Bay Street Wire
Tech & BusinessOpinion

The Quiet Migration: Why Boson AI is Betting on Toronto Over the Valley

Portrait of Priya Raman
Priya RamanToronto startups & VCNov 11AI

While the world watches the hype in California, Boson AI's expansion into Canada reveals a strategic shift toward predictability, talent accessibility, and sustainable growth.

In the high-stakes race to build foundation models for voice agents, the battle isn't just about who has the most GPUs—it's about where the people who build them actually want to live and work.

While Silicon Valley remains the default backdrop for AI hype, Boson AI is proving that the real strategic advantage right now might be found in Toronto. As BetaKit first reported, the well-funded research lab has quietly shifted its center of gravity northward. Since late 2023, Boson AI has grown a Toronto office that now houses roughly 20 of its 40 total employees, meaning half of the company's workforce is now based in Canada.

For Boson AI co-founder and CEO Alex Smola, the move wasn't just about geography; it was a calculated "Plan B." Speaking with BetaKit, Smola, a former vice-president and distinguished scientist at Amazon Web Services, noted that the company sought predictability in the face of erratic US policy. With the implementation of steep tariffs and drastic immigration restrictions under US President Donald Trump, Smola argues that Canada has become a far more stable environment for conducting business.

Opinion: In my view, this is the ultimate signal that the "Valley burnout" is becoming a liability. When a founder admits that the cost of living and high turnover in Silicon Valley make it difficult to scale, they are describing a systemic failure of the California ecosystem. Toronto isn't just a cheaper alternative; it's a talent sanctuary.

This is most evident in the war for talent. Smola told BetaKit that hiring AI experts from abroad has become "very difficult" in California due to work permit challenges. In contrast, Canada has offered a smoother path for recruitment. Smola highlighted that the Toronto office currently employs two Iranian staff members who are unable to travel to the US. The situation has become so skewed that Smola revealed he didn't even open a recent machine learning engineer position in the US, citing the prohibitive cost of living and immigration hurdles.

Beyond the people, the infrastructure is fueling the pivot. Boson AI has moved away from the "not sustainable" costs of big cloud providers, instead purchasing the majority of its computing power from eStruxture’s data centre in Barrie, Ontario. Smola noted that eStruxture was "just a lot nicer" to work with than a competing US provider that attempted to increase quotes and add extra fees.

With $70 million USD ($98 million CAD) in funding to challenge giants like Meta and OpenAI, Boson AI is leveraging Toronto's proximity to the University of Toronto's Master of Science in Applied Computing Program to source interns and elite researchers.

However, the window of opportunity may not stay open indefinitely. Smola warned BetaKit that immigration advisers have already indicated that requirements for work permits in Canada are increasing. While Canada currently presents a "massive" opportunity to recruit talent that would otherwise head to the US, the sustainability of this growth depends on whether Canadian authorities can maintain an environment that attracts and retains global talent.

Sources

More from Priya Raman