Qblox links NVIDIA DGX Spark to its quantum control stack in microseconds

[codeberg]

Quantum control company Qblox has connected NVIDIA's desktop AI computer, the DGX Spark, directly to its quantum control hardware. The Quantum Insider's Mohib Ur Rehman reports that the link uses NVIDIA's NVQLink technology

The key figure is latency. Qblox measured round trip times below 7.2 microseconds on the DGX Spark and below 4.0 microseconds on an RTX PRO 6000 Blackwell card. That is fast enough to put GPUs inside the real time control loop of a quantum computer

The aim is to let GPUs help while experiments are running. Calibration could update qubit settings on the fly, and error syndrome data could be sent quickly to GPU decoders for real time correction. The hardware uses Qblox's Q1 sequence processors inside its Cluster system, with NVIDIA's CUDA-Q software

The work builds on NVQLink announcements from October 2025 and Qblox's support for real time CUDA-Q in March 2026. Chief executive Niels Bultink and NVIDIA's Sam Stanwyck are quoted. The figures are vendor reported, as is usual with these announcements

Low latency links between classical and quantum hardware are becoming essential, as we saw with the SEEQC decoder work. A desktop sized AI computer controlling a quantum system is also a nice sign that the tools are becoming more accessible

Is real time GPU control the future of quantum hardware? And how small could quantum control systems eventually get?

Qblox Integrates NVIDIA DGX Spark With Quantum System Controllerthequantuminsider.com