Photonics, the technology that transmits data using light instead of electricity, is becoming essential to scaling AI data centers as they hit the limits of traditional copper wiring. The shift represents a fundamental reimagining of how AI infrastructure works, driven by the explosive power demands of training and running large language models (LLMs). Major players like Nvidia are betting billions that optical networking will solve the energy efficiency and heat dissipation problems that threaten to constrain AI growth.

Why Are Data Centers Switching From Copper to Light?

Traditional copper wire connections in data centers face a critical bottleneck. As AI models grow larger and require more computing power, the physical limitations of electrical signals become a problem. Copper cables generate excessive heat, suffer from signal loss over distance, and cannot keep pace with the bandwidth demands of modern AI systems. Photonics solves these problems by using fiber-optic systems to transmit data as light pulses instead of electrical signals.

The infrastructure components that make photonics work include lasers, optical fibers, modulators, and various connectors that work together to build enhanced fiber-optic networks. These systems are faster, more energy efficient, and allow data transmission over much greater distances than copper alternatives. For AI companies racing to build "gigawatt-scale AI factories," as Nvidia CEO Jensen Huang described them, photonics represents the key to scaling without hitting a wall of unsustainable power consumption.

How Is the Photonics Market Growing?

The investment opportunity in photonics has caught the attention of Wall Street. Goldman Sachs projects that the total addressable market for optical networking will expand dramatically, growing from an estimated $15 billion in 2026 to approximately $154 billion by 2028, a ninefold increase. This explosive growth projection has sparked a wave of new investment vehicles designed to capture the trend.

Roundhill Investments launched its Photonics and Optics ETF (LYTE) on August 5, 2026, becoming the second fund manager to bring a photonics-focused investment product to market. The move follows the success of Roundhill's Memory ETF (DRAM), which accumulated nearly $25 billion in assets since launching in April 2026. Tema ETFs had already launched the LAZR Photonics and Optics ETF on June 30, 2026, though that fund has attracted only approximately $18 million in assets to date.

Publicly traded photonics companies have seen dramatic stock price increases. Lumentum shares have risen over 100% in 2026, while Coherent is up over 66%, and Nlight has climbed roughly 100%. These gains reflect investor confidence that photonics will become a critical component of AI infrastructure.

What Role Is Nvidia Playing in the Photonics Shift?

Nvidia has emerged as the largest corporate investor in photonics technology, committing at least $6.5 billion to companies developing optical systems since March 2026. The chipmaker has taken stakes in major photonics firms including Coherent and Lumentum, and invested $2 billion in Marvell Technology, which designs chips that allow light to be used in place of electricity at the microchip level.

Nvidia's most ambitious photonics initiative is Feynman, a major GPU microarchitecture redesign scheduled for release in 2028. Feynman is designed to move away from copper-based connections entirely and embrace photonics for interconnectivity and networking within data centers. By Feynman's launch, system bandwidth requirements are expected to rise to more than four times current levels, with light moving deeper into the system to connect more than 1,000 GPUs simultaneously. At that scale, copper connections may struggle to deliver the bandwidth, distance, and power efficiency required.

"Photonics represents a way for Nvidia to scale their AI infrastructure without the energy costs that staying with electrical and copper will incur," explained Alvin Nguyen, senior analyst at Forrester. "By investing in photonics companies, Nvidia is making sure that advancements in photonics continue, and it will prevent them from hitting a scalability and performance wall that will occur if they remain on electrical and copper."

Alvin Nguyen, Senior Analyst at Forrester

How Are AI Companies Building Photonics-Powered Data Centers?

The photonics boom is not just theoretical; major AI infrastructure projects are already being built with optical networking in mind. Anthropic, the AI lab behind Claude, recently signed a $10 billion deal with Volta, an AI cloud startup, to secure compute capacity over six years. The facility will be located in Norway and will deliver 133 megawatts of capacity, powered by Nvidia's Vera Rubin GPU systems.

Volta is partnering with Bitdeer, a company with deep experience in data center operations, to develop the Norwegian facility. Bitdeer separately announced a $4.7 billion, 16-year lease agreement for its Tydal data center campus in Norway, which will deliver 121 IT megawatts of capacity configured entirely for Nvidia GPUs. The Tydal facility is designed to be among Norway's largest and most efficient AI data centers, with a Power Usage Effectiveness (PUE) rating of approximately 1.1 and powered entirely by renewable energy.

Steps to Understanding the Photonics Investment Opportunity

• Market Size Growth: The optical networking market is projected to grow from $15 billion in 2026 to $154 billion by 2028, representing a ninefold expansion driven by AI data center demand.
• Technology Advantage: Photonics systems transmit data using light pulses through fiber-optic cables instead of electrical signals, offering faster speeds, greater energy efficiency, and longer transmission distances than copper alternatives.
• Corporate Investment: Nvidia has committed at least $6.5 billion to photonics companies since March 2026, including major stakes in Coherent, Lumentum, and Marvell Technology, signaling confidence in the technology's future.
• Stock Performance: Publicly traded photonics companies have seen significant gains, with Lumentum up over 100%, Coherent up over 66%, and Nlight up roughly 100% in 2026.
• Infrastructure Deployment: Major AI labs are already building data centers designed around photonics technology, with projects like Volta's Norwegian facility incorporating Nvidia's latest GPU systems and renewable energy sources.

What Are the Risks to the Photonics Boom?

Despite the bullish outlook, photonics investments carry significant risks. The entire market growth is predicated on the AI buildout continuing without major disruptions or setbacks. A slowdown in AI adoption, regulatory changes, or unexpected technological breakthroughs could materially reduce demand for photonics infrastructure.

The LAZR photonics ETF has experienced limited investor interest since its June 2026 launch, accumulating only $18 million in assets despite the broader enthusiasm for the sector. This suggests that retail investors may be cautious about photonics as an investment theme, or that the market is still too nascent to attract significant capital flows.

"The development and commercialization of photonic and optical technologies may be contingent upon the broader adoption of related technologies, including artificial intelligence, data center infrastructure, and advanced telecommunications networks," noted Roundhill Investments in its prospectus filing for the LYTE ETF. "A slowdown or disruption in the growth of these related sectors could materially and adversely affect the revenues and profitability of Photonic and Optical Companies in which the Fund invests."

Roundhill Investments, LYTE ETF Prospectus

Despite these risks, industry experts view photonics as a sufficiently developed and distinct investment theme. The technology is no longer speculative; it is being integrated into real infrastructure projects backed by major financial institutions and deployed by leading AI companies. As data centers continue to consume more power and face mounting pressure to improve efficiency, the shift from copper to light appears not just likely, but inevitable.