Japan is embarking on an ambitious journey to enhance its AI sovereignty with the unveiling of its advanced ABCI 3.0 supercomputer. Spearheaded by the National Institute of Advanced Industrial Science and Technology (AIST), this upgraded infrastructure will incorporate thousands of NVIDIA H200 Tensor Core GPUs alongside a robust Hewlett Packard Enterprise Cray XD system, featuring NVIDIA Quantum-2 InfiniBand networking. This impressive configuration aims to drive forward Japan’s artificial intelligence research and development capabilities while fortifying its technological independence.
The ABCI 3.0 supercomputer showcases an evolution of Japan’s Open AI Computing Infrastructure, pivotal for boosting AI research and development in the country. Yoshio Tanaka, AIST Executive Officer, highlights the launch of the original ABCI in 2018, positioning it as a milestone in large-scale AI computing. With the latest upgrade, AIST, in partnership with NVIDIA and HPE, aspires to expand Japan’s generative AI research further.
Hirotaka Ogawa, Head of ABCI Operations, emphasises the urgency in cultivating AI research capabilities within Japan as generative AI emerges globally. This major upgrade of ABCI, through collaboration with NVIDIA and HPE, is set to elevate Japan’s position in global AI competitiveness, playing a foundational role in fostering future innovation within the nation.
Constructed and operated jointly by AIST and its business arm, ABCI Solutions, this advanced infrastructure is grounded in a significant initiative backed by Japan’s Ministry of Economy, Trade and Industry (METI). The initiative focuses on bolstering computing resources through a generous fund of $1 billion, encapsulating ABCI endeavors and cloud AI investments, thereby highlighting Japan’s dedication to enhancing its AI landscape.
NVIDIA’s active partnership with METI further solidifies the collaborative framework for research and educational advancements. Following a visit from Jensen Huang, the NVIDIA CEO, where prospects for AI, robotics, and quantum computing were discussed, Huang expressed firm commitment towards nurturing AI startups and enhancing AI training resources in Japan.
Promising support from NVIDIA situates ABCI as a bastion for developing AI and big data applications. Huang noted that “AI factories” represent the next epoch in data centres, essential for transforming vast data sets into actionable intelligence. These innovations position ABCI as an ultra-efficient infrastructure, expected to be operational by the year’s end, strategically located near Tokyo in Kashiwa.
With outstanding specifications, ABCI is poised to offer 6 AI exaflops of computing capacity along with 410 double-precision petaflops, a significant leap in computational prowess. The deployment of NVIDIA’s H200 GPUs marks a new era, offering unparalleled energy efficiency and ample memory, designed to fully accelerate generative AI workflows while reducing costs of ownership.
The novel integration of advanced NVIDIA Quantum-2 InfiniBand allows for high-speed, low-latency processing of intensive AI tasks, ensuring swift communication within the system—a vital trait for managing massive datasets in real-time. As the epicentre for collaborative AI research among industries, academia, and governments, ABCI stands as a key player in Japan’s strategic vision towards AI advancement, reinforcing its ambition to emerge as a leader in the global AI arena.
Japan enhances its AI sovereignty with the ABCI 3.0 supercomputer, integrating NVIDIA H200 Tensor Core GPUs and Cray XD systems. The initiative aims to boost AI R&D capabilities and strengthen technological independence, backed by a $1 billion investment from METI. With state-of-the-art specs and collaborative efforts from NVIDIA and HPE, ABCI 3.0 is expected to elevate Japan’s position in global AI competitiveness significantly.
The launch of the ABCI 3.0 supercomputer marks a significant milestone in Japan’s quest for AI sovereignty, reinforcing its commitment to advancing research and development capabilities. The collaboration with NVIDIA and HPE bodes well for propelling Japan’s generative AI capabilities, establishing the nation as a global competitive force in the AI landscape. This venture not only combines cutting-edge technology but also underscores Japan’s strategic investments aimed at reducing development costs and time for next-generation AI technologies.
Original Source: blogs.nvidia.com
