$1.7 billion. That is what Japan's government and corporate establishment have just committed to Rapidus Corporation, a chip startup founded barely four years ago with an audacious goal: leapfrog the global semiconductor industry and go straight to 2-nanometre mass production by 2027. The funding round, announced on 27 February 2026, is the most significant single act of industrial policy Japan has executed in the technology sector in a generation.
Let's break this down. The round totals 267.6 billion yen from the Japanese government and private sector companies, with 100 billion yen coming from the Information-Technology Promotion Agency (IPA), an independent agency under the jurisdiction of Japan's Ministry of Economy, Trade and Industry (METI). Private sector funding of 167.6 billion yen came from 32 companies, including Canon, the Development Bank of Japan, Fujitsu, NTT, SoftBank, and Sony Group. Private money, notably, came in faster than Tokyo had expected: the round exceeded the 130 billion yen private-sector target Rapidus had originally set for the fiscal year.
The governance structure attached to this money deserves close scrutiny. Under the arrangement, the government will initially hold roughly 10% of Rapidus's voting shares and the majority of its stake in non-voting stock, but retains the right to convert those shares to seize majority control if the company faces financial distress. The government also acquired "golden shares" granting it veto power over major corporate decisions, according to Tomoshige Nambu, an official at Japan's Ministry of Economy, Trade and Industry. In plain terms, Tokyo is not a passive investor. If Rapidus falters, the state can take the wheel.
The technical ambition behind this investment is real, not merely rhetorical. Rapidus opened its pilot line in April 2025 and demonstrated working 2nm gate-all-around (GAA) transistors, an architecture that surrounds the transistor channel on all four sides rather than three, improving current control and reducing leakage compared to older FinFET designs. The company started distributing process design kits to initial customers in early February 2026, the step that allows fabless chip designers to evaluate whether to commit production runs to Rapidus's line. The company uses High-NA EUV lithography equipment from ASML at the Hokkaido facility, and IBM is understood to have roughly 10 engineers stationed on-site as part of an ongoing technology collaboration.
On the demand side, CEO Atsuyoshi Koike is sounding confident. He told a press conference on 26 February that Rapidus is in active discussions with more than 60 companies looking to design chips for AI, robotics, and edge computing. "Since the start of the year, customers' demands for cutting-edge chips have surged," Koike said, adding that interest in 2nm specifically had intensified and that the company intends to push on to 1.4nm and 1nm nodes after that. With Gartner projecting global AI chip spending to triple from $45 billion in 2025 to around $135 billion by 2028, the market opportunity is not in question.
Here's the thing: 60 companies in discussions is not 60 companies with signed contracts. Critics question whether Rapidus can achieve stable yields at 2nm and convert prospective client interest into binding multi-year purchase orders. Distributing a process design kit and winning committed production volumes are very different commercial milestones. This is an ambitious plan for a four-year-old firm; TSMC and Samsung both started volume production using their own 2nm processes only in the fourth quarter of 2025. Catching, let alone competing with, those entrenched giants at the bleeding edge of fabrication is a formidable challenge.
The scepticism is not unfounded. Japan has a track record of government-backed semiconductor ventures failing; Elpida Memory went bankrupt in 2012 despite state support, and Japan Display has struggled for years. Japan once controlled more than half the global semiconductor market but can no longer mass-produce anything below 40 nanometres. The scale of the ambition, skipping straight to 2nm, is precisely what draws both admiration and scepticism in equal measure. Rapidus estimates it needs 4 trillion yen in total to achieve 2nm mass production, and has raised approximately 1.7 trillion yen so far, meaning this funding round, sizeable as it is, still leaves a very large gap to close.
Proponents of the Rapidus model argue, with some force, that the alternative, continuing to rely on TSMC and Samsung for the most advanced chips, carries its own strategic and economic risks. Supply chain concentration in Taiwan and South Korea has become a live geopolitical concern, not a hypothetical one. Chips at the 2nm scale are critical for high-performance computing, AI systems, advanced data centres, and next-generation communication networks, making domestic capacity in allied nations a genuine national security consideration, not simply an industrial vanity project.
From a fiscal responsibility standpoint, the government's structural protections, including the golden share and conversion rights, at least ensure that public money is not handed over without accountability strings attached. Whether those strings will prove sufficient to discipline a complex, capital-intensive deep-tech venture is another matter. What is clear is that Japan has made a considered bet, not a reckless one, and the involvement of Rapidus partners like IBM and ASML lends it technical credibility that earlier Japanese chip rescue efforts lacked.
Reasonable observers can disagree sharply on whether $1.7 billion of public and private capital chasing an extraordinarily difficult fabrication target is wise industrial policy or expensive optimism. The honest answer is probably both. The global chip race has no neutral sidelines; standing aside carries costs just as surely as stepping in. Tokyo has chosen to step in, with eyes open to the risk, and the world's semiconductor industry will be watching Chitose, Hokkaido very closely over the next 18 months to see whether the bet pays off.