Canada arguably invented modern AI. Geoffrey Hinton won the 2024 Nobel Prize in physics for foundational discoveries and inventions that enable machine learning with artificial neural networks. Canada has also been home to other pioneers who helped shape the field, including Richard Sutton, frequently credited as the father of reinforcement learning, and Yoshua Bengio, one of AI’s godfathers.
And yet the most important AI companies are almost all based in the United States. On Forbes’ 2025 AI 50 list, 42 of the 50 companies are U.S.-based, while Canada has just one. OpenAI, of course, was co-founded by Ilya Sutskever, a Canadian who studied at the University of Toronto under Hinton.
We were leading at the end of the second period, but we still lost Game 1 of the AI Stanley Cup Final.
That is the uncomfortable truth. Canada trained many of the world’s top researchers. We produced exceptional engineering talent. We helped build the intellectual foundation. But when the most valuable companies were formed, sadly, most of the ownership ended up elsewhere.
We are still early in this cycle, and that is when investments can produce outsized returns.
But losing Game 1 does not mean the series is over. Game 2 is already under way, and, just like Game 1, Canada is leading early again.
I recently delivered a keynote at the Engineering Deans Canada annual meeting on this exact point. I told the deans that they are among the most underleveraged forces in Canada’s innovation economy. Many of the next generation of tech giants are taking shape in front of them, in their classrooms, labs and research ecosystems. Game 2 is being played in quantum, robotics, physical AI, outer space, advanced manufacturing and smart energy.
Unlike Game 1, these are not businesses that can be spun up quickly with a few engineers and a laptop. The physical infrastructure is much harder to relocate. That is what gives Canada a real chance to win Game 2.
Why? Here are a few concrete examples. Waterloo’s Quantum Valley, Institute for Quantum Computing, and Perimeter Institute are where top researchers, physicists and ecosystem partners come together to advance quantum research in their world-class facilities. DistriQ in Sherbrooke, Que., has emerged as that province’s quantum innovation zone. Nearby Bromont is a critical world-class semiconductor packaging hub. And in Ottawa, the NRC’s Canadian Photonics Fabrication Centre is not just another research facility. The NRC and recent industry coverage describe it as North America’s only end-to-end pure-play compound semiconductor facility, and the federal government has announced plans to privatize it to unlock its potential.
In other words, this is not just an important lab. It is a strategic infrastructure for industries that will shape the next decade. These technologies are foundational to AI compute, photonics, quantum, telecom, autonomous systems and other critical industries globally.
That is how family offices should look at this next wave.
If Game 1 was dominated by software, Game 2 will be built much closer to the physical world. It will depend on specialized infrastructure, technical teams, scientific depth, manufacturing capability and other deep domain expertise. These companies take longer to build. They often require more capital. They can be harder to evaluate early. But they are also harder to move, harder to copy, and potentially more durable once they scale.
If Game 1 was dominated by software, Game 2 will be built much closer to the physical world.
That is what makes deep tech so relevant now. Most family offices are not only built around quarterly market cycles. They invest across asset classes, and their long-term private allocation is often about ownership in important companies across different stages of growth. We are still early in this cycle, and that is when investments can produce outsized returns, if you own the right companies.
Private investors are often first in line to see these opportunities, long before they become obvious to the broader market. In deep tech, the most important signals often appear early in research communities, university labs, technical founder networks and commercialization efforts that may not yet look fundable to a generalist investor. That is a real edge.
Many of the most important companies of the last few decades were founded and led by scientists and engineers, not by traditional businesspeople: Google, Nvidia, OpenAI and SpaceX, just to name a few. That reinforces a point investors often miss. Before they are founders, many of the most important future entrepreneurs are researchers. The next generation of category-defining companies is often formed much closer to the lab than to the boardroom.
Many of the next generation of tech giants are taking shape in classrooms, labs and research ecosystems.
We know this because in our own family office, we practise what we preach. Through our venture fund, Two Small Fish Ventures, in which our family office is one of the anchors, we have spent years investing in the places where many of these companies actually begin: universities, research institutes and labs, long before the broader market sees them as companies.
I say this not from a distance, but from years spent building and investing at the research layer, first as a serial entrepreneur who turned Wattpad into an AI company more than a decade ago, before AI became a household name, and now through Two Small Fish, where we have spent years working with universities, scientists and research labs as deep tech companies take shape.
Not surprisingly, we are now seeing more researchers who want to commercialize firsthand. We are also seeing more technical founders emerge from the lab, and some of the most consequential capital conversations happen years before those companies are visible to the broader market.
At the same time, AI is reducing the cost and time required to build many products whose defensibility used to come mostly from product functionality alone. Those businesses are less investable than they once were, as AI is putting pressure on traditional SaaS (Software as a Service) models. In other words, value is moving closer to the hard things again.
That shift creates an opportunity for investors who know where to look. I have long told people that Canada does not really have an auto industry. We have an auto supplier industry. That is not the same thing. The owner captures the margin, owns the customer and sets the rules. The same logic applies in deep tech. The recent pullback in EV battery plans, as automakers revisited their timelines, is a good example. The opportunity here is not simply to participate in the next wave. It is to own part of it.
Family offices are unusually well-positioned to benefit from it. They understand long-term ownership. They invest across asset classes. They can back important companies across stages. And in this case, they are physically close to ecosystems that they can actually access and understand.
That is what makes this, first and foremost, a generational investment opportunity. And by acting on it, family offices can also help shape a more ambitious Canadian ownership story.
Because in the next era of deep tech, the biggest opportunity may not be in supplying the winners, but in owning them.
Allen Lau is co-founder and Operating Partner of TSF Ventures, an early-stage, deep-tech venture firm that invests in the next frontier of computing and its applications.
Please visit here to see information about our standards of journalistic excellence.