TOKYO (Kyodo) -- U.S. chip giant Nvidia will join a project to develop a successor model to Japan's Fugaku supercomputer, which was once the world's fastest, Japan's state-backed Riken research institute said Friday.
Fujitsu, Riken and Nvidia hope FugakuNEXT will offer world-leading capabilities. They plan to start operating it around 2030, with Nvidia taking charge of graphics processing units, or GPUs, to achieve fast computing for artificial intelligence.
The new model is expected to contribute to fields such as drug development, automobile design and measures against climate change.
Using Fujitsu's central processing units, or CPUs, the new model is expected to be five to ten times more powerful than its predecessor and be capable of conducting 1 sextillion calculations per second for development of AI applications.
The former Fugaku, named after an alternative word for Mt. Fuji, earned first place in the TOP500 ranking of supercomputers in terms of computing speed in 2020 but lost the position in 2022. It ranked seventh in June.
FugakuNEXT will be located in Kobe in western Japan, the same site as its predecessor.
