Nvidia commits $6.5B to photonics in AI infrastructure push

Nvidia ↗2.95% has poured at least $6.5 billion into photonics companies since March, according to CNBC, aiming to overcome energy bottlenecks in AI data centers.cnbc
Investments include $2 billion each in Coherent, Lumentum ↘2.30%, and Marvell ↗7.27%, plus $500 million deals with Corning ↗11.13% and startup Ayar Labs.nvidia
Analysts say large-scale photonics deployment likely won’t begin until around 2028, as the technology must still prove it can replace copper at production scale.moomoo

Nvidia Invests $6.5 Billion in Photonics to Address AI Energy and Bandwidth Demands

Nvidia has committed at least $6.5 billion to photonics technology companies since March 2026, assembling a portfolio of partnerships aimed at replacing copper-based data connections with light-based alternatives as its AI infrastructure ambitions outpace existing capacity.cnbc

A Rapid Investment Spree

The spending began in early March with $2 billion investments in each of two optical networking firms, Coherent and Lumentum, under nonexclusive multiyear strategic agreements that include multibillion-dollar purchase commitments and capacity access rights for advanced lasers and optical networking products. On March 31, Nvidia announced a $2 billion stake in Marvell Technology, a deal covering custom chips, silicon photonics, and NVLink-compatible networking. In early March, Nvidia-backed startup Ayar Labs raised $500 million in a Series E round led by Neuberger Berman, with participation from ARK Invest and the Qatar Investment Authority, reaching a $3.75 billion valuation. Then on May 6, Nvidia invested $500 million in Corning as part of a partnership to build three new fiber optics manufacturing plants in North Carolina and Texas, a project expected to create at least 3,000 jobs and expand Corning’s U.S. optical manufacturing capacity tenfold.optica-opn

Why Photonics Matters for AI

Photonics uses light rather than electrical signals to move data between GPUs, memory, networking chips, servers, and data centers. The technology promises lower energy consumption at a time when power demands threaten to constrain AI deployment at scale. At Nvidia’s GTC conference in March, CEO Jensen Huang highlighted co-packaged optics as central to the company’s Spectrum-X Ethernet networking platform and said “the amount of silicon photonics technology capacity that we need is substantially higher than the world has today”.moomoo

A Long Road Ahead

Despite the scale of investment, analysts caution that widespread adoption remains years away. Forrester senior analyst Alvin Nguyen told CNBC that photonics gives Nvidia “a way to scale its AI infrastructure without significantly increasing power consumption,” ensuring the technology advances before electrical solutions hit their limits. IDC analyst James Patience noted that large-scale deployment is likely to begin around 2028, meaning investors will need patience as commitments convert into production capacity.moomoo