On August 15, the SEC received a 13F filing that should have been a footnote. Instead, it became a stress test for the entire AI infrastructure thesis. Leopold Aschenbrenner's Situational Awareness LP, a fund known for its macro-thematic bets on AI, revealed a portfolio transformation that reeks of conviction bordering on systemic risk. As of June 30, 2026, the fund had abandoned its long-short structure and gone all-in on a narrow set of AI hardware and infrastructure names. Micron's position surged from $5.86 million to $5.574 billion. SanDisk jumped from $724 million to $5.674 billion. Together, these two storage chip makers accounted for 55% of the fund's publicly disclosed equity portfolio. Bloom Energy, TSMC, Nebius, and a suite of crypto-adjacent miners—CoreWeave, Core Scientific, Applied Digital, IREN, Riot—rounded out the exposure. The puts that had hedged the fund in Q1 were gone. The message was clear: this is a directional bet on the AI compute supercycle, and the fund is not hedging against its own conviction.
To understand what this means, we need to step back and map the global liquidity landscape. The Q2 2026 filing captures a snapshot before the July sell-off that hammered the Philadelphia Semiconductor Index into a rare monthly decline. Since then, SanDisk, Micron, and CoreWeave have rebounded on cooling inflation and AI earnings sentiment, but the volatility has exposed the structural fragility of concentrated positions. This is not a story about one fund. It is a case study in how macro allocators are mispricing correlation risk in the AI infrastructure trade. And for anyone managing digital assets, the lessons are directly transferable.
The Concentrated Core: A Liquidity Audit
We do not predict the wave; we engineer the hull. Let me apply the same framework I used during the 2020 DeFi liquidity stress tests to Aschenbrenner's portfolio. The first red flag is the sheer size of the Micron and SanDisk positions relative to the fund's total disclosed holdings. At 55%, any sector-level drawdown translates directly into portfolio-level distress. The second red flag is the absence of hedges. The Q1 filing showed significant put positions on SMH, NVIDIA, Broadcom, AMD, Oracle, Micron, and TSMC. By Q2, those puts were largely gone. This is a fund that went from managing tail risk to ignoring it entirely.
Based on my experience auditing protocol failures in 2022, I can tell you that the most dangerous moment in any cycle is when conviction replaces risk management. The Terra-Luna collapse was not a surprise to those who understood the liquidity mechanics. The same logic applies here. When a portfolio is 55% concentrated in two correlated names—both tied to the same demand curve for AI chips—any disruption in the supply chain, a shift in capital expenditure cycles, or a regulatory crackdown on data center power consumption can trigger a cascade. Add leverage, which is likely given the fund's size and the nature of these positions, and a simple correction becomes a liquidity crisis.
The July sell-off was a warning shot. The Philadelphia Semiconductor Index fell, and the fund's top holdings were hit simultaneously. The rebound in August does not erase the structural risk. It only delays the reckoning. The question is not whether the AI infrastructure thesis is correct. The question is whether the market has priced in the fragility of the execution path.
The Crypto Connection: Mining, Storage, and the Energy Arbitrage
Now, let me connect this to the digital asset world. The fund's exposure to CoreWeave, Core Scientific, Applied Digital, IREN, and Riot is not accidental. These are companies that sit at the intersection of AI compute and Bitcoin mining. They are the physical backbone of the most capital-intensive sectors in the crypto economy. Block rewards depend on hashrate. Hashrate depends on energy. Energy depends on infrastructure. And infrastructure depends on the same semiconductor supply chain that Micron and SanDisk serve.
We do not predict the wave; we engineer the hull. In my 2017 ICO audit work, I learned that the most scalable systems are the ones that standardize their inputs. The AI infrastructure trade is doing the opposite. It is concentrating inputs into a handful of vendors and asset classes. The result is a system that looks efficient in bull markets but brittle in drawdowns. For crypto miners, the correlation to AI hardware stocks is a double-edged sword. When the narrative is hot, funding flows freely. When the narrative cools, margin calls ripple through the entire ecosystem.
Consider the energy component. Bloom Energy is a fuel cell company that provides clean power to data centers. Its stock is a proxy for the electrification of AI compute. But the same regulatory and operational risks that apply to Bloom apply to Bitcoin mining facilities. In both cases, the profitability depends on the cost of electricity and the price of the output (AI compute or Bitcoin). The correlation is not perfect, but it is high enough to create systemic risk when the macro environment shifts.
The Contrarian Angle: Decoupling or Collapse?
The conventional wisdom is that AI infrastructure is a long-term secular trend that will continue regardless of short-term volatility. The contrarian view, which I hold, is that the current concentration of capital is a sign of the top of the cycle. The decoupling thesis—that crypto assets will decouple from traditional AI stocks—is plausible only if the underlying liquidity remains intact. But if the AI hardware trade undergoes a correction, the miners and infrastructure companies that depend on the same capital flows will suffer. The decoupling will not happen because the fundamentals are the same.
We do not predict the wave; we engineer the hull. The true test of the AI infrastructure thesis is not the next earnings report. It is the ability to withstand a 30% correction in Micron without triggering a forced liquidation that drags down the entire portfolio. The 13F filing shows that Aschenbrenner's fund is not designed for that scenario. It is designed for a smooth upward trend. That is not engineering. That is gambling.
The Macro Takeaway: Positioning for the Next Cycle
What does this mean for the digital asset manager reading this? The answer is simple: look at your own portfolio and ask whether you are concentrated in any single narrative. The AI infrastructure trade is the most crowded it has been since the 2021 NFT mania. The same emotional FOMO that drove Bored Ape prices to absurd levels is now driving capital into storage chips and mining stocks. The difference is that the scale is larger, and the leverage is hidden.
In my own fund, I have been reducing exposure to miners and AI hardware proxies. Instead, I am focusing on assets that benefit from the next phase of the cycle: liquidity normalization and regulatory clarity. The 2024 ETF approval was a step toward institutionalization. The next step is a correction that cleans out the weak hands. The question is not whether it will happen. The question is whether you will be positioned to survive it.
The 13F filing from Situational Awareness LP is a warning. It is not a prediction of doom. It is a data point that every macro watcher should analyze. The market is telling us that the AI infrastructure trade is stretched. The storage chip positions are the canary in the data center. Listen to the canary. Engineer your hull accordingly.
