According to a recent LinkedIn post from Quantum Machines, Alice & Bob has decoded its LDPC-cat quantum error-correcting codes on graphics processing units using NVIDIA’s NVQLink, Quantum Machines’ OPX1000 control system and the Open Acceleration Stack. The post describes the result as the first time these codes have been decoded on GPUs through NVQLink, while indicating that real-time decoding with live data remains the next step.
The technical focus is decoder latency and throughput: incoming error syndromes must be processed quickly, and non-Clifford operations may require decoder outputs before a computation proceeds. The post suggests that scaling quantum machines will increase these demands, making accelerated decoding infrastructure an important component of practical quantum-error-correction architectures.
For Quantum Machines, the collaboration provides evidence that its hardware and software stack can participate in a developing GPU-accelerated quantum-computing workflow, potentially supporting customer adoption and ecosystem partnerships. However, the update does not disclose contract terms, revenue contribution, deployment scale or performance benchmarks, limiting near-term conclusions about financial impact.
More broadly, GPU decoding could help address a bottleneck in fault-tolerant quantum computing, although the cited demonstration does not establish production readiness. Progress toward live-data, real-time operation and the ability to sustain required latency and throughput will be key indicators for assessing commercial significance.