Anyon Computing has released an open-source, real-time control system built on NVIDIA's NVQLink fabric that welds superconducting quantum processors to CPU and GPU accelerators into a single machine. The Pasadena, California startup unveiled the design on September 15, 2026.
One supercomputer, not a quantum box bolted on
Anyon's architecture treats QPU nodes, their control cluster and GPU/CPU nodes as co-processor components sharing one RDMA-over-Ethernet fabric and a unified programming model. That lets error decoding, calibration, adaptive control and hybrid AI-quantum workflows execute across all components at microsecond latency. "A quantum supercomputer has to be one machine, not a quantum box bolted onto a classical one," said co-founder and CEO Dr. Roger Luo.
CUDA-Q, OpenQASM and QIR out of the box
The stack runs CUDA-Q, OpenQASM and QIR, and Anyon says it was developed almost entirely by autonomous AI agents and verified through simulation and hardware-in-the-loop testing before release. Publishing the design as open source lowers the barrier for other quantum hardware groups to plug into NVIDIA's growing quantum-classical stack.
Momentum around NVQLink
Anyon's release lands alongside a wave of NVQLink adoption across the industry. It follows partner news like the Altera–Riverlane QECI collaboration and Finland's IQM–Lumi logical qubit rollout, and complements NVIDIA's parallel push into a fault-tolerant quantum software layer.
Reporting based on coverage from The Quantum Insider, HPCwire and Quantum Computing Report.