Nvidia has announced an acceleration of its quantum computing initiatives at supercomputing centers worldwide through its open-source platform, Nvidia CUDA-Q.
Supercomputing sites in Germany, Japan, and Poland will leverage this platform to enhance the quantum processing units (QPUs) in their Nvidia-accelerated high-performance computing systems. Additionally, Nvidia revealed that nine new supercomputers globally are utilizing the Nvidia Grace Hopper Superchips to expedite scientific research and discovery, collectively delivering 200 exaflops—200 quintillion calculations per second—of energy-efficient AI processing power.
QPUs serve as the core of quantum computers, utilizing particle behavior, such as that of electrons or photons, to perform calculations much faster than conventional processors.
At Germany's Jülich Supercomputing Centre (JSC), a QPU developed by IQM Quantum Computers will complement the Jupiter supercomputer, which is powered by the Nvidia GH200 Grace Hopper Superchip.
Japan's ABCI-Q supercomputer, located at the National Institute of Advanced Industrial Science and Technology (AIST), aims to advance the nation’s quantum computing efforts. It, too, is powered by the Nvidia Hopper architecture and will incorporate a QPU from QuEra.
In Poland, the Poznan Supercomputing and Networking Center (PSNC) has installed two photonic QPUs by ORCA Computing, which are integrated with a new supercomputer partition accelerated by Nvidia Hopper.
“Quantum computing will thrive through the integration of quantum and GPU supercomputing,” stated Tim Costa, Nvidia’s director of quantum and HPC. “Our platform enables institutions like AIST, JSC, and PSNC to push the boundaries of scientific exploration.”
With the QPU integrated into ABCI-Q, AIST researchers will explore quantum applications in AI, energy, and biology, utilizing Rubidium atoms controlled by laser light as qubits—identical atoms utilized in precision atomic clocks, paving the way for scalable, high-fidelity quantum processors.
“Japan’s researchers will advance practical quantum computing applications with the ABCI-Q quantum-classical accelerated supercomputer,” remarked Masahiro Horibe, deputy director of G-QuAT/AIST.
PSNC’s QPUs will facilitate research across biology, chemistry, and machine learning using two PT-1 quantum photonics systems. These systems employ single photons at telecom frequencies as qubits, enabling a modular quantum architecture built from standard telecom components.
“Our collaboration with ORCA and Nvidia has fostered a unique environment for developing a new quantum-classical hybrid system at PSNC,” stated Krzysztof Kurowski, PSNC CTO. “The seamless integration and programming of multiple QPUs and GPUs are essential for developers, opening doors to a new generation of quantum-accelerated supercomputers.”
The integration of a QPU with Jupiter will allow JSC researchers to innovate in quantum applications for chemical simulations and optimization challenges, highlighting how quantum computing can enhance classical supercomputers. This QPU operates with superconducting qubits, behaving as artificial atoms at low temperatures.
“Hybrid quantum-classical supercomputing makes quantum computing more accessible,” said Kristel Michielsen, head of JSC’s quantum information processing group.
CUDA-Q stands out as an open-source and QPU-agnostic quantum-classical accelerated supercomputing platform, preferred by many organizations deploying QPUs.
Nvidia’s Grace Hopper Superchips are set to supercharge scientific research across nine supercomputing centers. New additions include EXA1-HE in France, Helios in Poland, and Alps in Switzerland, among others.
“When harnessed for AI, Grace Hopper systems are crucial for accelerating climate research, drug discovery, and breakthroughs in various fields,” noted Ian Buck, Nvidia’s vice president of hyperscale and HPC.
Furthermore, Isambard-AI and Isambard 3 from the University of Bristol, alongside facilities at Los Alamos National Laboratory and the Texas Advanced Computing Center, are part of an expanding network of Nvidia Arm-based supercomputers utilizing Grace Superchips and the Hopper architecture.
As countries recognize the significance of sovereign AI, investments in domestically controlled data and infrastructure are accelerating the development of efficient AI-driven supercomputers.
Using Nvidia's NVLink-C2C interconnect technology, the GH200 serves as the powerhouse for numerous scientific supercomputing centers globally, facilitating swift transitions from installation to operational science.
Isambard-AI's initial phase features an HPE Cray Supercomputing EX2500 equipped with 168 Nvidia GH200 Superchips, marking it as one of the most efficient supercomputers to date. With additional 5,280 Nvidia Grace Hopper Superchips coming this summer, performance is expected to increase significantly.
“Isambard-AI positions the U.K. as a leader in AI, enhancing open science innovation both domestically and internationally,” stated Simon McIntosh-Smith from the University of Bristol. “Our collaboration with Nvidia has allowed us to deliver phase one swiftly, leading to major advancements in data analytics, drug discovery, and climate research.”