated deleveraging.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.