Crowded trades represent the peak of market consensus and serve as a warning signal for accumulating risks. According to Morgan Stanley's latest Q1 2026 institutional holdings report on large-cap U.S. technology stocks, based on 13F filings, the true positioning of active Wall Street institutions reveals a highly bifurcated landscape. The current technology sector is not characterized by uniform crowding but by extreme structural imbalances. The report data shows that the AI storage and semiconductor equipment supply chain, represented by SanDisk, has become the most crowded institutional holding across the entire market. SanDisk's over-allocation magnitude reached +2.16%, increasing by 57 basis points sequentially this quarter, with its crowding degree continuously hitting new highs. Meanwhile, widely recognized AI leaders such as NVIDIA, Apple, and Microsoft are collectively underweighted by institutions, with deviations ranging from -2.39% to -1.24%, forming a stark "reverse crowding" phenomenon.
This structural divergence has direct and profound implications for the market. The extreme crowding in the AI hardware chain suggests that the room for further accumulation is nearing exhaustion. With positions highly concentrated, any shift in expectations could trigger rapid pullbacks. Conversely, the systematically underweighted tech giants and software sector possess ample potential buying power. Any fundamental catalyst could prompt position rebalancing, creating clear upward momentum. Overall Structure: Systemic Underweighting of Giants, High Structural Imbalance Morgan Stanley's report tracked 28 large-cap U.S. technology leaders, covering the entire industry chain from semiconductors, software, and the internet to hardware and equipment. It used the deviation of active institutional holdings relative to their S&P 500 weight as the core metric for measuring crowding. The data presents clear polarization. In Q1 2026, the "mega-cap tech seven" — comprising NVIDIA, Apple, Microsoft, Amazon.com, Alphabet, Meta Platforms, Inc., and Tesla Motors — had an average institutional underweighting of -125 basis points. Although this narrowed by 12 basis points from -137 bps in Q4 2025, it remains in a clear underweight territory. Meanwhile, the other 21 large-cap tech stocks were collectively over-allocated by 32 basis points, flat quarter-over-quarter.
This contrast reveals the core contradiction in current tech sector positioning: trillion-dollar market cap leaders are systematically underweighted, while mid-to-small cap tech stocks are concentratedly over-allocated. Capital is not evenly distributed but highly focused on a few niche directions. The report also notes a statistically significant relationship between active holding deviations and future stock price performance. Once institutional over-allocation reaches an extreme, stock price sensitivity to negative news increases sharply. Conversely, significantly underweighted assets, due to ample potential buying power, are more prone to subsequent technical rallies. The Most Crowded Trade: AI Storage and Equipment Chain, Bought to Historical Extremes Among all tracked stocks, SanDisk topped the list with a +2.16% over-allocation magnitude, making it the most crowded institutional holding across the entire market. Its over-allocation is more than double that of the second-ranked Seagate Technology PLC (+0.84%). SanDisk's over-allocation increased significantly by 57 basis points sequentially this quarter, indicating accelerating capital inflows. SanDisk's crowding is not accidental. Since its re-listing in Q1 2025 and inclusion in the S&P 500 in Q4 2025, the stock has become the core vehicle for Wall Street's AI storage thesis. Institutions widely believe that the explosive growth of AI data centers will continue to drive demand for NAND Flash and enterprise SSDs, with an upward industry cycle inflection point established, leading to a nearly unanimous concentrated crowding. Following SanDisk, Seagate Technology PLC, Western Digital, and Micron are all in significantly over-allocated territory. Semiconductor equipment leaders KLA-Tencor, Lam Research, and Applied Materials are also heavily held by institutions. These stocks collectively form the core camp of the current institutional crowded trade: the "picks and shovels" supply chain for AI computing infrastructure. Structurally, crowding in this sector exhibits three typical signals: extremely strong consensus among institutions with almost no divergence; a single niche absorbing the most incremental capital within the tech sector; and a self-reinforcing trend of buying more as prices rise, continuously strengthening the crowding behavior. Reverse Crowding: AI Leaders and Software Sector Face Collective Exodus Mirroring the extreme crowding in AI hardware is the collective underweighting of another set of assets, forming a "reverse crowding." NVIDIA, with a -2.39% deviation, is the most underweighted large-cap tech stock by institutions. Despite its robust AI demand and high earnings growth, active institutional holdings are significantly below its S&P 500 weight, narrowing only slightly by 18 basis points sequentially this quarter and remaining near historical underweight extremes. Apple (-2.32%), Microsoft (-1.86%), Amazon.com (-1.24%), and Alphabet are all in deep underweight territory, creating a significant divergence of "perceived overweight, actual underweight" — while the market discusses these AI leaders daily, actual institutional positions remain persistently low. The software sector has also been collectively abandoned. Leading software stocks like IBM, Oracle, Adobe, ServiceNow, and Palo Alto Networks generally have institutional holdings below their S&P 500 weights. The institutional logic is that the commercial impact of AI on software businesses remains unclear, with slow and uncertain earnings realization, prompting continued capital outflow towards more certain hardware segments. Thus, the current "crowding map" of the tech sector is clearly defined: the most crowded direction focuses on AI storage and semiconductor equipment (SanDisk, Seagate Technology PLC, KLA-Tencor, Lam Research); the extremely underweighted direction covers AI computing leaders, consumer tech giants, and general software (NVIDIA, Apple, Microsoft, IBM, etc.). Causes of Crowding: Triple Logic Driving Concentrated Capital Inflows Institutions abandoning consensus leaders like NVIDIA, Microsoft, and Apple to crowd into the storage and equipment sector is driven by three layers of highly consistent Wall Street logic. First, the extreme pursuit of earnings certainty. Amid rising macroeconomic uncertainty and interest rate volatility, institutions are only willing to pay for the "most certain" direction. AI computing infrastructure is the most certain capital expenditure trend in the global tech industry. Storage and semiconductor equipment, as essential components of this infrastructure, offer visible orders, measurable demand, and realizable earnings, providing a clear certainty advantage over software AI transformation and consumer electronics innovation. Second, the classic recurrence of the "picks and shovels" logic. Regardless of which cloud provider or application company ultimately wins, they must purchase chips, storage, and equipment. This "non-aligned, fully beneficial" business model is most favored by institutions, driving collective capital inflows into the upstream hardware chain. Third, high valuations and large market caps of leaders, with capital seeking alternative alpha. Giants like Microsoft, Apple, and Alphabet have massive market capitalizations and are not cheaply valued, leaving limited room for institutions to add significantly. Storage and equipment stocks offer moderate market caps, greater potential upside, and are further boosted by industry tailwinds like the bottoming of NAND Flash prices and booming HBM demand. They have become the preferred choice for overflow capital, further reinforcing crowding behavior. Risks and Opportunities: Late-Stage Crowding Signals Emerge, Underweight Leaders Hold Repair Potential The current crowding in the AI hardware chain exhibits typical high-risk characteristics. The over-allocation magnitudes for stocks like SanDisk and Seagate Technology PLC have reached historical highs. After most institutions have completed their positioning, the space for incremental capital entry is nearing exhaustion. In this state, further stock price gains depend heavily on earnings exceeding expectations; merely meeting expectations could lead to a "sell the news" reaction. Simultaneously, with positions highly concentrated, any negative developments—such as a slowdown in AI capital expenditure, NAND price volatility, capacity releases, or geopolitical policy changes—could trigger synchronized capital outflows, causing rapid pullbacks. In contrast, the extremely underweighted leading assets harbor clear opportunities. According to Morgan Stanley's statistical patterns, assets significantly underweight relative to the S&P 500 have a higher probability of future technical rallies. Stocks like NVIDIA, Apple, Microsoft, and Alphabet are currently in a state of divergence where they are "perceived overweight but actually underweight." Once a fundamental catalyst emerges and capital rebalancing begins, position replenishment could generate sustained upward momentum. The same applies to the software sector. As AI features are implemented and earnings gradually materialize, the process of institutions moving from underweight to neutral or even overweight could drive valuation repair. For investors, the core implication of the current positioning structure is: while AI hardware may have short-term momentum, the risk-reward profile has deteriorated significantly, making blind chasing inadvisable; the extreme underweighting of tech giants and the software sector offers more substantial repair potential from positioning divergence; and the market's shift from "extreme hardware crowding and persistent leader underweighting" towards structural rebalancing may become the core theme of the next phase.
Comments