Lightmatter, a start-up founded by three MIT alumni, is using photon technology to revolutionize chip communication and calculation. Limitations in the capacity for transistors on a chip have been seen as hurdles to the growth of modern computing. However, the company has rethought this process, utilizing light for data processing and transportation alongside electricity.
This method is applied in Lightmatter’s two products: a chip specialized in artificial intelligence operations, and an interconnect for efficient data transmission between chips. “The two problems we are solving are: How do chips talk, and How do you do these AI calculations,” explains Nicholas Harris, Lightmatter’s co-founder and CEO.
Lightmatter raised just over $300 million in 2023, valuing the company at $1.2 billion. It is now demonstrating its technology to some of the world’s leading tech firms, aiming to reduce the considerable energy demand of data centers and AI models. According to Harris, they aim to “enable platforms on top of our interconnect technology that are made up of hundreds of thousands of next-generation compute units.”
The company’s journey originates from Harris’s time at Micron Technology, where he observed the stalling of traditional computing advancement methods. He recognized photonics and quantum computing as potential for future growth, leading to his work on photonic quantum computing at MIT. This research resulted in integrated photonic chips that processed information using light, not electricity, leading to dozens of patents and many high-profile research papers.
The start-up’s first product, the Envise chip, combines electron-based memory with photon-based calculations, which allows for multiple simultaneous computations using different color lights. The second product, Passage, uses light’s latency and bandwidth benefits to link processors, similar to fiber optic cables, and facilitates chips as large as entire wafers to function as a single processor.
Both products aim to increase the energy efficiency of computing to meet the growing demand without a massive increase in power consumption. Harris points out the urgency of the issue, stating that “By 2040, some predict that around 80 percent of all energy usage on the planet will be devoted to data centers and computing, and AI is going to be a huge fraction of that.”
Source: Startup accelerates progress toward light-speed computing.
