Rapidus officially launched its IIM-1 semiconductor fab in Chitose City, Hokkaido, on April 1, marking a major milestone in Japan's ambition to produce cutting-edge 2nm chips. However, company executives caution that improving yield rates and securing future customers remain key challenges, reports Nikkei and Bloomberg.
At a Tokyo press conference, Rapidus President Atsuyoshi Koike emphasized that while the company is collaborating with 30 to 40 potential customers, the final number of long-term partners will likely be fewer than ten due to Rapidus's limited scale. The first trial wafers for 2nm chip prototypes are set to enter production in April, with initial prototype chips expected between mid and late July.
Currently, around 150 Rapidus engineers are operating over 200 machines at the new facility. The company plans to scale its workforce to between 1,000 and 2,000 employees by 2027 while expanding its equipment base. However, Koike warned against overestimating the feasibility of mass-producing 2nm chips by 2027, describing current efforts as only the "first round" of development.
Rapidus, which acquired 2nm semiconductor technology from IBM, is working on refining mass production techniques, including chiplet technology. Advanced process trials will begin in late 2025, with mass production of chiplets expected between late 2027 and early 2028.
Yield and production hurdles
Producing 2nm chips remains an immense technical challenge. Kazuyoshi Saito, a senior analyst at IwaiCosmo Securities, noted that IBM's 2nm technology is still in the research stage. Moreover, operating extreme ultraviolet (EUV) lithography equipment is time-intensive, making rapid production unlikely.
Even industry leaders like Samsung Electronics and Intel have struggled with yield issues in their previous 2nm process nodes, while TSMC has needed continuous refinements to ensure stable 2nm production. Given the complexity of this technology, experts suggest that a 10% yield rate for Rapidus's early prototypes would already be a significant achievement.
Despite these concerns, Rapidus has set an ambitious initial target of achieving a 50% yield rate for its 2nm chips, with the long-term goal of reaching 80% to 90%.
Saito suggested that if Rapidus encounters difficulties in mass production or customer acquisition, a partnership with TSMC could be a viable strategy to improve its outlook. Meanwhile, Seiki Soramoto, a member of the Japan Innovation Party, raised the possibility of shifting focus to simpler 5nm or 7nm production if 2nm yield rates remain low, ensuring a more practical path for Japan's semiconductor ambitions.
Article edited by Jack Wu