Rapidus Goes Global: Japan’s 2nm Chip Dream Reaches Europe

A small chip has become a national story

Semiconductors do their work invisibly. They sit inside phones, cars, factory robots, bank servers, medical scanners, satellites, power grids and the huge machines that train artificial intelligence. That is why losing semiconductor capacity is not merely a supply-chain inconvenience. It can mean losing leverage over the future.

That is the deeper reason Rapidus matters. In June 2026, the Japanese chipmaker announced memoranda of understanding with the UK Semiconductor Centre and Italy’s Fondazione Chips-IT. On paper, these are cooperation agreements. In context, they are much more. Japan, once a global semiconductor power, has spent decades outside the front rank of advanced logic manufacturing. Rapidus is Japan’s attempt to re-enter that conversation through IBM technology, a new Hokkaido fab, trillions of yen in government support and now European research links.

This is not a clean comeback story. It is a hard climb. TSMC, Samsung and Intel are already far up the mountain. ASML controls the most crucial lithography machines. U.S. design firms dominate much of the chip-design ecosystem. AI customers move fast. Foundry customers want proof, not patriotism. Rapidus must turn a national ambition into working silicon, yields, process design kits, customer trust and delivery discipline.

Why the UK and Italy matter

Rapidus’ memorandum with the UK Semiconductor Centre creates a framework for future semiconductor manufacturing cooperation. The United Kingdom is not a giant volume foundry base. Its strengths lie elsewhere: chip design, intellectual property, compound semiconductors, research institutions, universities and startups. In other words, Britain matters less as a place of mass production and more as a place of ideas, tools and specialized engineering.

The Italian agreement with Fondazione Chips-IT is also meaningful. Italy has been working to strengthen semiconductor design, research, training and industrial coordination as part of Europe’s broader chip strategy. For Rapidus, Europe is not only a market. It is a pool of engineers, research laboratories, design communities and potential customers in AI, automotive, industrial systems, communications, space and edge computing.

In chips, building a factory is not enough. Designers must want to use the process. Software tools must support it. IP blocks must be available. Prototypes must come back quickly. Yields must be understandable. The path to volume must be credible. The UK and Italian links are therefore part of Rapidus’ effort to turn a Hokkaido fab into a practical option for global customers.

Chitose: the snow-country fab chasing the world’s frontier

Rapidus’ main stage is not a Tokyo skyscraper. It is Chitose, Hokkaido. Its IIM-1 facility, short for Innovative Integration for Manufacturing, is intended to manufacture advanced logic semiconductors of 2 nanometers or below. Rapidus says construction began in September 2023, the pilot line was scheduled for April 2025, and mass production is expected in 2027.

The Hokkaido location matters. It brings land, water, airport access, regional development and a chance to build a new industrial cluster away from Japan’s old Pacific industrial belt. If Rapidus works, Chitose will not simply be a factory address. It will become one of the most important places on Japan’s industrial map.

But romance does not run a cleanroom. Advanced fabs need stable power, ultrapure water, constant maintenance, equipment logistics, supplier networks, engineers, housing and a city that can support long-term talent. A leading-edge fab is a giant living system. Rapidus must build not only a production line, but an ecosystem around it.

From IBM’s 2nm research to Japanese manufacturing

The technological root of Rapidus’ plan is its IBM partnership. In December 2022, IBM and Rapidus announced a strategic partnership to build advanced semiconductor technology and an ecosystem in Japan. Rapidus engineers would work with IBM researchers at the Albany NanoTech Complex in New York, one of the world’s most advanced semiconductor research centers.

This detail matters. Rapidus is not conjuring 2nm technology from nothing. It is trying to translate world-class research into manufacturable process technology in Japan. A transistor that works in a research environment is not the same as a profitable, repeatable, high-yield manufacturing process. Between the two sit equipment recipes, materials, metrology, defect analysis, data, human judgment and thousands of process decisions.

Rapidus has reported progress at IIM-1, including work on 2nm gate-all-around transistor operation on 300mm wafers. GAA, or gate-all-around, is a key transistor architecture for the next stage of scaling because it improves control over current at extremely small dimensions. But the industry knows the difference between a milestone and a business. Prototypes open the door. Customers, yields and delivery schedules decide whether the door stays open.

¥2.354 trillion: too much, or still not enough?

Reuters reported in April that Japan’s industry ministry approved an additional ¥631.5 billion to support Rapidus, bringing total government R&D assistance to ¥2.354 trillion. That is a huge number. It is fair for taxpayers to ask whether Japan is placing too much money behind one company.

Yet advanced semiconductors are no ordinary market. Taiwan, South Korea, the United States, Europe and China are all using public money, tax incentives or industrial policy to shape chip production. A leading-edge fab costs trillions of yen. The race involves national security, export controls, AI infrastructure and supply-chain resilience. Pretending this is a simple free-market contest would be naive.

The real question is not only whether the support is large. It is what Japan gets from it. If Rapidus creates manufacturing know-how, process engineers, design infrastructure, supplier experience, regional jobs, European relationships and a domestic option for advanced chips, then even the accounting becomes more complex. The subsidy is not just for wafers. It is for capability.

The 1980s memory and the long fall

Japan’s semiconductor history carries glory and warning. In the 1980s, Japanese companies dominated major parts of the memory-chip market, especially DRAM. NEC, Toshiba, Hitachi, Fujitsu and Mitsubishi became symbols of quality manufacturing, process discipline and integrated electronics power. Japan did not merely participate in semiconductors. It frightened competitors.

Then the model changed. DRAM became brutally price competitive. Korean firms rose. Taiwan built the foundry model around TSMC. U.S. fabless companies focused on design. The global industry split into specialized layers: design, manufacturing, tools, materials, packaging, software and IP. Japan’s vertically integrated electronics giants, so strong in one era, struggled in the next.

The 1986 U.S.-Japan Semiconductor Agreement remains a central memory in this history. To the United States, Japan’s rise threatened a strategic industry. To Japan, political pressure collided with industrial momentum. Then came the burst bubble, underinvestment, slow corporate restructuring and the rise of new foreign champions. Japan stayed powerful in materials and equipment, but it lost the leading-edge logic manufacturing spotlight.

Rapidus is powerful because it speaks to that memory. It is a new company, but it carries an old question: can Japan return to the front line?

Rapidus cannot simply copy TSMC

Rapidus will not beat TSMC by becoming a smaller TSMC. The Taiwanese giant has deep customer relationships, enormous scale, process maturity, manufacturing discipline and an ecosystem built over decades. Samsung and Intel also bring huge resources. Rapidus is late, small and ambitious. That combination can be dangerous unless the strategy is genuinely different.

That is why Rapidus’ talk of fast turnaround, co-creation with customers, single-wafer processing, advanced packaging and short-cycle development matters. The company’s best path may not be the world’s largest volumes. It may be high-value customers who need fast iteration, specialized performance, security, edge AI, robotics, automotive reliability or small-to-mid volume advanced chips.

The AI era may create exactly that kind of demand. Not every customer needs the same chip. Data centers, robots, cars, medical systems, satellites, telecommunications equipment and factory machines each have different power, heat, reliability, volume, security and timing requirements. Rapidus’ opportunity may be to serve the most demanding, most complicated customers — the kind Japan has historically been good at serving.

Semiconductor diplomacy

The UK and Italy agreements should also be read through geopolitics. The semiconductor world is being reorganized by U.S.-China rivalry, Taiwan Strait risk, export controls, AI security and subsidy competition. Chips are no longer merely products. They are instruments of national strategy.

Japan is tying itself to IBM and Albany in the United States, to European research and design communities, and to its own domestic strengths in materials, equipment, automobiles, telecommunications and precision manufacturing. Rapidus’ European outreach is therefore a form of semiconductor diplomacy. It tells partners that Japan wants to be part of the trusted advanced-chip network, not merely a buyer at the end of the chain.

Europe brings research, talent and customers. Japan brings materials, equipment, manufacturing discipline and a strategic location in Asia. The United States brings IBM’s technology base and a larger security framework. Rapidus sits at the intersection of those interests.

The risk is real

None of this guarantees success. The 2027 target is aggressive. Yield improvement is difficult. Tool operations are difficult. Customer acquisition is difficult. Design enablement is difficult. Recruiting enough engineers is difficult. Every part of the project is difficult, and the global leaders will not wait politely.

Even if Rapidus begins mass production, customers will ask hard questions. What is the yield? What is the price? How fast is the turnaround? Are the tools ready? Is the process design kit mature? Are enough IP blocks available? Can the company deliver reliably over years? Will it still be financially supported if the ramp is slow?

National ambition can open doors. It cannot ship wafers by itself.

Why Japan may still have to try

So should Japan walk away from the risk? Probably not. Semiconductors have become too important to outsource completely. AI, robotics, cars, defense, telecom, space, medicine, finance and energy all depend on advanced chips. A country that wants industrial independence cannot treat leading-edge semiconductors as someone else’s permanent problem.

Rapidus may succeed. It may stumble. It may become a specialized foundry rather than a giant. It may build capabilities that matter even if the original business plan changes. But the attempt itself reflects a sober truth: Japan cannot simply remember the 1980s. It has to build something new.

The 2nm line is invisible to the human eye. But on that line sit Japan’s next industrial policy, its AI competitiveness and its desire to remain a country that can still make the future.

From Hokkaido to Albany, from Chitose to Britain and Italy, Rapidus is trying to turn a national dream into working silicon. The name means speed. The question now is whether Japan can move fast enough.