Nvidia is placing a $4 billion bet on the future of computing, splitting the investment equally between Lumentum and Coherent—two companies pioneering photonics technology that could replace traditional silicon chips in the next generation of AI systems.
The move signals that even Nvidia, the undisputed king of AI chips with a market cap exceeding $3 trillion, recognizes the physical limits of Moore's Law are approaching. Traditional chip manufacturing is hitting walls at the 3-nanometer node, and photonics—using light instead of electricity to transmit data—represents the most promising path forward.
Here's what photonics does: instead of pushing electrons through silicon circuits, these chips use photons (light particles) traveling through optical waveguides. Light moves faster, generates less heat, and consumes dramatically less power—critical advantages when you're building data centers that already consume as much electricity as small countries.
The numbers make sense. Nvidia's current H100 GPUs generate roughly 700 watts of heat per chip. Data centers packed with thousands of these chips spend nearly as much on cooling as they do on electricity. Photonic interconnects could reduce power consumption by 50-70% while increasing bandwidth by an order of magnitude.
Jensen Huang, Nvidia's CEO, has been publicly discussing the "end of Moore's Law" for two years. This investment puts money behind those words. Lumentum specializes in optical transceivers—the components that convert electrical signals to optical ones and back. Coherent focuses on indium phosphide substrates, the material foundation for photonic chips.
By investing in both companies, Nvidia is securing access to the entire photonics supply chain. It's the same vertical integration strategy the company used when it bought Mellanox for $7 billion in 2020 to control networking technology. That acquisition now looks prescient—Mellanox's InfiniBand technology is essential to every major AI cluster.
The strategic calculus is straightforward: Nvidia generates roughly . If photonics extends the AI boom by even three years, this $4 billion investment pays for itself many times over. If photonics becomes the dominant technology, Nvidia locks in its position as the infrastructure provider for the next computing era.
