An AI hardware unicorn is alighting in Toronto, looking to hire talent in a city whose researchers and startups are advancing both fields. Here’s what you need to know.

The firm: Boston-headquartered Lightmatter has developed software, servers and  chip-connectors it says require less power than conventional systems for tasks like training AI models or running machine-learning applications. Instead of using wires to move electrons, the firm uses photonics—optic fibers that carry particles of light.

On Tuesday, Lightmatter announced it had raised US$155-million in investment in a Series C-2 funding round co-led by GV, an arm of tech giant Alphabet, and asset manager Viking Global Investors. The investment valued Lightmatter at US$1.2 billion. 

The city: With the new cash in hand, 150-person Lightmatter is looking to add about 75 more to its headcount in 2024. The firm plans to set up a Canadian subsidiary and hire about 20 workers in Toronto in the first half of next year, said Beth Keil, vice-president of people; it’s already got two employees there. Over time, “we’re seriously considering building out more of the team in Canada.”

Keil cited as draws the strength of Toronto’s tech startup and research ecosystems, and labour costs that are lower than in U.S. hubs. Lightmatter is already in talks with some photonics engineers in the city, and will recruit technical staff with circuit and design expertise there. “There’s a lot of other big tech companies that are in the Greater Toronto Area, including some AI research labs,” Keil said, noting that it’s easier to bring skilled foreign workers into Canada than the U.S.

Lightmatter will also look to build research partnerships with the University of Toronto and University of Waterloo, as well as hire interns from the schools. And the firm will seek to take advantage of government subsidies like the scientific research and experimental development (SR&ED) tax incentive program.

The challenge: Lightmatter isn’t the only startup promising a full-on photonic supercomputer that can do AI faster with less energy. But the company and its peers are still proving that their products have practical uses for which buyers will pay. (Companies like Ottawa-based Ranovus are successfully selling photonics-based components into existing data-centre hardware.)

Many Canadian tech executives won’t welcome another deep-pocketed foreign arrival poaching skilled workers, getting government incentives and using both to generate IP that will benefit shareholders across the border. But Keil hopes Lightmatter’s prospects will help draw people to it. “What we’re doing will help make things go faster, but also run cooler.”