IBM Research Automates QEC Decoder Chip Design Using Open AI Agent Factories at 11.5x Lower Inference Cost
IBM Research has built an AI agent factory that automates the architectural design of quantum error correction decoder chips, using autonomous multi-agent search and open-weight models. The system reports up to 204x hardware speedup and an 11.5x reduction in AI inference cost, making decoder chip design more economically viable for fault-tolerant quantum systems.
Within two years, this could enable custom QEC decoder chips to be designed and fabricated on the timescale of a single research cycle, allowing experimental fault-tolerant systems to test real-time decoding for larger code distances.