Nvidia pours $6.5 billion into photonics to solve AI's copper bottleneck problem

The Tech Download: How chip companies are looking to use light to solve this major AI bottleneck

The AI boom is in many ways a hype cycle like no other. Sure, there are comparisons to draw between the dotcom surge of the late 90s and the mobile revolution of the noughties, but in terms of capital invested and lofty predictions on it causing huge societal shifts, it stands ahead of the rest. The speed of that progress comes with big hurdles. AI builders are having to grapple with constraints like access to the energy that will power the huge data centers, a memory chip crunch and, increasingly, the efficiency of transferring data between AI chips and systems. An emerging technology, known as photonics, offers a route to solving for the latter. Photonics can be used in AI infrastructure by using light to move data between graphics processing units (GPUs), memory, networking chips, servers and data centers, instead of relying on electrical signals running along copper.Some photonics tech is already in use, including in fiber optic connectivity. But much of the connectivity inside AI servers and racks currently travels along copper wires, limiting speed and increasing energy costs. "One of the main bottlenecks for the performance of AI models is the speed of communication between chips and between chip servers," said Gil Luria, head of technology research at D.A. Davidson. "The faster the communication, the faster the user can get their answer or their task executed," he added. "By moving the connections between chips and between servers to optical, the performance of the models can improve significantly."

Nvidia spends $6.5B on photonics to fix AI's copper bottleneck

Nvidia has committed at least $6.5 billion to photonics companies since the beginning of March, making it the largest single investor in the technology that many in the industry believe will replace copper wiring as the backbone of AI data centres. The spending spree reflects a calculation that copper, the standard medium for moving data between chips, is approaching its physical limits just as AI training clusters are demanding exponentially more bandwidth. Photonics uses light rather than electrical signals to transmit data. It offers substantially higher bandwidth at lower power consumption, two constraints that become critical when thousands of GPUs need to operate as a unified system. The problem is that the photonics supply chain is not yet built to the scale AI infrastructure requires. Nvidia's broader investment strategy in 2026, which now exceeds $40 billion across AI equity bets, is designed to fix that. Where the $6.5 billion went The bulk of the spending went to three established optical component makers. Nvidia invested $2 billion each in Coherent and Lumentum in early March, with both deals including multi-billion-dollar purchase commitments and funding for new US fabrication capacity. A further $2 billion went to Marvell, which acquired photonics startup Celestial AI in December 2025 and is developing silicon photonics for AI networking. Nvidia then invested up to $3.2 billion in Corning, the glass and fibre optic manufacturer, through a combination of $500 million in equity warrants and multi-year purchase agreements. Corning will use the funding to increase its US-based optical connectivity manufacturing capacity by 10 times, expand fibre production by more than 50%, and build three new advanced manufacturing plants in North Carolina and Texas, creating more than 3,000 jobs. Nvidia also participated in Ayar Labs' $500 million Series E alongside AMD and MediaTek, valuing the co-packaged optics startup at $3.75 billion. Ayar Labs develops silicon photonics chiplets that can be integrated directly with processors, a technology called co-packaged optics that represents the next step beyond the discrete optical modules the larger deals target. Why copper cannot keep up The core problem is physics. Copper interconnects lose signal integrity and consume more power as data rates increase. Inside a single rack of GPUs, copper can still handle the bandwidth at acceptable power levels. But when AI training clusters span multiple racks, which they increasingly must, the distance between chips exceeds what copper can serve efficiently. Nvidia's next-generation Vera Rubin platform illustrates the split. The Vera Rubin Ultra NVL576, a 576-GPU supercomputer spanning eight racks, uses copper within each rack and optical interconnects between racks. Jensen Huang has called the platform the largest product launch in Taiwan's history, with each system containing nearly 2 million parts built through 150 ecosystem partners on the island. The transition from copper to optics is not a future event. Nvidia launched its Quantum-X and Spectrum-X Photonics platforms in March 2025, the first commercial-grade co-packaged optics networking switches, built with TSMC, Coherent, Lumentum, Corning, and Foxconn. The $6.5 billion in investments is designed to ensure the supply chain can produce these components at the volumes Vera Rubin will require. A supply chain Nvidia is trying to lock up The scale of Nvidia's photonics spending has raised concerns among competitors. TechTimes reported that Nvidia's purchase commitments to Coherent and Lumentum could effectively lock up the global supply of high-end laser components through 2027, pushing rival chipmakers and data centre operators to the back of the queue. AMD and MediaTek have responded by co-investing in Ayar Labs, but neither has matched the scale of Nvidia's photonics commitment. The investments also carry geopolitical weight. Huang has said that Chinese competitors running frontier AI on Huawei chips would be a damaging outcome for the US, and securing domestic photonics manufacturing is part of the same strategic logic. Other companies in the space include Lightmatter, valued at $4.4 billion, which is developing a 3D-stacked silicon photonics engine called Passage. Its L20 module, announced in March, achieves 6.4 terabits per second in each direction and is expected to begin sampling in late 2026. Broadcom, Intel, and Cisco are also developing optical interconnect products, but none has made the kind of ecosystem-level investment Nvidia has. The financial context Nvidia reported first-quarter revenue of $44.1 billion and guided to $91 billion for the second quarter, authorising another $80 billion in share buybacks. The company's market capitalisation stands at roughly $4 trillion. The $6.5 billion it has spent on photonics in three months is a rounding error on its balance sheet, but it represents a substantial fraction of the entire photonics industry's annual revenue. The pattern across Nvidia's investments is consistent. Capital flows to companies that either build the components Nvidia needs or buy Nvidia GPUs at scale. The photonics deals follow the same logic, securing supply of a technology that will determine whether Nvidia's next generation of AI platforms can ship on time and at scale. If copper is the bottleneck, and the physics says it is, then the company that controls the photonics supply chain controls the pace of AI infrastructure deployment. That is the bet Nvidia is making with $6.5 billion of its cash.

Nvidia is investing billions into this emerging technology that could change the AI industry

Nvidia has committed at least $6.5 billion into companies developing photonics technology in the past three months, as the company races to invest in solving one of the major bottlenecks to the rollout of AI. Photonics, the use of light to transmit data, is an emerging technology considered to be a more efficient alternative to the current process of transferring data using electricity. Electrical data transfer consumes more energy -- a factor which is increasingly seen as a blocker to the broader deployment of AI. Since the beginning of March, Nvidia has announced $2 billion investments into Lumentum, Coherent and Marvell, all of which are developing photonics tech. The chip giant also said it would invest $500 million into Corning to develop advanced optical connectivity solutions, and participated in optics startup Ayer Labs' $500 million Series E funding round. "Photonics represents a way for Nvidia to scale their AI infrastructure without the energy costs that staying with electrical and copper will incur," Alvin Nguyen, senior analyst at Forrester, told CNBC. "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." Photonics can be used in AI infrastructure by using light to move data between graphics processing units (GPUs), memory, networking chips, servers and data centers, instead of relying only on electrical signals running along copper. While copper is the main connectivity standard today because of its lower cost and high reliability, photonics will become more prominent in AI infrastructure over time, Brian Colello, senior equity analyst at Morningstar, told CNBC. "Nvidia's roadmap of next generation AI rack-scale solutions will require an increasing amount of optical connectivity to process the exponentially rising bandwidth with new models and higher usage," he said. The chip giant has already made some photonics tech available as part of its networking solutions offering, with the company announcing tools that it said will enable AI factories to connect millions of GPUs across sites while drastically reducing energy consumption and operational costs. "When you look upstream, you come to the conclusion that we're starting to scale our silicon photonics technology," Nvidia CEO Jensen Huang said at the GTC in March, pointing to Nvidia's ethernet networking platform used to connect AI factories and GPU clusters. He also said the company was starting to add photonics to its GPU-to-GPU interconnect technology. "Which means the amount of silicon photonics technology capacity that we need is substantially higher than the world has today," he added. "So we work with the supply chain to make sure we can help them build up the capacity in advance of that." Lumentum's stock has risen 134% since the start of the year, while Coherent is up 96%. Marvell has seen its shares increase by 122% in 2026 and Corning 111%. Nvidia is one of the many AI stakeholders recently making the move to funnel cash into photonics tech. Chipmaker AMD joined Nvidia in the Ayer Labs round, as well as acquiring startup Enosemi in 2025, alongside making equity investments in Teramount and Celestial AI. Alphabet and Microsoft venture arms backed nEye in an $80 million Series C in April. But deploying photonics tech across the AI infrastructure stack at scale comes with its own challenges. "The technology is sound, production scale is the harder problem," Nick Patience, AI lead at the Futurum Group, told CNBC. "Manufacturing yield on complex co-packaged optical assemblies remains a challenge because precise alignment of optical and silicon components is unforgiving, and when something goes wrong in the packaging process, the assembly typically can't be reworked," he said. "So the transition is underway, but it's still early," Patience added. "I would expect us to see large-scale adoption from 2028 onwards." Choose CNBC as your preferred source on Google and never miss a moment from the most trusted name in business news.

A bright future: Why photonics is the new stockmarket darling

Much like the transition from ADSL to fiber optics in everyday use, artificial intelligence is currently undergoing a similar transformation, but within its very infrastructure. It is no longer just the chips that matter, but rather the interconnects that link them. When the network becomes the bottleneck With the surge in AI models, the volumes of data to be processed and exchanged are exploding. These flows must circulate ever faster, with minimal loss and controlled energy consumption. However, in many cases, these exchanges still rely on electrical signals traveling through copper cables. At ultra-high speeds, typically between 400 and 800 Gbits/s, these signals reach their physical limits. They degrade, generate heat, and ultimately slow down the entire system. The problem is now clearly identified: the network has become a bottleneck, sometimes even more restrictive than the GPUs themselves. Silicon photonics: A key technological response Faced with this constraint, a solution is gradually gaining ground: replacing electrical signals with light signals. This is the principle of silicon photonics, often referred to by the acronym SiPh. The concept is simple on paper: using light instead of electricity to transport information. In practice, the benefits are substantial. Data transfers become significantly faster, signal losses are reduced, and energy efficiency improves markedly. This is a crucial point at a time when data centers are showing titanic levels of consumption. Integration ever closer to the processors Beyond the nature of the signal, the challenge also lies in how these technologies are integrated around the processor, a process known as packaging. Several approaches coexist, notably CPO (Co-Packaged Optics) and NPO or OBO (Near-Packaged Optics/On-Board Optics) architectures. All pursue a common goal: to bring optical components as close as possible to the processors. By reducing the distance between computation and transmission, these solutions limit losses, decrease heat, and improve the overall performance of the systems. These innovations are shaping the future of ultra-high-speed computing, making it more efficient and better suited to the demands of modern artificial intelligence. An industrial supply chain already in place This transformation is accompanied by a massive mobilization of technology players. Nvidia, at the forefront of AI, is investing heavily in photonics, both internally and through strategic partnerships with specialists such as Lumentum and Coherent, both recognized for their laser technologies. Marvell is following the same momentum with the acquisition of Celestial AI, a player entirely dedicated to optics for artificial intelligence. Meanwhile, Intel is already mass-producing 400G optical chips for data centers, while Cisco provides a large portion of the 800G optical modules used by hyperscalers. Broadcom integrates these technologies into its network switches, notably the Tomahawk range, which is widely deployed in infrastructure. Other players play a key role in the value chain: GlobalFoundries manufactures photonic chips for several companies, while Ayar Labs develops optical chiplets like TeraPHY, which can be directly integrated into processors. IBM is also contributing with components dedicated to short optical links, essential in CPO architectures, while STMicroelectronics is exploring applications ranging to optical sensors for autonomous vehicles. Europe also in the race While American giants dominate the landscape, several European companies are holding their own. STMicroelectronics develops and industrializes silicon photonics platforms for high-speed interconnections. Soitec provides the substrates essential for these circuits, while Nokia designs coherent optical solutions integrating these technologies. Toward a new technological race After years focused on computing power, the industry is entering a new phase. Competition is no longer played out solely on chip performance, but on the speed and efficiency of the connections that link them.