Build North
Subscribe
← All articles
AIOct 12, 20267 min read

Vancouver has the AI talent. Where are the AI-native products?

By Jatin Rajvanshi, product manager in Vancouver

Vancouver ranks among North America's top tech talent cities, yet its best AI product companies keep raising and incorporating elsewhere. Here's what's missing, and what product people can do about it.

Here's a pattern worth sitting with. A pair of McGill classmates start an AI agent tool as a side project in a Mount Pleasant condo. They get into Y Combinator, raise in the US, move to San Francisco, and in March 2026 close a US$50M Series B led by Benchmark. Then they open a Vancouver office, because, in the CEO's words, the talent here was too good to pass up.

That's Gumloop, and it's a happy story for Vancouver. It's also a clue. The city is good enough at producing AI builders that US-based companies come here to hire them. What it hasn't done yet is produce, and keep, the companies that employ them as their home base.

If you build products in Vancouver, this matters for your career, your next job and your next company. So let's work out why it happens.

The TL;DR:

The talent story is true, and it's also the easy part

The numbers are good. CBRE's recent reports put Vancouver among Canada's top three AI-specialty markets, alongside Toronto and Montréal, which together hold about 62% of the country's AI workforce in that edition. The same coverage says 44% of Vancouver's AI-skilled people work inside tech companies, among the strongest concentrations in North America, and that 71% of its software engineers work directly for tech companies.

Notice what that measures: engineers and ML specialists. People who can build. It says nothing about whether the company around them was founded here, is headquartered here, or makes the product decisions here.

Canada has been in this spot before. The Logic's 2023 series on AI talent noted Canada ranked first in the G7 in AI research publications per capita, and quoted Radical Ventures' Salim Teja warning that without enough domestic employers to absorb graduates, talent will leave. Three years on, the research engine still works. The employer question is still open.

The company is leaving earlier than the people are

Here's the more uncomfortable data. A Leaders Fund study, reported by The Globe and Mail, found that just 32.4% of Canadian-led "high-potential" startups launched in 2024 were headquartered in Canada, against more than 67% for the 2015 to 2019 classes. Nearly half of the 2024 class is headquartered in the US. ("High-potential" there means the company has raised at least US$1M and has mostly Canadian-educated senior leaders.) That's national, not Vancouver-specific, and we haven't found a city-level cut.

Gumloop's CEO was blunt about why. U.S. venture capital let them scale massively, he said. Our read: the move wasn't about talent, because they came back for talent. It was about the cheque size, the investors and the early customers.

That's the pattern. The people can be here. The company tends to be wherever the Series A lead sits.

AI-native product companies need things that talent alone doesn't buy

Here's our argument. A traditional software company can grow slowly out of a local market. An AI-native product company mostly can't, for three reasons.

1. The feedback loop is the product. AI products improve through fast iteration on real usage and evals. Early users who'll tolerate rough edges are concentrated in US tech companies. A small home market means a slower loop.

2. The cost curve is brutal. Inference, GPUs and model access eat margin early. Raising a large round fast matters more than it did for a typical SaaS startup.

3. The scarce skill is product judgment, not model skill. Knowing what to build with AI, how to evaluate it and when it's good enough to ship is learned by shipping. Mitacs-sponsored panellists at a Techcouver event argued that the Bay Area has kept experienced operators close to mentor younger talent better than Vancouver has. That's one side's view, from a talent non-profit's event, but it fits the pattern: plenty of strong engineers, fewer people who've run an AI product from zero to scale.

The Canadian angle: the incentives are better than the outcomes

Canada isn't short on help. SR&ED credits still make early R&D cheaper, IRAP supports technical work, and the dollar means two Toronto engineers can cost about the same as one in Silicon Valley, per a recruiter quoted in the 2023 Logic series. Vancouver's CBRE pitch includes the same value argument, and the city ties Waterloo for value to employers in the 2026 coverage.

But cheap talent is a reason for US companies to hire here, not a reason for Canadian founders to stay. That's the gap. We're an attractive back office for AI. The ambition is to be the headquarters.

There are signs of people trying to close it. In Toronto, venture studio AXL wants to launch 50 AI companies in five years and keep them in Canada, though even BetaKit's framing was that one studio won't fix this alone. Cohere is opening a third Canadian office in Ottawa, and co-hosted Gumloop's Vancouver launch night, which is a decent sign that Canadian anchor employers are starting to seed the ecosystem.

We don't know yet whether that adds up to a trend. One honest caveat: we couldn't find clean 2026 data on how many AI-native companies are founded or headquartered in Vancouver, so treat the "few" in our argument as an observation, not a measured fact.

What would actually change it

Not another report on talent. Three things, in order:

  • Anchor employers that run AI product work here, not just engineering. Product leaders who decide what ships have to sit in Vancouver.
  • Growth-stage capital that writes the first big cheque locally. Seed is fine. The leak is between Seed and Series B.
  • More people who've shipped AI products and stay to teach. That's the part individuals control.

What you can do tomorrow: everyone has a different job

A city doesn't become a hub through one group. It happens when students, newcomers, experienced operators and founders each do the thing only they can do. Here's our take on who does what. These are opinions, not findings.

Students in high school or early university Someone once told me that being a student is the only time, other than running your own company, when you have no boss. I think about that a lot. Use it. Be relentless, be creative, and treat everything as a problem to solve.

Vancouver can look perfect from a distance. Up close, like every city, it has problems: housing, access, the gap between research and products. Those are product briefs. Students can pick one, build for it, and see if it travels. The best outcome isn't only a better Vancouver, it's technology born here that works for all of Canada.

  • Build small, real things with AI and show them. A shipped project beats a certificate.
  • Take co-op seriously and aim it at product teams, not only engineering. Panellists at the Techcouver event credited co-op programs and industry-academic partnerships with preparing students for industry.
  • Start now on what's scarce: writing clearly, talking to users, judging whether an AI output is good.

New grads and career switchers

  • Pick one AI product problem and own its evals, even on the side. Product judgment on AI is learned by shipping.
  • Join early-stage teams and say yes to scope. Two years of broad responsibility at a small company teaches more than a narrow role.
  • Go to the demo nights. Gumloop's Vancouver launch drew more than 200 people, and rooms like that are where first jobs and co-founders come from.

Immigrants and newcomers

  • You may be the missing link. Local companies often sell first to the US and other markets, and people who know those markets, languages and buyers are an advantage here, not a gap to apologize for.
  • Don't wait for a Canadian title before you build. Start a project or a small company, and use programs like SR&ED and IRAP from the beginning.
  • Hiring managers: the Techcouver panellists say companies here already recruit Canadians returning from abroad plus international and out-of-province hires. Make that a deliberate policy, with real onboarding and not just visa support.

Experienced people (senior PMs, engineers, operators)

  • You hold the scarcest asset: you've shipped at scale. The panellists' complaint was about knowledge transfer from experienced operators. Mentor two people this year, and be specific about it.
  • Stay, or come back. The biggest thing a returning operator can do is make the next company here look normal.
  • Angel-invest or advise early companies. The leak is between Seed and Series B, and your cheque and your introductions both help.

Founders

  • Decide on purpose where your company lives and why. Hiring in Vancouver while selling and raising south is a legitimate hybrid, as Gumloop shows. Just don't let it happen by accident.
  • Check SR&ED and IRAP eligibility before your first round, so you extend runway before you dilute.
  • Hire juniors and teach them. You're building the next generation of operators, which is the thing this city lacks.

Vancouver won't become an AI product hub because a report says its talent is good. It gets there when each of these groups does its part, and enough of them stay long enough to teach the next ones.