Pasqal Introduces New Integration with NVIDIA CUDA-Q to Enhance Its Hybrid Quantum Computing Environment for HPC - The Quantum Insider

NVIDIA's quantum strategy does not require a single new qubit. The Pasqal-CINECA integration this week is the latest proof.

The French neutral-atom company wired its quantum processors into the Leonardo pre-exascale supercomputer in Italy using QRMI — the Quantum Resource Management Interface. QRMI abstracts the QPU into a schedulable compute resource that Slurm can dispatch alongside CPUs and GPUs. The practical effect: quantum code runs as a native job step inside a supercomputing workflow. Pasqal is a member of NVIDIA's startup Inception program. NVIDIA CUDA-Q is the runtime layer, and it is the real story.

This is not altruism. NVIDIA is building the integration plumbing for hybrid quantum-classical computing — a market that does not yet exist at scale but that it intends to own regardless of which qubit modality eventually prevails. CUDA-Q is open-source. The integrations are incentivized. But interoperability between quantum hardware and HPC infrastructure is a genuine friction point, and NVIDIA is filling it.

IBM published the QRMI reference architecture on arXiv this month, co-developed with Pasqal and STFC Hartree Centre. The paper frames QPUs as heterogeneous compute alongside CPUs and GPUs — a tool in the toolbox, as QuantWare put it. That framing is now operational, not aspirational. The architecture defines three tiers: quantum runtime with specialized classical accelerators, co-located GPU/CPU systems linked via low-latency interconnect (NVQLink, ROCE, Ultra Ethernet), and partner scale-out systems. IBM is already running it with RIKEN and Fugaku.

Scaleway, France's Iliad cloud subsidiary, announced CUDA-Q compatibility for its Quantum-as-a-Service platform the same week — single codebase targeting GPU clusters or IQM and AQT hardware without code changes. Scaleway's GPU emulator simulates 38 noiseless qubits on eight Blackwell Ultra GPUs. That is a GPU hitting its memory ceiling, not a quantum advantage demonstration.

Pasqal is pursuing a $2 billion SPAC listing via Bleichroeder Acquisition Corp. II. Hybrid HPC-quantum as standard supercomputer infrastructure is exactly the narrative that listing requires — and exactly the claim that has not been independently benchmarked. CUDA-Q is real infrastructure. Whether it is the foundation of genuine quantum advantage or the most expensive middleware ever written is a question neither announcement answers.