AI-native startups are growing ten times faster than the rest of tech, but we’ve read a similar story before.
The annual Global Startup Ecosystem Report (GSER) was released by Startup Genome and the Global Entrepreneurship Network at VivaTech Paris last week.
Startup Genome CEO and founder JF Gauthier noted at the beginning of the launch presentation on June 18, that the parallel to 20 years ago is hard to miss.
Just as software ate the world then, AI is eating the world now.
At this point, AI has moved from tech sub-sector to a general-purpose technology reshaping where capital flows, talent concentrates, and startup value is created.
The report covers more than 300 ecosystems worldwide. It exists, in part, to make sure policymakers don’t miss the memo, which does come with a pretty direct warning for Canada and the rest of the non-U.S. world.
“Governments are spending their scarce budget on compute infrastructure and corporate AI adoption — importing U.S. solutions,” Gauthier explained in a press release ahead of the launch presentation. “Leaving too little to fund updated policies that accelerate the building of a local AI-native startup ecosystem that produces exports and ensures their future prosperity.”
Global Ecosystem Value is up almost 40%, according to the report, but three U.S. ecosystems (Silicon Valley, Los Angeles, and New York City) are responsible for two-thirds of that growth.
“After a decade advising forward-looking governments to develop and execute powerful innovation policies to further our mission of decentralization of tech, we are seeing a concerning re-concentration,” he added in the press release.
Canada’s largest ecosystem, Toronto-Waterloo, posted a strong showing at 13th globally, up seven positions and tied with Paris. But it’s Calgary that’s increasingly pulling the spotlight as an emerging ecosystem.
And they’re not even trying to be Silicon Valley.
Calgary is a Canadian outlier
Calgary’s tech ecosystem is growing more than four times faster than any other Canadian market in Canada, based on GSER data.
The data also found that Calgary’s ecosystem has grown almost 40% since 2021, above the 9.6% Canadian average, with a healthy sector mix that includes fintech, cleantech, agtech, energy, and defence. Total venture capital funding from 2021 to 2025 reached $3.4 billion, and the city recorded 151 exits over the same period, above the global average of 103.
“Calgary has changed the game,” Gauthier said.
Gauthier said Startup Genome began working with the city in 2018 and 2019, helping Calgary change create community, funding, and program policies, with a focus on top-of-the-funnel actions like pre-accelerators and building local connectivity.
After this collaboration, seed funding to AI-native startups in Calgary grew 70%. Montreal, Toronto-Waterloo, and Vancouver all saw declines.
“When you do put a billion dollars in [compute infrastructure] and have no money to create AI and support AI-native startups, you don’t reap the benefits,” said Gauthier.
The four U.S. cities with the most compute infrastructure near frontier tech ecosystems (Richmond, Las Vegas, Charlotte, and Atlanta) are not transitioning any faster into AI-native startup hubs, he said.
Similarly, corporate AI adoption helps established companies stay on top, but it doesn’t generate new winners.
“It’s very true in Canada,” Gauthier said. “[They] neglected the top of the funnel, trying to do scale-up policies, but all cities in Canada, except Calgary, has been falling in the ranking, 10, 20 ranks down.”
Toronto is recovering now, he noted, because of a handful of AI successes, but the structural problem remains.
The 1,000 startup threshold
Based on 15 years of research, Startup Genome has identified the level at which an ecosystem starts producing consistent $100 million exits and billion-dollar outcomes.
For a city of 1 or 2 million, 1,000 AI-native startups is “when the magic starts happening,” Gauthier said.
Of course population size needs to be taken into account.
Gauthier points to a place like Seoul, a city of almost 10 million, where the tipping point would be closer to 2-3,000. A smaller Scandinavian city, by comparison, would see change in the 750-800 range of AI-native startups.
Roughly a 100-kilometre radius around a centre point (the distance within which founders can share learnings in real time), makes investors feel comfortable visiting portfolio companies, he said, and talent can move between startups without relocating.
A dispersed regional cluster just doesn’t produce the same effect.
Silicon Valley, in a surprise to no one, sits at approximately 20,000 AI-native startups. New York is at 6,000, with London and Singapore in a similar range. Dubai and Abu Dhabi together sit at roughly 500.
“You need to get there within five years,” Gauthier said. “And we’re already three years into this era. The lagging ecosystems risk declining ecosystem value.”
Chris Haley, Startup Genome’s head of research, described a concept the team calls funding velocity.
AI-native startups are being funded roughly a year faster than the rest of tech, and at larger round sizes. Large venture capital funds are now writing checks to the tune of $20 to $50 million at seed stage to lock in top AI-native startups early, compressing the traditional funding ladder.
The path to 1,000 startups, Gauthier argued, shouldn’t include slapping on a London or Singapore template. Those cities have thousands of startups on startups on startups already, and their policies reflect that scale, with Series A support, fund-of-funds structures, mid-to-late-stage infrastructure.
Smaller ecosystems copy those models, and end up funding the wrong stage.
“If you’re in a frontier city, organic transition is okay, but if you’re in a small ecosystem, you must boost creation through policy,” he said.
What works at this stage is top-of-funnel moves. For example, pre-accelerators that help founders test and vet demand first, early community programs, and funding designed to get more companies started.
Start more companies, testing demand early, and building the local connectivity that lets founders learn from each other before they read about it in a press release two years later.
More companies getting started, more founders testing demand early, more local density compounding over time toward that 1,000-startup mark.
The agentic AI opening
“In New York last year, there were 120 times more agentic AI startups getting at least $1 of venture capital than model startups,” Gauthier said.
While there are now 80 times more agentic AI startups, the foundational model category remains effectively closed to new entrants, dominated by Anthropic, OpenAI, and a small number of peers with billions in capital.
Specialized models are stable, but agentic AI is where the volume is. These startups, said Haley, build on top of third-party models and apply it to specific applications.
“The new competitive advantage in the AI era is actually industry expertise, job expertise, [and] great affinity with what AI can do,” Gauthier said.
Calgary’s workforce may be better positioned for this than most.
The city’s dominant industries (energy, agriculture, defence) are where deep job expertise produces the kind of specific, hard-to-replicate industry knowledge Gauthier is describing.
A supply chain manager in Calgary or a compliance officer at a mid-sized energy firm are the people who understand their domain in ways that a San Francisco startup just doesn’t.
Identifying potential founders is only part of the policy problem.
They need someone to sell to. Specifically, governments need to stimulate local corporate demand through open innovation strategies and venture clienting, structured programs that connect early-stage startups with corporations willing to be their first customers.
Enter, the technology leader. The decision to pilot a Canadian agentic AI startup, instead of waiting for a polished U.S. vendor, is the demand the whole model depends on.
“You need to connect your corporations with your agentic AI startups in your regions right now,” Gauthier said. “So that when the U.S. agentic AI startups come to your country in two, three years max, you already have a large community of AI-native startups.”
Calgary has shown that policy and deep institutional support works. Ottawa climbed more than ten positions in the emerging ecosystem rankings, with strong AI-native early-stage funding momentum. Edmonton ranks in the top ten in North America for funding runway.
Gauthier’s critique has a Canadian test case. Ottawa’s $2.3 billion “AI for All” strategy leans heavily on sovereign compute and corporate adoption. In the launch presentation, he called these moves defensive. They’re good for incumbents, but not fueling the engine for new startups.
The top of the funnel is the part it funds the least.
Canada has the talent and the infrastructure. The trouble is this isn’t Field of Dreams. You can build it, and they still won’t come, not unless someone funds the part where the startups really get started and someone agrees to be their first customer.
Final shots
- The magic number, according to 15 years of Startup Genome research, is 1,000 AI-native startups within a city. That is where consistent large exits and billion-dollar outcomes start appearing.
- The agentic AI window is open now, and Gauthier was explicit that ecosystems have two to three years before U.S. competitors arrive with polished products built on the same industry expertise that currently sits inside Canadian companies.
- Canada has the research base, the talent, and the sector depth. GSER’s data shows that the only thing missing is the policy decision to point all of it at startup creation instead of startup adoption.