After the U.S.–Canada Talks Collapsed, Did the Timing for Chinese EVs in Canada Change?

🌍 中文版：美加谈判破裂之后，中国电动车进加拿大的时机变了吗

On the night of August 21, U.S.–Canada trade talks fell apart just before the deadline. The United States put a 50% tariff on roughly $20 billion of Canadian goods, effective at midnight. Prime Minister Mark Carney said Canada would match it dollar for dollar starting September 8. The break came over medium- and heavy-duty trucks: Washington says Ottawa added a last-minute demand for tariff relief on that category; Ottawa says Washington pulled it from the list at the last minute. The framework on the table would have cut the U.S. tariff on Canadian autos from 25% to 15%. That same afternoon, Global Affairs Canada updated its report on the import quota for Chinese-made EVs: 15,063 of the first period’s 24,500 permits had been used, 61.5%. Two weeks earlier the figure was 51.1%.

Put those two facts side by side and you have the situation a Chinese automaker now faces in Canada: the door is shaking, and at the same time people are walking through it. Two voices tend to surface. One says the further the U.S. pushes Canada away, the more Canada needs China, so go in now. The other says all of North America is a minefield, so pull back. My view is that both are moves made at the wrong moment. August 21 did not bring the moment to act any closer. It only made waiting harder to sit through.

Yale’s Ben Polak teaches this with a game. Two people stand facing each other, each holding a wet sponge. They take turns, and on each turn a player either steps forward or throws. Hit, and you win. Miss, and your hands are empty while the other person walks up to you one step at a time. When should you throw? Polak’s answer is an inequality: the first time “my probability of hitting now” plus “the opponent’s probability of missing next turn” reaches 1, whoever’s turn it is must throw. Before that point, stepping beats throwing. He has run the game with students, MBAs, alumni, and deans. In theory, a throw at the critical distance should land about half the time. Real people land about one in ten. They throw too early, systematically. Polak offers two reasons: people overestimate their own odds, and there is an urge, which he half-jokingly calls an American disease, to always be doing something.

To apply the model to Canada, first be clear about what the sponge is. It is not selling cars. The sponge is the commitment you cannot take back: a plant or a joint venture, a dealer network across the country, a large inventory. Stepping is a different class of action: certification, a handful of dealer contracts, service and parts, a small batch of cars brought in under the quota, enough presence that local buyers know who you are. Miss with the first kind and you lose the money and several years. Get the second kind wrong and you can stop, at limited cost. The “hit probability” is the chance a large commitment survives, and it depends on demand, policy, and whether your network exists yet. The “opponent” is the weakest part of the fit. No one stands across from you holding a sponge; the closest equivalent is some mix of rival automakers and the policy environment. The model does not transfer cleanly. What it does is turn one unanswerable question into two answerable ones. How high is my hit rate right now? If I wait one more turn, who takes what from me?

After August 21, both answers moved a little.

On hit rate, there are two paths to separate. The first is Canada as a springboard into the United States. That path’s odds just got worse. Back in January, Trump said that if Carney thought he could turn Canada into a “drop off port” for Chinese goods headed to the U.S., he was “sorely mistaken.” After August 21, the U.S. tariff on Canadian autos is still 25%, and what the talks finally broke over was a vehicle category; Flavio Volpe of Canada’s auto parts makers’ association said the American motive was to get auto manufacturing out of Canada. Building in Canada to serve the U.S. is a problem even for companies that already have plants there: the example Carney gave was Ford’s Ontario plant, whose F-350s were left out of the tariff relief. The second path is Canada as a market in its own right. Its odds did not move much, or improved slightly. Carney’s government has more reason than it did two months ago to keep the China channel open; in June, Industry Minister Mélanie Joly said the Chinese EV quota was “already set” and “no surprises.” Utilization is climbing. But a trade war hurts Canadian consumers, and the annual quota of 49,000 vehicles, as Foreign Minister Anita Anand put it, is less than 3% of the Canadian market. This was never a wide road.

On the opponent side, the composition of the quota is the most direct evidence. A 61.5% utilization rate sounds busy. But by MobileSyrup’s late-July breakdown, the large majority of used permits were Tesla’s Shanghai-built Model 3, launched in Canada on May 1. The only Chinese brand actually delivering was Lotus, with the C$119,900 Eletre. In the weeks since, I have not seen another Chinese brand announce that deliveries have begun. BYD, Chery, Dongfeng, and the Geely brands other than Lotus are all still stepping: certification, site selection, hiring. Nobody has thrown. So the answer to “who takes what from me if I wait one more turn” is, for now: nobody I can see.

Add the two terms and the sum is well short of 1. Polak’s practical point is that in real life you do not need to calculate the critical distance. You only need to know whether you are still far from it or already there. In this game, it is clearly the former.

That also settles which kind of reasoning to use. Polak’s course introduces two tools. Backward induction works from the end of the game back to the present, and it requires that you know the opponent’s curve and that the opponent is running the same analysis. The dominance argument is cruder: if my best move this turn is the same no matter what the opponent does, I make it without guessing. Backward induction is precise and fragile; one irrational move by the other side and the whole chain is wrong. Dominance is rough and sturdy. Doing backward induction on Washington right now means using the most fragile tool for the most expensive decision. Nobody knows whether talks resume in September, whether the U.S. auto tariff moves, or how hard the U.S. will press on the Chinese EV quota in the CUSMA annual review. You do not need to know. Whether Canada and the U.S. are back at the table next month or escalating further, “step, don’t throw” is the answer on every branch. That is where the dominance argument earns its keep: far from the critical distance, you are allowed not to guess the opponent.

Waiting is not standing still. In Polak’s model every turn requires a step forward; without it, the critical distance never gets closer. The second quota window opens September 1. Under the first-period notice, permits are first-come, first-served and unused volume from the first period rolls into the second. The new notice Global Affairs Canada promised before the second period begins has not been published as of today. This step is worth taking: bring in a small batch under the quota and run the whole chain on real deliveries, from certification to dealers to service, insurance, parts, and warranty claims (the July piece was about that chain). Sign a few dealers, not a few dozen. Build the local team. These actions share one property: if the door does close, they remain assets. A plant does not.

One objection deserves a direct answer. This is a window, the argument goes; miss it and it is gone. The U.S. pushed Canada’s door open, and it can push it shut. I do rate the chance that Ottawa tightens the quota to buy American concessions higher than I did two months ago. But a closing door cuts volume, not the steps already taken. And if the door closes, the party that loses most is the one who already threw. The window argument supports stepping.

So my view narrows from “wait” to something more specific. August 21 did not change what Chinese automakers should be doing in Canada. It changed the psychology, making a step, which is correct but unsatisfying, feel like doing nothing. Two nearer things deserve more attention than the swings in U.S.–Canada talks: whether the September 1 rules for the second quota period shift to per-manufacturer allocation or shrink the total, and the day the quota starts being used up by Chinese brands rather than Tesla. That is when the inequality actually moves. Until then, keep stepping.

Adam Yang | Based in Toronto, with 10+ years in China-to-global expansion. He writes about how Chinese brands build trust in North America.

Sources: NPR on the collapse of U.S.–Canada talks, the 50% tariffs, and Canada’s September 8 retaliation; Fortune on the trucks dispute, the 25%→15% auto tariff framework, and Flavio Volpe’s comments; CNN: Carney says the U.S. “asked too much, offered too little”; Government of Canada quota utilization report, updated August 21, 2026, 15:00 ET; Notice 1162 on first-come-first-served permits and carry-over into the second period; MobileSyrup on the composition of quota usage, July 27, 2026; The Globe and Mail: Joly says Chinese EV quota “already set,” June 22, 2026; Global News: Anand on the quota being under 3% of the market, July 17, 2026; CBC on BYD’s plan for six dealerships and compliance work on two models, July 15, 2026; Bloomberg on Chery’s trip with Canadian dealers to the Beijing auto show, April 24, 2026; Electric Autonomy on Dongfeng preparing sub-C$35,000 EVs for Canada, July 23, 2026; Fox News on Trump’s “drop off port” remarks, January 25, 2026; Ben Polak’s Yale open lecture on game theory (the sponge duel). Policy and market information checked August 22, 2026.