Japan's National Institute of Advanced Industrial Science and Technology (AIST) is set to bolster its AI sovereignty with the launch of the AI Bridging Cloud Infrastructure 3.0 (ABCI 3.0) supercomputer. This state-of-the-art system, featuring thousands of NVIDIA H200 Tensor Core GPUs and Quantum-2 InfiniBand networking by HPE Cray XD, aims to propel Japan's research and development capabilities in artificial intelligence.

ABCI 3.0 builds on Japan's pioneering efforts in AI infrastructure, following the world's first ABCI launch in 2018. AIST Executive Officer Yoshio Tanaka highlighted the upgrade's strategic importance, aiming to advance research in generative AI and strengthen domestic technological independence.

The project, supported by Japan's Ministry of Economy, Trade and Industry (METI), is part of a broader$1 billion initiative to enhance computing resources and cloud AI capabilities. NVIDIA, pivotal in this collaboration, is committed to advancing AI research and education in Japan, as articulated by founder and CEO Jensen Huang during his recent visit.

Huang underscored the transformative potential of AI factories, which are critical for processing vast data sets efficiently. ABCI 3.0, located near Tokyo, promises unmatched computing performance with 6 AI exaflops and 410 double-precision petaflops, setting new standards in AI-specific and general computing capabilities.

The integration of NVIDIA's advanced H200 GPUs with Quantum-2 InfiniBand technology enhances computational efficiency, making ABCI 3.0 a cornerstone for accelerating AI research and development in Japan and beyond.