Recently, the U.S. photonic computing startup Lightmatter announced that it has raised $155 million in Series C-2 financing to accelerate the team's growth and expand the deployment of photonic chips, and to meet the growing demand for artificial intelligence and high-performance computing.
Lightmatter has raised $420 million to date and is currently valued at over $1.2 billion. This round will drive accelerated growth for the company to meet the growing demand for high performance computing (HPC) from AI innovators.
Accelerating growth! The photonic computing "unicorn" raises $155 million in new funding.
Passage testbed: an 8" x 8" computer chip that allows heterogeneous chips to communicate optically via programmable integrated photonic links. (Image credit: Lightmatter)
The Boston-based company plans to use the funding to increase its headcount (it currently has about 150 employees) and open a new office in Toronto.Lightmatter plans to expand its global team and offices, as well as appoint Danner Stodolsky as VP of Data Center Architecture and Colin Sturt as general counsel.
Specifically, Lightmatter said Tuesday local time that it successfully raised $155 million in a deal led by GV (formerly Google Ventures) and Viking Global Investors. The round is a continuation of a $154 million Series C round raised in May this year.
"Lightmatter is positioned as a key driver to advance the next generation of computing systems, which will further drive AI innovation. Through photonic technology, Lightmatter is ensuring a steady progression in computing performance, despite increasing power consumption challenges and slow progress in transistor scaling."
Lightmatter builds novel ways for computer chips to communicate with each other and accomplish computations. Its technology uses silicon photonics and waveguides, which are tiny structures that change the direction of light. And in the past, chips typically used wires to do these things.
The idea of using light in computing isn't new, and it can speed up the process while consuming less energy. But building the components needed to make it work has long been tricky.
Lightmatter sees recent advances in AI as an advantage: tech companies have been bringing high-powered chips to market that can meet the computational demands needed for tasks such as training large language models like those of OpenAI's ChatGPT chatbot, as well as a growing number of competitors.
Nick Harris, co-founder and CEO of Lightmatter, said, "With AI driving so much progress in computing right now, it's important to find new ways to continue to improve the speed and energy efficiency needed for these systems."
The company's products include a device called Passage, which can: hold a variety of different types of computer chips, allowing them to share data optically. The fibers also allow multiple channels to be connected together, building them into supercomputers - a capability the company is working with a number of customers to develop.Harris wouldn't name specific customers, but said they include some of the largest chipmakers and cloud service providers.
Harris said such supercomputers can train powerful AI models faster and with less power consumption than traditional methods.
Erik Nordlander, GV general partner, said, "In the rapidly evolving field of generative AI, the need for new computing and on-chip communication solutions has never been greater.Lightmatter is harnessing the power of silicon photonics to meet this challenge, unlocking performance bottlenecks, increasing bandwidth, and allowing AI models of size and scale to increase dramatically."
