Nvidia’s New Quantum Layer: Accelerating Design by Months
Nvidia has introduced CUDA-Q Logical, an open-source orchestration layer for its quantum platform, significantly streamlining the development process for fault-tolerant quantum computing.
September 14, 2026 – 2:22 pm
Credit: Nvidia
Nvidia’s CUDA-Q Logical, integrated into their open-source quantum platform, targets the challenging aspect of designing for a non-existent fault-tolerant quantum computer. Traditionally, this involves intricate guessing and iteration between algorithms, error-correction codes, and hardware specifications.
Fermilab demonstrated the power of this new layer by completing design work in just three weeks, a process that typically takes approximately five months. This acceleration was achieved with the collaboration of several organizations: Sandia National Laboratories, Infleqtion, IonQ, Iceberg Quantum, IQM Quantum Computers, QCDesign, and Quantum Motion.
A notable achievement attributed to Iceberg Quantum in partnership with Diraq is the potential to access 1,000 logical qubits with only 150,000 physical ones—a significant reduction compared to previous estimates.
The announcement also highlights Sandia’s development of QUOPS, a benchmark designed to evaluate progress in quantum computing across various platforms, providing a much-needed industry standard.
Phasecraft, a company known for its THRIFT algorithm, is also a user of Nvidia’s cuQuantum toolkit, as revealed in their concurrent launch of a molecular database.
Nvidia’s strategic focus on providing the classical layer for quantum computing, while fostering open-source tools, has implications for the entire industry. This approach lowers adoption barriers and encourages long-term investment in quantum technology.
In a broader context, quantum computing is making inroads into mainstream infrastructure. Oracle is installing a Quantinuum machine within its data center, marking a shift towards on-site quantum computing access rather than remote access through cloud brokers.