Israeli quantum technology company Quantum X Labs Inc. (Nasdaq: QXL) said on August 21, 2026 that its updated AI-driven quantum error correction (QEC) decoder improved on Google's own published benchmarks when evaluated on the same public surface-code dataset from a real quantum-hardware experiment.
Beating PyMatching In One Configuration
Working from Tel Aviv, QXL evaluated its updated decoder on a public surface-code configuration using the same cross-validation approach Google used for its own decoder comparisons. In that test, the QXL model outperformed matching-family benchmarks, including Google's published correlated-matching and PyMatching results for the identical configuration. Crucially, the QXL model was trained exclusively on synthetic samples and never on real hardware shots from the Google dataset.
Synthetic-To-Real Generalization
"These results are important because they bring us closer to the point where AI-driven quantum error correction can be evaluated against real hardware behavior, not only simulation," said Prof. Nir Sharon, Chief Quantum Technology Scientist at Quantum X Labs. He added that the team remains "disciplined" because the improvement was demonstrated on a single benchmark configuration, and the next step is to reproduce it across additional device centers and code geometries. The QXL decoder combines quantum-code structure, syndrome information and AI-based error weighting, and is designed for GPU acceleration.
NVIDIA CUDA-Q And IQCC Roadmap
QXL's roadmap wires the decoder into workflows with NVIDIA accelerated computing and NVIDIA CUDA-Q, with planned syndrome experiments through the Israeli Quantum Computing Center (IQCC) to push toward low-latency and eventually real-time decoding. Fault-tolerant quantum computing is widely viewed as a prerequisite for scaling quantum machines from lab curiosities into practical systems for drug discovery, cryptography and materials science.
Part Of A Wider AI-Plus-Quantum Push
QXL's result adds to a summer of tighter AI-and-quantum coupling, including IBM's QOBLIB launch for optimization benchmarking, IBM's modular cryogenic link toward Starling and Q-CTRL's 100-qubit QFT on IBM Heron. QXL now plans to extend its cross-validation to additional device centers and surface-code configurations before pushing toward IQCC hardware runs.
Reporting based on coverage from GlobeNewswire, Stock Titan and Quantum X Labs.