Quantum Campus shares the latest in quantum science and technology. Read by more than 2,000 researchers, we are always looking for news from across the country. Advertising and sponsorship opportunities are available.
This story is sponsored by IEEE Quantum Week 2026,13-18 September, Toronto.
Don’t think of gaming when you think of NVIDIA. Don’t even think AI, according to Sam Stanwyck, senior product manager for quantum computing software at NVIDIA.
“We were mostly known as a gaming company. Now, we are known as an AI company or a GPU company. But none of that is quite accurate,” he said. “The truth is, we’re an accelerated-computing platform company. We are constantly seeking to expand the set of problems that computers can solve and do it in a way that helps the rest of the world be more successful.”
At IEEE Quantum Week 2025, he described how that approach applies to quantum computing, despite the fact that NVIDIA is not developing a quantum processor itself. The field is already starting to see the “symbiotic relationship” between AI and quantum computing with applications running on hybrid systems that rely on CPUs, GPUs, and quantum processors. Stanwyck also highlighted the promise of GPUs being a part of the crucial work of error correction, calibration, and control in quantum-based systems.
NVIDIA’s strategy goes back to CUDA, a programming model and API that NVIDIA built for its graphics processing chips more than 20 years ago. The company is now taking a similar tack with what they are calling quantum-accelerated supercomputing.
That’s the exciting thing about IEEE Quantum Week. The fundamental researchers at universities and national labs meet the applied researchers in industry who are building systems that expand and democratize the promise of quantum computing, networking, and sensing.
Stanwyck discussed three elements of NVIDIA’s roadmap that reflect the strategy they have implemented with CUDA over the decades.
CUDA-Q is an open-source, qubit-agnotic, kernel-based platform, built for “broad integration with all.” It supports Python and C++ and allows for GPU-accelerated simulation in quantum computing.
“It’s really critical to offer low-level control and programmability to everything we can,” Stanwyck said. “It’s too early to say exactly what machines 10 years from now will look like and the modality as they scale up. So what we want to do as we design software now is give full flexibility, full control, to the developer. That’s what we’ve done with CUDA-Q.”
CuQuantum is part of CUDA-Q and also integrates with other frameworks for quantum circuit simulation such as Cirq, Qiskit, Pennylane, and Kibo. It provides GPU-accelerated primitives for state vector, tensor network, and density matrix dynamics simulations.
CUDA-QX is the library layer. “It won’t all be built by us,” he said. “We started with the first two for quantum error correction and hybrid solvers. We’re going to be developing more, and we expect our partners will be developing many more as well.”
Stanwyck also talked about collaborations where those sorts of early developments are taking place, including partnerships with the University of Toronto and St. Jude’s Children’s Research Hospital to develop generative pretrained transformers for quantum-accelerated chemistry research and with Quantum Machines to build early quantum-AI hybrid reference systems.
Partnerships and success like these will be on full display at IEEE Quantum Week 2026 in Toronto. Expect more than 1,750 colleagues, world-class keynote speakers, hundreds of technical papers, and dozens of panels and talks 13-18 September. Registration is now open.
