Rapidus, the Japanese government-backed foundry trying to bring leading-edge chip manufacturing back to Japan, has yet to produce a working SRAM cell on its 2nm process, according to Bloomberg. The report, published on Tuesday (10/11) by Takashi Mochizuki, says the milestone was due by the summer of 2026 and that the program is several months behind schedule. A day later, on Wednesday (11/11), the Jiji Press news agency reported that the company is considering delaying mass production by about a year, from the second half of fiscal 2027 to fiscal 2028. None of this has been confirmed by Rapidus — and the company publicly maintains its original target.
Why SRAM is the test that decides everything
SRAM (static random-access memory) is the standard test vehicle for a new logic process: the six-transistor cell is the densest, most regular structure on a chip, sits in the cache of every processor, and is brutally sensitive to manufacturing variation. A process that cannot show a working SRAM cell has not yet proven it can manufacture advanced logic at scale — no matter how well individual transistors have performed. That was exactly Rapidus's position in July 2025, when the pilot line at its IIM-1 fab in Chitose, Hokkaido, produced the first gate-all-around (GAA) transistors with sound electrical characteristics. The next step — turning transistors into working memory arrays — is what is reportedly stuck.
What the reports say
According to Bloomberg, which cited people familiar with the matter, beyond the missed SRAM milestone Rapidus is also working to tighten dust and particle control in the IIM-1 cleanroom: at 2nm, a speck of contamination can ruin features far smaller than the human eye can see, and dense memory arrays are the first to suffer. Jiji Press, meanwhile, pointed to difficulty connecting transistors and integrating them with other components — which would hurt chip performance — and slow progress on wafer yield, the share of good chips per wafer.
A follow-up published this Saturday (21/11) by East Asia Brief claims that 128-megabit SRAM test chips missed their yield targets, and that milestones from Japan's Ministry of Economy, Trade and Industry (METI), updated on November 5, require a functional 128-Mb SRAM array with a cell of 0.0187 square micrometers or less. That information comes from a single source and has not been independently confirmed — but it points in the same direction as the November 10 and 11 reports.
What Rapidus and the Japanese government say
A Rapidus spokesperson told Bloomberg the company remains fully committed to starting mass production in the second half of fiscal 2027 — a window that closes on March 31, 2028 — and declined to discuss technical details, citing confidentiality. In the Jiji coverage, the company said it will continue to do everything possible to meet the target. In mid-November, METI Minister Ryosei Akazawa characterized the project's progress as "on schedule". A government-commissioned expert panel had reported in September that development was progressing, and chairman Tetsuro Higashi said the previous month that the mass production plan was on track.
The schedule math, however, is tight: from a working SRAM cell to acceptable SRAM yield, then process qualification and risk production — normally about a year before high-volume output — starting that sequence in late 2026 or early 2027 leaves very little slack before March 2028. Bloomberg also noted that a serious setback could push the company's IPO back to 2032.
The money at stake — and the race with TSMC and Samsung
The stakes are high: the Japanese government has already committed more than ¥2.35 trillion (about US$15 billion) to Rapidus, and METI announced on October 19 an additional ¥400 billion in equity for fiscal years 2027 to 2029. Total support is projected to reach around ¥3 trillion by the end of fiscal 2027. Founded in August 2022 by eight major Japanese companies — including Toyota, Sony, NTT and SoftBank — with 2nm technology derived from IBM, Rapidus is Japan's attempt to reduce its dependence on Taiwan for advanced chips.
Meanwhile, the competition is not waiting: TSMC's 2nm process already accounted for 8% of the company's wafer revenue in the third quarter of 2026, with capacity set to more than double in 2027, and Samsung is also targeting commercial 2nm GAA production. According to Reuters, Rapidus has formed design partnerships with 17 companies, but prospective customers are reluctant to leave TSMC. A one-year slip would keep Japan dependent on Taiwan for the most advanced nodes until at least 2029 — which is why the next decisive signal to watch is a formal Rapidus announcement about working SRAM, not another behind-the-scenes report.
Read also: Rapidus CORE brings together 17 companies to design 2nm chips in Japan.
Image: Rapidus's IIM-1 fab under construction in Chitose, Hokkaido, photographed from New Chitose Airport in March 2024 (photo: どういたしまして, via Wikimedia Commons, CC BY-SA 4.0).
Sources: Bloomberg; Electronics Weekly; CASRAI; Hardware Busters; East Asia Brief; Rapidus — IIM.
Transparency note: this article was produced with the assistance of artificial intelligence, cross-checking public posts on X and sources on the internet, and reviewed before publication. Check the original sources to confirm the information.
Por Muse