NVIDIA is teaming up with Japan’s leading national research institute Riken to deploy two next-generation supercomputers designed specifically for AI-driven scientific discovery and quantum computing research.
Announced at the Supercomputing 25 conference in St Louis, these systems will feature a combined total of 2,140 NVIDIA Blackwell GPUs, interconnected by NVIDIA’s Quantum-X800 InfiniBand networking.
The first system is dedicated to AI for science, with 1,600 Blackwell GPUs accelerating research in high-impact fields including life sciences, materials science, climate and weather forecasting, manufacturing automation, and laboratory automation.
The second system, equipped with 540 Blackwell GPUs, will be devoted to quantum computing applications such as quantum algorithm development, hybrid quantum-classical simulations, and quantum error correction research.
“Integrating the NVIDIA GB200 NVL4 accelerated computing platform with our next-generation supercomputers represents a pivotal advancement for Japan’s science infrastructure,” said Satoshi Matsuoka, Director of Riken Center for Computational Science.
“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 challenges,” said Ian Buck, Vice President of Hyperscale and High-performance Computing at NVIDIA.
This initiative aligns closely with Japan’s broader supercomputing ambitions, including the development of FugakuNEXT — a co-designed supercomputer by Fujitsu and NVIDIA, slated for deployment by 2030.
FugakuNEXT aims to deliver 100 times the application performance of current systems by combining Fujitsu’s MONAKA-X CPUs and NVIDIA’s advanced GPUs, linked via NVIDIA’s new NVLink Fusion silicon technology, which enables ultra-high bandwidth between CPUs and GPUs.
NVQLink adoption gains momentum
NVIDIA has also announced that its NVQLink open and universal interconnect architecture will be embraced by more than a dozen supercomputing centres worldwide, including Riken, Singapore’s National Quantum Computing Hub, and Australia’s Pawsey Supercomputing Research Centre to propel regional quantum computing research.
NVQLink enables large-scale quantum-classical workflows via the NVIDIA CUDA-Q platform, delivering ultra-low latency and high throughput (400 Gb/s, sub-4 microseconds). This facilitates the union of quantum computing’s natural simulation strengths with the massive parallelism and programmability of GPUs to position supercomputers as quantum-GPU systems for scientific discovery.
The technology’s first real-world milestone includes integration with Quantinuum’s Helios quantum processor, which recently achieved the world’s first scalable real-time quantum error correction decoder implementation. It leveraged NVQLink’s microsecond latency to deliver protected, hybrid quantum-classical computation to showcase NVQLink as a vital gateway to unified quantum and classical supercomputing systems.
In addition to its broad adoption across Asia, NVQLink now counts supercomputing centres across Europe, the Middle East, and US national laboratories among its early adopters.
Entelechy Asia 