[News] Fujitsu Reportedly Weighs Outsourcing 1.4nm CPUs to Rapidus for 2029 Launch of Fugaku Successor

While Fujitsu announced in early October that it would co-develop AI chips with NVIDIA to power Japan’s data centers by 2030, new details have now surfaced. According to Nikkei XTECH, the company is reportedly considering Rapidus as a potential manufacturing partner for its 1.4nm-class CPUs designed for AI and high-performance computing, with practical deployment targeted for 2029.

Nikkei XTECH reports that Fujitsu, in collaboration with NVIDIA and Japan’s RIKEN research institute, is developing the Fugaku successor, code-named “Fugaku NEXT.” The supercomputer reportedly aims for a 100-fold performance boost over Fugaku and is scheduled to go online around 2030.

To power Fugaku NEXT, Fujitsu is reportedly developing 1.4nm-class CPUs, known as “FUJITSU-MONAKA-X.” The MONAKA-X is the successor to Fujitsu’s FUJITSU-MONAKA data center CPU, which is targeting practical deployment in 2027, the report notes.

Notably, the report points out that while MONAKA is manufactured using TSMC’s 2nm process, MONAKA-X will be further refined to 1.4nm, with key specifications expected to be finalized by March 2026.

As Nikkei XTECH reports, Fujitsu is considering both TSMC and Rapidus as potential manufacturing partners, emphasizing the importance of aligning the project with Japan’s domestic semiconductor push. The company is also reportedly planning an investment in Rapidus, positioning the foundry as a candidate for MONAKA-X and other future products.

After Rapidus reached a major milestone in July, announcing that prototyping for its 2nm gate-all-around (GAA) chips has begun at the IIM-1 foundry in Chitose, Hokkaido, CEO Atsuyoshi Koike, according to Hokkaido Shimbun, identifies U.S. firms IBM and chip-design startup Tenstorrent as the leading prospective customers.

For now, the Japanese foundry is eyeing 2nm mass production in 2027. Meanwhile, Intel has already begun 18A mass production—unveiling its Panther Lake chip built on the node earlier this month—while TSMC plans to kick off 2nm production by year-end.

Read more

• [News] Rapidus 2HP Reportedly Surpasses Intel 18A Logic Density Impacted by BSPDN, Rivals TSMC
• [News] Rapidus Unveils 2nm Progress: Prototypes Begin, EUV Exposure Done 3 Months after Delivery

(Photo credit: Rapidus)