FugakuNEXT Zettascale Supercomputer

In August 2025, RIKEN launched the FugakuNEXT International Initiative with Fujitsu and NVIDIA to co-design Japan's next flagship supercomputer, succeeding the Fugaku system that held the Top500 crown from 2020-2022. FugakuNEXT will use Fujitsu's FUJITSU-MONAKA-X Arm-based CPUs combined with NVIDIA GPU accelerators, targeting zettascale (10^21 FLOPS) performance — 1,000x current Fugaku. The development budget exceeds ¥110 billion ($750 million), with basic design completed in FY2025 and operation targeted for 2030.

FugakuNEXT is designed specifically for the convergence of AI and high-performance computing (HPC) — supporting both traditional scientific simulation (climate, materials, molecular dynamics) and large-scale AI training and inference. Two interim NVIDIA-based supercomputers will be operational at RIKEN in spring 2026 as bridge systems.

The strategic significance is maintaining Japan's position at the frontier of scientific computing. Fugaku demonstrated that Japanese-designed processors (Fujitsu's A64FX) could compete with US designs, and FugakuNEXT continues this with the MONAKA-X architecture. The NVIDIA partnership acknowledges the importance of GPU computing for AI while preserving Japanese CPU design capability. Japan is also investing in the US-Japan Genesis AI supercomputing initiative, ensuring interoperability with allied computing infrastructure.