Nvidia Expands Photonics Push With Billions In New AI Infrastructure Investments

Nvidia has committed at least $6.5 billion in investments tied to photonics and optical networking companies since March, widening its effort to address one of the biggest infrastructure constraints facing the artificial intelligence industry: the movement of massive volumes of data inside data centers.

The spending comes as AI companies, cloud providers and governments continue pouring billions into AI infrastructure amid broader geopolitical competition around computing power, semiconductor manufacturing and energy security.

The rapid buildout of AI systems has intensified concerns around electricity consumption and hardware bottlenecks, particularly as countries race to secure domestic chip and data center capacity following supply disruptions linked to trade tensions and ongoing conflicts affecting global technology supply chains.

Photonics uses light instead of electrical signals carried through copper wiring to move data between chips, servers and networking systems, according to CNBC. The technology is increasingly being adopted in AI infrastructure because it consumes less power and can transfer data at higher speeds.

Nvidia has announced $2 billion investments each in photonics companies Lumentum, Coherent and Marvell since March, according to multiple reports, while also committing $500 million to Corning for optical connectivity systems development. The company also participated in optics startup Ayar Labs’ $500 million Series E funding round.

Reuters reported earlier that Nvidia’s investments in Lumentum and Coherent were aimed at improving data center chip performance and securing future access to optical networking and laser technologies. The report also noted that the agreements included purchase commitments and manufacturing expansion plans in the United States.

Industry analysts have increasingly identified power use and data transfer limitations as major issues for AI infrastructure operators. AI systems require thousands of graphics processing units, or GPUs, to communicate simultaneously across servers and data centers, creating rising pressure on networking hardware.

Brian Colello, senior equity analyst at Morningstar, told CNBC that Nvidia’s next-generation AI rack systems would require larger amounts of optical connectivity to manage increasing bandwidth demands tied to larger AI models and heavier usage.

Nvidia Chief Executive Jensen Huang also highlighted the company’s growing demand for silicon photonics technology during Nvidia’s GTC conference in March. Huang said Nvidia was expanding the use of photonics across its networking platforms and GPU interconnect systems, adding that required manufacturing capacity was “substantially higher than the world has today.”

Reuters separately reported in late March that Nvidia invested another $2 billion in Marvell Technology to strengthen collaboration on AI chips and optical interconnect systems used in data centers. The partnership focused on integrating Marvell’s optical networking technology with Nvidia’s AI infrastructure offerings.

Corning also became part of Nvidia’s broader optical connectivity expansion earlier this month. In May, Nvidia committed billions of dollars in funding tied to new Corning manufacturing facilities in addition to an equity investment valued at up to $3.2 billion. Corning’s fiber-optic cable systems are widely used in large-scale data centers.

The growing focus on optical networking reflects rising concern across the industry over electricity demand linked to AI deployment. Data centers supporting AI models consume significantly larger amounts of power compared with traditional cloud computing systems, pushing companies to look for technologies that reduce heat generation and energy use.

Despite growing investment activity, large-scale deployment of photonics technology still faces manufacturing challenges. Nick Patience, AI lead at Futurum Group, told CNBC that production scaling remains difficult because optical and silicon components require extremely precise assembly processes.

Shares of photonics-related companies have surged alongside Nvidia’s investment announcements. Lumentum shares have risen more than 130% since the beginning of the year, while Coherent and Marvell have also posted sharp gains amid increased investor interest in AI infrastructure suppliers.