Introduction
NVIDIA and RIKEN, Japan's premier research institute, have announced the integration of advanced supercomputing technologies aimed at enhancing Japan's capabilities in artificial intelligence (AI) and quantum computing. This collaboration marks a significant step towards establishing Japan as a leader in scientific research and industrial innovation, leveraging NVIDIA's GB200 systems and quantum networking technologies to support various research fields.
New Supercomputers Overview
The collaboration introduces two new supercomputers that will be pivotal for RIKEN's scientific initiatives. The first supercomputer is designed specifically for AI applications in scientific research, equipped with 1,600 NVIDIA Blackwell GPUs. This system will utilize the GB200 NVL4 platform and will be interconnected using NVIDIA's Quantum-X800 InfiniBand networking technology. Its primary focus will be on advancing research in life sciences, materials science, climate and weather forecasting, manufacturing, and laboratory automation.
The second supercomputer will be dedicated to quantum computing, featuring 540 NVIDIA Blackwell GPUs, also built on the GB200 NVL4 platform. Similar to the AI-focused system, it will use NVIDIA Quantum-X800 InfiniBand networking to facilitate research in quantum algorithms and hybrid simulation techniques, bridging classical and quantum computing methods.
Significance of the Partnership
The partnership between NVIDIA and RIKEN represents a crucial advancement in Japan's scientific infrastructure. Ian Buck, NVIDIA's vice president for high-performance computing, emphasized that this collaboration will lay the groundwork for sovereign innovation, enabling breakthroughs in tackling complex scientific and industrial challenges. Satoshi Matsuoka, director of the RIKEN Center for Computational Science, highlighted that these supercomputers will create a unified platform for AI, quantum, and high-performance computing, significantly enhancing research capabilities across various domains.
Future Developments: FugakuNEXT
This initiative builds on an earlier collaboration announced in August between Fujitsu and NVIDIA, which aims to co-design a new flagship supercomputer, tentatively named FugakuNEXT, intended to succeed the renowned Fugaku supercomputer. The FugakuNEXT system will utilize FUJITSU-MONAKA-X CPUs and NVIDIA technologies, promising a dramatic increase in application performance—up to 100 times greater than existing systems. This supercomputer is also expected to incorporate production-level quantum computing capabilities in the future, further solidifying Japan's position in cutting-edge research.
Software Innovations Supporting Scientific Advancements
In addition to the hardware developments, NVIDIA is collaborating with RIKEN to create floating point emulation software that will optimize the performance of NVIDIA Tensor Core GPUs for scientific computing. This software aims to maximize the potential of GPUs for AI and high-performance computing applications. Furthermore, RIKEN plans to utilize NVIDIA CUDA-X™, a suite of over 400 optimized GPU-accelerated libraries and tools, to enhance its HPC applications, thereby advancing its initiatives in AI and quantum computing.
Conclusion
The introduction of these new supercomputers by NVIDIA and RIKEN signifies a transformative phase in Japan's scientific landscape, reinforcing the country's commitment to innovation in AI and quantum computing. With operational timelines set for spring 2026 for the new supercomputers and 2030 for FugakuNEXT, this collaboration not only positions Japan at the forefront of scientific research but also reflects a broader trend of nations investing in advanced computational technologies to drive future discoveries and industrial advancements.