Fujitsu Launches 2nm Monaka CPU and Sovereign AI Server for Secure Inference

The MONAKA CPU utilizes a 3D-stacked architecture combining 2nm core technology with 5nm cache and I/O sections, achieving a 3.8GHz clock speed and 8800MT/s memory transfer rates. By incorporating dedicated hardware acceleration for matrix operations and SVE2 vector instructions, Fujitsu claims the processor doubles AI inference throughput compared to competitors. This efficiency reportedly halves the number of servers and power consumption required for equivalent workloads, enabling practical on-premises AI inference even in locations lacking specialized cooling infrastructure.

The server lineup includes 1U and 2U rack-mount models equipped with CDI/CXL technology to mitigate memory constraints. Manufactured at Fujitsu’s Kasashima Plant, the systems emphasize supply chain transparency and support confidential computing via Arm CCA. Fujitsu plans to integrate these servers with its Kozuchi AI platform and Takane generative AI tools, creating a vertically integrated sovereign AI stack for financial, telecommunications, and manufacturing industries.