We label the ketchup and the eggs. We cannot label the search engine, the maps or the video calls, because there is almost nothing Canadian on that shelf. Here is why the empty shelf is an opening.
At the grocery store, I read the labels. The ketchup is Canadian, the eggs are Canadian, the produce is Canadian. Apps scan barcodes to rule on a product's citizenship in seconds, and patriotic shoppers lean on little maple leaf stickers to support local producers.
Then I get home, open my laptop, and my elbows come down. The search engine, the operating system, the email, the maps and the video calls are all American. There are no labels to check because there is almost nothing Canadian on the shelf. That empty shelf might be the best deal Canadians get out of this trade war, because it lets us build our AI capacity on another country's dollar.
The dependence showed itself in August. On Aug. 27, Washington ordered Lake Ontario renamed Lake America in the U.S. federal names database, and by Aug. 30 the new name was live on Google Maps. Within days it appeared on Hydro One's outage map, the LCBO's store locator and GO Transit's station finder, because every one of those sites rents its map from Google.3 Google's fix was for Canadian companies to set Canada as their default region. Premier Doug Ford's fix was a sign on the Winona shoreline reading Lake Ontario, Now and Always. I liked the sign. A sign is also what you put up when you don't control the map.4
On Sept. 8, Canada's retaliatory tariffs landed on more than 700 American goods, our answer to Washington's 50 per cent duties.12 I support every dollar of it. But a tariff is a tool for goods that cross the border in trucks. The technology that runs our days crosses in milliseconds, tariff-free, and Canada builds almost none of it.
Yet in a trade war no Canadian wanted, knowledge workers have a rare opening.
The two countries that build frontier AI, the United States and China, are locked in a price war, and neither is charging what the technology costs to build and run. American tech giants are spending hundreds of billions on AI infrastructure to hold market share,5 and Chinese labs are undercutting them. DeepSeek lists its top model at 66 cents (off-peak) to $1.32 (peak) to read a million tokens of text.6 OpenAI lists its flagship at $4 for the same million.7 When two foreign giants compete to subsidize the same product, the customer wins.
For Canada, the answer to the empty shelf sounds like betrayal at the checkout counter. At the trading desk it is common sense: buy low, build high, and take full advantage of a price war paid for with foreign capital.
Using subsidized American and Chinese models to automate routine work and build the tools we cannot buy Canadian is capital efficiency, and the gains stay here. Every skill learned and every tool built is domestic capacity, and nobody can put a tariff on it.
Ottawa's new AI strategy wants business adoption to climb from about 12 per cent to 60 per cent by 2034, with a public supercomputer and sovereign compute behind it.8 Compute without people who know how to use it is an empty building. The workforce that will run Canadian infrastructure tomorrow has to learn on somebody's dime today, and right now Washington and Beijing are offering to pay.
So as offices reset their routines this September, put AI on the professional development plan and treat it as the one part of the counter-tariff that costs Canada nothing. We don't own the shelf yet. We can build on it with their tools.
Sources
Elvin Peters spent a decade in marketing, half of it in paid media, and worked at Google. He is the author of The Artificial Advantage and writes from Toronto.
The Artificial Advantage: the frameworks behind everything here, written for professionals, not programmers.