Japan's semiconductor revival at a crossroads: the strategic game between TSMC Kumamoto and Rapidus

In the 1980s, Japan once held half of the global DRAM market. After the 1990s, however, the twin pressures of misreading the market and U.S.–Japan trade friction caused Japanese semiconductor manufacturers to collectively fall behind before the 22/20nm node. This history of decline still forms the backdrop of Japan's industrial policy. In 2021, the Kishida Fumio government launched a semiconductor revival strategy under the banner of economic security, unveiling two flagship projects of very different natures: TSMC's Kumamoto fab and Rapidus. Two paths—one taking root inward, one climbing upward—reflect Japan's difficult balance between technological sovereignty and global integration.

The Kumamoto Model: Rebuilding Supply Chain Resilience with Foreign Investment

TSMC's Kumamoto fab is not a traditional "foreign investment attraction." It is a wafer fabrication facility jointly established by TSMC with Japanese companies such as Sony and Denso, and heavily subsidized by the Japanese government. Its primary goal is to produce mature-node chips—28nm and more mature nodes widely used in automobiles, industrial equipment, and consumer electronics. For Japan, which has lost momentum in the global advanced-process race, the significance of this "makeup-class" investment lies in rebuilding the physical foundation of domestic production.

From an industry-chain perspective, the greatest value of the Kumamoto fab lies not in the facility itself, but in how it has reactivated the interactive network among Japan's semiconductor equipment, materials, and downstream applications. Japan still has world-class competitiveness in semiconductor chemicals and manufacturing equipment, but the disconnect from chip design and mass production has weakened its system integration capability. The Kumamoto project allows Japanese companies to connect directly with TSMC's manufacturing system nearby, bringing materials and equipment innovations straight into mass-production verification processes—something vital to restoring the industry's dynamic learning capability.

Economist Jun Akabane noted at a Stanford APARC symposium that the strategic logic of the Kumamoto model is "to deeply integrate wafer manufacturing with existing industrial clusters by filling the domestic supply gap for mature chips." This choice is highly consistent with the current trend of regionalizing global supply chains—it does not pursue the most cutting-edge process numbers, but rather the completeness of the industrial chain. For the automotive industry, a stable and reliable supply of mature chips is far more practically meaningful than a dazzling 2nm node.

Rapidus: A Technological Gamble Driven by National Will

In stark contrast to Kumamoto's pragmatic path, Rapidus embodies Japan's ambition to return to the pinnacle of global advanced logic chips. Founded in 2022 and jointly funded by eight Japanese corporate giants—including Toyota, Sony, and SoftBank—the company partners with IBM and imec, Belgium's microelectronics research center, with the goal of achieving mass production of 2nm chips around 2027. The project carries a generation of Japanese people's lingering frustration over the "lost three decades."However, the challenges facing Rapidus are almost textbook. In terms of technology, Japan has been away from advanced-process R&D for more than two decades; even with IBM’s technology license, the leap directly from 14nm to 2nm process development remains extremely difficult. In terms of equipment, only ASML in the world can supply EUV lithography machines, and Japan has no experience yet with using EUV. More importantly, at the market level: almost all customers for advanced logic chips are concentrated in the United States and China, while Japan lacks large-scale domestic demand for 2nm chips. Even if Rapidus succeeds in mass production, how can it define its customer base?

Jun Akabane’s criticism of Rapidus focuses on the mismatch between resource allocation and industrial logic. He points out that the vast scale and cutting-edge equipment Rapidus requires may absorb resources Japan could otherwise use to strengthen its own core advantages. In the monopoly landscape formed by TSMC, Samsung, and Intel through enormous R&D capital, the historical window for a single country to achieve a point breakthrough may have already closed.

The internal tension and external risks of the two routes

Kumamoto and Rapidus represent two different views of economic security. The Kumamoto model emphasizes “reliable supply”—that is, through cooperation with China’s Taiwan, ensuring a stable inflow of the chips needed now, while enhancing Japan’s domestic manufacturing capabilities in the process. It accepts the reality of the global division of labor and tries to capture more value from it. Rapidus, by contrast, holds the ideal of “strategic autonomy,” trying to regain a place in the world’s most advanced technology fields so as not to be “choked off” in extreme scenarios such as a Taiwan Strait crisis.

Both approaches have their own blind spots. The main criticism of the Kumamoto model is that large amounts of Japanese public funds flow into a factory with foreign shareholding, while key technological decisions remain in TSMC’s hands. Once overcapacity occurs or the geopolitical situation changes, foreign capital may scale back its investment in Japan, leaving behind only an “island factory” highly dependent on external process technology. Rapidus’s risk is more direct: whether it can achieve 2nm mass production by 2027 is already the focus of industry skepticism. More seriously, even if its technological breakthrough succeeds, Rapidus may fail to reach an economically viable scale of production and be reduced to a “national collectible.”

In addition, the irreversibility of policy dependence is also causing concern. Deep government intervention in specific projects may squeeze the resilience of non-governmental innovation. Historically, MITI’s industrial policies created brilliance, but they also made companies lose their sensitivity to the market during the memory-chip transition. To bet the nation’s fate once again on an extremely narrow technology track—is that repeating the same mistake?

Industrial choices under the shadow of geopolitics## Industrial Choices Under the Geopolitical Shadow

No discussion of Japan's semiconductor strategy can ignore the framework of US-China confrontation. America's technology blockade against China has created strategic room for maneuver for Japan: Japan is both America's most loyal supply-chain partner and has the chance to play a "pivot" role in semiconductor equipment and materials. TSMC's decision to build a plant in Japan is part of the "friend circle" restructuring tacitly approved by the United States. The Japanese government's calculation is that by welcoming TSMC, it can consolidate its position as a junior partner in the US-led system, while Rapidus tries to compete for the most cutting-edge technological position inside the tent of the US-Japan alliance.

But geographic dependence remains a sword hanging overhead. TSMC's Kumamoto plant still has its core intellectual property and advanced process R&D base in Taiwan. Should the situation in the Taiwan Strait change, even a fully operational Kumamoto plant would not be enough to replace Japan's entire chain of dependence on TSMC. Rapidus, for its part, obtained a license from America's IBM, but the political winds in Washington could adjust technology export rules at any moment. Japan appears to have two legs, yet in reality each one is planted firmly in the soil of other nations' politics and capital.

The Long-Term Picture: The True Coordinates of Japan's Semiconductor Revival

Measuring the success or failure of these two projects should perhaps not use "ability to achieve mass production" as the sole criterion. If the Kumamoto plant succeeds in building a closed supply-demand loop within the archipelago—materials, equipment, design, manufacturing, and end-use applications working in full synergy—then even if the process node remains at 28 nanometers, it could provide more than a decade of strategic buffer for Japan's high-end industry and automotive society. As for Rapidus, even if its 2-nanometer plan later proves economically unviable, the semiconductor engineering teams and patent assets cultivated during its R&D could become bargaining chips for Japan's future acquisitions or for checking powerful rivals.

Of course, the highest state of industrial policy lies in finding a dynamic balance between "security" and "efficiency." Japan's current semiconductor strategy—more precisely, this attempt at a "two-track parallel" approach—itself embodies that realist predicament. No single path can satisfy both political symbolism and industrial-economic logic. Kumamoto and Rapidus are precisely the two answers Japan is trying out in the cracks. Neither is a perfect solution, but at least they show that Japan is attempting to discover a new way of being from its "lost technological era."

In the years ahead, when the smoke clears from the 2-nanometer battlefield, the world may see Japan rediscovering its own industrial philosophy at that moment—not pursuing the highest peak, but the deepest valley. In the world of silicon, the best defense may not be possessing the strongest spear, but possessing a shield that can heal itself.

*This article is based on an analysis of an academic lecture by Akabane Jun, released by the Freeman Spogli Institute for International Studies (FSI) at Stanford University. Original report reference: https://fsi.stanford.edu/news/tale-two-approaches-revitalize-japans-semiconductor-industry*