The EU Didn't Drop the Tariff. It Confiscated the Price War.

"The EU dropped its tariffs on Chinese EVs." Every headline said so in January.

Walk into a BYD showroom in Germany after the deal, though, and you notice something. The price on the car hasn't moved.

What no headline mentioned is that, legally, it now can't come down.

The tariff isn't being removed. It is being replaced by a price floor, a minimum import price, or MIP. A tariff is a cost. A floor is a different kind of thing altogether. A cost you can absorb, spread across volume, engineer away. A floor you cannot push through.

So this piece isn't about the news that "the tariffs are gone". It is about what a floor does to different Chinese carmakers, because it does very different things to different ones.

The whole argument in two sentences, because the rest is just working it out:

The MIP hands the tariff money back to the carmakers who weren't fighting on price, and forces a price rise on the ones who were. Either way, the price war is over.

It's not a tax cut

The reflex reading of the MIP is "a softer tariff". That reading is wrong, and the error matters.

A tariff sits on top of your cost. You pay a percentage at the border, then do what every manufacturer does with a cost: try to dilute it. More volume, local production, a leaner bill of materials.

A floor does the opposite. It doesn't add to your cost; it takes away your freedom on price. Under the MIP you commit to a minimum price, and the load-bearing detail most coverage skips is this: the floor is pegged to the net price to the first independent consumer in the EU. Not the price to the importer. Not the price to the dealer. The price the customer actually pays.

That single design choice is the whole story. The floor can't be quietly undercut a layer down the chain. The dealer can't rebate through it. The carmaker can't discount through it.

So stop asking how much tax anyone saved. Ask the only question that matters for a Chinese brand in Europe: can they still fight on price?

"Equivalent effect": who gets a refund, and who gets a bill

This is the part that decides everything, and it turns on a single phrase in EU trade law.

A price undertaking is not a discount on the tariff. By law the floor must have "equivalent effect to the duty": it has to remove the same injury the duty removed. Read that carefully. The floor is set so the consumer price lands roughly where the duty would have put it, had the duty been passed through to the buyer.

So the real question was never whether the MIP refunds the tariff. It is whether you were passing the tariff through in the first place.

Take the BYD ATTO 3: a volume C-segment SUV, sold into the most price-sensitive part of the market, just the kind of car that got Chinese brands into Europe on price. It lists in Germany at €38,990.

I don't know BYD's real landed cost, so pick a range. Say the car lands at the border somewhere between €18,000 and €22,000. At a combined duty of about 27% (BYD's 17% countervailing rate plus the standard 10%), that is roughly €5,000 to €5,900 a car. Put in your own landed cost; the direction won't change.

Now watch what that figure becomes.

If BYD was passing the duty through, having already lifted the price to carry the tariff, the MIP hands that €5,400 or so back. Same price to the customer; the money stays with BYD instead of going to EU customs. A clean refund.

If BYD was absorbing the duty, eating it to keep the car cheap, which is just what fighting on price means, then its current price sits below the floor and the MIP refunds nothing. It forces the price up, by roughly that same €5,400.

Those are the same sentence read from both ends. The refund and the forced rise are not two outcomes for two companies; they are one mechanism, and the side you land on is settled by whether price was your weapon.

A fair caveat about the size of the swing. The floor can be set two ways: built up from the exporter's own import cost plus the duty, or benchmarked against a comparable European-built car. The second can land higher than the duty alone would, so for some brands the bill is larger than the tariff they were dodging, and even a brand that was passing the duty through can face a small rise. That changes how hard the bill lands, not who is on which side of it.

For the brands that collect the refund, it matters more than it would for a Western incumbent. BYD's net profit fell by nearly a fifth in 2025, then by 55% year on year in the first quarter of 2026, its steepest drop in six years, all while it sold more cars than ever. For a company squeezed like that, the money is oxygen. For the brands that get the bill instead, the floor is the tariff finally reaching the till.

There is good news in here. It just isn't general. It goes to whoever wasn't using the weapon the MIP takes away next.

The weapon is confiscated

Consider that same ATTO 3 again.

The Comfort version listed in Germany at €44,625, then was cut by nearly 15%, to about €37,900, in a stretch the German trade press ran under a blunt headline: "Der Preiskrieg geht weiter", the price war continues. The base car sits at €38,990 today. However you measure it, the same model has swung by the better part of €7,000. That swing is the weapon: not the low price itself, but the freedom to move it, hard and fast, to take share when the moment calls for it.

Under the MIP that move is a breach of contract. Because the floor is pegged to the net consumer price, you can't cut in the showroom and you can't rebate through the dealer; the guidance asks for vehicle-level traceability from export to final sale, and it watches for cuts to the net price afterwards. The knife isn't blunted. It is out of the drawer.

One point of precision, because it is the difference between a slogan and the truth. The MIP is negotiated model by model; the Commission accepted its first BEV undertaking, covering a single model, in February 2026. So this is not a global switch. A brand can leave one model under the tariff and keep fighting on price there, while placing others under the floor for the refund. What it cannot do is both at once on the same car: you cannot collect the refund and keep the price weapon on one model. The weapon isn't destroyed. It is a line on the budget you spend model by model, and every model you point at the refund is a model you have disarmed.

Here I will flag my own uncertainty, because a price war doesn't necessarily die when you freeze the list price. It can move. The floor sits on the vehicle's net price; it says nothing about the monthly payment. You can already watch the fight relocating. Leapmotor, a second-tier Chinese brand riding into Europe on Stellantis's network, is leasing its T03 city car in Germany for €49 a month. That is not a list price, it is a lease structure, and the battlefield moves with it, from the figure on the windscreen to the figure on the monthly invoice, where the floor doesn't obviously reach. "Obviously" is the word doing the work. Subvented leases and generous trade-ins are just the sort of indirect discount the EU's ban on cross-compensation exists to catch, which I'll come to. So the honest version isn't that the weapon is gone. It is that the obvious weapon is gone, the next one is contested, and the regulator is already standing near it.

The second tier won't die: it moves sideways

It would be too easy to leave the weaker brands for dead. They aren't, and writing them off that way would be lazy.

Start with what they are not, which is cheap and nothing else. Dacia Spring rides on Renault's dealer network, brand trust and residual values; MG trades on a reactivated British name SAIC paid real money for. Price was the wedge that got them through the door, not the only thing in the house.

But the wedge is what the MIP removes, and without it the move isn't death, it is sideways. You can see it already. That same Leapmotor T03 lease, sitting on Stellantis's 850-odd European outlets, isn't a price war; it is a fight over channel and finance, aimed at fleet and middle-market lease buyers rather than at the list price. Fleet and lease-only channels, regional niches, white-label and supplier deals: that is the sideways move, and the model-by-model menu is how you fund it, by pulling the entry car or walling it off under the tariff where it can't scale.

That is a real move. It is also a confession. You don't restructure around a weapon you have lost; you retreat around it. The cheapest slice of the market, the one they came in through, stops being theirs to take.

The only real exit cuts deeper than the floor

There is one move that escapes both the tariff and the floor at once: build the car in Europe. A locally built car gets EU market access without the extra import duty. It is the genuine exit, and it is far slower and dearer than the "China speed" reputation suggests. For most of the field it isn't an option at all.

Look at BYD's own timeline. Its €4 billion plant in Szeged, Hungary, with a nominal capacity of around 300,000 units, has slipped: through 2026 it is expected to make only tens of thousands of cars, with the real ramp pushed into 2027. Meanwhile BYD has shifted weight to a $1 billion plant in Turkey, due on stream by late 2026, because Hungarian wages and energy made the original plan dearer than planned.

What the timeline alone doesn't tell you is the deeper cost. Localisation doesn't merely add years; it dismantles the economics that made the export model work. European labour, local-content rules, energy, unions, a slower iteration cadence: clear all of those and you haven't kept your cost advantage, you have spent it. You slowly become another European carmaker. The exit leads out of the very thing that made you dangerous.

The obvious rebuttal is Tesla, which localised in Shanghai and stayed lethal: scale, cost and iteration all intact. True, and it is the exception worth handling with care. Tesla localised at a scale that turned one plant into a global export hub, on a narrow model range that amortises cleanly. BYD has the volume to attempt something similar; whether a twenty-model range can amortise a plant the way Tesla's handful of models does is another question, and not a settled one. What is clear is the floor beneath: a brand selling tens of thousands of cars a year can't spread a €4 billion plant at all. For them, localisation isn't a slower version of BYD's escape. It is no escape at all.

The price war separated the brands with a margin cushion from those without one. Localisation cuts deeper: it separates the brands big enough to build from the brands that will only ever import.

The third wall the MIP can't touch

There is a wall beyond the floor that the MIP doesn't even reach: the subsidy layer.

It is separate from the tariff, drawn nation by nation, and increasingly tied to where the car is built, so a deal struck in Brussels doesn't touch it. France is the sharpest example today. Its 2026 eco-bonus scores the car's production carbon footprint across its whole industrial life, and sets the bar (60 points out of 80) high enough to shut most made-in-China EVs out of the subsidy: the MG4, the China-built Dacia Spring, the very BYD ATTO 3 from the sum above, and the Shanghai-built Tesla Model 3. The French-built Renault 5 clears it; so does a Berlin-built Tesla Model Y, while its Shanghai-built sibling does not. Clean-transport groups say plainly that this is the design, not an accident.

And the point doesn't need France to spread anywhere, because every country draws this gate on its own axis. Germany scrapped its EV subsidy outright in 2023, then brought a different one back on 1 January 2026: up to €6,000, graded by household income and capped by vehicle price. Nothing to do with where the car is built, but a separate gate all the same, and one a higher MIP floor can push a car clean out of by lifting its price past the cap. France screens on production carbon, Germany on income and price, the next market on something else again. Subsidy eligibility is a second wall, national and drawn however each capital likes, and the MIP opens none of it. Clear the floor and you can still lose several thousand euros at the point of sale: in France for building the car in the wrong place, in Germany for pricing it over the cap.

An operator's view: the same instrument, run before

There is a layer here you only see from inside the operation, because it isn't a question of policy but of enforcement.

A MIP is not something you accept. It is something you enforce, in your own systems, every day. The floor is set model by model and configuration by configuration, pegged to the net consumer price, so every configuration and every transaction has to clear a floor of its own. That is a problem for your pricing engine and your dealer management system: hold the floor by configuration, keep it consistent across every dealer, watch what each car actually sells for, and prove to Brussels that you are holding the line. Miss it on a single configuration and the penalty is binary: the whole undertaking can be withdrawn and the full tariff returns.

I know this is hard, because the same instrument has run once before.

This is not Europe's first minimum-price undertaking. It is the first on cars, and the template is a decade old.

From 2013 to 2018 the EU ran a minimum-price undertaking on Chinese solar panels. It did not fail on price. It failed on circumvention, and the car undertaking is built against the same three moves, whether or not anyone consciously learnt from solar or the template simply never changed:

• Canadian Solar ran cross-compensation, bundling covered and uncovered products to one customer as a back-door discount, and was expelled → The law makes mitigating cross-compensation an explicit condition of any undertaking, so finance and bundling can't be used to discount through the floor.
• ReneSola's network made provenance unmonitorable and its undertaking "impracticable", and it was expelled → The floor is pegged to the consumer price, not the dealer's, so a distribution layer can't hide the real transaction price.
• With floors set product by product, exporters split one product into near-identical versions to slip beneath them → The car floor is set configuration by configuration, so trim-splitting only meets another floor.

I should be honest about what I can prove here, because the temptation is to tell a tidy story. All three defences are real and written into the guidance: the cross-compensation condition, the consumer-price reference point and the configuration-level floors are stated outright. What I cannot show is the causation, whether they were put there to close these particular solar holes or are simply how this instrument has always been shaped. And a step further: "the EU learnt" is itself a story I would be choosing. A price undertaking is a standard legal tool with a fixed architecture, and these clauses may be nothing more than its boilerplate, present on solar too and circumvented anyway. The mapping in that list is mine. The defences are not: they are real, and that is what decides whether you can game them. You can't.

But the solar story has an ending that ought to correct the tone of everything I've just written. In 2018 the EU let those measures lapse, not because they had failed but because it had decided cheap Chinese panels now served its own renewable-energy targets, with the installers who wanted them pushing hard and Europe's own panel makers objecting in vain. Pause on that. An institution that drops a trade defence the moment a bigger policy goal and a louder lobby pull the other way is not a strategic machine with a thirteen-year memory; it is a fragmented, reactive system that happens to deploy instruments with sharp edges. So when I say the car MIP "closes" the solar holes, read that as a fact about the shape of the tool, not about a steady hand on the tiller. The edges are real. The grip is not.

So the real test isn't whether you accept the floor. It is whether you have the configuration-level enforcement chain to live inside it honestly: one more line that sorts the field, this time on the maturity of your systems.

What the floor reveals

Put it together and the same shape keeps appearing. Surviving the MIP, not signing it but surviving it, takes several things at once: a margin cushion, a range that doesn't live or die on the cheapest segment, the scale to localise, and systems good enough to prove compliance. Few Chinese carmakers have all four.

That is the same thing I argued last time, in a different form. The Chinese push into Europe was never one story. It is a sorting, a divergence happening inside the group rather than between China and the West. The MIP doesn't change that. It brings it forward, and cuts the line deeper.

Let me set out what I'm sure of and what I'm betting on, in three tiers rather than two.

One, a fact: the price war is over. The floor is real and the law is clear.

Two, my reading: it sorts the field along the four lines above, margin, portfolio, scale and systems. That is a judgement, not a law of nature; a recession, a collapse in demand or a breakthrough in batteries could shuffle it again.

Three, the bet beneath the bet: that the regime lasts long enough for any of this to play out. That is not a question of industrial policy but of politics, and EU trade politics is fragmented, member-state-driven and reactive, the same instincts that let the solar floor lapse in 2018. Which points to a move the four sorting lines miss. If the regime itself may not survive a political cycle, the sharpest play for a second-tier brand may not be to sink a few hundred million into localising for a floor that could soften within three years. It may be to take the forced price rise, bank the cash the refund hands back, and ride the regime out on the smallest irreversible spend it can manage, betting on Brussels' own record of letting these instruments lapse. I am not certain that is the right call. I am certain it belongs on the table, and the reflex to localise now leaves it off.

But the direction holds.

A tariff is a cost everyone carries together. A floor is a test only some of them can pass.

Sources

• EU publishes guidance on minimum price mechanism with China — electrive, January 2026
• Battery Electric Vehicles from China: a timely update on the EU's anti-subsidy duties and the Commission's new guidance on price undertakings — CMS Law
• BYD ATTO 3 (MY25) (2025-2026) price and specifications — EV Database
• Definitive duties adopted by the EU on Chinese battery electric vehicles — Cleary Trade Watch, October 2024
• BYD Atto 3 cut to €37,900 ("Der Preiskrieg geht weiter") — InsideEVs.de, 2024
• Commission accepts price undertaking from Chinese electric car producer — European Commission Trade, February 2026
• BYD's 2025 net profit drops 19% as domestic price war bites — CnEVPost, March 2026
• BYD first quarter profit falls 55.4% — Reuters / Yahoo Finance
• Leapmotor offers T03 lease in Germany for €49/month — CnEVPost, May 2026
• BYD shifts focus from Hungary to Türkiye for European EV production — electrive, July 2025
• Tesla Gigafactory Shanghai delivered over 3 million cars in less than 5 years — Global China EV
• Bonus écologique 2026 : barème, conditions et modèles éligibles — mondial.paris / ADEME
• Germany cuts all zero-emission vehicle subsidies with immediate effect — electrive, December 2023
• Germany plans targeted EV subsidies for low- and middle-income households — EU Alternative Fuels Observatory
• European Commission wants 3 Chinese manufacturers removed from minimum price agreement — CleanTechnica, March 2015
• Three Chinese manufacturers to be removed from minimum price agreement — pv magazine, March 2015
• EU lifts tariffs on Chinese solar panels — Clean Energy Wire, September 2018