Real-Time GPU–QPU Integration for Photonic Quantum Machine Learning using NVQLink

Quandela proposes a hardware architecture using NVIDIA’s NVQLink to directly couple photonic quantum processors (QPUs) with GPUs in a colocated, low-latency environment for real-time quantum machine learning inference. The piece argues conventional cloud quantum stacks impose at least 5 seconds of end-to-end latency per datapoint due to queueing and treating the QPU as a remote resource, which the colocated approach aims to eliminate.