Optical technology from Lumentum and Coherent could be crucial for connecting GPUs in AI data centers.
Nvidia is investing $2 billion each into Lumentum and Coherent, both working on photonics technology for data centers, such as optical transceivers, circuit switches, and lasers for high-speed, long-distance data transfers. This technology could enhance energy efficiency, data transfer speeds, and bandwidth in AI data centers, following Nvidia’s 2020 acquisition of Mellanox for improving NVLink and GPU data movement.
Lumentum’s nonexclusive multiyear agreement includes a “multibillion purchase commitment and future capacity access rights for advanced laser components,” along with R&D and manufacturing support. Coherent’s deal has a similar structure with a “multibillion-dollar purchase commitment and future access and capacity rights for advanced laser and optical networking products.”
The expansion of AI systems like Anthropic’s Claude Cowork and Microsoft’s Copilot Tasks is increasing AI data center bandwidth needs for rapid task execution. Photonics could offer a solution with higher bandwidth and lower latency than copper cables, using less power.
Nvidia is not alone in pursuing photonics. Last month, DARPA issued a call for research proposals to enhance photonic computing for AI applications. Nvidia’s competitor AMD acquired silicon photonics startup Enosemi last year to boost its optics innovation for AI systems.