Nvidia intensively invests in photonic technology: at least $6.5 billion spent in three months

```

Nvidia is betting on photonic technology at an unprecedented pace, seeking to fundamentally solve the energy consumption bottleneck that restricts large-scale AI deployment.

Since March this year, Nvidia has pledged to invest at least $6.5 billion in multiple photonic technology companies, making it one of the most proactive strategic buyers in this emerging field. Photonic technology uses light instead of electrical signals to transmit data, and is regarded by the industry as a key path to breaking through the bottleneck of AI infrastructure expansion.

Boosted by the intensive entry of tech giants like Nvidia, the share prices of related listed companies have generally doubled so far this year: Lumentum is up 134%, Marvell up 141%, Coherent up 104%, and Corning up 108%.

Nvidia CEO Jensen Huang stated clearly at the GTC conference in March: "The capacity of silicon photonics that we need far exceeds the current global supply level." He also revealed that Nvidia has begun integrating photonic technology into GPU interconnection solutions and is working with supply chain partners to proactively expand capacity.

Frequent Moves: $6.5 Billion Layout in the Photonics Track

Within three months, Nvidia has successively announced several major investments. According to CNBC, Nvidia invested $2 billion each in Lumentum, Coherent, and Marvell, all of which are deeply involved in photonic technology research and development; in addition, Nvidia invested $500 million in Corning to develop advanced optical connection solutions, and also participated in a $500 million Series E funding round for optical startup Ayer Labs.

Behind these moves is Nvidia’s strategic judgment on the path to expanding AI infrastructure. Alvin Nguyen, a senior analyst at Forrester, told CNBC: "Photonic technology provides Nvidia with a path to expand AI infrastructure without greatly increasing energy consumption. By investing in photonic companies, Nvidia ensures this technology continues to advance, thus avoiding hitting scalability and performance ceilings of electrical and copper cable solutions."

Photonic Technology: The Next-Generation Connectivity Standard for AI Infrastructure

The core application of photonic technology in AI infrastructure is to use optical signals instead of electrical signals over copper cables to transmit data between GPUs, memory, network chips, servers, and data centers. Compared to copper cables, photonic solutions consume less energy, but copper cables remain the mainstream standard for their cost advantage and high reliability.

Brian Colello, a senior equity analyst at Morningstar, told CNBC: "The roadmap for Nvidia's next-generation AI rack-level solutions will require increasing amounts of optical connectivity to handle the exponentially growing bandwidth demands brought by new models and higher usage."

Nvidia has already incorporated some photonic technologies into its networking solutions product line and announced the launch of related tools, claiming they can help AI factories connect millions of GPUs across sites while significantly reducing energy consumption and operating costs.

Nvidia is not fighting alone. Chipmaker AMD also participated in this round of financing for Ayer Labs and will acquire startup Enosemi in 2025, while also investing in Teramount and Celestial AI. In April this year, the venture capital arms of Alphabet and Microsoft jointly participated in nEye’s $80 million Series C funding round.

The concentrated entry of multiple tech giants reflects a high degree of market consensus on the strategic value of photonic technology—against the backdrop of explosive growth in AI computing power demand, energy consumption of data transmission has become an unavoidable systemic constraint.

Mass Production is Still the Core Challenge, Large-Scale Adoption May Have to Wait Until 2028

Despite strong capital enthusiasm, the large-scale commercialization of photonic technology still faces significant engineering obstacles. Nick Patience, head of AI at Futurum Group, pointed out: "The technology itself is feasible, but mass production is a much harder problem to solve."

He further explained that manufacturing yield rates of complex co-packaged optical components are a major challenge, because the precision alignment tolerance between optical and silicon-based components is very low—once there is a problem in the packaging process, components usually cannot be reworked or repaired.

"The transformation is underway, but still in the early stages," Patience said. "I expect large-scale adoption will gradually begin starting in 2028." This means that although Nvidia’s strategic layout is already in full swing, investors still need to be patient for investments to translate into actual capacity and products.

Risk Warning and DisclaimerThe market is risky, and investment needs to be cautious. This article does not constitute personal investment advice, nor does it take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Investment according to this is at your own risk. ```