Wall Street financial giant Bernstein's latest "Global Semiconductor Equipment Tracker" research report shows that the investment wave in semiconductor stocks tied to the AI computing theme is spreading from AI computing resource procurement by frontier AI labs and cloud computing giants to semiconductor manufacturing equipment stocks that are critical to AI computing infrastructure.
Bernstein's analyst team stated that the stock market investment wave surrounding AI semiconductors is accelerating its spread to the upstream of chip manufacturing, and semiconductor equipment companies are expected to become an important force in承接 the latest round of massive AI computing capital expenditure and semiconductor profit growth cycle.
Bernstein cited that Japan market semiconductor equipment supplier sales reached 545 billion yen in August, up 50% year-over-year, with front-end manufacturing, assembly, and testing equipment growing approximately 37%, 22%, and 98% respectively; the three-month moving average sales continued to rise, indicating that equipment demand expansion has increasingly strong mid-term demand support.
Meanwhile, Micron's latest fiscal quarter revenue grew approximately 379% year-over-year to $54.229 billion, with next quarter revenue guidance further rising to $61.5 billion, plus or minus $1.5 billion, showing that storage demand is translating into strong revenue, and Micron management stated that it will greatly expand production capacity in the future to drive storage chip supply scale expansion; on October 2, the Philadelphia Semiconductor Index rose 2.4%, reflecting the market's continued attention to semiconductor profit growth against the backdrop of surging US Treasury yields.
Since the beginning of this year, technological advances in the world's most frontier AI agents/AI large models, represented by Meta Muse, OpenAI Astra, and Anthropic Claude, have provided an important technical foundation for the large-scale commercial expansion of AI applications across industries and the continued surge in AI computing demand—especially the expansion of agent application scope, which is expected to simultaneously increase demand for AI core infrastructure resources such as AI GPUs/TPUs, high-performance CPUs, data center high-performance HBM/DRAM/NAND storage chips, and high-speed optical interconnect components.
From the perspective of AI inference system architecture, GPUs and specialized accelerators handle model computation, CPUs are responsible for converting inference results into actual operations, and memory and storage are responsible for saving and retrieving task states. Long context and multi-turn calls increase prefill computation and KV cache demand; browsers, code sandboxes, retrieval, and task orchestration increase server CPU load; files, databases, persistent memory, and cache tiering extend demand to server DRAM, enterprise SSDs, and high-speed networking. Nvidia's technical materials have already described AI agent inference as a vast systems engineering endeavor spanning GPU HBM, CPU DRAM, local NVMe and remote storage, as well as internal high-speed optical interconnects critical for data transmission.
In the view of Anthropic management, storage chip components in AI data center server clusters, as well as AI GPUs, remain the clearest supply bottlenecks at the AI computing industry chain level. Market research firm TrendForce estimates that server DRAM contract prices will cumulatively rise about 270% in 2026, and enterprise SSD prices will cumulatively rise about 235%; HBM contract prices may still rise 70%–140% in 2027, continuing to show doubling growth. These data reflect the combined effect of continued AI computing demand expansion and storage chip price increases. TrendForce calculations also show that NVL72 rack shipments covering Blackwell and Vera Rubin platforms are expected to grow more than 50% year-over-year in 2027; its market research chart shows that related system output value is expected to rise from about $226 billion in 2026 to $711 billion in 2027, a sharp year-over-year increase of 214%.
From an engineering logic perspective, agents expand a single question-and-answer into planning, retrieval, tool calling, code execution, and result verification, and multi-turn reasoning increases computing demand. Management of US semiconductor equipment giant Applied Materials has previously clearly pointed out that agent applications urgently need vast AI accelerators such as AI GPUs as well as more CPU-intensive computing architectures, and increase DRAM and NAND demand, providing additional growth momentum for wafer fabrication equipment.
Therefore, the semiconductor equipment segment can be said to benefit from the combined advancement of "capacity expansion demand" and "process upgrading." The gate-all-around structure of advanced logic chips, high-layer 3D NAND, and HBM stacking with advanced packaging have raised technical requirements for precision deposition, selective etching, chemical mechanical polishing, and defect control. Continued demand growth may both drive new production lines and increase upgrade demand for existing lines.
These leading semiconductor equipment companies headquartered in Japan are favored by Bernstein fundamentally because their product advantages cover key processes in AI chip manufacturing. For example, one of the global semiconductor equipment leaders—Tokyo Electron, Applied Materials' strongest competitor—has a broad front-end equipment layout, benefits from storage and advanced logic investment, and yen depreciation may also enhance its pricing competitiveness. The reason Japan semiconductor equipment has become a focus of Bernstein's observation is that Japanese suppliers account for about one quarter of the global wafer fabrication equipment market and hold prominent competitive positions in multiple key processes. In addition, Bernstein continues to be bullish on several semiconductor equipment super leaders in the US and China markets, including Applied Materials, KLA, Lam Research, Naura Technology, and AMEC.
For the AI computing theme, Bernstein stated that the core investment value of semiconductor equipment companies comes from the shared demand for advanced manufacturing capabilities across multiple chip routes: competition among GPUs, custom AI chips, HBM, and server memory can jointly increase investment in deposition, etching, processing, inspection, and testing. Bernstein therefore maintains "Outperform" ratings on 11 companies, forming a semiconductor sector investment portfolio and stock selection framework of "global capacity expansion + rising process complexity + share gains by advantaged manufacturers." The table below fully lists the stock investment ratings, target prices, and corresponding 12-month upside potential for these stocks as given by Bernstein's analyst team; the calculation baseline uniformly adopts the closing prices listed in the report for October 1, 2026, rather than real-time stock prices.
Japan semiconductor equipment sales surge 50%: the AI investment wave is reaching wafer fabs. What Bernstein's analyst team highlights is a set of equipment sales data with global industry observation value, rather than unfulfilled procurement intentions. The report uses August statistics released by the Semiconductor Equipment Association of Japan (SEAJ) on September 30, and simultaneously observes single-month and three-month moving averages to distinguish short-term fluctuations from trends: August Japan supplier semiconductor production equipment (SPE) sales were 545 billion yen, up 50% year-over-year in yen terms and down 4% month-over-month; up 39% year-over-year in dollar terms. The three-month average sales rose 47% year-over-year and 7% month-over-month in yen terms, and rose 35% year-over-year and 10% month-over-month in dollar terms, continuing the upward cycle that began in mid-2023. Bernstein stated that these data positively indicate that growth is not merely due to yen conversion, nor should a single-month month-over-month decline be taken as evidence of a reversal in prosperity.
By segment, front-end wafer fabrication equipment grew 36.6% year-over-year and fell 8.3% month-over-month; assembly equipment grew 22% year-over-year and 5.9% month-over-month; testing equipment grew 98% year-over-year and 17% month-over-month, with testing equipment's three-month average rising about 14% month-over-month. Front-end maintaining high year-over-year growth, packaging continuing to expand, and testing nearly doubling constitute the most important industry chain structure signal in this report.
Bernstein's analyst team's latest expectations show that the firm forecasts the global wafer fabrication equipment (WFE) market to grow 26.3% in 2026 and 32.6% in 2027; multiplying these two years of forecasts together, the 2027 market size will expand by about 67.5% compared with 2025, meaning that the hundreds-of-billions-of-dollars semiconductor equipment industry faces two consecutive years of significant expansion. Bernstein emphasized that Micron's strong business outlook, showing fiscal 2026 net capital expenditure reaching $27.37 billion and the latest disclosed fiscal 2027 first-half capital expenditure plan of about $25 billion, provides concrete support for storage manufacturers to continue expanding manufacturing investment.
The reason Bernstein's semiconductor equipment research focuses on the Japan market is mainly that large-scale global wafer fab capacity expansion has greatly benefited Japanese semiconductor equipment suppliers, which occupy a high weight in the Japanese stock market and have important influence in the global semiconductor industry chain. Bernstein stated that Japanese equipment companies have a strong business match with the recovery in storage capital expenditure: Tokyo Electron is the world's fourth-largest and Japan's largest semiconductor production equipment supplier, covering six major product areas, benefiting from DRAM and advanced logic investment, and may expand share and profit margins through pricing competitiveness after yen depreciation; DISCO has about 85% share in grinding and dicing equipment, with recent growth coming from HBM and CoWoS, and can subsequently benefit from hybrid bonding, 3D stacking, and backside power delivery related processes; Kokusai's batch atomic layer deposition (Batch ALD) is mainly used in NAND and is expected to expand adoption with advanced processes such as gate-all-around (GAA).
In addition, Lasertec has about 50% share in mask inspection and holds an exclusive supply position in actinic inspection; penetration of the new A200HiT equipment into wafer fab applications is expected to expand the serviceable market, and Bernstein is very bullish on the A200HiT new product expanding EUV wavelength inspection applications; Advantest, with the report's stated about 65% share in HBM testing equipment and its supply position in Nvidia AI GPU testing, benefits from increased testing intensity, higher average selling prices, and product mix upgrades. From an engineering perspective, the more complex the chip structure, the more stacked layers, and the higher the total value after packaging, the stricter the requirements for precision processing, defect detection, and test coverage during manufacturing, so equipment spending can benefit simultaneously from capacity increases and higher process investment per unit of capacity.
From capacity expansion dividends to "complexity dividends": semiconductor equipment leaders take over the AI investment wave. The coverage of Bernstein's semiconductor equipment investment recommendations further illustrates that this round of semiconductor equipment investment opportunities has cross-regional and cross-process market extension characteristics, and presents the "complexity dividend" uniquely possessed by semiconductor equipment within the semiconductor sector—that is, the more complex the architecture, stacking, and packaging of AI infrastructure chips such as AI chips/storage chips/optical chips, the higher the equipment capabilities required for manufacturing and verification; expensive chips and multi-die packaging also raise failure costs, making more sufficient inspection and testing more economically valuable.
In the US market, the case for Applied Materials includes long-term WFE growth, serviceable market expansion, service business growth, and capital returns; Lam Research (LRCX) benefits simultaneously from technological transitions such as GAA, advanced packaging, HBM, and NAND upgrades; KLA receives valuation premium support due to structural growth in process control, durable competitive advantages, relatively lower China domestic substitution risk, and disciplined capital allocation.
In the China market, Bernstein stated that Naura Technology covers physical and chemical vapor deposition, dry etching, thermal processing, and cleaning, and serves logic, DRAM, and NAND customers; AMEC focuses on dry etching and is expanding into deposition areas such as ALD, LPCVD, and epitaxy; Piotech, based on multiple types of thin-film deposition equipment, is further expanding into wafer-to-wafer and die-to-wafer hybrid bonding equipment. The common driver for all three is domestic substitution and share gains, so the growth of Chinese equipment companies includes localization factors independent of global total demand growth. The report maintains only a "Market Perform" rating on SCREEN (7735.JP), listed in the Japanese stock market: despite a lower valuation, cleaning intensity has not clearly increased, competition is strong, and a sharp decline in China revenue share may affect profit margins. All of this positively indicates that Bernstein places greater emphasis on earnings elasticity brought by technological upgrading and share gains, rather than giving a uniform bullish judgment simply based on the "semiconductor equipment" label.
How does the agent convert token demand into equipment spending? Meta's Muse continuously executes tasks on independent cloud virtual machines, while OpenAI's GPT-6 Astra strengthens computer operation and multi-step professional work capabilities; together they point to AI shifting from single question-and-answer to continuously completing tasks; notably, in the view of Wall Street analysts, these latest bases for judgment supporting agent application expansion do not need to use "AGI has been achieved" as a premise for argument. From the underlying engineering mechanism, a task repeatedly undergoes planning, reasoning, tool calling, code or browser execution, feedback verification, and retries, expanding the total amount of input, output, and reasoning tokens, and increasing simultaneously running execution environments; GPUs and AI ASICs handle model computation, CPUs handle scheduling and tool execution, HBM handles high-speed model data access, server DRAM carries runtime states, and SSDs and networking support data, persistent memory, and tiered caching.
Subsequently, rising sustained utilization and growing customer usage drive cloud vendors to purchase more servers, and chip manufacturers then increase wafer, advanced packaging, and testing capacity based on demand visibility, ultimately transmitting into equipment procurement. A real-world example is Anthropic signing a seven-year $11.6 billion agreement with Akamai, explicitly for CPU workloads; Akamai expects related capital expenditure of about $5.5 billion and an additional about $1.7 billion prepurchase in 2026 for key components including memory. Therefore, the agent's "task scale × execution depth × concurrency demand" can be said to be the core variable connecting application prosperity and equipment expansion; tokens do not correspond to computing power at a fixed ratio, but as long as total task volume and concurrency demand growth exceed efficiency improvements in single-task efficiency, computing power and manufacturing investment will still expand. The growth in AI chip/storage chip capacity investment, large-scale expansion of advanced packaging, and near-doubling of semiconductor testing equipment observed by Bernstein are consistent with this AI capital transmission chain of accelerated semiconductor equipment spending expansion.
Statistics from another Wall Street financial giant, Goldman Sachs, show that although US hedge funds sold most industries in September, semiconductor equipment and software received strong inflows; however, in the week ending September 30, broad technology funds saw net outflows of $2.63 billion, while financial and utility funds saw net inflows of $1.13 billion and $468 million respectively, enough to show that global capital allocation directions have diverged. Goldman Sachs' latest view also clearly mentions that investor interest is comprehensively expanding toward AI deployment tools, cybersecurity, neutral core AI data infrastructure and agent commercial applications, and semiconductor equipment benefiting from AI computing demand expansion.
The investment insights from Bernstein's report can be said to echo the fund flow statistics from Goldman Sachs—AI semiconductors still have solid industrial growth support, the semiconductor equipment industry chain is beginning to become one of the important links for institutions to capture manufacturing investment growth, and the comprehensive diffusion of the application layer provides an incremental source for the next round of computing demand. In its research report, Bernstein's estimated earnings per share for Applied Materials rose sharply from $12.76 in 2026 to $18.30 in 2027, with the corresponding price-to-earnings ratio falling from 40.1x to 27.9x; DISCO's estimated earnings per share rose from 1,830.28 yen to 2,336.94 yen, with the corresponding price-to-earnings ratio falling from 33.0x to 25.9x. These latest upward earnings revisions highlight valuation digestion brought by earnings growth under strong stock price and profit forecasts. Bernstein maintains "Outperform" ratings on 11 leading semiconductor equipment companies covering the China, US, and Japan markets, forming an investment portfolio and stock selection framework of "global capacity expansion + rising process complexity + share gains by advantaged manufacturers."
Comments