OAKLAND, Calif., May 31, 2010 (FBC) — Boston-based Lightmatter for AI computing raised $154 million on Wednesday and said it has tripled its valuation as customers buy faster and more energy-efficient computers for AI work. His systems.
“We use light to connect computer chips together, and we use computations for deep learning,” said Nick Harris, founder and CEO of Lightmatter. “The reason we’re getting these customers and data center-scale deployments with our relationship is because the generative AI boom is driving high-end chips like crazy.”
Light has been used in the telecommunications industry for decades to move data around the world over fiber optic cables. Now, large technology companies are looking to use light-based technology in their data centers to reduce energy costs. The challenge is to reduce the size of the devices used to create or control light.
The more advanced the chips are, the more power they consume. “Getting the data center’s power to the chips can be about half of the total operating cost of the data center,” Harris said, noting that light transmits data more efficiently than electrical signals traveling over wires.
The chip, called Passage, which connects other chips to photons and back, will be mass produced and used in data centers next year, Harris said. A big part of the money will help to release this product.
A more futuristic product, Invis, is a system that uses light instead of electrons to make a computer. Harris says Invis can build large AI models, but not train them. This year, customers said they will test the system.
GV (Google Ventures), the venture capital investment arm of Alphabet Inc ( GOOGL.O ), and Fidelity Management & Research Company are among the investors that participated in this latest round, the company said.
Reporting by Jane Lanhee Lee; Editing by Lisa Shumaker
Our Standards: The Thomson Reuters Trust Principles.