NVIDIA, RIKEN Propel Japan’s AI and Quantum Computing with New Supercomputers
NVIDIA and RIKEN have announced a significant advancement in Japan’s supercomputing capabilities. Two new supercomputers will be unveiled, dedicated to artificial intelligence (AI) and quantum computing, powered by NVIDIA’s latest technologies.
NVIDIA’s Contribution to Japan’s AI and Quantum Initiatives
The first supercomputer will utilize 1,600 NVIDIA Blackwell GPUs on the GB200 NVL4 platform. It aims to facilitate research in various scientific fields, including:
- Life Sciences
- Materials Science
- Climate and Weather Forecasting
- Manufacturing
- Laboratory Automation
This system will employ NVIDIA Quantum-X800 InfiniBand networking to enhance performance across these domains. The second supercomputer, designed specifically for quantum computing, will feature 540 NVIDIA Blackwell GPUs, also utilizing the GB200 NVL4 platform. This system aims to accelerate research on quantum algorithms and hybrid simulation methods.
Strategic Partnerships and Innovations
Ian Buck, NVIDIA’s vice president of hyperscale and high-performance computing, emphasized RIKEN’s role in leading a new era in computing. He stated, “Together, we’re helping Japan build the foundation for sovereign innovation.”
Satoshi Matsuoka, director of the RIKEN Center for Computational Science, highlighted that the new supercomputers represent a crucial development for Japan’s scientific infrastructure. They will create a unified platform for AI and high-performance computing, fostering breakthroughs in both scientific research and industrial applications.
FugakuNEXT: The Next Generation of Supercomputing
These new systems come after a collaboration announcement between Fujitsu and NVIDIA in August, aimed at developing the FugakuNEXT supercomputer. This upcoming system is expected to surpass the renowned Fugaku supercomputer with enhanced performance and capabilities.
FugakuNEXT will integrate FUJITSU-MONAKA-X CPUs optimized through NVIDIA technologies, promising up to 100 times the application performance of current CPU-based systems. Future iterations will also incorporate production-level quantum computing components.
Software Solutions for Accelerated Achievements
In addition to hardware advancements, NVIDIA is collaborating with RIKEN to create floating point emulation software. This will enable extensive use of NVIDIA Tensor Core GPU capabilities in traditional scientific computing.
RIKEN plans to implement NVIDIA CUDA-X
, which comprises numerous optimized GPU-accelerated libraries and tools. This will enhance their HPC applications, thus advancing AI and quantum computing efforts in Japan.
Operational Timeline
The new supercomputers are slated to become operational by spring 2026, while FugakuNEXT is scheduled for completion by 2030. This collaboration marks a pivotal step in establishing Japan as a leader in AI and quantum computing research.
