.jpg)
Japan's scientific research capabilities are set for a significant leap forward through a new collaboration between NVIDIA and the RIKEN research institute. The partnership will see the deployment of two new supercomputers, powered by a total of 2,140 NVIDIA Blackwell GPUs, dedicated to advancing AI for science and quantum computing, solidifying Japan's position in sovereign AI and next-generation research.
RIKEN is building two new supercomputers powered by NVIDIA's GB200 NVL4 platform.
The systems will use a total of 2,140 NVIDIA Blackwell GPUs and Quantum-X800 InfiniBand networking.
One system, with 1,600 GPUs, is dedicated to AI for science in life sciences, climate, and materials.
The second system, with 540 GPUs, will accelerate quantum computing algorithm research.
These systems will also act as proxy machines to codesign the future FugakuNEXT supercomputer.
Both supercomputers are scheduled to become operational in the spring of 2026.
This initiative represents a major expansion of the partnership between NVIDIA and RIKEN, building on the previously announced FugakuNEXT supercomputer project. The two new GPU-accelerated systems are not only powerful research tools in their own right but will also serve as development platforms for the hardware, software, and applications destined for FugakuNEXT. This future system, developed with Fujitsu, is planned to feature FUJITSU-MONAKA-X CPUs connected to NVIDIA architecture and is expected to deliver a 100x application performance increase.
The first of the two new systems, equipped with 1,600 GPUs, will drive RIKEN's "AI for science" initiative. It is designed to accelerate breakthroughs in critical fields such as life sciences, materials science, climate and weather forecasting, and manufacturing automation. The second system, dedicated to quantum computing, will use its 540 GPUs to advance research in quantum algorithms, hybrid simulation, and quantum-classical computing methods.
The collaboration extends beyond hardware to include critical software development. NVIDIA is working with RIKEN to create floating point emulation software that leverages NVIDIA Tensor Core GPU performance for traditional scientific computing. Furthermore, RIKEN plans to utilize NVIDIA CUDA-X, a collection of hundreds of GPU-accelerated libraries and tools, to boost its cutting-edge HPC applications. This unified software and hardware approach aims to create one of the world's leading platforms for AI, quantum, and high-performance computing.
About NVIDIA
NVIDIA is the world leader in AI and accelerated computing.
TechIntelPro empowers B2B technology decision-makers with the latest industry trends, deep insights, leadership perspectives and actionable intelligence to drive both organizational and professional milestones—and thrive on the tech pulse.