NVIDIA and Alice & Bob Are Building Hybrid Quantum‑GPU Supercomputers

Although the size of NVentures’ individual investment has not been disclosed, the funding reflects NVIDIA’s growing interest in the quantum computing ecosystem and its strategy of supporting hardware builders that can integrate with its accelerated computing stack.

Why Quantum Computers Need GPUs

Modern quantum systems cannot operate in isolation. They require extensive classical computing infrastructure to compile quantum circuits, simulate algorithms, process measurement results, and run error‑correction routines.

NVIDIA’s strategy is to connect quantum processing units (QPUs) with AI‑class supercomputers built on GPUs. According to NVIDIA, useful quantum computing will likely emerge from hybrid systems that tightly integrate QPUs with classical accelerated computing resources.

This hybrid model allows GPUs to:

• Simulate and validate quantum algorithms before running them on real hardware
• Optimize circuits and compile quantum programs
• Run real‑time control and decoding tasks during quantum computations

The collaboration with Alice & Bob is designed to build exactly this kind of architecture.

CUDA‑Q: The Software Layer for Hybrid Quantum Programs

A key component of the partnership is NVIDIA CUDA‑Q, an open‑source quantum development platform designed for hybrid quantum‑classical workloads.

CUDA‑Q allows developers to write programs that run across CPUs, GPUs, and QPUs simultaneously using a unified programming model. The platform is also QPU‑agnostic, meaning it can integrate with multiple quantum hardware technologies, including Alice & Bob’s cat‑qubit processors.

This software layer enables researchers and HPC centers to experiment with large‑scale quantum workflows even when real quantum hardware remains limited.

cuQuantum and GPU‑Accelerated Quantum Simulation

Before algorithms run on real quantum hardware, they often need to be simulated at scale. NVIDIA’s cuQuantum software development kit accelerates these simulations on GPUs, allowing researchers to study quantum circuits with performance beyond what current quantum processors can achieve alone.

This capability is especially valuable during the early stages of quantum hardware development, when simulation helps validate architectures, algorithms, and control systems.

Dynamiqs: Simulating Cat‑Qubit Physics on GPUs

Alice & Bob’s open‑source Dynamiqs library focuses on simulating the physics and control dynamics of quantum systems. The company has integrated CUDA‑Q into Dynamiqs, enabling GPU acceleration for large‑scale simulations of cat‑qubit systems.

Early benchmarks indicate the integration can improve simulation efficiency by up to 75×, making it easier for researchers to model quantum devices before deploying real hardware.

These simulations help refine control protocols, optimize error‑correction techniques, and guide hardware design.

NVQLink: Connecting GPUs and Quantum Processors

At the system architecture level, Alice & Bob and NVIDIA are collaborating on NVQLink, an open platform designed to connect GPUs directly with quantum processors.

NVQLink tightly couples a conventional supercomputing host with a quantum system controller, enabling GPUs to participate in quantum workloads in real time.

In practical terms, this allows GPUs to handle tasks such as:

• Circuit compilation
• Real‑time decoding for quantum error correction
• Dynamic calibration of qubits
• Feedback control during quantum computations

Low‑latency communication between the quantum processor and GPU infrastructure is critical for fault‑tolerant quantum computing, where control systems must react rapidly to measurement results.

Toward Hybrid Quantum‑Classical Supercomputers

The broader goal of the NVIDIA–Alice & Bob partnership is not a standalone quantum machine but hybrid quantum‑classical supercomputing platforms suitable for high‑performance computing (HPC) centers.

In such systems:

• Quantum processors perform specialized quantum operations.
• GPUs handle simulation, orchestration, and real‑time control tasks.
• Hybrid software platforms like CUDA‑Q coordinate workflows across all computing resources.

By integrating cat‑qubit processors with NVIDIA’s accelerated computing ecosystem, the companies aim to make quantum hardware easier to deploy within the same infrastructure used by modern AI and HPC systems.

If successful, this approach could help move quantum computing from isolated laboratory devices toward practical hybrid supercomputers capable of solving real‑world problems.