The demonstration features Wiwynn’s compute architecture integrated with GUC’s multi-chip packages and Ayar Labs’ TeraPHY optical engines, which are powered by the SuperNova light source. Unlike previous proof-of-concept tests, this setup provides real-time telemetry, allowing operators to monitor signal quality and thermal drift as the system runs. The integration validates that optical connections can maintain stability under live workloads, a prerequisite for large-scale data center deployment.

By incorporating system-level requirements directly into the chip design phase, the collaboration aims to shorten development cycles for hyperscalers. The system also introduces serviceability features, such as detachable fiber units and in-chassis routing, designed to fit into standard ORV3 compute environments. For data center operators, the ability to track metrics like eye height and wavelength drift in real time addresses long-standing concerns regarding the reliability of optical I/O in dense, high-power AI environments.