Ask a management team what happens if the tariff lands and you get the same answer everywhere. We localise. It is on slide fourteen. There is a map on it, with a pin in the market, and the pin is usually a different colour from the other pins.
Nobody in the room asks what a pin costs.
The slide
BYD had that slide. It announced its Hungarian plant in December 2023 and signed for the land the following month. Trial production began in late January 2026. Series production has now slipped to the back end of this year. The Turkish plant, a billion dollars announced in 2024, has not broken ground and has no timeline. The company has struggled to recruit senior managers and dealers who understand the market. Its site contractors have been hauled in front of the European Parliament over labour conditions.
This is a vertically integrated manufacturer with its own battery business, a cash pile, a government behind it and every commercial incentive on earth to move fast. Three years in, it is still ramping.
If the best-capitalised and most motivated localiser on the planet needs three years, the localisation slide in the diligence pack is a work of fiction.
Not a lie. A fiction. The people who wrote it believe it. That is what makes it dangerous.
The wall that moved
It is worth being precise about what the tariff wall actually did, because the consensus reading is wrong.
BYD's cars still land in Europe carrying roughly 27 per cent in duty, and China-made vehicles still took a record 9.3 per cent of new car sales in the bloc last December. The duty did not stop the shipping. It never does. Ships kept sailing round the Cape when the Red Sea closed, and they will keep sailing through a tariff.
What the wall did was reprice the decision to remain a shipper. Chery, SAIC, Leapmotor and BYD are all now putting steel in European ground, not because the container stopped working but because the container stopped being defensible.
That is the shift. It is not about cost. It is about position.
The missing row
Here is why capital keeps mispricing this, and it is a structural problem rather than a stupidity problem.
Distance behaves beautifully in a model. Freight, duty, insurance, currency exposure. It is a line item. It is measurable, hedgeable, insurable and reversible inside a quarter. If the number moves you reroute, renegotiate, or eat it.
Proximity behaves like nothing in the model at all. It arrives first as capex, then as several years of ramp losses, then as an asset that does not exist until the day it works. The discounted cash flow punishes it at every stage of its life except the last one. There is no row for a supply base that has to be created, or a plant director who cannot be hired locally, or a regulator who has never met you.
Worse, the market prices the announcement. A localisation plan is worth something on the day it is declared and nothing thereafter, because nobody goes back three years later to check whether the steel arrived. There is no post-mortem culture around industrial promises. There is barely a post-mortem culture around industrial results.
The precedent
This is not a novel situation and the returns are a matter of record.
Washington squeezed Japanese car exports in 1981. The Japanese response was not a lobbying campaign. It was concrete. Honda in Marysville by 1982. Nissan in Smyrna by 1983. Toyota in Georgetown by 1988.
Those plants became a forty-year annuity and a market position that nobody has dislodged since. The capital that funded them was underwriting something that could not be repriced by a tariff, a freight rate or a strait. The firms that stayed shippers into that market were squeezed until they were nothing, and their export paperwork is landfill.
The lesson was never that protectionism works. The lesson is that when trade politics forces a choice between buying distance and building proximity, the builders take the next four decades.
The word is the tell
There is no good word for what these companies are actually doing, which is why it keeps getting waved through.
Localisation is a logistics word. It smells of translation memory and regional pricing. It sounds like a project, with a workstream and a Gantt chart and an end date. Nothing in the word suggests payroll, works councils, a regulator, a workforce that does not exist yet and a supply base that has to be grown from nothing.
What to underwrite
The diligence question is not whether a company intends to localise. Everybody intends to localise. Intent is free.
The question is what evidence exists that it can. Ask for the last one. How long from land purchase to series production, actually, not as planned. Ask whether the company has ever hired a plant director in the target country without importing one from home. Ask whether a qualified local supply base exists or whether it has to be created, and at whose cost. Ask who on the executive team has run a factory they did not grow up in.
Then look at what the answers do to the ramp curve, and price that, rather than pricing the slide.
The interface
Manufacturing Engineers qualify supply bases before the money moves. They walk the plant and read the ramp curve against the capex schedule. They know what a first-off part looks like when the process is not yet capable, and what it costs to find out late. They can tell the difference between a factory that is late and a factory that is not going to work.
That is not a diligence checklist. It is a discipline, and it is the one missing from the room when the localisation slide goes up.
Kaipability works at this interface, between the capital that is being asked to fund proximity and the industrial reality of building it. If your portfolio has a localisation slide in it, the conversation starts with the last one.
Questions this dispatch answers.
Written to be quoted by AI assistants and search engines. Self-contained answers, verdict first.
