Situational Awareness’s 13Fs show a shift from chip puts to AI stocks

Correction — September 12, 2026: This article previously treated put-option values as evidence of the fund’s net short exposure and described absent options as closed trades. Form 13F does not establish either conclusion. We corrected the reported-value labels, clarified the assumptions behind the illustrative July loss calculation, replaced the holdings chart and removed a chart whose title asserted causation that its price data did not establish.

Situational Awareness's March 31 filing disclosed large put positions on the chip trade before the fund's reported July losses. Its five largest rows were puts on the VanEck semiconductor ETF, Nvidia, Oracle, Broadcom and AMD. Put rows accounted for 61.9% of the $13.7 billion in reported 13F value. That percentage measures the value of the underlying shares, not option premiums or the share of fund capital committed to bearish bets; it does not establish the fund's net exposure.

By June 30, the disclosed mix had changed. The quarterly filing it submitted in August shows $20.2 billion in reported holdings value across 26 line items, down from 42. Share positions, rather than option rows, made up 99.6% of that value, and SanDisk and Micron accounted for 55.6%. Micron and Taiwan Semiconductor appeared as share holdings in June; the March filing had included puts on both companies.

The fund reportedly lost roughly 67% in July, and JPMorgan has now ended its lending relationship with it, Reuters reported on Friday, citing the Financial Times. The filings show a substantial change in disclosed instruments and concentration. They cannot establish a complete reversal of the fund’s net exposure or explain its actual trading losses.

What the fund actually owned

The disclosed June holdings were concentrated in memory and the power behind it. SanDisk was 28.0% of reported value, at 2,495,344 shares. Micron was 27.5%, at 4,828,786 shares. Bloom Energy’s share position came next at 9.4%, then Taiwan Semiconductor’s share position at 6.2% and Nebius at 6.1%. Below that sat a long tail of AI-infrastructure names TECHi readers will recognise: CoreWeave, Core Scientific, Applied Digital, IREN, Riot, CleanSpark, Bitdeer and Hive.

Form 13F is a quarter-end snapshot of reportable securities, including certain US-traded foreign shares and listed options. It omits short positions and does not provide a complete account of derivatives, borrowing or net exposure. Option rows report underlying-share value rather than option premiums. The useful finding here is narrower: SanDisk and Micron together represented more than half of June’s reported 13F value.

What sat underneath them is its own argument. Bloom Energy, a fuel-cell maker selling on-site power to data centers, was the third-largest line at 9.4%. Nebius, the neocloud spun out of Yandex, was fifth at 6.1%. Add CoreWeave, Core Scientific, Applied Digital, IREN, Riot, CleanSpark, Bitdeer, Hive, Solaris Energy Infrastructure and Keel Infrastructure, and roughly a fifth of the book was a single wager that AI compute is constrained by electricity and the buildings that hold it. That is a coherent thesis, and TECHi has made a version of it repeatedly. The reported margin pressure shows why concentration and financing need to be assessed together. The filing alone does not quantify how much this exposure was leveraged.

The put rows for Nvidia, Oracle, Broadcom, AMD, ASML, Intel and the VanEck semiconductor ETF appear in March but not in June. Their absence does not tell us whether contracts were sold, expired, exercised or replaced by exposures outside the filing. Micron and Taiwan Semiconductor appeared as share holdings in June after March put disclosures; that change does not, by itself, establish a switch in the fund’s net position. June also included 10,000 Cerebras Systems shares with reported value of $2.21 million, or 0.01% of the total.

Holding the June 30 share counts constant gives an illustrative measure of the price shock. SanDisk fell 46.6% in July and Micron fell 28.7%. On that unchanged-holdings assumption, the two positions would have declined by $4.24 billion by July 31. The sum of declines to each stock’s separate July low is $5.14 billion; it is not necessarily a simultaneous portfolio loss. Actual trades and hedges could materially change the result. Other disclosed holdings exposed the fund to related AI-infrastructure risks. The stock declines alone do not establish a deterioration in the businesses: SanDisk is still up more than 1,700% over twelve months and Micron more than 500%.

View full-size image (opens in a new tab)

Source: SEC Form 13F-HR information tables for March 31 and June 30, 2026; TECHi calculations. Percentages divide each instrument category’s reported value by that quarter’s total reported value. Options use underlying-share value, not premiums. The chart does not measure NAV or net exposure.

How leverage can amplify concentrated losses

Reporting on the collapse has put the fund’s peak assets near $45 billion, falling to roughly $10 billion, with leverage of up to 400%. The meaning of that ratio matters: at four times assets to equity, a 15% asset decline would reduce equity by about 60%, before financing costs and trading. It would not mechanically wipe out all equity, although lender margin requirements could force sales sooner. The 13F itself does not disclose the fund’s financing terms.

The sequence that followed is the one prime brokerage is designed to produce. Margin calls arrived from multiple lenders. The fund sold roughly $16 billion of listed equities to Citadel, which bought the positions at a discount of about 10% and has since unwound more than 80% of that risk through nearly 100 block trades worth over $4 billion. Ken Griffin told clients the deal "could not have been completed without the extraordinary cooperation of the trading and prime brokerage teams at the banks serving both firms," according to Disruption Banking.

That cooperation is now the subject of an investigation. The SEC has subpoenaed Goldman Sachs, JPMorgan, Citigroup and Bank of America, examining the timing of the trades that triggered the margin calls and the banks' communications with the fund about its use of leverage, Reuters reported in August after the New York Times broke the story. The fund said it expects regulators to examine funds that "produce significant returns, or have particularly dramatic drawdowns" and will cooperate. Leopold Aschenbrenner, the former OpenAI researcher who founded it in 2024, wrote to investors that the firm "came closer to permanent capital impairment than is acceptable to us."

Why this is a memory-stock story

The damage did not stay inside the fund. Forced selling of a $16 billion book, then weeks of block trades unwinding it, lands on the order book of whatever the book contained. TECHi has tracked both ends of that: SanDisk was the S&P 500's best stock of the first half before this, and Micron passed its $41 billion test while the stock failed anyway.

From July 31, SanDisk is up 34.5%, Micron 18.5% and CoreWeave 24.0%, using the article’s price snapshot. Those rebounds are consistent with selling pressure easing, but the price series alone cannot establish the cause or rule out changes in fundamentals and other market flows. For anyone holding Micron or SanDisk through that window, the risk is that forced liquidations can add pressure even when the long-term investment case is unchanged. The filings do not establish how much of either stock’s price movement this fund caused.

The bid that is disappearing

JPMorgan’s reported withdrawal has forward consequences because prime-brokerage credit helps finance concentrated positions. The fund has reportedly moved toward smaller brokers, but the cited information does not establish the size, pricing or collateral terms of its replacement credit lines.

If other prime brokers tighten lending against concentrated AI portfolios, that could reduce demand for memory, neocloud and data-center power names. JPMorgan’s decision does not prove an industry-wide reduction in credit. TECHi also tracks financing and liquidity conditions in adjacent markets, including Bitcoin's golden cross and Zcash. Those markets have different drivers and are not evidence of the fund’s borrowing terms. The neocloud names and CoreWeave's debt-funded backlog illustrate why access to capital matters to the investment case. A broader tightening remains a risk to monitor, rather than a result established by these filings.

What the fund did next

The most revealing disclosure is the most recent. On Aug. 28, Situational Awareness filed a Schedule 13D reporting a 21.1% stake in SharonAI Holdings, a Nasdaq-listed AI cloud company: 5,396,127 Class A shares plus prefunded warrants for 2,674,823 more, reported with shared voting and dispositive power by the fund, its related entities, Leopold Aschenbrenner and Carl Shulman and declared as an investment in stock the filers believed was undervalued.

The June 13F listed 5,396,127 SharonAI shares at $456.8 million. The August Schedule 13D reports that share count plus 2,674,823 shares underlying prefunded warrants, and says the only transaction in the preceding 60 days was a June 30 warrant exercise. That supports continuity through the filing date, not a claim about holdings today. Against the article’s snapshot of roughly 1.8 million shares traded daily, the disclosed beneficial position is several days of trading volume; that comparison suggests a potential exit constraint, not a guaranteed time to liquidate.

SharonAI is not a shell. The Australian neocloud listed on Nasdaq in February, raising $125 million, and has since signed a five-year, $1.25 billion contract with ESDS Software Solutions and a $950 million agreement with an unnamed Asia-Pacific technology company, while expanding planned data-center capacity from 70MW to 100MW and announcing a $350 million convertible note. It is a credible business and an illiquid stock at the same time, which is precisely the combination that a levered fund cannot own at size and an unlevered one can.

Read alongside the reported drawdown, the disclosure shows that the fund still held a substantial AI-cloud position at the August filing date. It does not establish that financing was the only weakness in its strategy.

The number that decides what happens next

The next 13F, covering Sept. 30 and expected in mid-November, should provide another snapshot of reportable holdings. It can show whether disclosed memory concentration persisted and which AI-infrastructure positions were reported at quarter-end. It will not show replacement borrowing, undisclosed hedges or the complete path of trades since June.

The filings establish concentrated disclosed holdings before the reported drawdown. Reports of margin calls and JPMorgan’s withdrawal show financing pressure, but the public record does not isolate every cause of the losses or establish the fund’s current leverage. For investors in memory and neocloud stocks, concentration and financing remain risks to examine separately from the underlying business thesis.

Calculation sources: March 31 Form 13F; June 30 Form 13F; SEC Form 13F instructions. Percentages use each filing’s total reported value. Option rows use the value of underlying shares, not option premiums or fund capital.