Lumilens, a Silicon Valley optical interconnect startup, emerged from stealth Wednesday with more than $700 million in new Series C funding and $900 million total capital, valuing the company at $5.51 billion. The round was co-led by Atreides Management, Bain Capital Ventures, Meritech, Seligman Ventures, and Spark Capital, with participation from Qualcomm Ventures, Peak XV, Redpoint, and others. Founded in 2024, Lumilens designs silicon photonics-based optical transceivers and co-packaged optics (CPO) that replace copper interconnects inside AI data centers, addressing a critical bottleneck as GPU density grows faster than conventional networking capacity.
Lumilens' pitch is stark: the constraint on AI has shifted from acquiring GPUs to connecting them efficiently. Within two years of founding, the company shipped its first product into production AI data centers under a multi-billion-dollar customer agreement. A 400,000-GPU data center requires over 2.4 million optical transceivers and five million fiber strands; conventional copper faces physical limits that constrain scale. Lumilens manufactures both scale-up optics (GPU-to-GPU within racks) and scale-out fabrics (between clusters), built on its proprietary LumiCore platform combining silicon photonics, mixed-signal ICs, and optical systems.
For operators deploying massive AI clusters, Lumilens represents the manufacturing side of solving the GPU-connection bottleneck that's become as acute as HBM shortage. The company's competitive edge rests on early production traction, high-volume manufacturing automation (robotics + AI), and support from investors like Qualcomm (ecosystem access) and Spark Capital (who led Series B in 2025). Expect aggressive margin pressure as Chinese competitors (Eoptolink) chase the same $100+ billion market, but Lumilens' lead customer commitment and module-swap capability give it runway to achieve scale dominance.