Nvidia has partnered with Japanese research institute Riken for deployment of the chipmaker’s GB200 NVL4 systems in two new supercomputers.

One system will be an AI supercomputer dedicated to scientific research, while the other will be a quantum computer.

In total, 2,140 Blackwell GPUs will be deployed across the two machines – 1,600 in the AI supercomputer and 540 in the quantum computer. Both will feature Nvidia’s GB200 NVL72 platform and be interconnected by Quantum-X800 InfiniBand networking.

“Riken has long been one of the world’s great scientific institutions, and today it stands at the forefront of a new era in computing,” said Ian Buck, VP of hyperscale and high-performance computing (HPC) at Nvidia.

“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.”

The announcement builds on the collaboration first announced by Riken and Nvidia in August 2025, under which the chip giant will work alongside Fujitsu to co-design the successor to the Fugaku supercomputer, currently known as FugakuNEXT.

That system will feature Fujitsu-MONAKA-X CPUs, Arm-based CPUs developed by Fujitsu and set for release in 2027. Built on 2nm technology, Fujitsu-MONAKA chips integrate with GPUs and other accelerators, and use a microarchitecture optimized for advanced 3D packaging and ultra-low voltage circuit operation.

In FugakuNEXT, Riken will connect the Fujitsu-MONAKA-X and Nvidia GPUs with NVLink Fusion, an interconnect technology capable of bringing together custom silicon and CPUs with Nvidia GPUs via chiplet-based UCIe interconnect.

The two new supercomputers will be operational in spring 2026, while FugakuNEXT is aimed for operation by 2030.

The expanded collaboration with Riken was just one of a number of announcements Nvidia made during the Supercomputing 2025 event, which took place in St. Louis, Missouri, this week.

In other quantum-related news, Nvidia said that more than a dozen supercomputer centers across the globe had now adopted its NVQLink interconnect to link quantum processors with HPC systems to support hybrid quantum and classical computing workloads.

The chipmaker also revealed this week that it had secured Arm as a partner for its NVLink Fusion offering.

In a statement, Arm said that by integrating its Neoverse technology with NVLink Fusion, it expects to further strengthen its architecture and “deliver intelligence per watt, shaping the architecture of the AI era.”

On November 20, Nvidia posted record earnings for its third quarter, exceeding analyst expectations and reporting a revenue increase of 62 percent year-on-year to $57 billion.

Despite growing concerns of an AI bubble, data center-specific sales rose 66 percent to $51bn, with Nvidia CFO Colette Kress stating that the company has “visibility to $0.5 trillion in Blackwell and Rubin revenue from the start of this year through the end of calendar year 2026."