Ayar Labs, Wiwynn Partner on CPO AI racks
Partnership aims to deploy co-packaged optics in rack-scale AI infrastructure for hyperscale data centers.
Ayar Labs and Wiwynn have announced a strategic partnership to develop optically connected, rack-scale AI systems designed to support next-generation hyperscale workloads.
The collaboration integrates Ayar Labs’ co-packaged optics (CPO) technology with Wiwynn’s rack-level system design and manufacturing capabilities to address the bandwidth and power limitations of traditional copper interconnects.
The joint solution combines Ayar Labs’ TeraPHY optical engines with its SuperNova remote light source, enabling high-bandwidth optical I/O integrated directly with AI accelerators.
By replacing copper interconnects with optical links, the companies aim to deliver improved scalability, performance, and energy efficiency for large-scale AI clusters.
Designed for hyperscale environments, the rack-scale architecture can support systems of 1,024 AI accelerators or more, with each accelerator capable of delivering over 100 Tbps of optical connectivity.
The platform also incorporates a liquid-cooled infrastructure, advanced fiber management, and serviceable system designs tailored for high-density AI data centers.
Wiwynn contributes extensive experience in rack-level integration and large-scale manufacturing, having deployed servers to more than 750 data centers worldwide.
By combining this expertise with Ayar Labs’ silicon photonics-based CPO technology, the companies aim to accelerate the transition from component-level optical innovation to deployable AI infrastructure.
The partners will showcase their joint AI CPO architecture at the Optical Fiber Communication Conference 2026 in Los Angeles, where they will preview a liquid-cooled, HVDC-enabled rack-level AI system designed for next-generation hyperscale data centers.
