A custom chip can now fix quantum computer errors in 550 nanoseconds, acting like a high-speed reflex for a fragile machine.
Hardware-integrated neural networks can perform quantum error correction fast enough to keep a superconducting processor stable. Quantum computers usually fail because their internal states collapse before a human or a normal computer can fix them. This new decoder operates with a reflex speed that allows the system to correct itself while it is still running. This milestone is a critical step toward building quantum computers that can actually finish a long calculation without breaking. It proves that real-time AI is the key to making quantum hardware practical for the real world.
Real-time Surface-Code Error Correction Using an FPGA-based Neural-Network Decoder
arXiv · 2605.04892
Quantum error correction (QEC) is essential for achieving low error rates required for fault-tolerant quantum computation. In stabilizer-based codes such as the surface code, errors are inferred from repeated syndrome measurements and corrected by a classical decoder. To prevent error accumulation, decoding must be performed with both high throughput and low latency to keep pace with the QEC cycle and enable real-time feedback for universal logical operations. Here we report a hardware-integrate