NVIDIA and RIKEN Propel Japan’s AI and Quantum Computing with New Supercomputers

NVIDIA, in collaboration with RIKEN, is set to significantly enhance Japan’s research capabilities by introducing two state-of-the-art supercomputers. These systems are designed to advance scientific AI and quantum computing, according to NVIDIA Newsroom. The deployment of NVIDIA GB200 systems in these supercomputers positions Japan as a leader in the next generation of industrial research and quantum computing.

Enhancing Japan’s Scientific Infrastructure

The new supercomputers will incorporate a total of 2,140 NVIDIA Blackwell GPUs. This integration supports Japan’s sovereign AI strategy, aiming to expand its secure domestic infrastructure and bolster leadership in science, industry, and technology. The first system, designed for AI applications, will utilize 1,600 NVIDIA Blackwell GPUs interconnected by NVIDIA Quantum-X800 InfiniBand networking. This system is expected to propel research in diverse fields such as life sciences, materials science, climate and weather forecasting, and manufacturing.

Advancing Quantum Computing

The second supercomputer focuses on quantum computing, featuring 540 NVIDIA Blackwell GPUs. This system will accelerate research in quantum algorithms, hybrid simulation, and quantum-classical computing methods. Ian Buck, NVIDIA’s vice president of hyperscale and high-performance computing, highlighted the partnership’s potential to establish a foundation for sovereign innovation, addressing complex scientific and industrial challenges.

Strategic Partnership and Future Developments

This initiative follows a previous announcement of a collaboration between Fujitsu and NVIDIA to co-design the FugakuNEXT supercomputer, the successor to the renowned Fugaku supercomputer. The new GPU-accelerated systems will serve as proxy machines for developing hardware, software, and applications for FugakuNEXT, which is expected to deliver 100x greater application performance compared to existing systems. The future integration of quantum computing capabilities is also planned.

Software Innovations for Scientific Advancements

NVIDIA is working with RIKEN to develop floating point emulation software, leveraging NVIDIA Tensor Core GPU performance to enhance traditional scientific computing. RIKEN plans to use NVIDIA CUDA-X, a suite of GPU-accelerated libraries and tools, to advance its high-performance computing applications. These efforts are part of Japan’s broader initiative to lead in AI for science and quantum computing.

The two supercomputers are slated to become operational in spring 2026, with FugakuNEXT anticipated to be operational by 2030. This collaboration underscores Japan’s commitment to innovation and NVIDIA’s role in strengthening the country’s computational infrastructure.

For more information, visit the NVIDIA Newsroom.

Image source: Shutterstock