The Pulse
Fujitsu Sets November Sales for MONAKA CPU and Sovereign AI Server
Fujitsu plans to begin sales of its FUJITSU-MONAKA processor and MONAKA Server in November 2026, with wider shipments in Japan and Europe scheduled from April 2027.
AI.info Team ·
Fujitsu is putting domestic control of AI infrastructure against the performance and supply-chain advantages of established processor vendors. The Japanese company announced on September 14 that it will begin global sales of its FUJITSU-MONAKA processor in November 2026, alongside a Japan-designed and Japan-manufactured server aimed at sovereign AI deployments.
Fujitsu says MONAKA delivers twice the AI inference throughput of other CPUs while reducing the servers and power required for equivalent workloads. Those are company claims; the announcement does not identify the comparison processors or publish test conditions. The products therefore mark a commercial launch and a set of Fujitsu performance claims, rather than independently verified benchmark results.
Fujitsu ties MONAKA to Japan’s supply chain
FUJITSU-MONAKA is an Arm-based processor designed and developed in Japan. Fujitsu describes it as a 2nm 3D-stacked CPU, with 5nm process technology used for its cache and I/O components. The chip reaches a maximum operating frequency of 3.8GHz and a memory transfer speed of 8800MT/s, according to the company.
Fujitsu says the processor combines dedicated hardware instructions for matrix operations with Arm SVE2 vector processing and software optimization for AI inference. The company also says MONAKA supports confidential computing, using hardware encryption to protect data and applications held in memory, including in multi-tenant cloud environments.
The processor is intended to be sold as a standalone product to cloud operators, data-center companies and server vendors. Fujitsu plans to provide MONAKA to those customers from the fourth quarter of fiscal 2026, with global availability including the United States and Asia-Pacific.
In a 2024 Fujitsu research article, Dr. Priyanka Sharma, director of software engineering and head of the MONAKA Software R&D Unit at Fujitsu Research of India Private Limited, described the company’s work on the software ecosystem.
“Fujitsu has a strong legacy of contributing towards the Arm ecosystem and we have extensively contributed towards building the Arm software stack.”
MONAKA Server targets CPU-based AI inference
Fujitsu positions the MONAKA Server as a sovereign AI platform that can run inference using the CPU system alone. The company says its space-saving and power-efficient design is intended for deployment in air-cooled data centers, allowing customers to expand AI capacity without specialized cooling infrastructure.
Domestic manufacturing is part of the product’s pitch. Fujitsu says the server will be designed, developed and manufactured in Japan at its Kasashima Plant, allowing the origin of components and manufacturing history to be visible and trackable. The company links that traceability to supply-chain transparency and reliability for sovereign applications.
The announcement also connects the hardware to Fujitsu’s domestic AI software, including the Fujitsu Kozuchi AI platform and the Takane enterprise generative AI system. Fujitsu says it will combine those products with industry-specific models in a vertically integrated sovereign AI offering.
Rack systems for enterprise, academic and HPC users
Starting in November 2026, Fujitsu plans to sell 2U servers with two CPUs and 1U systems containing one or two CPUs in Japan and Europe. The 1U models use CDI and CXL technologies to allocate memory and accelerators beyond a single server’s physical boundaries. Fujitsu says this can mitigate memory shortages and improve resource utilization as models and workloads grow.
The company says its cooling design can support air cooling at an ambient temperature of 40 degrees Celsius and water cooling at a water temperature of 45 degrees Celsius. Fujitsu claims the 1U systems can reduce server-cooling power consumption by up to 80%, though the announcement does not provide a test configuration or baseline for that figure.
A separate 2U multi-node system will target academic and high-performance computing users. Each node contains two CPUs, with four nodes in the system. Fujitsu lists surrogate modeling and fluid analysis among the intended workloads.
November sales precede wider server shipments
Fujitsu says the processor and MONAKA Server will enter sales channels in November 2026. It plans to offer the server to selected customers in finance, telecommunications and manufacturing during the second half of fiscal 2026 to expand use cases. Shipments in Japan and Europe are then scheduled to begin sequentially from April 2027.
The schedule makes the announcement both a product launch and a commercial roadmap. Fujitsu is offering the CPU to external infrastructure companies while using the server to present a domestic stack for organizations that place a premium on control of hardware, data and operations.
The immediate test will be whether Fujitsu can translate its stated throughput and power figures into repeatable results across real AI-inference workloads. The first products are scheduled to reach customers in November 2026, with wider server shipments planned from April 2027.