According to recent documents filed with the U.S. Securities and Exchange Commission (SEC), hedge fund Leopold Ashenbrenner's Situational Awareness AI, before its liquidation in July which wiped out 67% of the fund's value in just one month, had almost fifty percent of its U.S. stock portfolio invested in SanDisk and Micron.
For the entire AI sector, this data indicates where investments have shifted. Shares of companies whose stocks have plummeted, particularly those involved in situational awareness, are also among those benefiting from the AI-focused data center boom, serving as a cautionary tale about how limited this business is becoming.
Where the Money Was Before the Crash
Situational Awareness LP filed a 13F-HR report with the SEC on August 14, detailing assets as of June 30, prior to the July liquidation. SanDisk and Micron accounted for approximately 50% of the reported equity assets, making the fund highly vulnerable to one aspect of the AI supply chain.
This was not the fund's only significant investment. Another quarterly 13F report for the quarter ended March 31, which was published by Cryptopolitan on July 30, listed a U.S. stock portfolio worth $5.52 billion and put options worth $8.7 billion related to chip manufacturing companies. The largest long position was Bloom Energy, with an investment of $879 million, followed by SanDisk and CoreWeave.
The March 31 Situational Awareness report shows $13.68 billion in 13F securities, but this figure should not be confused with the fund's total assets or its overall economic risk exposure.
Why Memory Chips Became the Core of AI Trading
The demand stimulus from SanDisk and Micron was accompanied by an actual shift in demand. Counterpoint Research analysts noted that as AI tasks transition from training to inference, enterprise SSD drives accounted for 48% of total NAND flash shipments in Q2 2026.
Prices are expected to follow this trend. According to TrendForce, NAND Flash sales revenue is projected to grow 10-15% quarter-on-quarter in Q3 2026, while DRAM prices will rise 13-18% due to high demand for AI data processing and purchases for major data centers. Micron also reported a record Q3 fiscal year profit of $41.46 billion, significantly exceeding the $9.30 billion profit from the same period last year.
The Lesson Learned from the July Training
Understanding the right trends did not protect the fund from losses. Situational Awareness informed its investors that the fund recorded an unaudited loss of 67% in July, while year-to-date returns had been close to 80%. Ashenbrenner noted that after exhausting liquidity, the fund sold part of its public portfolio to reduce leverage. Reuters confirmed the 67% loss for July and the sale of most public assets.
Ashenbrenner directly acknowledged the risk in his letter to investors:
"We welcome volatility. But it should by no means jeopardize the fund."
He also wrote that the fund took steps to "continue the fight."
On July 28, an anonymous Trac user estimated Ashenbrenner's portfolio losses at $600 million due to declines in Bloom Energy and SanDisk stock. However, Cryptopolitan could not verify this information; furthermore, 13F reports do not specify cash, short positions, or personal investments.
Ashenbrenner remains strongly committed to AI. His investment firm has put $400 million into Source Foundry, a semiconductor equipment startup valued at nearly $5 billion that specializes in lithography—a bottleneck in chip manufacturing dominated by ASML.
What Does the $7 Trillion Build-Out Depend On?
This comes amid forecasts pointing to colossal AI investments. Goldman Sachs believes that between 2026 and 2031, nearly $7.6 trillion will be spent on computing power, data centers, and energy.
The Semiconductor Industry Association and Deloitte assert that semiconductors account for over 95% of the cost of an AI server rack, and data center infrastructure investments could potentially reach $4 trillion by 2028.
It is precisely because of this scale that the fund's failure matters beyond just one manager. Confidence in AI growth does not guarantee that all links in its supply chain will develop in sync, or that leveraged, concentrated positions will survive a liquidity crunch. The documents show an investor who correctly assessed the requirements but still got caught out by the way he financed his bet.
