最相信 AGI 加速的多頭，Q1 買了數十億的名義 put？

During this U.S. stock 13F filing season, one of the funds drawing the most market attention is not Bridgewater, nor 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 titled "Situational Awareness: The Decade Ahead," which included a very radical core judgment, explicitly stating that AGI might arrive faster than most people imagine, and that what will truly be scarce in the future isn't just model capability itself, but computing power, electricity, data centers, chips, storage, and the national-level resource competition surrounding the AI arms race.

Two years later, time has proven him right.

Leopold internalized a set of judgments about the next decade of AGI, then mapped these judgments onto the capital market. Precisely for this reason, from its inception, Situational Awareness has never looked like an ordinary tech fund, but rather like translating the AGI roadmap directly into an AI infrastructure investment map.

This is also why, in the AI investment field, every move it makes attracts significant market attention. The latest 13F filing reveals that this most knowledgeable AI bull seems to be quietly building a large-scale position in put options.

1. SALP: A Fund Built on AGI Conviction

Public information shows that Leopold founded an investment company focused on AGI, backed by prominent Silicon Valley figures including Patrick Collison, John Collison, Nat Friedman, and Daniel Gross.

According to market reports, Situational Awareness achieved a net return of approximately +47% after fees in the first half of 2025, significantly outperforming the S&P 500 and tech hedge fund indices over the same period. Its uniqueness lies in the fact that it is not simply bullish on "tech stocks," but is highly concentrated on AI infrastructure, betting on where AI capital expenditures will ultimately flow.

As mentioned at the outset, his underlying logic is that if AGI truly accelerates, the companies to be revalued first might not be application-layer companies, but those possessing computing power, electricity, data centers, storage, optical communications, semiconductor equipment, and energy resources. Therefore, its high returns do not rely on simply buying an index, but on gaining an edge through a group of high-beta AI infrastructure targets, such as Bloom Energy, Sandisk, Lumentum, CoreWeave, and Core Scientific.

First, we need to explain what 13F is.

The 13F is a quarterly filing of holdings submitted to the SEC by U.S. institutional investment managers, typically used to observe changes in the quarterly holdings of large funds in U.S. stocks, ETFs, and related options. However, it is essentially just a snapshot at the end of the quarter, telling the market "what was disclosed at a specific point in time," without fully reconstructing the fund's entire trading strategy. Especially for the options portion, the 13F does not reveal strike prices, expiration dates, whether they are paired with other positions, nor can it directly deduce the fund's true net exposure.

This is also where interpretations of this document are most prone to misunderstanding.

The reporting date for this Q1 13F was March 31st. Last10K shows the document was filed on the evening of May 15th Eastern Time, but the SEC acceptance time was May 18th. This means it wasn't simply a case of "not filing"; there's a time lag between the filing and the market actually seeing the disclosed results. Hence, there was much discussion on social media platforms about "waiting for Leopold's 13F."

More critically, the disclosed results of this 13F don't entirely match market expectations. Many originally thought Leopold would continue to heavily increase positions in core AI assets like NVIDIA, Broadcom, AMD, TSMC, and ASML. However, the reality is that SALP established a large number of new PUT option positions, covering a range of core AI and semiconductor targets including the SMH semiconductor ETF, NVIDIA, Oracle, Broadcom, AMD, Micron, TSMC, ASML, and Intel.

This has led the market to rethink a question: Why is the person most convinced of AGI's accelerated arrival starting to buy insurance on AI leaders?

Simply attributing it to "being bearish on AI" is too crude. What's truly worth analyzing is the macroeconomic backdrop against which he made this move and the changes in the AI trading structure it reflects.

2. Deciphering SALP's Latest 13F: From Betting on AI to Managing AI Volatility

The most striking action revealed in this 13F filing is undoubtedly the large number of new put option positions established by SALP:

• The largest is the SMH Semiconductor ETF PUT, with a disclosed value of approximately $2.043 billion;
• Followed by the NVDA PUT, at approximately $1.568 billion;
• Then the ORCL PUT, at approximately $1.073 billion;
• The AVGO PUT, at approximately $1.006 billion;
• And the AMD PUT, at approximately $969 million;
• In addition, it also established new positions like MU PUT, TSM PUT, ASML PUT, and INTC PUT;

On the surface, this looks like being bearish on AI leaders. However, the issue is that a PUT does not necessarily represent outright short selling — after all, the option amounts in a 13F are more often disclosed based on the notional value of the underlying securities, not the actual premium cost paid by the fund. More importantly, the 13F doesn't show strike prices, expiration dates, whether they are paired with other positions, or the true net exposure within the portfolio.

Therefore, directly concluding that Leopold is "fully bearish on NVIDIA and semiconductors" is inaccurate. A more reasonable interpretation is that he is buying 'insurance' for his long AI infrastructure portfolio. Many of the targets originally held by SALP are themselves high-beta, high-volatility companies sensitive to interest rates, such as Bloom Energy, CoreWeave, Core Scientific, IREN, Applied Digital, and Sandisk mentioned earlier. Their long-term logic is tied to AI infrastructure, but their short-term stock prices are often heavily dependent on risk appetite and valuation environment.

Once the market starts de-risking due to rising oil prices, persistent inflation, higher interest rates, or geopolitical conflicts, these high-beta assets are often the first to be sold. This context is also related to the macroeconomic backdrop at the end of March: on one hand, the Middle East situation and the risk of US-Iran conflict pushed up oil price expectations; on the other hand, rising oil prices can exacerbate inflation stickiness, diminishing market confidence in interest rate cuts.

For high-valuation growth stocks, this represents a "double squeeze": oil prices push up inflation, inflation suppresses rate cuts, interest rates remain high, and the valuations of long-duration tech assets come under pressure.

Viewed against this backdrop, Leopold's action of establishing a large number of PUTs becomes easier to understand. It is not a rejection of AI, but an acknowledgment that no matter how strong the long-term logic of AI is, it cannot completely ignore macroeconomic headwinds.

Especially for a fund like SALP, whose portfolio contains many high-beta assets. If it only holds offensive positions, the portfolio's net value would fluctuate significantly during a systemic market pullback. By buying PUTs on highly liquid, representative core AI assets like SMH, NVDA, AVGO, AMD, and ORCL, it can use relatively standardized tools to hedge against systemic pullback risk across the AI trade.

The real implication behind this is that Leopold hasn't turned from an AI bull into an AI bear. Instead, he has switched from 'aggressively going long on AI with a single-minded focus' to 'continuing to bet on AI infrastructure, but starting to manage path volatility.'

This represents a more mature portfolio management approach.

3. So, Where is Leopold's Offensive Direction?

If establishing new PUTs addressed the "defensive issue," then the list of added, reduced, and closed positions tells us where Leopold's offensive direction truly lies.

Based on the disclosure, SALP still retains and has added to a number of AI infrastructure-related targets. For example, shares of Sandisk were slightly increased, and positions in CoreWeave, IREN, Applied Digital, Riot Platforms, CleanSpark, Bitfarms, Bitdeer, etc., were also on the addition list. The significant long positions currently retained include Bloom Energy, Sandisk, CoreWeave, IREN, Core Scientific, and Applied Digital.

This indicates that it hasn't abandoned AI. On the contrary, it continues to bet on the same long-term logic: AI capital expenditure will continue to flow downstream. The real beneficiaries are companies that control electricity, data centers, storage, computing capacity hosting, and infrastructure bottlenecks.

This is very close to MSX's Q2 main theme judgment. We emphasized in "AI Infrastructure Rallied Throughout Q1, Who Can Sustain 'High Valuations' in Q2?" that the trading focus of AI has shifted from simple GPUs to networks, storage, and electricity. The market now cares more about where the continuously expanding CapEx of big tech ultimately flows in terms of orders, revenue, and profit. Equipment, networks, storage, and electricity are favored not because they are sexier, but because they better match the market's current preference for deliverability.

From this perspective, SALP's long positions are quite representative: Bloom Energy corresponds to electricity and independent energy supply; CoreWeave, Applied Digital, Core Scientific, and IREN correspond to data centers, computing hosting, and infrastructure capacity; positions related to Sandisk, Micron, and TSM correspond to storage, semiconductor manufacturing, and the hardware supply side.

In other words, Leopold hasn't stopped buying AI. He is more concerned about where the AI money is ultimately spent and who can turn that money into revenue on their financial statements.

Looking at the reductions and liquidations, they are equally informative. SALP closed positions in INTC CALL, Lumentum, Cipher Mining, and reduced positions in CoreWeave CALL, Bloom Energy, Core Scientific, etc. The most noteworthy aspect here is that he isn't simply withdrawing from a certain direction; he is reducing some positions that have fully appreciated, are highly volatile, or have stronger leverage characteristics.

For example, with CoreWeave, he reduced CALLS but still holds common stock. This suggests he hasn't completely given up on CoreWeave, but has switched from a more aggressive options expression back to a relatively more controllable common stock expression. Similarly, for Bloom Energy and Core Scientific, reducing positions doesn't necessarily mean the logic is invalid; it's more likely portfolio-level risk control and profit-taking.

The liquidation of Lumentum is more intriguing. In the MSX Q1 new review, AI hardware and optical communications were the two strongest themes, with AXTI, AAOI, LITE, and LWLG all achieving gains of over 100%. The strength of optical communications essentially stemmed from the explosion in demand for optical interconnects, optical modules, and network links driven by AI data centers. However, the problem is that the more a theme runs in Q1, the more susceptible it becomes to crowded trades and a deteriorating risk-reward ratio as it enters Q2.

So Leopold liquidating LITE and reducing some high-beta AI infrastructure positions doesn't necessarily mean he is bearish on the direction. It might be a more pragmatic acknowledgment that: The most successful trades of Q1 are not necessarily the most cost-effective trades of Q2.

This is the most crucial takeaway from this portfolio adjustment. He is not denying AI; he is proactively making a structural shift. He is moving from buying anything on the AI chain to only retaining assets that can better absorb long-term capital expenditure, possess stronger infrastructure attributes, and better navigate macroeconomic volatility.

What he is abandoning is not AI, but the linear fantasy that "all AI will rise together."

This 13F is essentially just a snapshot as of March 31st. It doesn't mean Leopold still holds exactly the same positions in May. However, it still offers strong insights for the current market conditions.

First, the long-term AI theme isn't over, but the trading structure has changed. In the future, not all AI stocks will rise. It's about who can deliver, who commands a premium, who is too crowded, and who needs hedging.

Second, in an environment of high oil prices, high interest rates, and high volatility, the truly effective strategy is not simply all-out offense or all-out defense, but playing offense with defense — allocating core positions to certainty, marginal positions to elasticity, while using hedging tools to control portfolio drawdown. Leopold's recent moves essentially demonstrated this logic using actual portfolio positions.

Third, this also confirms a major shift in the U.S. stock market for 2026: Index Beta weakens, while structural Alpha strengthens. In the past, simply buying the Magnificent Seven or NVIDIA might have guaranteed success. But now, the market is more discerning; it will scrutinize every company: Can your AI story ultimately translate into orders? Into revenue? Into profit? If not, even high valuations will be compressed.

This is why AI Infrastructure 2.0 becomes important. Future capital won't just follow GPUs; it will move down the chain of Computing → Interconnect → Storage → Electricity → Data Center Infrastructure to find segments that can truly deliver.

Final Thoughts

If you only look at the surface, the most eye-catching part of this 13F is the string of massive PUTs.

But if you truly review the entire portfolio, you'll find that Leopold hasn't merely "turned from an AI bull into a bear." Instead, he has undergone a more mature upgrade: still betting long on AI infrastructure, but beginning to acknowledge the volatility risks of high-valuation, high-beta assets in the short term.

This is the most important takeaway from this 13F. It tells us that the direction of AI might still be correct, but the path to that direction will certainly not be a straight line.

For a true fund manager, the key is never just to bet on the destination correctly, but also to survive the volatility on the journey.

For ordinary investors, the greatest insight from this 13F is also clear: The AI trade in 2026 has shifted from 'buying stories' to 'buying deliverables'; from 'buying leaders' to 'finding bottlenecks'; from 'all-out offense' to 'playing offense with defense.'

This is the most interesting signal, and one that absolutely should not be ignored.