NVIDIA Powers Japan's AI and Quantum Supercomputing Push

NVIDIA has announced a significant expansion of its partnership with RIKEN, Japan's premier national research institute, to deploy two new supercomputers powered by the company's latest Blackwell GPU architecture. The systems, scheduled to become operational in spring 2026, represent a major investment in Japan's computational infrastructure for scientific research.

The first system will focus on AI-driven scientific research, deploying 1,600 NVIDIA Blackwell GPUs using the GB200 NVL4 platform. This configuration will be interconnected with NVIDIA's Quantum-X800 InfiniBand networking technology, creating what the partners describe as a unified platform for advancing research in life sciences, materials science, climate modeling, and manufacturing automation.

A second system dedicated to quantum computing research will feature 540 Blackwell GPUs using the same GB200 NVL4 platform. This installation aims to accelerate work on quantum algorithms, hybrid simulations, and quantum-classical computing s, positioning Japan at the forefront of quantum research infrastructure.

Ian Buck, NVIDIA's vice president of hyperscale and high-performance computing, emphasized the strategic importance of the collaboration. "RIKEN has long been one of the world's great scientific institutions," Buck stated. "Together, we're helping Japan build the foundation for sovereign innovation that will drive breakthroughs to solve the world's most complex scientific and industrial s."

The RIKEN deployment follows NVIDIA's August announcement of a collaboration with Fujitsu to codesign FugakuNEXT, the successor to Japan's renowned Fugaku supercomputer. The two new RIKEN systems will serve as development platforms for FugakuNEXT, which is planned to feature Fujitsu's MONAKA-X CPUs paired with NVIDIA technologies through NVLink Fusion connections.

Satoshi Matsuoka, director of the RIKEN Center for Computational Science, described the integration as "a pivotal advancement for Japan's science infrastructure." The partnership aims to create what Matsuoka called "one of the world's leading unified platforms for AI, quantum and high-performance computing," enabling accelerated discoveries across basic sciences and industrial applications.

Beyond hardware, NVIDIA is working with RIKEN to develop specialized software, including floating point emulation technology that leverages Tensor Core GPU performance. The research institute also plans to implement NVIDIA's CUDA-X platform, which provides hundreds of optimized GPU-accelerated libraries and tools for high-performance computing applications.

The FugakuNEXT system, representing the next phase of Japan's supercomputing evolution, is targeted for operation by 2030 and is expected to deliver significantly enhanced performance for scientific applications while integrating production-level quantum computing capabilities in future phases.