Nvidia Provides GPUs and Software for ABCI-Q: The World’s Largest Quantum Research Supercomputer, Powered by CUDA-Q.
In Japan, ABCI-Q, the world’s largest supercomputer for quantum research, has been put into operation. The system runs on Nvidia GPUs and is closely integrated with the CUDA-Q software platform. Nvidia announces this at Computex.
The ABCI-Q supercomputer was developed by the Japanese research institute AIST and is deployed in the G-QuAT research center. The system contains 2,020 Nvidia H100 GPUs, connected via Nvidia’s Quantum-2 InfiniBand network. The infrastructure supports hybrid workloads combining classical GPU computing power with quantum processors from Fujitsu, QuEra, and OptQC, among others.
read also
From CUDA to CUDA-Q: Nvidia Wants Quantum Computers After All, as Long as They Contain GPUs
The project uses CUDA-Q, an open-source platform from Nvidia for hybrid quantum-GPU computing. It allows developers to build applications that run simultaneously on classical and quantum hardware. ABCI-Q is intended to provide a test environment for the development of practical quantum algorithms and systems.
CUDA-Q Strategy
The introduction of ABCI-Q fits within a broader strategy of Nvidia. Until early 2025, CEO Jensen Huang was still skeptical about the practical usability of quantum computing. Since then, however, the company has made a clear change of course. Nvidia wants to help shape quantum computing, as long as GPU accelerators continue to play a central role.
For this purpose, the Nvidia Accelerated Quantum Research Center (NVAQC) was previously established near Boston. Together with partners such as QuEra and Quantum Machines, the company is working on architectures and algorithms where GPUs and quantum chips collaborate. The goal is not to view quantum computers as a replacement for GPU systems, but as a complement. CUDA-Q is intended to form a central link in this.
With this approach, Nvidia is trying to build a software ecosystem around quantum computing, similar to what it did earlier with CUDA for GPUs. As long as CUDA and now also CUDA-Q remain a standard for developers, Nvidia maintains its strong position in the data center and AI ecosystem.
With ABCI-Q, this strategy now gets a concrete showcase, where researchers can build the bridge between classical and quantum computing power.