Cassino is not being rescued. It is being let out.

When a Chinese manufacturer turns up at an idle European plant, the coverage calls it a strategic partnership. It is closer to a tenancy, and the difference decides who pays for the building.

Stellantis in Cassino is where that difference is currently being negotiated in public. The plant is the clearest case in Europe of capacity that was built, staffed and then left with nothing to make, and it is being offered to Chinese partners while the same companies ask Brussels for a rule to keep Chinese cars out. Neither half of that is irrational. Both have to be read at once.

Europe did not lose the demand. It built for demand that never arrived.

Analysis by AlixPartners, reported by Bloomberg in March 2026, puts Stellantis at roughly 6.5 million vehicles of annual capacity in Europe, running at an average of 46%. That is around 3.5 million cars of capacity with nothing to build, and 14 of 24 plants below the 50% line usually treated as the point where a site stops paying for itself. The same analysis puts European utilisation close to 71% in 2017.

Read that as a demand story and you reach the wrong conclusion, which is that volume comes back and fills the buildings. Read it as a capacity story and you reach the right one. The buildings were sized against a forecast of a European market that has not turned up, and nothing in the current mix argues that it is about to.

Stellantis does not dispute the arithmetic. At a parliamentary hearing on 17 June 2026, chief executive Antonio Filosa said the group aims to exceed 80% utilisation in Europe by 2030, and named three routes to it: more production, converting some plants, and manufacturing partnerships with other carmakers. The third route is the one that matters here. The company's own plan for the empty half of its European footprint involves filling it with somebody else's cars.

Germany is working through the same arithmetic and reaching a different answer, and we take that apart separately.

Three different counts say three different things, and they have to be kept apart

Three numbers circulate about this plant and the group behind it. They do not contradict each other. They measure different objects, and whoever quotes one of them as "the" utilisation figure is about to be wrong in a room where being wrong costs money.

The first is a group ratio. 46%, AlixPartners, across all 24 Stellantis plants in Europe. It tells you how much of a continental industrial base is working. It tells you nothing about any single site, because an average of 46% is equally consistent with every plant at 46% and with half of them full while half sit dark. In this case it is much closer to the second.

The second is a plant ratio. The research office of Fiom-Cgil, in an analysis published in May 2026, put Cassino's 2025 output at 19,539 vehicles against a nominal capacity of 285,000. That is under 7%. Different object: one site, one year, measured against what the site was built to do.

The third is not a ratio at all. It is a calendar. The Fim-Cisl quarterly count for the group put Cassino at 2,916 cars in the first quarter of 2026, down 37.4%, built across 16 working days in three months. Its half-year count put the plant at 6,700 vehicles, down 36.2%, across 39 days, with production described as running five or six working days a month and a full-year expectation of around 13,000 units. That does not describe a plant running slowly. It describes a plant that is mostly shut, with a workforce attached to it.

One caution before any of these travels further. Even the denominator is contested: the same plant is quoted at 285,000 units of capacity in the union analysis and at 300,000 in the Financial Times reporting of August 2026.

The reason to be pedantic is not tidiness. All three numbers get used as opening positions, and each one argues for a different deal. A group ratio argues for a European restructuring. A plant ratio argues that this particular site is finished. A calendar argues that the site is alive, staffed and idle, which is the only one of the three that supports a tenant.

April, May, August: not an industrial plan, an auction in progress

In early April 2026, Dongfeng visited Rennes and Madrid along with sites in Italy and Germany. Italian reporting at the time had four Stellantis plants on the table, Cassino among them, and the Italian government signalled it would not stand in the way.

On 20 May the answer came, and it was not Cassino. Stellantis and Dongfeng announced a European joint venture, 51% Stellantis and 49% Dongfeng, headquartered in Europe, covering sales, distribution, production, purchasing and engineering for the Chinese group's electric vehicles, with at least one Voyah model to be built at Rennes. Italian worker representatives were told the group was not yet ready to put an operating plan for Cassino in front of them.

On 5 August the Financial Times put Cassino back on the table, this time with Leapmotor or Dongfeng, and alongside the Maserati brand.

Three positions in four months, with the order of the sites changing twice. An industrial plan names a plant, a model and a date. This names a shortlist. What happened between April and August is price discovery, and the asset being priced is not the building. It is the combination of a building, a trained workforce, a supplier base within driving distance, and a national government with a political reason to help.

BYD's refusal explains the mechanism better than the acceptances do

In May 2026, Bloomberg reported BYD in talks with Stellantis and others about taking on underused European plants. What BYD would not take is the structure. Stella Li has been explicit that the company prefers to run its own operations: "It's very hard to partner and ask permission from another person. We prefer to run everything on our own." Asked about a jurisdiction where a joint venture was being sought, her answer was "I don't think a JV will work."

That inverts the usual framing, and it is the most useful thing in the whole file. The question at an idle plant is not whether a Chinese manufacturer will come. It is which one, and what it needs from you.

A manufacturer that needs a European building, a European workforce and a European supplier base will pay for the use of them, and will take 49% of a company to get it. One that does not need them will buy the asset outright, build its own, or keep importing and price above whatever floor Brussels sets. Those are three different counterparties and three different outcomes for everyone standing around the plant. The first keeps the site running and the ownership where it is. The second changes the owner. The third does nothing for the site at all, whatever the announcement says.

Telling them apart before a term sheet exists is not a press-reading exercise. It is the distinction a deal thesis has to survive before anything else, and it is the one written down first on a buy-side or sell-side mandate, because it decides whether you are selling, letting, or being used as leverage in somebody else's negotiation.

Asking Brussels for a European content rule while negotiating with Dongfeng is a double track, not hypocrisy

On 12 June 2026, Volkswagen, Stellantis and Renault, together accounting for roughly 60% of European car output, sent a joint letter to members of the European Parliament. It asked for a "Made in Europe" rule built on two thresholds: 70% of vehicles sold in the EU carrying 70% of their value from inside the bloc, plus Norway, Iceland and Liechtenstein, measured across the full value chain from engineering through to manufacturing, with the remaining 30% left open. The press shorthand for it is the 70/70 threshold, which is a description rather than an official name, and the letter was reported as a request to simplify and loosen a stricter Commission draft tying subsidies to assembly inside the EU plus 70% local components. It is the same rule that decides which Chinese plants in Europe are actually being built and which stay on paper.

Three weeks earlier, the same Stellantis had signed a 51/49 joint venture with Dongfeng to build Chinese electric models in a French plant.

Stated like that it reads as a contradiction. It is not one. Read the threshold again and notice what passes it. A Voyah assembled at Rennes inside a company Stellantis controls, with European engineering counted in, is a European car by that definition. A container of kits assembled by an importer is not. The three carmakers are not asking for a wall. They are asking for a wall with a door, and for the door to be the shape of the arrangements they have already signed.

It is worth naming plainly, because the public argument is conducted in the language of protection and the private one in the language of utilisation, and both are sincere. It is also the same substitution Italian car retail has already been through, one floor down: there, the brand on the door stays and the name on the share register changes. Here the badge on the bonnet stays European and the decision about what runs down the line moves.

For a fleet buyer, the badge on the bonnet is not the question

Whoever signs for several hundred cars is not underwriting a flag. They are underwriting what the vehicle is worth in 36 months, and that is set by used demand, network coverage and parts, not by the postcode of the assembly line.

On that specific point there is data, and it is not comfortable: figures from DAT, the German vehicle valuation body, reported in May 2026, show Chinese electric and plug-in hybrid models depreciating at roughly twice the market average, with the gap widening. Who carries that fall inside a leasing contract is a separate question with a separate answer, and we take it apart on its own.

What matters here is narrower. Assembling a Chinese model in Cassino does not solve that problem. It relocates it. The things that would actually move the number, a dense service network, parts availability, a certified used programme that gives the second owner a reason to bid, are not delivered by a contract-assembly arrangement, and a joint venture built to fill a line has no particular reason to build them first. A European compliance plate is not, on its own, a residual value argument.

The only party that can move now is the supplier base, and its window is shorter than the plant's

The plant will survive under somebody's name. That is the honest reading of everything above, and it is also the reading least useful to the companies around it, because a supplier does not need the plant to survive. It needs the plant to order.

The interval is the problem. Fim-Cisl's expectation for Cassino in 2026 is around 13,000 vehicles. The relief actually scheduled is the next Maserati Grecale, expected at the plant from 2027, with the current Giulia and Stelvio reported as staying on the lines to the end of that year. The Stellantis plan of €5 billion for Italy to 2030 assigns Cassino to the medium-to-upper and luxury segment, which is a mission rather than a date: Melfi has a named Alfa Romeo C-SUV, Pomigliano a new generation of electric city cars from 2028, Atessa the next commercial vehicles. Cassino has a category.

So a supplier has to cross 2026 and most of 2027 on a customer working five or six days a month, then meet whatever follows with a different product mix, possibly a different brand, and possibly a partner whose purchasing organisation sits in Wuhan or Hangzhou. Three of those four variables are unknown today. The fourth, the calendar, is the only one already fixed.

A supplier that decides to move rather than wait has to know who is on the other side of that table. In this market the buyer is usually closer than the customer.

That is one question, not a mandate. What share of the order book depends on this line, what each partner scenario does to it, and which of them leaves a European supplier in the bill of materials. It is the shape of a commissioned piece of research: one question, the base rebuilt from registrations, filings and capacity, and a written answer to act on or ignore.

The conclusion for anyone sitting at a table around this plant is not that Europe stops building cars. It is that the terms have changed. A buyer pays for an asset and inherits its problems. A tenant pays for the use of it, for as long as the use suits him, and leaves the building where he found it. Cassino is being offered on the second set of terms. The numbers should be run that way: not what the plant is worth, but what the lease is worth, and how long it runs.