Optical interconnect startup Ayar Labs has partnered with server maker Wiwynn to develop optically-connected rack-scale AI infrastructure.

The offering will combine Ayar Labs’ co-packaged optics technology with Wiwynn’s rack-level system design, the partners said in a statement, creating a “new class of rack-scale AI infrastructure that’s not constrained by the bandwidth and reach limitations of copper.”

The optical interconnect uses Ayar Labs’ TeraPHY optical engines, which are powered by the company’s SuperNova remote light source.

The rack-scale infrastructure can scale to 1,024 AI accelerators and beyond. It has been designed to allow the accelerators to operate as a single, unified system across multiple racks, and the companies claim the system also supports the delivery of more than 100Tbps of optical connectivity across each accelerator.

The offering additionally incorporates a liquid-cooled architecture, which has been optimized for high-power operations, and provides support for external laser small form factor pluggable (ELSFP) light sources, advanced fiber management, and serviceable system designs necessary for hyperscale environments.

“AI infrastructure is outgrowing the limits of copper, and hyperscalers need a fundamentally new approach to scale,” said Mark Wade, CEO and co-founder of Ayar Labs. “Optically connected racks eliminate the interconnect bottleneck and unlock the next order of magnitude in performance and efficiency. By combining Wiwynn’s global system and manufacturing leadership with Ayar Labs’ co-packaged optics expertise, we are delivering pioneering, rack-scale architectures purpose-built for optically connected scale-up AI networks.”

William Lin, president and CEO of Wiwynn, added: “Silicon photonics is reshaping how AI infrastructure is built. With Ayar Labs’ leadership in co‑packaged optics and Wiwynn’s strengths in rack‑level integration and manufacturing, we accelerate the shift from silicon‑ready innovation to system‑ready solutions. Together, we are enabling advanced optical I/O that delivers greater scalability and energy efficiency for cloud and hyperscale customers, powering next‑generation AI data centers.”

Founded in 2015, California-based Ayar Labs develops optical interconnect solutions based on open standards, optimized for AI training and inference workloads. Earlier this month, the company announced it had raised $500 million in a Series E funding round that saw participation from Nvidia, AMD, MediaTek, and Alchip.

Wiwynn was founded in 2012 in New Taipei City, Taiwan. It is a wholly-owned subsidiary of the Winstron Group, a technology service provider.