Build the footprint for the tariff after next

In the spring of 2025 the United States imposed tariffs on imported vehicles and automotive parts, with relief for parts that qualified under the rules of origin of the United States–Mexico–Canada Agreement, and raised duties on imported steel and aluminium. Within weeks, suppliers with Mexican plants were being asked by customers, investors and local politicians what they intended to do. Several announced that they would move production north.

Tallisford assembles about 61 per cent of its heating and air-conditioning modules in two plants in Coahuila and ships them to vehicle assembly plants in the United States. We were asked the same question, including by our largest customer's purchasing team, and our chief executive's first instinct was to answer it with an announcement. I recommended that we stay, and I want to explain the reasoning, because I think it applies well beyond one company.

A plant outlives a tariff schedule

A module line is a twenty-year asset. Moving one from Ramos Arizpe to Ohio would have cost about US$165 million, taken eighteen months to requalify with customers and raised our cost by roughly US$38 a module, most of it in labour. Those figures become fixed the moment the decision is made. The tariff that prompts a decision like that does not. North American trade policy has changed sharply three times since 2018: the steel and aluminium duties of that year, the renegotiated agreement that entered into force in 2020, and the 2025 measures. The agreement's six-year joint review falls this year and could change the rules again.

A footprint designed around one tariff schedule is a bet that the schedule will last longer than the plant's payback period. For a module line paying back over six or seven years, I could not find anyone, inside the company or outside it, willing to make that bet with confidence. That did not mean doing nothing. It meant spending money first on the things that help under almost any regime.

Content you can prove is worth more than content you claim

The most useful discovery of 2025 was that our problem was largely one of evidence. Most of what our Coahuila plants shipped already qualified under the agreement's rules of origin, because the aluminium tube, the blowers and the control heads came mostly from North American suppliers. But qualification has to be demonstrated, with supplier certifications and records that trace regional content through the bill of materials. When the tariffs arrived, we could document only 58 per cent of the module value we shipped as qualifying. The rest probably qualified, but probably is not a customs position.

We spent US$27 million over nine months closing that gap: certifications from 340 suppliers, a traceability system for the bill of materials, and the requalification of aluminium fin-stock and brazing-sheet supply from Asian mills to North American ones, which costs us more per tonne. By the second quarter of 2026, 91 per cent of the module value shipped from Mexico was documented as qualifying, and the tariff cost we were absorbing had fallen from about US$9.4 million a month at its peak to US$2.1 million.

We also renegotiated with customers. Three of our four largest agreed to share tariff costs on parts with documented content, on the reasonable ground that a qualifying part from Mexico still costs them less than an equivalent made in Ohio. Over the year they reimbursed about US$31 million. The fourth moved one module programme, worth about US$64 million a year, to a competitor with a plant in the United States. That was the price of not making an announcement, and I had told the board to expect it.

Three regimes, one footprint

The method I now use for any footprint decision is to test it against three trade regimes rather than one forecast. In the first, today's tariffs remain. In the second, they are removed or substantially relieved for qualifying content. In the third, the rules tighten, perhaps with higher regional or US content requirements after the review. A move that pays only in the first regime is a wager on policy. A move that pays in two of the three, or improves our position in all three, is a strategy.

The test also changed how we talk to customers. Instead of asking a purchasing team to trust that our modules qualify, we now send the origin documentation with each quotation, together with the share of value we expect to source within the region over the life of the programme. Two customers have since told us that this evidence weighed more in their sourcing decisions than an announcement would have, because it reduced a risk on their own books rather than flattering ours.

Under that test, moving module assembly north failed. Documenting our content passed in all three regimes. Re-sourcing aluminium to North American mills passed in two. One capacity decision did pass outright: when we added battery-thermal capacity, we placed it in Tennessee rather than Coahuila, because the customers for those parts wanted supply close to their battery and vehicle plants for reasons that had nothing to do with tariffs.

I have heard the opposing argument many times, and it deserves respect. Customers and governments notice who invests in their country, goodwill counts in sourcing decisions, and a supplier that waits for certainty may wait for ever. I accept all of that. My answer is that an announcement can be made about investments that make sense anyway, and that a plant built to make a political point will still be on the balance sheet long after the point has been forgotten.

There is also a cost to my approach that I should not hide. Documentation is invisible. Nobody photographs a traceability system, and for several months in 2025 our customers and some investors read our silence as complacency. A strategy officer who recommends patience has to explain it more often, and in more detail, than one who recommends a groundbreaking.

Directors can ask three things of any footprint proposal. How much of our current content is documented as qualifying, rather than believed to qualify? How does the proposal perform if the tariff that prompted it is removed? And what will we do with the plant we are leaving? Management that has answered all three is planning for the tariff after next. Management that has answered only the first is reacting to this one.