Lightmatter, a photonic computing startup, secured $154 million in new funding as it makes its debut in the rapidly growing AI computation market. Lightmatter’s hardware-software combination solves computational processes like matrix vector products through the use of optical flow, using arrays of microscopic optical waveguides to allow the light to perform logic operations. It is potentially faster and more efficient than traditional silicon chips, which are approaching the limits of speed and density. Lightmatter’s approach is a potential new solution to the escalating power consumption and wastage of heat generated by current systems. It aims to offer speed-ups and efficiency improvements by a magnitude, rather than incremental percentages, and could reduce the cost and unwieldiness of developing AI.
Lightmatter CEO and founder, Nick Harris, started the company based on research on optical computing that he and his team did at MIT and secured $11 million in seed funding in 2018. The latest round of funding comes from several investors, including GV, SIP Global, and HPE Pathfinder. With Lightmatter’s software stack, Envise (computing hardware), Passage (interconnect), and Idiom (software platform), the workflow for machine learning developers should remain the same as the neural networks build in industry-standard applications and import libraries. Mass production is planned for 2024.
Lightmatter’s photonic AI hardware is an alternative to traditional computing that uses optical flow. The company uses microscopic optical waveguides to allow the chips to perform logic operations when light passes through them. Prioritized speed-ups and efficiency improvements make Lightmatter’s approach a potential new solution to escalating power consumption and heat generation issues current systems experience.
Nick Harris is the founder and CEO of Lightmatter. Harris and his team focused on optical computing research at MIT before Harris established the computing startup.
