Quandela and NVIDIA have published a technical white paper outlining an architectural framework for integrating photonic quantum processing units into existing artificial intelligence and high-performance computing infrastructure, according to a corporate release issued by the companies. The newly proposed framework combines central processing units, graphics processing units, and quantum processing units, utilizing NVIDIA CUDA-Q, cuQuantum, and NVQLink technologies to support quantum algorithm development, GPU simulation, and low-latency QPU integration.
Architecture and Low-Latency Integration Mechanics
The hybrid framework relies on the functional complementarity of CPUs, GPUs, and QPUs, as detailed in the technical documentation. According to the companies, the CPU handles workflow orchestration, while the GPU remains central to AI processing by managing intensive computations and simulations. The QPU functions as a specialized hardware accelerator for workloads where quantum computing provides measurable value. Experimental validation presented by Quandela at ISC 2026 demonstrated a low-latency connection between a photonic QPU and an NVIDIA GPU host via an FPGA-based Quantum System Controller using NVIDIA NVQLink, moving beyond traditional cloud APIs and asynchronous job queues.
Pro Tip: According to Quandela Chief Technology and Product Officer Jean Senellart, collocated quantum acceleration inside HPC environments allows photonic QPUs to be treated less like remote experimental instruments and more like accelerators deployed alongside GPUs.
The Three-Step Deployment Path for Organizations
Organizations can explore quantum applications progressively by relying on existing infrastructure without replacing current hardware, according to the white paper. The collaboration outlines a three-step adoption path comprising Access, Integration and Discovery, and Scaling. Access enables researchers and organizations to experiment with quantum computing without owning dedicated infrastructure. Integration and Discovery involves embedding the QPU into established CPU and GPU environments to simulate, test, and identify relevant workloads. Scaling targets the expansion of applications that demonstrate performance potential toward larger quantum systems.
Software Platforms and Target Quantum Workloads
Development and simulation within the framework rely on the NVIDIA CUDA-Q platform, the cuQuantum software development kit optimized for GPU-based quantum simulation, and Quandela’s MerLin framework, according to the companies. Initial target workloads focus on photonic Quantum Machine Learning, including quantum reservoir computing, quantum feature maps, and hybrid neural-network architectures. Quandela noted that photonic circuits can remain configured during inference while handling lightweight updates for new data points, making system-level latency a decisive performance factor addressed by the NVQLink interconnect.
Frequently Asked Questions
What is the primary purpose of the Quandela and NVIDIA white paper?
According to the companies, the white paper outlines an architectural framework for integrating photonic quantum processing units directly into existing AI and high-performance computing environments using NVIDIA technologies.
How do GPUs and QPUs interact in this new architecture?
As stated in the technical documentation, GPUs handle intensive processing and simulation while communicating with an FPGA-based Quantum System Controller through NVIDIA NVQLink to provide low-latency data exchange with the photonic QPU.
What are the first target workloads for this integration?
According to Quandela, the initial target workloads involve photonic Quantum Machine Learning applications, such as quantum reservoir computing and hybrid neural-network architectures.
Where is the new architecture being presented?
According to the release, the hybrid architecture will be presented at IEEE Quantum Week 2026 in Toronto on September 17 at the NVIDIA booth.
Join the Conversation: How do you see hybrid quantum-classical computing impacting your industry’s workflow over the next few years? Leave a comment below to share your perspective.