Photonic computing company Lightmatter has launched the Open Silicon Photonics for AI Systems initiative and published a 300-page white paper alongside the coalition’s founding members in an effort to standardize silicon photonics-ready infrastructure for AI.

The initiative, which will operate as an official workstream within the Open Compute Project, was first announced in March 2026 and had ten founding members, including Celestica, Corning, Dell Technologies, Flex, Foxconn Interconnect Technology, Hyve Solutions, Keysight, Qualcomm Technologies, and Quanta Cloud Technology.

Nine additional companies have now joined the project.

According to Lightmatter, the whitepaper, called ‘Architecture Vision: Open Silicon Photonics for AI Systems,’ aims to guide the development of a shared co-packaged optics (CPO) blueprint to scale Modular Hardware System (MHS)- and Open Rack v3 (ORv3)-compliant AI clusters from 72 to more than 1,024 nodes.

Unlike pluggable optics, where the transceivers are kept separate from the main chips, CPO integrates optical and electrical components, such as optical engines and switching ASICs, on a single packaged substrate.

The reference architecture will offer hyperscalers a shared framework to support the transition from copper to photonics, scalable and interoperable across multiple SerDes generations, switch, and XPU design cycles.

"Interconnect has become as fundamental to AI infrastructure performance as compute," said Nick Harris, founder and CEO of Lightmatter. "The industry's shift to silicon photonics will unlock the next major leap in frontier model capabilities. Aligning 19 companies on a system-level architectural vision was an enormous undertaking, and I'm proud of what this coalition has accomplished. This white paper lays the foundation upon which this growing OCP community will develop a shared MHS-based CPO rack architecture that fits into hyperscalers' build flows. The work begins now."

California-based Lightmatter specializes in photonics-enabled technologies that use light instead of electrical signals for computing, meaning its chips don’t experience the same heat or resistance as traditional chip architectures. In March 2026, the company launched its Passage L20 interconnect, the successor to the company’s Passage M1000 and L200 offerings, which were launched in April 2025.