NVIDIA Expands Quantum Computing Frontier with CUDA-Q Platform
Jessie A Ellis Mar 20, 2025 12:26
NVIDIA's CUDA-Q platform enhances quantum computing research, offering improved performance and scalability for hybrid quantum supercomputers, according to NVIDIA Developer Blog.
NVIDIA is making significant strides in the realm of quantum computing with the introduction of its CUDA-Q platform, designed to streamline both software and hardware development for hybrid, accelerated quantum supercomputers. According to the NVIDIA Developer Blog, this platform allows users to write code once and test it on any quantum processing unit (QPU) or simulator, significantly accelerating the workflow and enabling researchers to focus on scientific breakthroughs.
Enhanced Capabilities with CUDA-Q v0.10
The latest version, CUDA-Q v0.10, brings enhanced features and performance. It supports eight QPU backends across four different qubit modalities and includes compatibility with the state-of-the-art NVIDIA Blackwell GPUs. This version also introduces support for the NVIDIA GB200 NVL72 and its fifth-generation multinode NVLink capabilities, further pushing the boundaries of performance as demonstrated by standardized Quantum Economic Development Consortium (QED-C) benchmark applications.
Quantum Research and Collaboration
In a bid to standardize quantum benchmarking, NVIDIA is collaborating with the QED-C. Tom Lubinski, founding chair of the QED-C Standards and Performance Metrics Technical Advisory Committee, emphasized the need for transparent and unbiased metrics in quantum benchmarking efforts. The QED-C benchmarking GitHub repository now includes CUDA-Q, allowing users to assess its simulation performance on standard applications.
The power of CUDA-Q is showcased through QED-C benchmarks for simulating Hamiltonians from the HamLib dataset, covering a range of problems from chemistry to optimization. A 33-qubit state vector simulation using a single NVIDIA GB200 chip is 34 times faster than a 192-core EPYC CPU and twice as fast as the previous generation NVIDIA GH200 Grace Hopper Superchip.
Accelerating Quantum Chemistry and AI Applications
NVIDIA, in partnership with IonQ, Amazon, and AstraZeneca, has developed an accelerated quantum chemistry workflow using CUDA-Q within Amazon Braket. This workflow aims to model the nickel-catalyzed Suzuki–Miyaura Cross-Coupling reaction, a significant process in drug molecule synthesis. The integration of quantum and classical computing techniques allows for the exploration of broader chemical reactivity questions.
Furthermore, CUDA-Q is facilitating the development of AI for quantum applications, such as the generative quantum eigensolver (GQE), which extends to generating circuits for combinatorial optimization problems.
Industry and Academic Engagement
CUDA-Q's kernel-based programming model enables easy scaling across multiple GPUs, attracting industry and academic partners to transfer workloads to CUDA-Q. Companies like Aramco are leveraging CUDA-Q for hybrid workflows in image processing applications, which aim to efficiently identify object boundaries in three-dimensional images.
The CUDA-Q MQPU backend allows algorithms to be parallelized using multiple GPU simulated QPUs, dramatically shortening development cycles. This is crucial for projects like those with Hewlett Packard Enterprise (HPE), which explore methods for distributing large quantum circuits across accelerated quantum supercomputers.
Educational Initiatives and Future Prospects
NVIDIA is also fostering a skilled quantum workforce through the CUDA-Q Academic initiative, partnering with over 25 leading universities. Workshops based on the NVIDIA Quick Start to Quantum Computing series have been effective in teaching advanced quantum concepts to beginners.
The CUDA-Q platform is poised to become the industry standard for developing hybrid applications, with performance and flexibility accessible to a broad audience. As NVIDIA continues to expand its software and hardware integrations, the potential for accelerated quantum computing applications grows ever larger.
Image source: Shutterstock