However, SALP still increased its holdings in a number of AI infrastructure funds; in other words, it abandoned the illusion of a "broad-based rise in all AI stocks." During this US 13F filing season, one of the most watched funds in the market wasn't Bridgewater or Berkshire Hathaway, but a fund with a very unique name—Situational Awareness LP. Its manager, Leopold Aschenbrenner, is not a traditional Wall Street veteran, but a former member of the OpenAI Superalignment team. In 2024, he published a lengthy article, "Situational Awareness: The Decade Ahead," with a very aggressive core judgment: he stated that AGI might arrive sooner than most people imagined, and that the real scarcity in the future wouldn't just be model capabilities themselves, but computing power, electricity, data centers, chips, storage, and the national-level resource competition surrounding the AI arms race. Two years later, it has proven he was right. Leopold internalized a set of judgments about AGI over the next decade and then mapped these judgments onto the capital market. Because of this, Situational Awareness, from its inception, has not resembled an ordinary technology fund, but rather a direct translation of the AGI roadmap into an AI infrastructure investment map. This is why its every move in the AI investment field attracts considerable market attention. The latest 13F filing reveals that this AI-savvy bull seems to be quietly building a large position in put options. I. SALP: A product that turns AGI belief into a fund. Public information shows that Leopold founded an investment company focused on AGI, backed by Silicon Valley heavyweights such as Patrick Collison, John Collison, Nat Friedman, and Daniel Gross. According to market reports, Situational Awareness's net return after fees was approximately +47% in the first half of 2025, significantly outperforming the S&P 500 and technology hedge fund indices during the same period. Its unique feature lies in its focus not simply on "tech stocks," but on AI infrastructure, betting on where AI capital expenditure will ultimately flow. As mentioned at the beginning, its underlying logic is that if AGI truly accelerates, the first to be revalued will not necessarily be application-layer companies, but rather those that control computing power, electricity, data centers, storage, optical communications, semiconductor equipment, and energy resources. Therefore, its high returns do not rely on simply buying indices, but rather on a group of highly volatile AI infrastructure stocks that differentiate themselves: such as Bloom Energy, Sandisk, Lumentum, CoreWeave, and Core Scientific. Here, it's necessary to explain what 13F is. 13F is a quarterly disclosure of holdings filed by U.S. institutional investment managers with the SEC, typically used to observe quarterly changes in large funds' holdings in U.S. stocks, ETFs, and related options.However, it is essentially just a snapshot at the end of a quarter, only telling the market "what was disclosed at a certain point in time," and cannot fully reconstruct the fund's entire trading strategy, especially the options portion. The 13F report doesn't show the strike price, expiration date, or whether it's coordinated with other positions, let alone directly deduce the fund's true net exposure. This is the most easily misinterpreted aspect of this document. The reporting date for this Q1 13F was March 31st. Last10K data shows the document was submitted on the evening of May 15th (Eastern Time), but the SEC accepted it on May 18th. This means it wasn't simply "not submitted," but rather there was a time lag between submission and the market actually seeing the disclosure results. This is why there were many discussions on social media about "waiting for Leopold's 13F." More importantly, this 13F disclosure result is not entirely the same as what the market originally expected. Many people originally thought that Leopold would continue to significantly increase its holdings in core AI assets such as Nvidia, Broadcom, AMD, TSMC, and ASML. However, the reality is that SALP has established a large number of new put option positions, covering a number of core AI and semiconductor stocks, including the SMH Semiconductor ETF, Nvidia, Oracle, Broadcom, AMD, Micron, TSMC, ASML, and Intel. This has led the market to rethink a question: why are those who most believe in the accelerated arrival of AGI starting to insure AI leaders? Simply attributing it to "being bearish on AI" is too simplistic. What truly deserves analysis is the macroeconomic context in which they made this move, and how this reflects a change in the AI trading structure. II. Understanding SALP's Latest 13F: From Betting on AI to Managing AI Volatility The most striking move revealed in this 13F filing is SALP's creation of a large number of new put option positions: the largest is the SMH Semiconductor ETF PUT, with a disclosed value of approximately $2.043 billion; followed by the NVDA PUT, approximately $1.568 billion; then the ORCL PUT, approximately $1.073 billion; the AVGO PUT, approximately $1.006 billion; and the AMD PUT, approximately $969 million. In addition, it also created positions in MU PUT, TSM PUT, ASML PUT, and INTC PUT. On the surface, this seems like a short position on AI leaders, but the problem is that a PUT doesn't necessarily represent a one-sided short position—after all, the option amount in the 13F is more about the nominal value disclosed based on the size of the underlying security, not the actual premium cost invested by the fund. More importantly, the 13F doesn't show the strike price, expiration date, whether it's linked with other positions, or the true net exposure of the portfolio.Therefore, it's inaccurate to say that Leopold is "completely bearish on Nvidia and semiconductors." A more reasonable interpretation is that he's buying "insurance" for his long AI infrastructure portfolio. Many of the companies SALP already holds are inherently highly volatile and interest rate-sensitive, such as Bloom Energy, CoreWeave, Core Scientific, IREN, Applied Digital, and Sandisk mentioned above. Their long-term logic is related to AI infrastructure, but their short-term stock prices are highly dependent on risk appetite and valuation conditions. Once the market begins to reduce risk due to rising oil prices, recurring inflation, higher interest rates, or geopolitical conflicts, these highly volatile assets are often the first to be sold off. This is also related to the macroeconomic background at the end of March: on the one hand, the situation in the Middle East and the risk of conflict between the US and Iran pushed up oil price expectations; on the other hand, rising oil prices would exacerbate inflation stickiness, reducing market confidence in interest rate cuts. For high-valuation growth stocks, this amounts to "double pressure": oil prices push up inflation, inflation suppresses interest rate cuts, and if interest rates don't fall, the valuation of long-duration technology assets will be compressed. In this context, Leopold's creation of numerous new puts becomes easier to understand. It's not a denial of AI, but rather an acknowledgment that even with strong long-term AI logic, macroeconomic headwinds cannot be completely ignored. This is especially true for funds like SALP, whose portfolios contain many high-beta assets. If they only hold aggressive positions, the portfolio's net asset value will fluctuate wildly in the event of a systemic market pullback. By buying highly liquid and representative core AI asset puts like SMH, NVDA, AVGO, AMD, and ORCL, they can use relatively standardized tools to hedge against the systemic pullback risk of the entire AI trading portfolio. The real meaning behind this is that Leopold hasn't gone from being a bull on AI to a bear on AI, but rather has shifted from "aggressively going long on AI" to "continuing to bet on AI infrastructure, but starting to manage path volatility." This is a more mature portfolio management approach. III. So where is Leopold's offensive direction? If creating new puts solves the "defensive problem," then the list of adding, reducing, and liquidating positions truly tells us where Leopold's offensive direction lies. According to the disclosures, SALP has maintained and increased its holdings in a number of AI infrastructure-related stocks. For example, it slightly increased its holdings in Sandisk stock, and also increased its holdings in CoreWeave stock, IREN, Applied Digital, Riot Platforms, CleanSpark, Bitfarms, Bitdeer, and others. The important long positions it has maintained at present also include Bloom Energy, Sandisk, CoreWeave, IREN, Core Scientific, and Applied Digital.This indicates that it hasn't abandoned AI; on the contrary, it's still betting on the same long-term logic: AI capital expenditure will continue to be passed down, and the real beneficiaries are companies that control bottlenecks in electricity, data centers, storage, computing power, and infrastructure. This is very close to MSX's Q2 main theme judgment. In our article "AI Infrastructure Surges Throughout Q1, Who Can Support 'High Valuations' in Q2?", we emphasized that the focus of AI trading has shifted from pure GPUs to networks, storage, and electricity. The market is now more concerned about where the orders, revenue, and profits from the continued expansion of capital expenditure by major companies will ultimately go. The reason why equipment, networks, storage, and electricity are more advantageous is not because they are more attractive, but because they better fit the current market's aesthetic for realization. From this perspective, SALP's long positions are very representative: Bloom Energy corresponds to electricity and independent energy supply; CoreWeave, Applied Digital, Core Scientific, and IREN correspond to data centers, computing power hosting, and infrastructure; and the positions related to Sandisk, Micron, and TSM correspond to storage, semiconductor manufacturing, and hardware supply. In other words, Leopold isn't against AI; he's more concerned with where the money will ultimately go and who can turn it into reported revenue. The reduction and liquidation of positions are equally informative. SALP liquidated its holdings in INTC Calls, Lumentum, and Cipher Mining, and reduced its holdings in CoreWeave Calls, Bloom Energy, and Core Scientific. Most notably, it didn't simply withdraw from a particular direction, but rather reduced positions in stocks that had already risen significantly, were highly volatile, or had strong leverage. For example, with CoreWeave, it reduced its Calls but still held common stock, indicating it wasn't completely abandoning CoreWeave, but rather shifting from more aggressive options to a more controllable underlying stock. Similarly, with Bloom Energy and Core Scientific, the reduction doesn't necessarily mean the logic failed; it's more likely a matter of portfolio-level risk control and profit realization. The liquidation of Lumentum is even more intriguing. In the MSX Q1 IPO review, AI hardware and optical communication were the two strongest performing sectors. AXTI, AAOI, LITE, and LWLG all saw gains exceeding 100%. The strength of optical communication essentially stemmed from the explosive demand for optical interconnects, optical modules, and network links from AI data centers. However, the problem is that the stronger a sector performs in Q1, the more likely it is to face crowded trading and a declining risk-reward ratio in Q2. Therefore, Leopold's liquidation of LITE and reduction of some high-elasticity AI infrastructure positions doesn't necessarily indicate a lack of confidence in this sector, but rather a more realistic acknowledgment that the most successful trades in Q1 are not necessarily the most cost-effective trades in Q2.This is the most important aspect of this portfolio adjustment. It's not about negating AI, but rather proactively shifting the portfolio structure, moving from buying any AI-related asset to retaining only those assets better suited to withstand long-term capital expenditures, possessing more infrastructure-like attributes, and being more resilient to macroeconomic fluctuations. What he's abandoning isn't AI itself, but the linear illusion that "all AI will rise together." This 13F report is essentially a snapshot as of March 31st and doesn't mean Leopold will still hold the same position in May, but it still offers strong insights into the current market. First, the long-term AI theme hasn't ended, but the trading structure has changed. In the future, it won't be about buying any AI and it will rise, but rather about who can realize profits, who can capture the premium, who is too crowded, and who needs hedging. Second, in an environment of high oil prices, high interest rates, and high volatility, a truly effective strategy is neither simple all-out offense nor all-out defense, but rather using defense to fuel offense—betting on certainty with core positions, betting on flexibility with marginal positions, and using hedging tools to control portfolio drawdowns. Leopold's actions this time essentially demonstrated this logic with real positions. Third, this also confirms a major change in the US stock market in 2026: the index beta is weakening, while the structural alpha is strengthening. In the past, simply buying the "Seven Sisters" (AI giants) or Nvidia could guarantee easy profits; but now the market is more discerning, questioning every company: Can your AI story ultimately translate into orders? Can it translate into revenue? Can it translate into profit? If not, even a high valuation will be compressed. This is why AI Infrastructure 2.0 is becoming important. In the future, funds will not only look at GPUs, but will also look down the chain of computing power → interconnection → storage → power → data center infrastructure to find the links that can truly generate returns. In conclusion: On the surface, the most eye-catching part of this 13F filing is the series of huge puts. But if you really look at the entire portfolio, you will find that Leopold is not "turning from AI bulls to bears," but rather a more mature upgrade: still betting on AI infrastructure in the long term, while starting to face the volatility risks of high-valuation, high-elasticity assets in the short term. This is the most important aspect of this 13F report: it tells us that while the direction of AI may be correct, the path to that direction will never be a straight line. For true fund managers, the important thing is never just betting on the right destination, but surviving the fluctuations along the way. For ordinary investors, the biggest takeaway from this 13F report is also clear: AI trading in 2026 has shifted from "buying stories" to "buying realized profits"; from "buying leading stocks" to "finding bottlenecks"; from "one-sided offense" to "offensive with a defense." This is the most interesting signal, and one that should never be ignored. [MSX Maitong]