Nvidia’s spending $4 billion on photonics to stay ahead of the curve in AI

Nvidia is making a massive $4 billion strategic bet on photonics technology to maintain its dominance in the AI hardware race. The company announced dual $2 billion investments in optical component specialists Lumentum and Coherent, securing multiyear purchase commitments and capacity rights for advanced laser components, circuit switches, and optical transceivers. This investment directly targets a critical bottleneck in modern AI infrastructure: moving vast amounts of data between GPUs within sprawling data centers. The push is driven by the rise of complex, multi-step agentic AI systems from companies like Anthropic and Microsoft, which require significantly higher bandwidth and lower latency to function effectively.

Photonics technology, which uses light to transmit data through optical fibers, promises a solution by offering substantially higher bandwidth and lower power consumption than the copper cables currently prevalent in data centers. Nvidia's move follows its successful 2020 acquisition of Mellanox, which enhanced its NVLink interconnect technology, and positions the company against rivals like AMD, which acquired photonics startup Enosemi last year. The strategic importance of this field is underscored by DARPA's recent call for photonic computing research proposals. For Nvidia, this investment is about future-proofing its AI systems, ensuring that the physical connections between its powerful GPUs don't become the weak link that slows down the next generation of AI applications.