Nvidia burns $4B to light up American photonics manufacturing
Summary
Nvidia has invested $2 billion each in Coherent and Lumentum, accompanied by non‑exclusive multibillion‑dollar purchase commitments, to secure supply of silicon photonics components. The move aims to lock in access to lasers, pluggable transceivers and optical networking products that Nvidia needs for scale‑out datacentre interconnects and multi‑datacentre distribution. Both vendors plan to expand manufacturing capacity in the United States as part of the deal.
The company has previously favoured copper for certain high‑power, scale‑up NVSwitch fabrics to save energy, but it embraces silicon photonics for scale‑out networks to cut power consumption and reduce the number of pluggable modules via co‑packaged optics in its next‑gen switches. The announcement comes amid surging AI investment and follows a large OpenAI funding round in which Nvidia is a major backer.
Key Points
- Nvidia will invest a total of $4 billion — $2bn each into Coherent and Lumentum — plus commit to multi‑billion dollar purchases.
- The investments secure supplies of silicon photonics components such as laser modules, transceivers and optical switches used in datacentre networking.
- Coherent and Lumentum will expand US manufacturing capacity as part of the strategic agreements.
- Nvidia prefers copper for some scale‑up fabrics due to power benefits, but is adopting silicon photonics for scale‑out networking and co‑packaged optics to reduce power and module counts.
- The move strengthens Nvidia’s position in the AI datacentre supply chain as bandwidth demands and distributed workloads grow.
- This comes shortly after a major OpenAI funding round in which Nvidia is a significant participant, underscoring wider industry momentum.
Author style
Punchy: This is a strategic, cash‑heavy play. Nvidia is not just buying components — it’s buying supply security and manufacturing scale in the US. If you’re tracking AI infrastructure, this is one of those moves that changes vendor dynamics and supply chains.
Why should I read this?
Short version: Nvidia just ploughed $4bn into optics suppliers so it won’t run short of the lasers and transceivers that keep huge AI farms talking to each other. If you care about AI datacentre costs, power bills, or who controls the supply chain, this is worth two minutes of your time.
Context and relevance
As AI models and datacentre fabrics scale, optical interconnects become critical to manage power and bandwidth. Nvidia’s purchase commitments and investment accelerate US photonics manufacturing and signal that large AI infrastructure vendors expect silicon photonics to play a central role in future scale‑out architectures. This affects networking suppliers, cloud operators and anyone planning large distributed training or inference deployments.