13F Insights: Where Smart Money Is Moving in U.S. Stocks

Key takeaways:

1. 13F is the most-watched quarterly "revealing of the cards" moment on Wall Street, but in essence it is a position adjustment map delayed by at least a month and a half. Right now, crowded trading in the AI sector is being unwound, there is sharp divergence among tech leaders, and top-tier capital has already shifted to focus on "cash flow, hard tech, and real moats".

2. Google has been simultaneously increased in position by Temasek, Bridgewater, Jinglin, Baillie Gifford and Berkshire Hathaway, and is becoming the most widely agreed-upon asset in this round; NVIDIA remains the core, yet some investors are taking profits while others are liquidating their positions, leading to a significant widening of divergence; semiconductor upstream players such as ASML, AMAT and optical communication are starting to gain more attention; SpaceX, on the other hand, has been listed as a core heavyweight holding by both Temasek and Baillie Gifford, representing the longer-term main line of "next-generation infrastructure".

3. The RockFlow investment research team believes that there are three common pitfalls to avoid when analyzing 13F filings: treating them as real-time transaction records, only focusing on what was purchased rather than the corresponding position sizes, and overlooking the style differences across different funds. The correct approach is to first examine the overall portfolio structure, then track marginal changes, followed by analyzing industry allocation shifts, then identifying the consensus and divergences among multiple institutions, and finally returning to valuations and financial reports to form an independent judgment. 13F filings are the footprints left by institutional investors, not a ready-made answer for your own investments.

Whenever the 13F disclosure season rolls around, the market invariably gets reflexively excited. Everyone is keeping a close eye on what Warren Buffett has bought, how much Bridgewater has adjusted its positions, or which top-tier institution is quietly bottom-fishing AI stocks.

However, many people overlook a fundamental premise when "copying others' homework": Form 13F is not an institution' s "real-time trading plan", but rather a "position adjustment map" delayed by at least one and a half months. Understanding it gives you plenty of extra research leads; but if you blindly follow the trades, you may end up taking over a position based on already outdated logic.

We are currently in the 13F disclosure window for the second quarter of 2026. Over the past month, the US stock market has been in a highly sensitive repricing phase, with consensus on interest rate cuts/hikes tugging back and forth, crowded trades in the AI sector being unwound, and sharp divergences emerging among tech leaders. The "easy win era" where stock prices would take off as long as a company was tangentially related to computing power, models or cloud services has come to an end.

By comparing the underlying positions of several leading institutions, you will find that these players have clearly become more selective, as they are making a full shift away from "AI fanaticism" and toward "cash flow, hard tech, and real moats".

In this article, the RockFlow Investment Research Team will walk you through a comprehensive review of the latest consensus and preference differences among well-known institutions that have disclosed their 13F filings to date, and point out common mistakes investors make when reviewing 13F filings as well as the correct five-step analysis framework.

Position adjustment signals from five core institutions

Temasek: Increasing Stakes in SpaceX, Google and ASML, Betting on Hard Technology

Temasek's recent move is highly indicative of market trends. The concentration ratio of its top 10 heavyweight stocks stands at 51.42%, which shows that its portfolio is not diversified and they are making active bets on core directions. The three most critical moves are as follows:

significantly increase holdings in SpaceX

In the second quarter, SpaceX rose to become Temasek's fifth-largest holding, with a stake of approximately 9.83 million shares worth around USD 1.68 billion. This reflects a very clear allocation strategy: betting on next-generation infrastructure. The significance of SpaceX goes beyond commercial aerospace; it also represents the convergence of multiple fields including satellite internet, low-orbit communication, defence-related science and technology, and global connectivity networks.

Heavily increase positions in Alphabet

Google has overtaken Visa to become Temasek's second-largest holding in US stocks, with a position valued at approximately USD 2.53 billion. This indicates that Temasek is targeting not only hard technology but also stable cash flow, and Google happens to deliver on both fronts: AI infrastructure, advertising cash flow, cloud computing growth, a global ecosystem entry point, and a relatively reasonable valuation. Assets of this kind are highly favored by long-term capital in the current market environment.

Buy ASML

ASML is a core player in the global advanced lithography machine sector. Temasek's increased stake in ASML indicates that it places greater emphasis on the scarce link at the very top of the semiconductor industry chain.

To sum up Temasek in one sentence: it sells off a portion of its high-flying tech holdings and acquires tech assets that are more fundamental, more hard-core and more scarce.

Bridgewater: Holding Tech Leaders on One Hand, Defensive Allocations on the Other

Bridgewater has not yet disclosed its latest Q2 holdings and changes, but its quarterly stock positioning stands at approximately 22.4 billion U. S. dollars. Its top five core heavyweight positions include: SPY, IVV, Amazon, NVIDIA, and Alphabet. This is fully in line with Bridgewater's style: macro-diversified, and never easily betting the entire portfolio on a single direction.

However, it also sent a clear signal this time that Bridgewater remains willing to embrace the cash flow of tech leaders. Among them, its position in Amazon surged by about 125% month-on-month, while its holding in Nvidia rose by around 21% month-on-month. This indicates that in Bridgewater's view, tech leaders are still the most important profit engine for US stocks.

On the other hand, Bridgewater has reduced its holdings in some IVV positions, reflecting a certain defensive adjustment. Therefore, Bridgewater is not simply "bullish on the technology sector"; instead, it is making more refined balancing moves: it will neither exit the main track of AI and technology, nor mindlessly chase the highs across the entire market.

Jinglin: Sharp contraction, liquidated positions in NVIDIA and Meta, shifted focus to equipment and optical communication

Jinglin took extremely aggressive actions in the second quarter. Its total position value in US stocks dropped from 3.88 billion US dollars to 2.19 billion US dollars, representing a decrease of around 43%. This indicates that they are clearly making a contraction in their risk exposure.

What has drawn the most attention is that Jinglin Asset Management fully liquidated its positions in NVIDIA and Meta, which the market could easily interpret as a "bearish stance on AI". However, a closer look at what it has bought yields a more intriguing conclusion: Jinglin has newly established positions in ASML, Applied Materials (AMAT), and Applied Optoelectronics (AAOI). Meanwhile, Alphabet has become its largest heavyweight stock, with a market value of approximately 647 million U. S. dollars.

Jinglin has not completely abandoned the technology sector; instead, it has adjusted its positioning by shifting from AI star stocks that have seen substantial gains to the upstream and more certain segments of hard technology. This is a typical transition "from beta to alpha" — that is, seeking segments in the industrial chain with more verified performance, rather than paying for the crowded AI premium.

Berkshire: Ending its Wait-and-See Stance, Makes a Rare Large-Scale Purchase of Google Shares

Berkshire's moves have been the most talked-about. After a prolonged period of standing pat with a "sell-only" stance, it shifted to net buying in the second quarter, purchasing roughly $23.5 billion worth of stocks. The most significant move came as follows: it substantially ramped up its stake in Alphabet, with a purchase size of around $10 billion.

What makes this significant is that Berkshire’s style is completely different from that of many tech funds. It is not an institution that chases hot trends. What Berkshire values are business models, Competitive Edge, cash flow, management teams, long-term capital returns, and margin of safety. If capital on the scale of Berkshire’s starts placing Google in a core position, the market will need to re-examine a question: Is Google an AI laggard, or an undervalued AI cash flow asset?

Of course, you shouldn't just blindly copy a purchase simply because Berkshire Hathaway made it. But this kind of signal is worth putting on your research list. Especially when multiple institutions are increasing their positions in the same direction, it is no longer just the preference of a single fund, but may indicate that a consensus is taking shape.

Baillie Gifford: SpaceX becomes its largest overweight position, continues to bet on long-term growth

Baillie Gifford has always had a preference for long-term growth and innovative assets. Its holding market value rebounded to approximately 110 billion U. S. dollars in the second quarter. The most notable change is that: SpaceX has become its largest heavyweight holding, with its positioning ratio rising to 8.66%.

Meanwhile, Baillie Gifford increased its holdings in NVIDIA and Google, while reducing its positions in Amazon, Netflix and Spotify. This indicates that within the long-term growth camp, capital is also undergoing reallocation: tilting more toward infrastructure, core AI assets and platform-level technology, and cutting exposure to certain content and consumer internet segments.

Several most important conclusions from this round of 13F filings

After compiling all institutional trading activities and submitting them to Bobby AI for in-depth analysis, we have derived several key conclusions from this round of 13F filings:

Google is emerging as a consensus asset among institutional investors

Temasek increased its positions. Bridgewater took heavy positions. It is the top holding of Jinglin. Baillie Gifford increased its holdings. Berkshire made large-scale purchases.

This is no coincidence. Google's appeal lies in the fact that it embodies three attributes simultaneously: a tech leader, an AI infrastructure provider, and an asset with stable cash flow. As AI trading enters its second phase, the market no longer rewards only imagination, but also starts to reward delivery capabilities. Google happens to stand right at this intersection.

Nvidia remains a core holding, but institutional investors have begun to scrutinize its crowded trade status

Some keep buying, some sell off a portion, while others liquidate their positions entirely. This shows that NVIDIA remains a key player, but it is no longer a one-sided consensus free of divergence. The issues NVIDIA needs to address next are: whether its performance can maintain high growth, whether its gross profit margin can be sustained, whether its customers' capital expenditure will continue, whether the competitive landscape will deteriorate, and whether its valuation can continue to be digested by earnings.

More long-term capital is gaining focus on the upstream of the semiconductor industry

ASML, AMAT, optical communication, and equipment chains are being included in the scope of attention by more institutions. This indicates that capital is shifting from "AI application speculation" to "the scarcity of AI infrastructure".

If the first phase of AI trading was about buying into narratives, and the second phase about buying computing power, then the third phase will likely be about buying into whoever has a grip on the bottlenecks of computing power.

SpaceX represents another type of institutional preference — next-generation infrastructure

Both Temasek and Baillie Gifford hold SpaceX in an extremely high regard. This indicates that global long-term capital is on the hunt for super growth assets beyond the secondary market. SpaceX is not a technology stock in the traditional sense; rather, it is a complex of aerospace infrastructure, communication infrastructure, defence-related science and technology assets, and a gateway to the global low-orbit network.

3 Most Common Mistakes Investors Make When Reviewing 13F Filings

Form 13F is a very useful tool, yet it is also highly susceptible to misuse. Particularly when encountering sensationalist headline clickbait news, many people will rush in directly to "copy others' homework"(i. e. blindly follow others' investment positions), which is extremely dangerous.

Mistake 1: Treating Form 13F as real-time transaction records

Form 13F filings come with a disclosure lag. What you see today reflects the fund's holdings as of the end of the previous quarter. By the time the filing is officially released, the fund may have increased its positions, reduced its positions, liquidated its positions, engaged in hedging, or even changed its market outlook entirely.

Therefore, don't rush to follow the moment you see "a certain institution bought a certain stock". First ask yourself several questions: How much has this stock risen since the end of the quarter? Is the current price already far away from the range where the institution might have built its position? Has the market already front-run the trade after the disclosure? Does the latest financial report still support the original buying logic?

13F filings serve as clues, not a buy button.

Mistake 2: Only focusing on "what was purchased" while ignoring "how much was purchased"

Adding a new stock is not necessarily important. What really matters is what its positioning proportion is, whether it enters the top ten heavy positions, whether positions have been increased for consecutive quarters, whether the position increase range is significant, and how much impact it has on the entire portfolio.

A 0.1% test position is completely different from an 8% core positioning. For instance, the fact that SpaceX has become Baillie Gifford's largest overweight holding conveys far more information than a regular newly added position. When Google enters the core positioning of multiple institutions, that is also far more noteworthy than the simple statement that "a certain fund bought Google".

Mistake 3: Ignoring Differences in Fund Styles

The same stock appearing in the 13F filings of different funds carries entirely different implications:

For Berkshire, it could be a long-term core asset. For Bridgewater, this may be part of macro Asset Allocation. For Jinglin, this is likely the core bet following the structural shift. For Baillie Gifford, this is likely a growth narrative spanning a decade. For event-driven funds, this may only be a short-term catalyst for trading.

Therefore, the first step in reviewing a 13F filing is to examine the fund itself. First, figure out what this institution relies on to generate returns: whether it focuses on long-term value, macro allocation, growth investment, or event-driven strategies, what its turnover rate is like, and how long its historical holding period for similar assets typically lasts.

If you fail to understand a fund's style, you will easily misinterpret its signals.

5-Step Framework for Properly Analyzing 13F

If you don't just want to watch the fun, you can use the following framework.

Focus on the position structure rather than news headlines

First, look at the portfolio as a whole, then at individual stocks. Find out what the top ten holdings are, what their concentration level is, whether the portfolio is more concentrated or more diversified, and whether any new core assets have emerged. Temasek's top ten holdings concentration reaches 51.42%, which means the changes in its heavyweight positions carry a great deal of information value.

Observe marginal changes

The core of 13F lies in "what has changed". Categorize the positions into four types: new additions, increased holdings, reduced holdings, and liquidated positions, then dig deeper: which assets have seen continuous position increases, which have been taken profit on, which have been completely abandoned, and where the capital is flowing from and to. This is far more useful than only looking at individual stocks.

Observe industry and thematic shifts

The real stance of institutional investors is often embedded in industry rotation. This round of rotation makes it quite clear: AI remains a focus but is no longer a target for indiscriminate buying, the upstream semiconductor sector is drawing more attention, Google has emerged as a representative of tech cash flow, SpaceX stands for next-generation hard tech infrastructure, and high-flying tech stocks are facing pressure to take profits.

Examine the Consensus and Divergence Among Multiple Funds

A single fund's buying activity is merely a clue; it is the simultaneous position buildup by multiple funds that merits in-depth research. However, consensus cuts both ways: the stronger the consensus, the more widely accepted the underlying logic; yet the stronger the consensus, the more crowded the trading may become. Google is a typical example of "strengthening consensus" in this round, while NVIDIA represents the typical case of "remaining important, yet with widening divergence".

Back to Stock Prices, Valuations and Financial Reports

Form 13F only tells you what has happened in the past, and cannot help you judge whether it is worth buying at present. Before making a real investment, you also need to verify whether the current stock price has skyrocketed, whether the valuation is still acceptable, whether the latest financial report supports the institutional logic, where the risk points lie, and under what circumstances it will prove that this judgment is wrong.

Conclusion: Form 13F is the footprint left by institutional investors

The RockFlow investment research team believes that one of the most dangerous moves in the market is taking others' positions as your own investment logic. When an institution buys a stock, there may be complex portfolio constraints, hedging arrangements, tax considerations, liquidity requirements and risk budgets behind the decision. What you see is just its footprints, not its complete thought process.

Yet these footprints remain valuable, as they reveal where top-tier capital is flowing, which assets are gaining consensus, which trades are starting to get crowded, which sectors are seeing capital rotation, and which stocks merit being added to your research watchlist.

Don't rush to copy others' homework. First, learn to spot the clues. What can truly help you weather cycles is the research framework you build on your own.