Nicholas HarrisCo-founder & CEO

Lightmatter was founded in 2017 after recognizing that Moore's Law and Dennard scaling were no longer delivering the improvements needed for future computing. The original vision was to accelerate AI computation using light instead of electrical signals.

As transformer-based AI models became dominant, the main challenge moved from computation to moving massive amounts of data efficiently. The company pivoted toward solving networking bottlenecks with silicon photonics.

The immediate priority is delivering production chips for customers while scaling manufacturing and preparing silicon photonics for high-volume deployment.

Unlike conventional semiconductors, photonic chips combine optical and electronic components. Testing and validating both systems simultaneously requires entirely new manufacturing processes.

The physical behavior of light naturally performs mathematical operations like addition and multiplication, which are fundamental to neural networks. This enables significantly lower energy consumption compared to traditional electronics.

Modern AI systems rely on thousands of processors working together. Efficient communication between chips is now as important as the chips themselves, making networking a critical performance bottleneck.

The company works with hyperscale cloud providers, GPU manufacturers, and custom AI chip developers that need advanced optical interconnect technology for large AI clusters.

Passage is Lightmatter's photonic interconnect platform that integrates optics directly into AI hardware, enabling processors to communicate at extremely high bandwidth.

A scale-up domain is a tightly connected group of AI processors that communicate fast enough to function almost like a single chip, improving training and inference performance.

Photonic communication depends on laser technology. Lightmatter's GUIDE laser platform provides the optical source required for high-speed photonic links.

As AI clusters continue growing, networking consumes an increasing share of system cost and power. Photonic networking becomes increasingly valuable as systems scale.

He expects silicon photonics to become mainstream inside AI infrastructure, with many leading AI models running on systems connected by Lightmatter's optical technology.