Nvidia Corp. said today it’s tackling one of the major headaches for developers of quantum applications with the launch of a new orchestration layer within its open-source CUDA-Q platform.
The announcement came at this week’s IEEE Quantum Week 2026 event that kicked off in Toronto today. It’s designed to accelerate the quantum computing industry’s shift toward newer fault-tolerant quantum systems.
Nvidia said CUDA-Q Logical is meant to expand the capabilities of CUDA-Q, which is a software environment that’s used by developers to build hybrid applications that span both classical and quantum computing architectures. With CUDA-Q, developers can create software programs that can orchestrate workloads across both traditional processors, such as central processing units and graphics processing units, and also quantum processing units, known as QPUs.
CUDA-Q is one of the most popular quantum programming platforms around, but Nvidia says the industry’s recent shift toward fault-tolerant QPUs has caused a lot of problems for developers. Fault tolerance is necessary if quantum computers are to become commercially viable.
These systems employ “logical qubits,” which are groups of physical qubits that coordinate with one another carefully to correct errors and prevent their calculations from being corrupted. But creating software for logical qubits is extremely difficult, because the error-correction code they use changes the underlying physical resources needed to execute an application, throwing everything out of kilter, Nvidia explained.
With CUDA-Q Logical, Nvidia is trying to get around this. It provides researchers with a programmable and verifiable environment in which they can simulate and test fault-tolerant quantum computing systems. It allows developers to model different algorithms, error-correction techniques and QPU architectures side-by-side. In doing so, they can quickly come up with the most optimal configuration before they set up the necessary hardware.
Timothy Costa, Nvidia’s vice president and general manager of quantum, said that because quantum computing is rapidly maturing with the “era of logical qubits,” researchers need an open and customizable programming platform that represents all aspects of fault-tolerant systems. “The addition of CUDA-Q Logical provides power and flexibility to explore fully integrated, co-optimized systems regardless of qubit type and architecture,” he said. “It drastically shortens the timeline to useful quantum-GPU supercomputing.”
The biggest advantage of CUDA-Q Logical is that it saves time. The Australian quantum startup Iceberg Quantum, which designs fault-tolerant architectures that make optimal use of QPU hardware, used the new capability to model a new architecture for silicon-based qubits developed by Diraq Pty Ltd. Its model revealed that it should be possible to create 1,000 logical qubits from just 150,000 physical qubits, which is 10 times less than what Diraq originally estimated.
The Fermi National Accelerator Laboratory has also been playing around with CUDA-Q Logical to evaluate new error-correction strategies, runtime requirements and algorithms across multiple different quantum computing architectures. Fermilab’s researchers were able to reduce the average development cycle of fault-tolerant algorithms from five months to just three weeks, meaning they can be built roughly seven-times faster than before.
“Fault-tolerant quantum computing is the path to unlocking new scientific discovery, but getting there will require researchers to co-design algorithms, error-correction, architectures and hardware together,” said Fermilab Chief Technology Officer Anna Grassellino. “Using CUDA-Q Logical, our team explored combinations of these resources in just three weeks, compared with what would have typically taken about five months of building specialized infrastructure.”
Nvidia also took the opportunity to talk about the broader quantum computing community is using its Quantum-GPU Supercomputing Platform, which is a cloud service that combines high-performance GPUs with quantum processors. It’s designed for experimental applications and workloads that are best solved by a combination of classical and quantum computers.
For example, Diraq, which designs quantum computers that use spin qubits based on modified silicon transistors, is using Nvidia’s open-source Ising models to calibrate its processors. Other startups, such as Anyon Computing and Quantum Machines, are using Nvidia’s NVQLink networking technology to link QPU clusters directly to GPU supercomputers. BlueQubit, IonQ Inc. and Qedema Quantum Computing have integrated their tools with CUDA-Q to deploy hybrid-quantum workloads in production.
