Alphabet’s venture arm GV has led an investment round in a Boston-based tech start-up Lightmatter that use light beams on chips to boost AI speed.
Instead of using common multiplier-accumulator or MAC unit, the start-up, in order to avoid speed limits that today’s chips have been facing, uses an optical component in their chips called Mach-Zehnder interferometer.
It will be the first time when Lightmatter succeeded to get backing from a corporate investor like GV, which also highlights the possibility of hardware that have a potential to develop super-fast AI, and this characteristic of the technology have been noticed not only by technology firms but also other industries like healthcare and retail.
The idea remained under the doubts by investors when they first came to know about our idea and one had asked me to prove that it’s not merely a science project and can actually work, Lightmatter’s co-founder and chief operating officer, Thomas Graham, said, recalling the sentiment. Investment by one of the top technology companies in the world has now validated the worth of our team, he added.
AI is an advance technology which requires huge processing powers to run the features which consumer uses like voice recognition or creation of a playlist. And transistor-based technology currently being in use is lacking in capabilities to handle the requirements of AI to be operated at higher data scales as it consumes a lot of energy as well as much of a time to process that learning process of AI using the existing computer processors.
And Lightmatter’s computer chips are based upon optical technology which solves this limitation issue by utilizing the properties of light to perform the calculation task at significant speed to handle faster and more energy efficient computers which will be part of next generation of AI.
Founded in 2017, the start-up raised $22 million in Series A-1 funding round, to which the existing investors Spark capital and Matrix Partners also participated and the whole Series A came to be raising a total of $33 million.