NVIDIA Bridges Quantum and Classical Supercomputing

29 Oct 2025 – NVIDIA is forging a tighter link between quantum and classical supercomputers. At its GTC Washington, D.C. event on October 28, NVIDIA unveiled NVQLink, a new high-speed interconnect and open architecture that couples quantum processors (QPUs) directly with GPU-accelerated supercomputers. The system supports a broad ecosystem – connecting 17 different quantum hardware platforms (from startups like IonQ, Pasqal, Quantinuum, Rigetti and others) and nine U.S. national laboratories via ultra-fast links. By enabling QPUs to plug in seamlessly with powerful GPU nodes, NVQLink creates a hybrid quantum-classical supercomputing platform.

The motivation is to overcome a key bottleneck in today’s quantum computing: qubits are fragile and error-prone, requiring continuous, low-latency classical control for tasks like error correction and calibration. Traditionally, quantum experiments have been limited by slow connections to classical computers. NVQLink solves this by providing a deterministic, high-bandwidth interface between QPUs and conventional HPC systems, so that error-correcting algorithms and feedback can run in real time. In practice, this means future “quantum-accelerated” supercomputers could tackle complex problems – in chemistry, materials science, optimization and more – faster than classical machines alone, by offloading parts to quantum co-processors.

NVIDIA CEO Jensen Huang likened NVQLink to a “Rosetta Stone” for computing: “In the near future, every NVIDIA GPU scientific supercomputer will be hybrid, tightly coupled with quantum processors… NVQLink is the Rosetta Stone connecting quantum and classical supercomputers”. He predicted this marks the onset of a new era of quantum-GPU computing. U.S. Department of Energy officials also praised the technology, calling it the “bridge to the next era of computing: accelerated quantum supercomputing” that will help solve grand scientific challenges.

Crucially, NVQLink is an open architecture. NVIDIA collaborated with a who’s-who of quantum hardware companies and national labs to ensure it works across many qubit modalitiesm – from superconducting circuits to trapped ions and neutral atoms. More than a dozen quantum startups (including Quantinuum, Alice & Bob, Oxford Quantum Circuits, Quandela and others) and major labs contributed to the design. This open approach means researchers can integrate any QPU into the NVQLink framework via NVIDIA’s CUDA-Q software platform. Developers will be able to write applications that seamlessly distribute workloads between CPUs, GPUs and QPUs.

The first NVQLink deployments will be at U.S. national labs. By signing up on NVIDIA’s site, supercomputing centers can begin accessing the architecture. It’s early days yet – NVQLink development kits are just rolling out – but the vision is clear. Much as GPUs revolutionized AI by working alongside CPUs, NVIDIA is betting that QPUs working alongside GPUs will revolutionize computing once again. With NVQLink connecting them, the hope is to unlock new performance in applications ranging from drug discovery to logistics that neither classical nor quantum computing could achieve alone. Jensen Huang’s bold GTC demo, backed by enthusiasm from quantum startups and government labs, underscores the momentum toward practical hybrid quantum-classical supercomputers.