Japan’s Semiconductor Rally Is Becoming a National Economic Story

The chip complex becomes a national market signal

When semiconductor shares rise in Tokyo, the move no longer represents just one industry. Kioxia, Advantest and Tokyo Electron now connect artificial intelligence, data centers, memory, manufacturing equipment, the yen, global interest rates, Korean and Taiwanese capital spending and Japanese industrial policy in a single market narrative.

By mid-2026, the concentration had become striking. A Reuters market analysis cited an estimate that Tokyo Electron, Advantest and Kioxia together represented roughly one quarter of the Nikkei’s value. Add Murata, Sony and Kyocera, and the technology-related share approached 35%.

That means a rising Nikkei increasingly reflects global confidence in AI infrastructure rather than uniform strength across the Japanese economy. The same concentration creates vulnerability. A change in U.S. chip shares, Nvidia spending expectations, memory prices, Treasury yields, the yen or export controls can move Japan’s headline index sharply.

Three companies, three layers of the supply chain

Kioxia, Advantest and Tokyo Electron are all labeled semiconductor shares, but they occupy different economic layers.

Together they illustrate the AI infrastructure chain. Advanced logic and memory are needed to run models. More complex chips require more intensive testing. Sustained demand leads manufacturers to order more production equipment.

Kioxia’s extraordinary comeback

Kioxia embodies both the failure and revival of Japanese semiconductors. Its predecessor, Toshiba Memory, traces its history to the invention of NAND flash in the 1980s. Toshiba’s financial crisis forced a sale to a Bain Capital-led consortium for roughly ¥2 trillion in 2018.

Memory is brutally cyclical. Producers expand when demand rises, then prices collapse when supply overtakes the market. Kioxia delayed listing plans several times and received a modest valuation when it finally went public in December 2024.

The picture changed in 2026 as AI demand expanded from training to inference. Training initially favored high-bandwidth DRAM. Large-scale inference requires enormous storage capacity for models, data, retrieval and caching. Competitors had prioritized DRAM and HBM, leaving less flexible NAND capacity.

Kioxia began shipping samples of 10th-generation BiCS Flash from Kitakami, developed with SanDisk. Its wafer-bonding approach supports better performance and power efficiency, while Fab2 provides room for expansion. Reuters reported that the company’s shares rose more than sevenfold in 2026 at one point, giving it a market value above Toyota.

Advantest is the quality gate for AI chips

Advantest becomes more valuable as chips become more expensive and complex. An AI accelerator may combine multiple chiplets, high-bandwidth memory and advanced packaging. A defect discovered after deployment can destroy an extremely costly component and disrupt a data-center system.

Testing therefore takes place at the wafer level, after packaging and before system integration. Engineers verify electrical behavior, speed, temperature tolerance and power use. Complexity can increase both the number of tests and the time required for each device.

Advantest benefits not only from higher chip volume but also from greater testing intensity per chip. In the age of chiplets, HBM and advanced packaging, testing is no longer a final inspection. It is part of the process that makes mass production possible.

Tokyo Electron reflects the world’s fab cycle

Tokyo Electron is the most globally exposed symbol of Japan’s chip revival. It supplies equipment for coating and developing, deposition, etching, cleaning and other essential fabrication steps.

Its results depend not only on Japanese factories but on TSMC in Taiwan, Samsung and SK Hynix in Korea, subsidized fabs in the United States and Europe, Chinese demand and the memory cycle. When Tokyo Electron rises, investors are usually expressing confidence in global fab spending.

Equipment has also become geopolitical. U.S.-led export controls restrict advanced technology sales to China, placing Japanese suppliers between a major commercial market and allied security policy.

Japan once ruled the chip world

In the 1980s, Japanese companies controlled roughly half of the global semiconductor market. NEC, Toshiba, Hitachi, Fujitsu and Mitsubishi Electric dominated DRAM and challenged U.S. producers.

The advantages were formidable: manufacturing quality, patient capital, integrated supplier networks, government-industry coordination and huge domestic demand from electronics and computing.

Success produced trade conflict. U.S. companies accused Japan of dumping and restricting market access. The 1986 U.S.–Japan Semiconductor Agreement constrained Japanese pricing and sought to expand the foreign share of Japan’s domestic market.

The agreement was not the only cause of decline. Yen appreciation, capital intensity, memory cycles and slow corporate decisions all mattered. Japan struggled to adapt when the industry’s architecture changed.

The world shifted from memory giants to foundries

During the 1990s, semiconductor production moved from vertically integrated corporations toward specialization. U.S. fabless firms focused on design. TSMC built the pure-play foundry model in Taiwan. Design, fabrication, equipment, materials, assembly and testing became globally distributed.

Many Japanese electronics groups retained integrated structures. Consensus-heavy decision making made it difficult to match the speed of the smartphone processor, GPU and foundry eras. DRAM operations were consolidated and abandoned. Elpida Memory failed in 2012 and was acquired by Micron.

Japan did not disappear. It remained powerful in silicon wafers, photoresists, specialty gases, ceramics, equipment, testing, power semiconductors and image sensors. It lost finished-chip market share while preserving critical positions inside the supply chain.

The foundation that Sony, Renesas and suppliers preserved

Sony built a leading global position in CMOS image sensors. Demand has moved beyond smartphone cameras into vehicles, factories, robots, security and spatial computing.

Renesas, descended from the chip operations of NEC, Hitachi and Mitsubishi Electric, maintained strength in automotive microcontrollers and embedded devices. When its Naka plant was damaged in the 2011 earthquake, the world discovered how much automobile production depended on one Japanese facility.

Shin-Etsu Chemical, SUMCO, JSR, Tokyo Ohka Kogyo, Disco, Lasertec, SCREEN, Tokyo Electron and Advantest retained global positions in wafers, chemicals, cutting, inspection, cleaning, lithography-related processes and testing. Japan’s revival rests on this installed industrial base.

Supply shocks changed the politics

Two crises transformed semiconductor policy. The 2011 earthquake showed how the shutdown of a specialized plant could interrupt global auto and electronics production. The pandemic-era shortage beginning in 2020 forced automakers to cut output for lack of relatively inexpensive chips.

At the same time, U.S.–China rivalry made advanced chips central to economic security, communications and military capability. Tension around Taiwan highlighted the concentration of leading-edge manufacturing.

Japan began treating semiconductors not as an ordinary corporate business but as infrastructure, alliance policy, regional development and national security. The scale of subsidies and political involvement increased accordingly.

TSMC Kumamoto builds a new cluster

The most visible practical success is TSMC’s move into Kumamoto. Japan Advanced Semiconductor Manufacturing, backed by TSMC, Sony Semiconductor Solutions and Denso, built a fab in Kikuyo.

The first plant serves mature and near-advanced processes needed for vehicles, industrial systems and image sensors. Government support also extended to a second fab. The objective is not only frontier technology; it is reliable domestic supply for industries Japan already leads.

The project transformed Kumamoto. Construction, housing, transportation, power, water and education demand surged. It also created pressure through land prices, traffic, groundwater concerns and labor shortages.

Rapidus is the boldest—and riskiest—bet

If TSMC Kumamoto is pragmatic supply-chain reinforcement, Rapidus is Japan’s high-risk attempt to return to leading-edge logic. Toyota, Sony, NTT, NEC, SoftBank, Denso, Kioxia and MUFG Bank are among its investors.

Rapidus is transferring 2-nanometer gate-all-around technology from IBM and working with imec and other partners. Its IIM-1 pilot line in Chitose began operations in April 2025. The company announced initial operation of 2nm GAA transistors in July 2025 and planned full-scale advanced-packaging pilot work in fiscal 2026, with mass production targeted for fiscal 2027.

Success could provide domestic advanced logic for AI, autonomous systems, quantum computing, communications and defense. Failure would leave enormous public and private investment without sufficient customers or manufacturing yield.

A foundry needs more than transistor technology. It must support customer design, EDA tools, intellectual property, packaging, yield, delivery, security and price. Rapidus must build in a few years the trust TSMC accumulated over decades.

A fivefold domestic-sales goal

In 2026, Japan’s government set a goal of increasing sales of domestically produced semiconductors fivefold by 2040. The policy places chips at the center of AI, data centers, robotics, vehicles, energy, defense, communications and healthcare.

Support extends to TSMC, Rapidus, Kioxia, Micron, power semiconductors, research and workforce development. METI’s strategy combines stable supply with an attempt to rebuild commercial competitiveness.

AI changes the meaning of chip demand

Traditional semiconductor cycles were heavily linked to PC, smartphone, vehicle and consumer-electronics shipments. AI adds a new demand architecture.

Training requires GPUs, accelerators, HBM and networking chips. Inference expands demand into storage, enterprise servers, vehicles, factories, robots and edge devices. AI also requires power systems, cooling, optical networking, testing and advanced packaging.

Japan may not own the dominant GPU platform, but it has strong positions in many surrounding layers. The rally is therefore not a search for a Japanese Nvidia. It is a bet that sustained global AI spending will pull through memory, equipment, testing, sensors and materials.

The yen is both tailwind and warning

A weak yen increases the yen value of overseas earnings for exporters such as Tokyo Electron and Advantest. In the short run, that supports profit and share prices.

But fabs require imported equipment, foreign intellectual property, energy and construction materials. Yen weakness also raises the cost of domestic investment. A revival based only on currency translation and foreign AI demand would be fragile.

Success must be measured in research, domestic suppliers, engineers, power capacity, productivity and wages—not only in market capitalization.

Five risks beneath the rally

Semiconductors are a structural growth industry and one of the most violent capital cycles in modern manufacturing. Prices often collapse not because demand disappears, but because supply expands too aggressively.

Japan.co.jp view: the real test is not the Nikkei

The semiconductor rally has become a national economic story because it combines the memory of 1980s dominance, decades of decline, earthquake and pandemic shortages, U.S.–China rivalry and the AI revolution.

Japan should not seek to recreate the 1980s. The industry is now globally specialized. Taiwan, Korea, the United States, the Netherlands, China and Europe each hold different strengths. A realistic Japanese strategy is not autarky but control of several technologies and production capabilities that allies and customers consider indispensable.

Kioxia represents memory, Advantest testing, Tokyo Electron equipment, Sony sensors, materials companies chemistry and wafers, and Rapidus advanced logic. If these layers reinforce one another, Japan can become a critical node without relying on a single national champion.

The real measure of revival is not a record Nikkei. It is whether fabs sustain production, engineers are trained, regional wages rise, Japanese firms build AI-era products and customers choose the ecosystem after subsidies fade.

The rally is an advance payment on that future. The national question is whether Japan’s industrial economy can deliver what its stock market has already priced in.