Direct import and Remote Fulfillment are two flows with opposite tariff exposure, and only one of them can be re-sourced. On direct import Amazon picks the vendor, so it picks the country of origin, and the March document shows what that buys [2]. Remote Fulfillment cannot be moved the same way: third-party sellers list on the Canadian marketplace while their inventory stays in US warehouses, and each parcel crosses the border only after a Canadian customer orders [6]. A duty on US consumer goods entering Canada lands on exactly those parcels [5], per order, on stock Amazon does not own. The remedy there is not a different supplier. It is inventory placed inside Canada, and the seller carries that cost.
The dates make the planning look brisk rather than careless. Trump announced an additional 50 percent tariff on certain Canadian imports on 20 July, and Amazon was still setting out expansion plans in late July [3]. Canada's answer arrived 50 days after that announcement [1]. Back in March, Amazon had judged the tariff hit smaller in Canada than in the US because of USMCA coverage, while warning that the risk of a Canadian recession was higher than usual because of US trade policy uncertainty [4]. The July tariffs cover some goods that had qualified for USMCA protection [3], so the assumption holding up the Canadian numbers was partly removed after the numbers were written. The documents call the forecasts preliminary and subject to change [16].
The competitive gap the planners described has figures under it, and they do not line up. Amazon's same-day reach was about 54.5 percent of Canadian Prime members, against competitors said to reach 70 to 85 percent of Canadian households within two to four hours [10]. Different denominators, different speeds, not a like-for-like comparison [11], but even the friendliest reading leaves Amazon 15.5 points behind [2]. Walmart has since launched Walmart+ in Canada with unlimited same-day delivery [12]. The plan answers with geography: 63 percent of Canadian shipments fulfilled within 160 miles of the customer by 2029, and 93 percent within 1,000 [13]. That is a siting problem rather than the throughput route taken by some European retailers, where Decathlon doubled output across seven sites with robotics instead of new buildings [17].
The rejected delivery stations are the tell. Third-party last-mile delivery in Canada costs roughly half what it does in the US, which strips out much of the financial case for Amazon running the last mile itself [14], and the twelve conventional stations that came back with negative five-year paybacks went to partner models instead [15]. So the fastest-growing market gets served on capacity Amazon rents, by the same instinct as the sourcing switch: change the counterparty, keep the plan. US duties on Chinese electric vehicles already showed tariffs redirect trade more reliably than they stop it [18].
Amazon has more than C$65bn invested in Canada since 2010 and over 46,000 people there [8], which is not a position anyone reverses over a duty schedule. Its spokesperson says prices in the Canadian store have not increased outside normal fluctuations, and that the company is monitoring the new tariffs [7]. That statement is about the store. It is not about who absorbs a charge levied at the border on a third party's inventory.
