Fujitsu launches made-in-Japan FUJITSU-MONAKA CPU for sovereign AI servers

Fujitsu will begin global sales of its next-generation CPU, FUJITSU-MONAKA, in November 2026, the company said on September 14. The processor, designed and developed in Japan, will be sold as a standalone product to cloud and data center operators and to server vendors, and inside a new Fujitsu MONAKA Server that the company positions as sovereign AI infrastructure.

What goes on sale

Fujitsu MONAKA Server is designed, developed and manufactured in Japan at the company's Kasashima Plant, which Fujitsu says improves traceability and supply-chain transparency. From November 2026 the company will sell 1U rack-mount servers with one or two CPUs and 2U servers with two CPUs, aimed at AI agents and AI inference, in Japan and Europe. Buyers include data center operators, enterprises, the academic and HPC sector, and the defense sector, which Fujitsu links to national security.

The chip

FUJITSU-MONAKA uses a 3D-stacked design that pairs a 2nm process for the cores with a 5nm process for the cache and I/O sections, a combination Fujitsu says improves cost efficiency and, together with its ultra-low-voltage operation technology, delivers high power efficiency. The CPU runs at a maximum operating frequency of 3.8GHz with memory transfer speeds of 8800MT/s, and provides hardware acceleration for the matrix operations behind AI inference through dedicated instructions combined with SVE2 vector operations and software optimization. Fujitsu rates it at twice the AI inference throughput of other CPUs, which it says halves the number of servers and the power drawn for an equivalent processing load.

Built for air-cooled data centers

The 1U server adopts CDI/CXL technology, which allows memory and accelerators to be allocated beyond the machine's physical boundaries and mitigates the memory shortages common in inference workloads. Drawing on Fujitsu's cooling technology, the 1U system can run at ambient air temperatures up to 40 degrees Celsius, or water temperatures up to 45 degrees Celsius with water cooling, without specialized cooling infrastructure — a fit for operators that cannot add liquid cooling.

Why it matters

Fujitsu ties the launch to the shortage of computing resources and the rapid growth in power demand that generative AI has created, and to rising interest in platforms with reliable, transparent supply chains. Beyond the hardware, it plans a vertically integrated sovereign AI stack that combines its Kozuchi AI platform and its Takane enterprise generative AI with industry-specific models. The CPU goes to cloud and data center operators and server vendors from the fourth quarter of Fujitsu's fiscal 2026, with global availability including the US and APAC; early MONAKA Server customers in finance, telecom and manufacturing arrive in the second half of the fiscal year, with broader shipments in Japan and Europe from April 2027.