Welcome to the now “annual” Canadian AI Ecosystem Map. What a year it’s been.
The report also goes to feed the excellent (and searchable!) directory at Canada.ai.
The point of creating this map was to emphasize that the strength lies in the Canadian AI Ecosystem, as opposed to just one city’s. This year, we’ve seen ties strengthen, but also some weaknesses exposed. What I see now is a corridor of cities, each remarkable in their own right, coming together into a cohesive form that is comparable in size and influence with the US East Coast or Paris and London.
To borrow from Startup Genome’s [Startup] Ecosystem Lifecycle Model, the Canadian AI ecosystem is transitioning from the “globalization” phase towards the “expansion” phase:*
Activation: low output of startups, limited local experience, and resource gaps/leakages
Globalization: large exits attracting international resources and resulting growth in startups, along with more established organizations and institutions starting to foster connections with the global ecosystem
Expansion: multiple large exits and unicorns, effectively making the world its resource pool, and the sheer number of startups starting to hit a critical mass to achieve a rhythm of successes
Integration: balanced in competition with other top ecosystems, solidified in its rhythm of emerging unicorns and deeply involved with the broader economy to sustain its competitiveness
This transition is powered by a combination of factors. A big highlight is the government’s commitment of $4 billion to fundamental research over the next five years, ensuring Canada’s knowledge infrastructure that feeds the rest of the ecosystem. This has come as a response to the Naylor Report, which showed the critical need to maintain Canada’s historical advantage in fundamental research. It’s these investments in long-term, basic research that have led to breakthroughs like deep learning.**
A good deal of this funding is going to our world-class AI labs across the country. With their decades of experience in cutting edge research, they recognize the importance of focusing these investments on the interdisciplinary efforts the tech demands. These groups, like the Montreal Institute for Learning Algorithms (MILA) and The Vector Institute, are also now fully independent legal organizations, freeing them to be more global in their thinking and activities.
Their prominence is part of what has attracted the number of large international actors in the AI ecosystem, with nearly three times as many landing here in the last year. AI labs from Google, Facebook, Amazon, DeepMind, Samsung, Uber, Huawei, Oracle, etc. are bringing in big-time resources, but can also put a strain on talent resources. [For more, see the full report.]
Big international labs are a double-edged sword
At first, big labs setting up shop in Montreal was an important milestone. It was a sign that the talent drain to Silicon Valley was slowing and even reversing. But if we’re not careful it could be a situation where it’s a matter of Silicon Valley’s talent pumps just reaching us directly.
Overall, these labs have lent us credibility and made our ecosystem overall more attractive for global investors. So far, they’ve been a net positive for our maturing process and added to the fanfare of AI in Canada thanks to visits from the Prime Minister. However, we can’t forget that the (virtually un-taxable) intellectual property they create will be directly funnelled back to Silicon Valley without any fiduciary benefit realized in Canada. Ultimately, we shouldn’t lose focus on our local growth-stage startups in welcoming the large international players.
In the medium term, it’s the growth-stage startups (those with more than 25 employees) that will drive the ecosystem’s maturation. They are the ones that can compete with the big labs for talent, and really capitalize on the new international attention. Early-stage startups (less than 25 employees) also rely on the growth stage startups to provide a buffer from the big labs. Without a balance of competition among a mix of startups at different stages, we’ll lose our local champions and potentially prevent the next growth-stage and early-stage startups.
Supporting our talent magnets
Attracting international talent (note: not organizations but individuals) is the most important place to channel our efforts. The rate of university graduates alone won’t be sufficient to meet demand.
So far, we’ve maintained a healthy pool of talent thanks in part to immigration. Canada’s independent AI labs working on impactful projects along with its general soft power have made our country quite competitive against Silicon Valley when it comes to attracting smart people from abroad. But it’s the progressive immigration policy that has kept the gates wide open for them to actually get here.
Today, it’s quite easy for ambitious, educated people to come and work in Canada. The Global Skills Strategy, started last November, has shortened the visa application process to 2 weeks for the much-needed talent (compare that to the 18 months for an H1B1 Visa in the US). Another incredible initiative is the automatic 3-year Post-Graduate Open Work Permit for anyone who has completed a university degree in Canada.
This is critical for the AI industry. Our talent report came up with only about 20,000 AI experts on earth. Even if we were so lucky to have all of them show up in Canada, we would not only have work for all of them, we would still hire every AI student graduating out of University.
Let’s continue to open up immigration to increase and sustain our talent pool growth.
Connect Canada’s AI Corridor
Canadian cities cannot complete the maturing process alone. Two weeks ago, 25 EU countries recognized that alone they would be left behind in AI and so they needed to band together to share resources and projects. In Canada, we need a national effort, and to build as many bridges as possible. We’ve had a healthy competition between cities, especially Toronto and Montreal, but we’re reaching a point where it will hurt us more than help us. Similarly sized hubs like London, Paris, or New York benefit from being self-contained, whereas Canada’s hubs have their relative strengths. Toronto’s strength is in funding; Montreal and Waterloo’s is research; Ottawa and Quebec’s is policy.
Our big, unaddressed disadvantage is geography. A 4-5 hour train ride between Montreal and Toronto is too long. VIA Rail’s research into a dedicated rail line for the Windsor-Quebec City Corridor is a great first step. This corridor contains 5 of the 7 hub cities our report covered (Waterloo, Toronto, Ottawa, Montreal, and Quebec City). The dedicated rail will potentially have trains running at up to 177km/hr every 45 min, dropping travel time by about 25%. Let’s do that, and let’s keep looking ahead. Imagine the possibilities of a 30-minute trip enabled by either a hyperloop or high-speed train.
A model for a more connected ecosystem lies in the ability of assets to flow between cities and allow many groups to grow, not just in one of the hubs. AI cuts across our economy and so must be treated a multi-disciplinary development. The needed scale for this type of cross-functional effort is difficult to achieve. Our institutions will take time to adapt, and we need to let them. But it will be much less painful if we create more opportunities for collaboration by opening up our country’s, and our cities’, borders.
Special thanks to: Yoan Mantha, Simon Hudson, Ahyoung Lee, Amanda Durepos, Wei-Wei Lin, Caroline Bourbonniere, and Genevieve Jacovella Remillard for all their efforts in putting this together.
*This model is an assessment of entire startup ecosystems, so the sizes of each stage I am considering for an AI ecosystem are relatively smaller than Startup Genome’s thresholds.
**These long-term investments in research also directly lead to a new generation of high-valuation startups homegrown in Canada. This last year saw multiple AI companies receiving rounds of $100M+, and overall the number of active AI-related startups has grown by 28%.