Details
- On August 22, 2025, RIKEN, Fujitsu, and NVIDIA announced the co-design of FugakuNEXT, the successor to Japan's landmark Fugaku supercomputer.
- The collaboration features RIKEN as lead research institution, Fujitsu as technology partner and prime contractor, and NVIDIA supplying AI acceleration and system architecture, with support from Japan's Ministry of Education, Culture, Sports, Science and Technology.
- FugakuNEXT will utilize Fujitsu's MONAKA-X CPUs, employing advanced 2nm process technology, connected with NVIDIA architecture through NVLink Fusion for cohesive hybrid AI and HPC operations.
- This initiative builds on Fugaku's 2020 launch (415.5 Linpack petaflops), with FugakuNEXT promising up to 1,000 times the performance of its predecessor on a budget exceeding $750 million.
- Key use cases include rapid scientific discovery, advanced manufacturing, earth science modeling, disaster prediction, and pharmaceutical research, with the initial design phase extending to February 2026.
Impact
FugakuNEXT signals Japan's commitment to AI leadership and its ambition to challenge the dominance of US and Chinese exascale supercomputers. Tapping into Fujitsu's cutting-edge semiconductor advances and NVIDIA's AI prowess, the project underpins Japan's sovereign computing aspirations and reflects a wider shift toward hybrid, quantum-capable scientific infrastructure on the world stage.