Written by Kenji Takahashi
The Structural Case for Japanese Equities
The Japanese equity market is undergoing a historic transformation, propelled by a confluence of structural tailwinds and a voracious global appetite for artificial intelligence infrastructure. The Nikkei 225 index’s ascent past the 62,000 mark represents more than a cyclical rally; it signals a fundamental re-rating of Japan’s role in the global technology supply chain. A decade ago, Japan’s market capitalisation leaders were dominated by traditional automakers and telecommunications conglomerates. Today, four of the top ten companies on the Tokyo Stock Exchange are semiconductor-related entities: SoftBank Group, Tokyo Electron, Kioxia, and Advantest. This seismic shift underscores Japan’s resurgence as an indispensable node in the AI hardware ecosystem.
The macro environment remains exceptionally supportive. The persistently weak yen provides a substantial earnings tailwind for export-oriented technology firms, amplifying their global competitiveness. Concurrently, the Tokyo Stock Exchange’s stringent corporate governance reforms—mandating that companies trading below book value disclose improvement plans—are forcing management teams to prioritise capital efficiency and shareholder returns, unlocking long-dormant value across the market. NAND flash prices have surged 63 percent in a single month, while legacy 2D NAND prices have rocketed 300 percent in six months as Samsung and Kioxia redirect capacity toward advanced AI-optimised products.
However, the most potent catalyst is Japan’s unique positioning within the AI hardware value chain. While the initial wave of AI investment heavily favoured High Bandwidth Memory (HBM) for training applications, the impending explosion of AI inference workloads starting in 2027 will exponentially increase demand for advanced storage solutions—AI solid-state drives, High Bandwidth Flash (HBF), and Storage Class Memory (SCM). This structural transition presents a compelling opportunity for investors to position themselves in the key Japanese beneficiaries before the inference cycle begins in earnest.
Kioxia Holdings Corp (TSE: 285A)
Since its initial public offering on 18 December 2024, Kioxia has experienced an unprecedented revaluation. The stock has surged approximately 359 percent year-to-date, with shares reaching an all-time high of ¥49,430 on 11 May 2026. Its market capitalisation now stands at ¥26 trillion ($240 billion), vaulting the company past Honda Motor to become the fourth-largest listed entity in Japan—a remarkable ascent from 43rd place at listing, when its market cap was below ¥0.8 trillion. Formerly known as Toshiba Memory, the company was spun off from its parent in 2018 and recapitalised by Bain Capital before its delayed but ultimately triumphant public listing.
Kioxia’s product roadmap is exceptionally well-aligned with the shifting demands of the AI ecosystem. The company’s CM9 series enterprise AI SSDs, built on 8th-generation BiCS FLASH TLC technology, have already sold out their entire annual production capacity, reflecting the intense demand for high-performance storage in hyperscale data centres. Furthermore, Kioxia is scheduled to commence mass production of its GP series SSDs by the end of 2026. These drives, capable of 10 million IOPS and based on XL-FLASH low-latency Storage Class Memory, are specifically designed to fulfil substantial orders from NVIDIA and other leading AI accelerator manufacturers.
Looking further ahead, Kioxia’s development of High Bandwidth Flash (HBF) represents a paradigm shift in storage architecture. The prototype chip boasts a 5TB capacity and a sustained bandwidth of 64 GB/s, while operating under 40W—yielding an impressive efficiency of approximately 1.6 GB/s per watt. This is achieved through a unique “daisy chain” ring topology utilising a serial interconnect, combined with a 128 Gbps PAM4 high-speed transceiver, flash prefetch logic, and a distributed independent controller architecture. HBF is purpose-built for the inference era, where massive model weights must be served at ultra-low latency.
Perhaps most intriguingly, Kioxia is advancing OCTRAM, an IGZO oxide semiconductor 3D DRAM technology. Having successfully verified 8-layer stacked transistors with ultra-low leakage control, OCTRAM possesses the potential to disrupt the existing HBM market by offering comparable bandwidth at significantly lower power consumption and cost. This positions Kioxia not merely as a storage company, but as a potential competitor in the memory hierarchy itself.
Financial Summary
| Metric | FY2025 (Actual/Guidance) | FY2026 (Market Estimate) |
| Revenue | ¥2.18T – ¥2.27T (+28–33% YoY) | — |
| Operating Profit | ¥709.5B – ¥799.5B | ~¥4.0T (~4x increase) |
| Net Profit | ¥453.7B – ¥513.7B | ~¥2.4T |
| Market Cap | ¥26T ($240B) | — |
Consensus analyst rating: BUY. Daiwa Securities has set a 12-month target price of ¥50,000. Given the robust technological pipeline and the impending surge in inference-driven storage demand, Kioxia represents a cornerstone asset in any technology-focused portfolio.
Verdict: BUY
TDK Corporation (TSE: 6762)
TDK has emerged as a stealth beneficiary of the AI revolution, experiencing a massive surge in its equity valuation driven by the insatiable demand for advanced electronic components. As the complexity of AI hardware increases—with higher power densities, more intricate thermal management requirements, and exponentially greater data throughput—the demand for high-performance passive components, precision sensors, and advanced energy devices scales proportionally. TDK’s dominant positions in multilayer ceramic capacitors (MLCCs), thin-film inductors, and battery technology make it an indispensable supplier for next-generation data centres and AI accelerator modules. The company’s ability to consistently deliver components that meet the rigorous thermal and electrical requirements of advanced AI systems ensures a durable competitive advantage that is difficult for competitors to replicate at scale.
Verdict: BUY
TOWA Corporation (TSE: 6315)
In the highly specialised field of semiconductor packaging, TOWA Corporation commands a near-monopoly in die-level compression molding equipment, holding approximately 66 percent of the global market share. The proliferation of advanced packaging techniques—essential for heterogeneous integration and HBM production—has dramatically accelerated demand for TOWA’s machinery. This dynamic is reflected in the company’s financial performance, with fiscal 2025 second-quarter orders growing 20 percent year-over-year, heavily driven by aggressive capacity expansion in the Chinese and Korean markets, which contributed 33.7 percent of incremental growth. Following a 400 percent stock surge in 2025, the valuation reflects significant optimism. While the underlying fundamentals remain exceptionally strong and the structural demand drivers are intact, the current share price already discounts much of the near-term upside, warranting a more measured stance until a more attractive entry point materialises.
Verdict: HOLD
DISCO Corporation (TSE: 6146)
DISCO Corporation dominates the global market for wafer thinning and dicing equipment, maintaining a market share exceeding 90 percent. The transition to advanced nodes and complex 3D packaging architectures inherently increases the intensity and precision required in the dicing process. Management has explicitly attributed the record-breaking ¥93 billion in shipments during the first quarter of fiscal 2025 directly to surging demand for HBM. As semiconductor manufacturers continue to scale production of AI-specific memory—with layer counts increasing from 12 to 16 and beyond—DISCO’s equipment remains a non-negotiable capital expenditure, ensuring robust and highly visible revenue streams for the foreseeable future. The company’s razor-and-blade model, with recurring consumable revenue from dicing blades, provides additional earnings stability.
Verdict: BUY
Resonac Holdings Corporation (TSE: 4004)
Formerly known as Showa Denko, Resonac has positioned itself as a critical materials supplier for advanced semiconductor packaging. The company currently controls approximately 50 percent of the global market for Thermal Compression Non-Conductive Film (TC-NCF) and standard NCF—materials that are absolutely essential for the stacking and bonding processes used in HBM manufacturing. This strategic dominance translated into a 5.8 percent year-over-year growth in core operating profit for its semiconductor materials segment in 2025. As HBM layer counts increase from current-generation 12-high stacks toward 16-high and beyond, the volume and specification requirements for Resonac’s materials will only intensify. The company’s deep process expertise and co-development relationships with leading memory manufacturers create formidable barriers to entry for potential competitors.
Verdict: BUY
Additional Names to Watch
While the aforementioned companies represent the core AI infrastructure play in Japan, several other firms warrant close monitoring as the cycle matures.
Shibaura Mechatronics has established itself as the global number two provider of advanced 16+ layer chip-to-wafer (C2W) hybrid bonding equipment, trailing only the Netherlands’ Besi. This technology is critical for future iterations of HBM, and Shibaura’s positioning ensures it will capture a significant share of the next wave of memory packaging capital expenditure.
Tokyo Electron (TEL, TSE: 8035) holds a strong secondary position in wafer-to-wafer (W2W) hybrid bonding equipment with its Synapse series, positioning it well for the upcoming HBF cycle. TEL’s fiscal 2026 fourth-quarter results beat expectations, and management has guided for 40 percent revenue growth, driven by AI investment in cutting-edge fabrication technology.
For investors seeking speculative, long-tail exposure, Helical Fusion offers a unique proposition. As a Japanese fusion energy startup utilising proprietary helical stellarator technology, it is the first domestic company to sign a fusion Power Purchase Agreement. With total funding of approximately ¥5.2 billion, partnerships with over ten industrial companies, and plans for a demonstration device (“Helix HARUKA”) in the 2030s followed by a pilot power station (“Helix KANATA”) in the 2040s, it represents a high-risk, high-reward venture at the frontier of energy technology.
Summary of Verdicts
| Company | Ticker | Thesis | Verdict |
| Kioxia Holdings | TSE: 285A | AI SSD + HBF + OCTRAM; inference cycle leader | BUY |
| TDK Corporation | TSE: 6762 | Indispensable AI component supplier | BUY |
| TOWA Corporation | TSE: 6315 | 66% molding equipment share; rich valuation | HOLD |
| DISCO Corporation | TSE: 6146 | 90%+ dicing share; HBM-driven record shipments | BUY |
| Resonac Holdings | TSE: 4004 | 50% HBM bonding materials market share | BUY |
────────────────────────────────────────────────────────────
Disclaimer: This article is for informational purposes only and does not constitute investment advice. The author and equitiesorbis.com may hold positions in the securities discussed. Past performance is not indicative of future results. Investors should conduct their own due diligence before making investment decisions.
