Latest statistics reveal that foreign investors have been aggressively purchasing Japanese stocks for eight straight weeks as of the week ending May 23. This sustained inflow is primarily driven by a significant drop in oil and natural gas trading prices, alongside a bullish sentiment for AI chip and semiconductor equipment stocks. These sectors, which are closely tied to AI computing infrastructure and hold substantial weight in Japan's benchmark Nikkei 225 index, are seeing rising prices due to strong prospects for AI computing demand. Stocks such as SoftBank, Socionext, Advantest, Tokyo Electron, Lasertec, and Disco, which collectively represent a significant portion of the Nikkei 225, have become the central narrative for the recent large-scale foreign capital inflows and are key contributors to the index's record-breaking gains. Nvidia's latest earnings report underscores the robust underlying demand for AI computing power. The company's quarterly revenue surged 85% year-over-year to $81.62 billion, with data center revenue nearly doubling to $75.2 billion. CEO Jensen Huang described AI data center construction as "the largest infrastructure expansion in human history." The sustained, powerful rally of Japan's AI chip and semiconductor equipment leaders can be seen as part of a broader revaluation of the AI computing supply chain, spurred by the record-breaking spillover demand for Nvidia's AI GPUs. Compared to the U.S. and South Korean markets, which are also dominated by the AI computing frenzy, Japan's uniqueness lies in its lack of super-dominant AI or memory chip giants like Nvidia, AMD, Micron, SK Hynix, or Samsung. Instead, Japan possesses a cluster of indispensable AI assets deeply embedded in the computing infrastructure supply chain, such as Tokyo Electron, Advantest, Disco, Lasertec, Socionext, and SoftBank. Consequently, foreign investors widely regard Japan as a "second front" in the AI computing infrastructure industrial chain. Nvidia's performance clearly highlights that the global frenzy for building AI computing infrastructure is far from over and is expanding from AI GPUs/ASICs to data center CPUs, high-performance networking infrastructure, enterprise-grade HBM/DRAM/NAND storage, server clusters at the system level, AI super factories, and large-scale enterprise AI cloud computing systems. Morgan Stanley predicts that by 2028, nearly $3 trillion in AI-related infrastructure investment will flow through the global economy, with over 80% of the spending still ahead. Analysts at Bank of America note that from an engineering perspective, the bottleneck for AI computing infrastructure is expanding from "sufficiency of AI GPUs/ASICs" to "system-level throughput delivery." GPUs/ASICs handle matrix calculations, CPUs manage scheduling and agentic AI workflows, leaders in HBM/DRAM/NAND storage chips address large model weights, KV cache, vector databases, inference data lakes, and data movement, while data center interconnect and networking chains like Marvell, Arista, and Astera handle cluster communications. Semiconductor equipment giants and EDA players like TSMC, Intel, ASML, Lam Research, KLA, Cadence, and Synopsys determine advanced process nodes, advanced packaging, and yield ramp-up.

Foreign investors poured approximately ¥1.08 trillion (about $6.77 billion) into Japanese equity assets during the week ending May 23, a nearly 14% increase from the previous week's net purchase of ¥948.4 billion. The chart below shows recent foreign capital inflows into Japanese stocks, in billions of yen. The chip sector attracted the largest share of foreign capital. This follows Nvidia CEO Jensen Huang's statement last week during an earnings call that demand for its flagship AI chips is unprecedentedly strong, and his characterization of AI data center construction as "the largest infrastructure expansion in human history." SoftBank Group Corp., a major Japanese investment firm focused on AI chips, saw its stock price surge 17.62% last week, with its market capitalization now exceeding $250 billion. Socionext, a semiconductor giant specializing in data center AI ASIC chip design, rose 12.26%. Semiconductor equipment firms Advantest, Tokyo Electron, and Lasertec also posted significant gains. Year-to-date, foreign investors have injected nearly ¥11.7 trillion into Japanese stocks, a stark contrast to the net purchase of only about ¥742.1 billion during the same period last year. Meanwhile, Japanese long-term government bonds, which have faced heavy selling pressure recently in Japanese financial markets, unexpectedly recorded net foreign purchases of about ¥1.35 trillion, compared to an outflow of roughly ¥1.03 trillion the previous week. As oil prices have recently fallen and market expectations for a final U.S.-Iran peace deal have turned optimistic, global bond selling has eased, and higher yields have attracted some foreign institutional investors like hedge funds. However, foreign investors broadly reduced their holdings of short-term debt instruments by about ¥2.22 trillion, the largest reduction since March 28. The chart below shows recent foreign capital inflows into Japanese debt securities, in billions of yen. In other developments, Japanese investors were net sellers of foreign stocks, withdrawing about ¥358.7 billion, marking the third weekly net sell-off in the past four weeks. However, with easing Middle East geopolitical tensions and falling oil prices raising expectations for the reopening of the Strait of Hormuz, they made small net purchases of ¥10.3 billion in foreign long-term bonds, extending their buying streak to four consecutive weeks.

According to Bank of America analysts, AI computing infrastructure is entering a more durable and broader capital expenditure cycle. Around the same time, a report from Morgan Stanley, another Wall Street giant, indicated that the AI computing arms race has entered a system-level expansion phase, with AI infrastructure demand showing a rare "inelastic" trend—meaning tech giants continue to ramp up AI data center construction regardless of cost curves. This "demand inelasticity" is expected to further strengthen U.S. economic resilience and the overall earnings growth of the S&P 500 index. A recent research report from JPMorgan Chase, a leading Wall Street commercial bank, suggests that driven by the tech stock bull market frenzy centered on the AI computing theme, the S&P 500 index could break through the seemingly distant milestone of 9,000 points within the next year. This implies a potential further rise of over 20% for the benchmark index, which has already surged more than 60% since 2024. Wall Street analysts are expanding the AI computing infrastructure narrative from a "GPU-dominated/single-core driven" story to a full-stack computing power revaluation encompassing "AI GPU/ASIC + CPU + HBM/DRAM/NAND storage chips + optics-interconnect-led data center high-speed connection systems." They predict global AI computing infrastructure spending will approach $3 trillion by 2028. In other words, the global popularity of AI agents combined with record-high capital expenditures from North American tech giants is shifting the market's AI investment focus from a "single-point computing power race around GPUs" to an "AI agent-driven full-stack computing power system." The next wave of excess alpha returns will no longer belong solely to the strongest leaders in the AI GPU/ASIC field but will systematically diffuse across the full stack of AI computing infrastructure layers, including data center high-performance CPUs, DRAM/NAND/HBM storage, AI PCBs, liquid cooling systems, data center optical interconnect systems, ABF substrates/glass substrates, and extensive wafer foundry services. SoftBank's stock price has soared since the beginning of the year and recently hit new highs, largely because it is the majority shareholder (holding nearly 90%) of Arm Holdings Plc. Arm's stock has surged recently due to exploding CPU demand. Arm's instruction set architecture technology is widely used across the consumer electronics industry and, in recent years, has increasingly served as the foundational architecture for hyperscale AI data center server CPU product lines. SoftBank's substantial investments in recent years in Arm, Graphcore, and Ampere Computing, along with its latest combined investments in OpenAI (of which SoftBank is already a major shareholder) and the U.S. "Stargate" AI infrastructure project, reflect its full-stack strategy spanning from the most fundamental AI hardware architecture to AI computing infrastructure clusters and up to the AI application layer. SoftBank founder Masayoshi Son has publicly stated that SoftBank aims to become the largest provider of AI computing and application-level foundational platforms in the era of "Super Artificial Intelligence" (ASI) over the next decade. He has adopted an almost "all-in" aggressive stance towards OpenAI, the developer of ChatGPT and a global leader in large AI models. Recent reports from several Wall Street financial giants indicate that the semiconductor equipment sector is one of the biggest winners amid soaring AI computing and storage demand. As the global hyperscale AI data center construction drive led by tech giants like Microsoft, Google, and Meta intensifies, it is comprehensively accelerating the expansion of 3nm and below advanced process AI chip production, CoWoS/3D advanced packaging capacity, and DRAM/NAND storage chip capacity expansion among chip manufacturing giants. The long-term bullish thesis for the semiconductor equipment sector is becoming increasingly solid. Japanese semiconductor equipment companies play a crucial role in the equipment value chain related to core AI computing infrastructure like AI GPUs/ASICs. For instance, the actinic EUV mask inspection technology focused on by Lasertec, whose stock price has skyrocketed 200% since 2025, is a vital and indispensable technical link in the global AI chip supply chain. Lasertec is arguably the core supplier for "mask inspection/metrology" within the most critical chain of chip manufacturing—the "EUV lithography chain." Using the same 13.5nm actinic wavelength as EUV to detect "printable defects" that only manifest under EUV lithography conditions ensures mask yield and mass production stability, making it a key link for the successful implementation of the EUV lithography process. The current super-cycle for AI chips and high-end storage is precisely expanding the application boundaries and capacity intensity for EUV and High-NA technologies—the more extensive and complex the mask coverage, the stronger Lasertec's scarcity, and the greater the elasticity in its orders and valuation. Compared to the two U.S.-based semiconductor equipment giants Applied Materials and Lam Research, Japan's Tokyo Electron holds the highest global market share in the coater/developer field. Tokyo Electron is also a formidable competitor to Applied Materials in areas such as ALD, CVD, PVD, RTP, CMP, etching, and ion implantation equipment. In TSMC and Intel fabs, the presence of Tokyo Electron and Applied Materials is ubiquitous. Unlike ASML, which remains focused on lithography, and Lam Research, which emphasizes etching, cleaning, patterning, and critical thin-film processes—particularly high-aspect-ratio (HAR) etching/deposition and related process capabilities required for advanced HBM storage—Tokyo Electron and Applied Materials provide high-end equipment that plays a vital role in almost every step of chip manufacturing. Their products cover crucial chip-making processes like Atomic Layer Deposition (ALD), Chemical Vapor Deposition (CVD), Physical Vapor Deposition (PVD), and Rapid Thermal Processing (RTP). Year-to-date, Tokyo Electron, a significant component of the Nikkei 225, has seen its stock price rise over 50%, with its market capitalization surpassing $150 billion, significantly outperforming the Nikkei 225 index.