Details
- Pasqal announced integration of NVIDIA's CUDA-Q platform with its QRMI runtime, enabling CUDA-Q workloads on neutral-atom quantum systems via standard Slurm HPC workflows on March 16, 2026.
- Involves Pasqal, NVIDIA, CINECA (integrating with Leonardo pre-exascale supercomputer), and QRMI collaborators including IBM, RPI, STFC Hartree Centre; follows Pasqal's path to public via Bleichroeder Acquisition Corp. II.
- QRMI exposes QPUs as schedulable resources in Slurm for secure authentication, allocation, monitoring; CUDA-Q unifies CPUs, GPUs, QPUs with tight interleaving of classical and quantum routines, hardware-agnostic design.
- Builds on prior 2025 CUDA-Q integration; first on-premises at CINECA with Leonardo for hybrid GPU-QPU workloads, already live on Pasqal cloud; contrasts with traditional quantum access requiring new models.
- Pasqal serves 25+ clients like IBM, Thales; next-gen Vela QPU (256+ qubits) launching 2026 for quantum advantage in optimization, simulation, AI; QPUs at Jülich, TGCC, hyperscalers like Azure Quantum.
Impact
This integration lowers barriers for HPC adoption of quantum acceleration, positioning neutral-atom QPUs as native components in supercomputing like Leonardo, rivaling GPU revolutions. It accelerates hybrid quantum-classical apps in AI and simulation amid Pasqal's scaling to 10,000+ qubits and public listing. Competitors like IBM lag in Slurm-native interoperability, enhancing Pasqal's edge in production-grade quantum-HPC.
