Miles Jamison 
- NVIDIA has added CUDA-Q Logical to its CUDA-Q software
- CUDA-Q Logical lets researchers compare algorithms and error-correction methods
- Fermilab has used the software to assess resources needed for fault-tolerant computing
NVIDIA has expanded its open-source CUDA-Q software with a new orchestration layer for developing applications designed for fault-tolerant quantum computers, the company said Monday.
What Does CUDA-Q Logical Do?
CUDA-Q Logical gives researchers a programmable way to design and evaluate components of fault-tolerant quantum systems built on logical qubits, according to NVIDIA. The approach allows researchers to evaluate different combinations of algorithms, error-correction methods and quantum hardware configurations.
Logical qubits are intended to help quantum processors handle errors that occur in physical qubits, a capability needed for larger workloads in fields such as drug discovery, financial modeling and materials development
Organizations like Fermi National Accelerator Laboratory, Sandia National Laboratories, Infleqtion, IQM Quantum Computers and QCDesign Quantum Motion are already using the tool.
NVIDIA introduced a related architecture, NVQLink, in October 2025 to connect GPU computing with quantum processors. The architecture links 17 quantum processor developers, five controller developers and nine national laboratories, including Fermilab and Sandia.
How Is Fermilab Using the Technology?
At Fermilab, researchers assessed how error-correction techniques and quantum hardware affected the resources required for fault-tolerant computing, using CUDA-Q Logical.
The resulting process gave researchers a consistent way to evaluate different fault-tolerant system configurations. NVIDIA said the work cut the time needed to develop and assess fault-tolerant algorithms from five months to three weeks.
The modeling also produced a lower hardware estimate for an Iceberg Quantum architecture based on Diraq’s qubits. NVIDIA said the design could support 1,000 logical qubits using 150,000 physical qubits, roughly one-tenth of Diraq’s earlier estimate.
What Is Sandia’s QUOPS Benchmark?
Sandia National Laboratories developed QUOPS as a hardware-agnostic benchmark for tracking quantum computing progress toward practical applications. NVIDIA said a reference implementation of the benchmark is now available in CUDA-Q.
Sandia’s quantum computing work has drawn other federal backing in recent years. In 2024, the Intelligence Advanced Research Projects Activity selected the laboratory for a research program on universal fault-tolerant quantum computing focused on potential applications for the intelligence community.
