As the boom in AI infrastructure investment continues to gain momentum, semiconductor, server, and networking equipment stocks have become market darlings, while investor concerns about overcrowding in AI‑related equities are growing increasingly acute.
Zhitong Finance APP has learned that, as the investment boom in artificial intelligence (AI) infrastructure continues to heat up, semiconductor, server, and network equipment stocks have become market favorites. However, investors' concerns about over‑crowded trading in AI‑related concept stocks are also growing. Bernstein's latest research report points out that, despite signs of capital outflows from some AI‑related stocks recently, overall trading congestion in the U.S. technology hardware, semiconductor, and network equipment sectors remains quite high. Among them, AMD (AMD.US),$Arista Networks (ANET.US)$and$Applied Materials (AMAT.US)$The crowding in equities is particularly pronounced, while…$Qualcomm (QCOM.US)$、$SanDisk (SNDK.US)$Meanwhile, the stocks are trading at relatively low levels.
In a report released on October 25, Bernstein noted that as investors increasingly recognize AI infrastructure providers as direct beneficiaries of the artificial intelligence boom, capital continues to flow into related equities, driving a marked rise in trading congestion across several technology‑hardware subsectors.
According to the firm's quantitative analysis, within the global technology, media, and telecommunications (TMT) sector, communications equipment is currently the most crowded sub‑segment, followed by semiconductors and electronic devices. Notably, network equipment and traditional hardware—historically among the least favored by investors—have now become areas where market capital is heavily concentrated, underscoring the profound impact of AI‑related investment themes on the landscape of industry‑wide capital allocation.
However, Bernstein emphasized that although the technology sector is currently the most crowded among global industries, its level of crowding has not yet reached exceptionally extreme levels from a historical perspective. By contrast, the financial and industrial sectors are currently experiencing crowding that stands out even more relative to their respective historical norms.
In the semiconductor sector, among the 12 semiconductor and semiconductor equipment stocks covered by Bernstein, seven are in the top 10% of trading‑crowding levels. Notably, the AI chip segment exhibits the highest degree of trading congestion, primarily involving…$Broadcom (AVGO.US)$, AMD and$NVIDIA (NVDA.US)$. The semiconductor equipment sector followed closely, including Applied Materials,$Lam Research (LRCX.US)$、$KLA Corp (KLAC.US)$and$PDF Solutions (PDFS.US)$Waiting company.
From a stock-specific perspective, AMD is one of the most crowded holdings among Bernstein's covered technology hardware and semiconductor stocks, with a standardized crowding score of 2.35. Arista Networks and$Lumentum (LITE.US)$They reached 2.20 and 1.91, respectively.$Coherent (COHR.US)$And the crowding level of Applied Materials is also at a relatively high level.
By contrast, Qualcomm has emerged as the most notable outlier in the semiconductor sector. The report notes that, since the beginning of this year, Qualcomm has spent most of the time in the bottom 10% of trading congestion, with only a brief deviation during April and May when its stock price rose on news related to its data center business.
In Bernstein's analysis, Qualcomm is not only one of the least crowded semiconductor stocks within its coverage universe, but also ranks among the lower end of the spectrum among global large-cap technology hardware and semiconductor companies.
There is also a clear divergence in the semiconductor equipment sector. Bernstein notes that, since June, KLA Corp has experienced significantly lower trading congestion compared to its peers, including Applied Materials, Lam Research, and PDF Solutions.
The firm believes this is primarily due to KLA Corp's relatively limited short-term investment appeal. With longer product delivery cycles, KLA Corp's growth in 2026 is expected to lag behind some peers, while rising DRAM costs for its products could put roughly 100 basis points of pressure on its gross margin. Moreover, KLA Corp's exposure to the memory‑chip supply chain is relatively limited, and its valuation is higher than that of Applied Materials and Lam Research, prompting some investors to shift toward other semiconductor equipment companies.
In the tech hardware sector, trading congestion for AI server and storage-related stocks is also at a high level. Bernstein data shows that Apple (AAPL.US),$Western Digital (WDC.US)$、$Seagate Technology (STX.US)$and$Dell Technologies (DELL.US)$All are located within the top 10% of trading congestion, while$HP Inc (HPQ.US)$and$Super Micro Computer (SMCI.US)$Then it falls within a lower congestion range.
Among them, Apple is one of the stocks with the most persistent trading congestion within Bernstein's coverage. Since the beginning of 2024, the stock has almost consistently remained in the top two tiers of highest congestion.
Bernstein noted that despite rising memory chip costs sparking market concerns about Apple's profitability, trading congestion has not declined significantly. The firm believes that increasing DRAM and NAND costs could pose downside risks to Apple's gross margin and earnings per share in the December quarter. If the upcoming earnings report falls short of expectations, highly concentrated investor positions could amplify short-term stock price volatility.
However, Bernstein maintains its "Outperform" rating on Apple, believing that Apple is poised to become a key gateway for consumers to access AI services, while avoiding the massive AI capital expenditures faced by some large cloud computing and software companies.
In the storage industry, Western Digital has remained in the top 10% of stock‑market trading congestion since June 2025, making it one of the longest‑running holdings within Bernstein's coverage to consistently exhibit such high levels of market activity.
The trading congestion for Seagate Technology has also risen markedly in the second half of this year. As market confidence in the investment rationale for hard disk drives (HDDs) has strengthened, the stock has remained in its most crowded range since July. Nevertheless, Bernstein continues to rank Seagate among its top picks and maintains an "outperform" rating. The firm notes that Seagate's nearline HDD capacity for the 2026 calendar year is already sold out, and most of the related capacity through 2028 has secured order commitments.
In addition, Seagate's leading position in heat-assisted magnetic recording (HAMR) technology is a key reason why Bernstein remains optimistic about its earnings prospects. The firm projects that by 2027, HAMR‑based products will account for roughly 70% of Seagate's nearline hard drive shipment capacity, while Western Digital's share is expected to be only around 5%.
In stark contrast to these popular storage stocks is SanDisk. Bernstein notes that, despite being a direct beneficiary of the NAND flash memory industry's upcycle and having seen its share price surge this year, SanDisk remains relatively under‑crowded in terms of trading activity. The firm believes this may reflect the market's underestimation of SanDisk's profitability and the sustainability of the current storage‑industry cycle.
Bernstein notes that the market continues to view SanDisk as a quintessential cyclical stock and remains concerned that the pace of average selling price growth for NAND products may slow. However, SanDisk's newly negotiated long-term supply agreements with its customers are expected to help mitigate this risk.
According to the report, more than half of SanDisk's storage‑chip shipments in fiscal year 2027 and approximately two-thirds of its shipments in fiscal year 2028 are covered by long-term agreements that include price caps and floors. These agreements involve eight data‑center and edge‑computing customers and are backed by $16.5 billion in financial guarantees.
Bernstein believes that these arrangements help enhance the predictability of SanDisk's future revenue and earnings, and, combined with the company's inherent operating leverage, its earnings growth potential may not yet be fully reflected in the market.
However, the firm also cautions that, given SanDisk's relatively short trading history since its relisting and the fact that certain crowded‑trade indicators require at least two years of historical data, the current assessment of SanDisk's crowded‑trade levels remains somewhat limited in reliability.
In the networking equipment sector, among the seven stocks covered by Bernstein, five are in the top 10% of trading congestion. Notably, companies such as Arista Networks, Lumentum, and Coherent are all among the most heavily concentrated holdings. Bernstein believes that while these companies are still expected to benefit from growing demand for AI infrastructure, their stock prices could face greater downward pressure if the market experiences a fresh cooling of popular trades.
$Celestica (CLS.US)$This has resulted in a different trend. The report notes that, although the stock's overall trading congestion remains above the industry average for technology hardware and semiconductors, it has declined markedly since this summer and now ranks among the less crowded companies in Bernstein's coverage of network‑equipment stocks.
Bernstein believes that, as market positioning becomes less crowded and valuation multiples have moderated, Celestica may be poised to benefit from more favorable investment conditions ahead of its upcoming Investor Day.
In terms of investment ratings, Bernstein maintains an "outperform" rating for AMD, NVIDIA, Broadcom, Applied Materials, Apple, Western Digital, Seagate, HP Inc., and SanDisk, while keeping a "market perform" rating for Qualcomm, HP Inc., and Super Micro Computer.
Bernstein emphasized that crowded trading does not necessarily indicate a deterioration in the fundamentals of the underlying companies, nor does it inevitably signal an imminent decline in stock prices. However, when a large number of investors simultaneously hold similar AI‑related securities, any shift in market sentiment could trigger coordinated liquidation, thereby amplifying price volatility.
The firm believes that, as AI‑infrastructure investment continues to draw capital, investors should place greater emphasis on the degree of crowdedness in individual stock holdings, treating it as a key risk‑management indicator. While assessing companies' growth potential, they should remain vigilant about the potential downside risks stemming from concentrated positioning in popular trades.
Editor/stephen