← Doing Business in Japan’s Semiconductor Industry

From industrial policy to national strategy

Japan has subsidised its semiconductor industry for decades, usually through modest, sector-specific programmes. What changed in the 2020s was framing. Semiconductors moved into the economic-security portfolio — treated alongside energy and critical minerals as a matter of national resilience rather than industrial competitiveness.

That reframing unlocked money at a different order of magnitude, and it changed the institutional architecture. Support no longer flows solely through METI’s industrial programmes; it now runs through the Economic Security Promotion Act framework, NEDO commissioned R&D projects, and direct government equity participation.

Under the Takaichi administration, the trajectory has steepened further. In the FY2026 initial budget, METI’s allocation for advanced semiconductors and AI rose to roughly ¥1.23 trillion — a near-quadrupling of prior support levels — with the ministry deliberately shifting from ad-hoc supplementary budgets toward stable, regular multi-year allocation. In March 2026 the government set a target of approximately ¥40 trillion in annual domestic semiconductor sales by 2040, against roughly ¥8 trillion currently, extending an earlier ¥15 trillion goal for 2030. A longer-horizon investment framework announced in June 2026 earmarks a substantial share of a multi-decade public-private investment programme for AI and chips.

Foreign readers should treat the headline figures with appropriate caution — different sources aggregate equity, subsidy and R&D commitments differently, and multi-year totals are not the same as disbursed cash. The direction, however, is unambiguous.

Rapidus: what it is and what it is not

Rapidus is the most-discussed and most-misunderstood element of Japanese policy.

What it is: a state-backed logic foundry venture, founded in 2022 with participation from major Japanese corporates and heavy government support, targeting 2nm gate-all-around production. Its process technology is being developed in partnership with IBM. Its facility, IIM-1, is in Chitose, Hokkaido.

Where it stands. The pilot line entered operation in April 2025. In July 2025 the company announced confirmed operation of its 2nm GAA transistors, and subsequent measurements have reportedly met planned electrical characteristics. NEDO approved the company’s FY2026 plan and budget in April 2026, covering both front-end 2nm integration and short-TAT manufacturing technology, and chiplet and advanced packaging development. A separate packaging R&D site, Rapidus Chiplet Solutions, has been established in Chitose within Seiko Epson’s facility. Mass production remains targeted for 2027.

What it is not: a TSMC competitor in the conventional sense. Rapidus’s stated differentiation is short turnaround time — rapid prototyping and fast design-to-silicon cycles for specialised, high-value applications — rather than volume cost leadership. Whether that positioning is commercially viable at 2nm economics is the open question, and reasonable analysts disagree.

What foreign companies should take from it. Two things. First, Rapidus is creating a Hokkaido supply-chain cluster essentially from scratch, which is an opening for equipment, materials and services vendors. Second, the scale of state commitment means Rapidus will not be allowed to fail quietly, which affects how you should model competitive scenarios.

TSMC Kumamoto: the counterpart bet

Alongside the domestic champion, Japan pursued the opposite strategy — attracting a foreign leader.

JASM, TSMC’s majority-owned Kumamoto subsidiary, is held with Sony Semiconductor Solutions, DENSO and Toyota as minority investors, with TSMC holding approximately 86.5%. The first fab began volume production at the end of 2024, running mature and specialty nodes for automotive, industrial and image-sensor applications.

The second fab has been the more strategically significant development. Originally announced for 6/7nm-class technology, the plan has shifted toward 3nm — confirmed by TSMC’s CEO during a Tokyo meeting with Prime Minister Takaichi in early 2026, and reflected in prefectural documentation. This changes the character of the site from a mature-node capacity play into a genuine leading-edge presence, with implications for the supplier ecosystem around it.

The Japanese government has provided substantial subsidy support to JASM, and has signalled openness to further support for the upgraded plan.

Two bets, one portfolio

Government officials have consistently framed Rapidus and JASM as serving distinct markets rather than competing. That is largely defensible: Rapidus targets short-TAT specialty logic, JASM targets volume foundry capacity for automotive, industrial and increasingly AI-related applications.

The underlying logic is portfolio construction. Japan cannot know which approach to domestic leading-edge capability will work, so it is funding both — plus a materials and equipment base that benefits regardless of which succeeds.

The economic-security layer

The subsidies do not exist in isolation. They sit inside a framework — anchored by the Economic Security Promotion Act — that designates certain goods as “specified critical materials,” semiconductors among them, and creates both support mechanisms and control mechanisms.

The control side is what foreign companies underestimate. Designation as a critical sector means:

• Prior notification obligations for inbound investment (see the regulatory page in this series)
• Expectations around supply-chain disclosure for companies receiving support
• Conditions attached to subsidies, including domestic production commitments and, in some cases, constraints on technology transfer

Accepting Japanese government support is not free money. It creates obligations that persist, and it can complicate a later exit — particularly a sale to a foreign buyer.

What this means if you are a foreign company

If you supply equipment or materials: the capital deployment is real and is creating procurement demand, particularly around Hokkaido and Kumamoto. Cluster-adjacent presence matters more than in most markets, because Japanese fabs weight supplier responsiveness heavily.

If you are considering a Japanese joint venture: subsidy eligibility is a genuine deal term worth negotiating, and it is not automatically available to foreign-controlled entities. Structure matters.

If you are acquiring a Japanese company that has received public support: examine the subsidy conditions in diligence. Clawback provisions, domestic production commitments and technology-retention conditions can survive a change of control and materially affect your post-deal plans.

If you are modelling Japanese competitive capability: be sceptical of both the enthusiasts and the dismissers. Rapidus has produced working 2nm prototypes on a compressed timeline, which is more than most observers expected. It has not yet demonstrated yield at production scale, which is where such ventures historically fail.

Note: policy figures and project timelines in this sector change frequently. Figures cited reflect publicly reported information as of August 2026 and should be verified against primary sources before use in any transaction.

Related: [Mapping Japan’s Semiconductor Clusters] · The Legal and Regulatory Landscape

Written and edited by Masanori Takano, Chief Executive Officer of TechMedix Institute, Inc.