About Lightmatter
Lightmatter was founded in 2017 by Nicholas Harris, Darius Bunandar, and Thomas Graham, MIT researchers pursuing a fundamentally different approach to computing hardware for AI workloads, using photonics, meaning computing and data transmission using light rather than traditional electrical signals, based on the argument that photonic computing can offer significant advantages in both computational speed and energy efficiency for certain AI workloads compared to conventional electronic chips, since light-based computation and data transmission can, in principle, move information with less energy loss and heat generation than pushing electrons through traditional electrical circuits, an increasingly significant consideration as AI data centers have become substantial consumers of electrical power.
The company has pursued two related but distinct photonic technology applications: photonic computing chips that perform certain AI computations, particularly the matrix multiplication operations that are computationally central to how neural networks process information, using light-based rather than purely electronic computation, and photonic interconnect technology that uses light to move data between chips within an AI data center much faster and more energy-efficiently than traditional electrical interconnects, addressing the same kind of data movement bottleneck between AI accelerator chips that has become an increasingly important constraint as AI computing clusters scale to enormous numbers of interconnected processors. Lightmatter's photonic interconnect products, in particular, have found commercial traction as a complementary technology that can be integrated alongside conventional electronic AI accelerator chips from companies like Nvidia and AMD, addressing the data movement and communication bottleneck between chips rather than replacing the core electronic computation itself. The company has positioned its photonic technology as an important architectural innovation for continuing to scale AI computing infrastructure given the physical and energy efficiency limits that purely electronic computing and data movement approaches increasingly face at very large scale. Lightmatter has raised very substantial venture funding from investors including Google Ventures, Fidelity, and HP Enterprise, reaching a valuation exceeding $4 billion.
