A joint white paper from JIJ and ORCA says their hybrid quantum-classical workflow beat classical decomposition baselines on bp-verified energy-grid optimization tests, tackling a unit commitment dataset with 25,755 variables and 48,939 constraints.
The system split the problem into classical and quantum layers, sending binary generator on/off decisions to ORCA’s PT-2 photonic processor while linear-programming solvers handled continuous dispatch variables and final refinement.
In a stress test with a 25% spinning-reserve increase, the quantum-assisted model raised committed generators from 60 to more than 74 and avoided the load-shedding penalties that hit static classical day-ahead schedules.
Current gains are tempered by speed limits—PT-2 still faces at least 300 ms sampling overhead and classical orchestration delays—but the paper projects ORCA’s mid-2026 PT-3 system will cut latency to 10 ms and outperform top classical solvers on both quality and runtime.