By Kentaro Okasaka and Sam Nussey
TOKYO, Oct 7 (Reuters) – Japan’s Rapidus, with $15 billion in state backing, is tying up with chip design firms as it seeks to answer a major question hanging over the country’s tech ambitions: can a company with no manufacturing track record begin producing cutting-edge chips in a year?
Only a handful of companies globally can make the 2-nanometre chips Rapidus wants to produce, which requires the industry’s most advanced manufacturing technology. Chips produced with cutting-edge processes offer faster, more energy-efficient performance than earlier-generation chips.
The leading companies are Taiwan’s TSMC, which spent decades refining its processes and dominates leading-edge chipmaking, South Korea’s Samsung Electronics, which began producing 2-nm chips last year, and Intel.
The Rapidus venture, set up in 2022, must secure customers ahead of the planned launch of 2-nm chip production in the second half of the next fiscal year, a project that Prime Minister Sanae Takaichi is counting on to rebuild Japan’s semiconductor industry.
Rapidus said on Monday it would partner with 17 companies including US chip design software firm Synopsys and India’s Infosys to help customers design chips, as the company seeks to secure manufacturing contracts.
“The biggest question for Rapidus is still who will actually fill the fab,” said Nori Chiou, investment director at White Oak Capital, using the industry terminology for a semiconductor fabrication plant. Rapidus’ plant will be based in Hokkaido.
While Monday’s announcement was not yet evidence of commercial traction, he said, it represented incremental progress.
Failure of the project would be a major setback for Japan’s industrial strategy and could reverberate across the country’s chip sector, as it seeks to strengthen supply chain resilience amid rising regional tensions and concerns over a possible Chinese move against Taiwan that could disrupt global supplies.
HEAVYWEIGHT COMPETITION
The unprecedented AI investment boom has fuelled demand for semiconductors and computing infrastructure, a trend Rapidus hopes to capitalise on.
TSMC “will continue to dominate the majority of the market. But there are plenty of customers in the remaining 20% who cannot get capacity from TSMC because they are low on its priority list or their orders are small,” said Akira Minamikawa, an analyst at Omdia.
Rapidus CEO Atsuyoshi Koike expressed confidence there was enough market demand to support an additional leading-edge manufacturer.
“One or two companies are nowhere near enough,” Koike told Reuters.
Some potential customers remain cautious.
“We’re already having TSMC make our chips, so we can’t switch everything to Rapidus. If it didn’t work out, we couldn’t go back to TSMC, so we have to be cautious,” said one executive who declined to be identified because the deliberations are private.
Customers may use Rapidus to diversify supply chains, while the presence of a leading-edge manufacturer could also benefit Japan’s materials and equipment makers, Minamikawa said.
Japan’s global semiconductor market share has fallen from about 50% in the 1980s to less than 10% today. Chipmakers operating in Japan include memory producer Kioxia. TSMC itself is also present through a partnership with Sony.
Other government-backed initiatives include Noetra, which is developing a foundational model for physical AI and robotics.
Japan’s industry ministry in April announced it had approved an additional $4 billion in assistance to Rapidus, bringing total research and development support to $15 billion.
TECHNOLOGICAL HURDLES REMAIN
Preparations for mass production were progressing on schedule and the environment was better than envisioned when Rapidus was founded in 2022, CEO Koike said.
Yet significant technological hurdles remain. Rapidus continues to work with IBM and has started pilot operations, but experts said commercial success remains far from guaranteed.
“Running a chip fab around the clock, maintaining stable, consistent yields and continuing production in a way that makes the business viable is extremely difficult,” said Kazuyoshi Saito, an analyst at IwaiCosmo Securities.
“Even Samsung … has struggled.”
Beyond proving its technology, Rapidus faces the longer-term challenge of building a profitable business. The company has targeted an initial public offering for around the financial year ending March 2032.
“Its success or failure won’t be known for 10 years and no one will be able to take responsibility” if it is not a success, said a government official.
Still, Rapidus’ backers have ambitious plans. A listing and manufacturing operations in the United States are possibilities, said Daishiro Yamagiwa, a lawmaker who heads the ruling Liberal Democratic Party’s parliamentary group on chip strategy.
(Reporting by Kentaro Okasaka and Sam Nussey; Additional reporting by Fanny Potkin; Editing by Miyoung Kim and Sonali Paul)
Comments