NICHOLAS HARRIS, PH.D., FOUNDER & CEO, LIGHTMATTER

Listen on your favorite podcast platform:

Nick Harris has had only one job since getting his Ph.D. in electrical engineering. Ever since co-founding Lightmatter in 2017 using prize money from business plan contests, he, his co-founders Darius Bunandar and Thomas Graham and their teams have been working on establishing light as the preferred way to communicate and compute in data centers. Before his receiving his doctorate from MIT on a merit scholarship, he worked at Micron, the semiconductor memory manufacturer based in Idaho. It was at Micron that Nick realized that scaling the electrons-and-transistor-based approach to electronics was increasingly challenged.  

Lightmatter has raised about $800 million dollars, most recently in a 2024 round that valued the company at $4.4 billion. Investors include Google Ventures, Fidelity and T. Rowe Price. With the initial goal of optical computing, the company is now focused on using light to speed communication between datacanter components and systems. Two main products help customers with this task. “Guide” chips are light sources that easily integrate into modern server architectures. “Passage” interconnect platform chips that route the generated light. Of the company’s many differentiators, two stood out to me. First, the implementation of the products in standard semiconductor processes, which significantly enhances reliability and manufacturability, frequent tormentors of optical products. Second, 3D architectures that allow Lightmatter’s chips to get larger without the typical performance sacrifices of size.  

Nick's view is that photonics won't stop at speeding up how chips talk to each other. Rather, it’s a stepping stone back toward computing itself running on light. He believes that as AI workloads shift from being interconnect-bound back to compute-bound over the next 5 to 10 years, the same optical technology that powers Lightmatter's interconnects today could revive the case for photonic computation, building on the world's first photonic computer (Envise) that his team demonstrated running real AI workloads and video games, and which now sits in the Computer History Museum. In his framing, light is the next fundamental substrate for computing.