Opinion

Simon Chong has had a revelation, and it is delightfully simple: Canada's industrial infrastructure—factories, metal, supply chains, and people who know how to build at scale—is not just a legacy asset gathering dust in the Rust Belt, but rather the secret competitive moat for the next generation of artificial intelligence. Physical AI, he explains, needs these things. Canada has these things. Therefore, Canada will dominate Physical AI. The syllogism is so clean you can practically hear the air being let out of the room when you finish reading it.

The argument, stripped of its promotional gloss, amounts to this: because factories exist and robots will eventually exist, the factories will win. This is what economists call "confusing input for output." Yes, Physical AI—robots, autonomous systems, hardware-integrated AI—theoretically requires manufacturing capacity, supply chain resilience, and skilled labor. Canada certainly possesses these, mostly inherited from decades of automotive, aerospace, and industrial production. But possessing the ingredients does not make you a chef. Possessing a steel mill does not automatically make you competitive in robotics any more than owning wheat fields makes you a baker. The leap from "we have factories" to "we will build globally competitive Physical AI companies" is where Chong's logic evaporates like morning frost.

This is not Canada's first rodeo of optimistic industrial reclamation. The country has spent the last thirty years watching its manufacturing base shrink, its talent migrate to Silicon Valley and Boston, and its tech ecosystem struggle to scale beyond software services and niche hardware plays. Venture capital has historically fled Canada for US markets despite similar infrastructure advantages. The 2010s saw numerous "reshoring" narratives, supply-chain localization stories, and advanced manufacturing initiatives that produced far more press releases than commercial victories. The pattern is consistent: infrastructure alone attracts neither world-class engineers nor venture dollars at scale. Geography and nostalgia do not substitute for ecosystem density, capital liquidity, or network effects.

Chong's thesis also embodies a peculiar form of cargo cult capitalism: the belief that proximity to Physical AI's raw materials—metal, electricity, labor—automatically translates to competitive advantage in the intelligence layer. The people who know how to build at scale in a steel foundry are often not the people inventing transformer architectures or solving embodied AI problems. These are different skill sets, different networks, different funding sources. Factories are necessary but not sufficient. Saying "we have factories, therefore we'll win at Physical AI" is like saying "we have sand, therefore we'll dominate semiconductor design."

What could go wrong with this strategy? Everything. Global capital will continue flowing to wherever AI talent clusters most densely. Early Physical AI leadership will likely consolidate in the US, with secondary clusters in China and Europe—not because they have better factories, but because they have denser networks of venture capital, world-class researchers, and customers willing to risk early deployments. Canada's factories, however robust, will likely become contract manufacturers for companies founded and funded elsewhere. The higher-margin design, software, and IP layers will remain concentrated in the hubs where capital and intelligence meet. Canada will be left with good manufacturing economics and minimal venture upside.

This pitch reveals a deeper anxiety haunting Canadian tech: the fear that organic innovation ecosystems cannot compete globally, so perhaps we should double down on our one perceived advantage—the physical stuff we already own. It's a comfortable narrative because it requires no fundamental risk-taking, no brain-drain reversal, no radical shift in capital allocation. Just unlock the value that was always there, waiting beneath the factory floors. The tech industry has heard similar stories from every region that faced displacement: we have infrastructure, therefore we will recover. Most of those predictions aged poorly.

The real question isn't whether Canada has factories. It's whether factories, in 2024, are where venture returns concentrate. The answer, historically, is no.