# What happened Quantum Machines ran the first end-to-end NVIDIA CUDA-Q program that executed across live qubits and a PPU classical processor connected by NVIDIA NVQLink. The demonstration used one program to dispatch quantum operations to a QPU and classical tasks to GPUs and CPUs, with the full measurement-control exchange completing in about a millionth of a second.
# Why it matters Today's quantum processors require tightly choreographed control and fast classical feedback for advanced workloads such as quantum error correction. This demo shows a path for developers to write hybrid quantum-classical programs in common languages (Python, C++ or QUA) without having to hand-code the low-level pulse sequences and timing that traditionally demand specialized hardware expertise. The result is a simpler software interface and a fast data path between qubits and classical processors.
# How it works
# Technical result reported Quantum Machines and NVIDIA demonstrated that measurement data can be delivered to classical processors and decisions returned to the quantum control system in microseconds. That microsecond-scale turnaround is the timing required for realtime interaction in workloads such as quantum error correction and other hybrid algorithms that need fast classical processing in the loop.
# Context and positioning The companies presented the live run at IEEE Quantum Week in Toronto in front of researchers and engineers. Quantum Machines described the integration as removing layers of complexity that have slowed deployment of quantum workloads, reducing time-to-value for research and commercial teams.
NVIDIA framed the result as an example of how QPUs can operate alongside GPUs and CPUs as unified quantum supercomputing systems. NVQLink is the microsecond-speed connection architecture that distributes work between conventional computing and the quantum processor.
# Practical implications for developers and researchers
- Developers can write hybrid code once with CUDA-Q and have it routed to the appropriate processor without reauthoring hardware-specific control flows.
- The integration reduces the need for specialist knowledge of pulse-level control when deploying hybrid workflows.
# Short summary Quantum Machines' demonstration shows a working integration of CUDA-Q and NVQLink that lets a single program coordinate live qubits, GPUs and CPUs with microsecond feedback. The result simplifies hybrid programming and provides the latency characteristics needed for realtime quantum-classical interaction.