Fujitsu said on Sept. 14 that it will begin selling its in-house next-generation CPU, FUJITSU-MONAKA, and introduced the Fujitsu MONAKA Server built around the chip. The company said the processor is being mass-produced on TSMC’s 2 nm process and draws on technology developed through Fugaku, the supercomputer jointly built by Fujitsu and Japan’s RIKEN. Fujitsu described MONAKA as the first CPU to combine a 2 nm process with a 3D stacked design, using 2 nm for compute cores and 5 nm for cache and I/O under a heterogeneous integration approach. The Arm-based chip supports SVE2, runs at up to 3.8 GHz, and offers a memory transfer rate of 8800 MT/s. In AI inference workloads, Fujitsu said dedicated matrix instructions allow MONAKA to deliver twice the inference throughput of competing CPUs in the same category. The company also highlighted Arm CCA confidential computing support for protecting data and applications in memory across cloud and multi-tenant environments. Fujitsu plans to ship the CPU globally from November to cloud providers, data center operators and server makers. It said 24 Japanese financial and telecom companies have already deployed test units, while more than 40 companies are in talks. Fujitsu also said it has signed a supply agreement with SuperMicro and plans to enter the U.S. and Asian markets in 2027.