In the last week of August 2026, three things happen in the automobile industry. BYD loads another seven thousand cars onto one of its own ocean carriers, part of a fleet it built because chartering ships had become a bottleneck. Tesla registers two hundred more robotaxis in Texas — the fleet has more than doubled since July, per TxDMV data — and the cars keep teaching themselves: to cross parking lots, to find their passengers, to run themselves through the wash. And in Wolfsburg, the supervisory board of Volkswagen — Europe’s largest industrial group, 600,000+ employees, ten brands — prepares to vote, tomorrow, on its own restructuring using three competing proposals: one from labor, one from management, one from the state of Lower Saxony. Not one of them can pass against the other two. The board has already failed once, before the summer break, to give its own executive team a mandate. The Handelsblatt put it in a sentence I cannot improve on: other companies have competitors who make their lives difficult. Volkswagen has its own governing bodies. The margin tells you how much time is left. The group earns 3.8 percent. On a Golf, that is roughly €1,300 — the price of the infotainment package. Volkswagen operates, by its own CEO’s admission, at costs 30 percent above comparable manufacturers, a gap of about €1.5 billion per year. And the gap is not primarily on the assembly line. It sits in the superstructure: middle management, cross-functional units, coordination layers — an apparatus, as one German commentator noted, next to which a state bureaucracy looks lean. Most coverage treats this as a cost problem, a leadership problem, or a union problem. It is none of these. It is a Tainter problem, and Tainter problems have a property that makes them different from cost problems: they cannot be solved by the system that has them. The frame: complexity as a solvent Joseph Tainter’s The Collapse of Complex Societies (1988) is an archaeologist’s answer to a question usually left to moralists: why do sophisticated civilizations fall apart? His answer needs three sentences, not a chapter. Societies are problem-solving organizations. Each problem they solve — defense, irrigation, legitimacy — adds a layer: administrators, institutions, rules. Every layer is rational at the moment of its creation, but each successive one yields less benefit at higher cost, until the society spends its surplus maintaining structure rather than producing anything — at which point collapse becomes, in Tainter’s coldest phrase, an economizing process.¹ Two details matter for what follows. First, complexity is almost never rolled back voluntarily, because every layer has a constituency that will defend it more fiercely than any beneficiary will attack it. Second, declining returns can be masked for decades by an external subsidy — Rome had conquest plunder, later empires had colonies or cheap oil. While the subsidy flows, the superstructure feels affordable. When it stops, the structure does not shrink. Only the income does. Wolfsburg as a complex society Now run Volkswagen through this machine. Every crisis in the company’s modern history was solved by adding a layer. The postwar settlement produced the VW Law and the state of Lower Saxony as a permanent shareholder. The labor conflicts of the seventies and nineties produced the densest co-determination architecture in industrial Europe, including a works council with power no American or Chinese executive would believe. Dieselgate produced a compliance and oversight apparatus. The electric pivot produced Cariad and a set of parallel software organizations. Each standoff between capital, labor and state produced new committees whose function is to manage the standoffs between capital, labor and state. Not all of the accretion was forced by crisis, and this is the part that cuts deeper: on the product side, complexity was chosen. Ten brands, several competing for the same customers; combustion, hybrid and electric drivetrains engineered in parallel; three vehicle platforms — MEB, PPE, SSP — under simultaneous development; multiple software architectures where competitors run one. Every one of these was an answer to a real problem. Together they are a museum of answers, each with its own staff — and because these exhibits were selected by strategy rather than forced by crisis, they are harder to defend and, in this structure, no easier to retire. Each layer was locally rational. Together they now consume the margin. That is the declining-returns curve, and Volkswagen has the rare distinction of publishing its position on that curve every quarter: 3.8 percent. And the external subsidy? For twenty years it was China. In 2019, the last full year before Covid, the Chinese joint ventures delivered a proportionate operating profit of €4.4 billion — more than the entire core VW brand earned that year (€3.8 billion). And because the ventures are equity-accounted, the money arrived below the operating line: profit without production, cash without overhead. Income of that quality makes any overhead feel affordable, which is why the question of whether Wolfsburg’s earned its keep never had to be asked. That income has now collapsed — local competitors took the market — and the apparatus is still standing there, invoiced monthly, with nothing underneath it. This is Tainter’s moment, transposed to the corporate register: the subsidy ends, the maintenance costs remain, and the system discovers it cannot vote its way to simplicity. The immune system Why not? Other companies restructure. Nokia amputated itself. General Motors went through bankruptcy and came out lighter. Because Volkswagen, uniquely, is legally armored against its own reform. The VW Law gives Lower Saxony, with about 20 percent of votes, a blocking minority — major decisions require 80 percent. Plant closures need the supervisory board, where labor representatives and the state together outvote any hard restructuring. Renault lives with a state shareholder too; the difference is that Paris holds no statutory veto, and Renault’s management does not face a works council with equivalent legal power. When Volkswagen’s management board began exploring a carve-out of the core brand — essentially an attempt to route around the blockade — IG Metall called it what, from inside the system, it is: an attempt to circumvent the VW Law and co-determination. The system detected reform as a pathogen and produced antibodies. Volkswagen is often described as a company distorted by politics. The description has it backwards. Volkswagen was constructed in 1960 as a social contract with an attached car factory. The three-party structure is not an impediment to the institution’s purpose. Over sixty years, it has become the institution’s purpose. Cars are the revenue model of the deal, and for as long as China paid, the deal never had to notice that the revenue model was decaying. There is a fourth party, and its silence is part of the architecture. Porsche SE — the holding of the Piëch and Porsche families — controls 53.3 percent of the voting shares, an outright majority, and is the only actor at the table whose interest is value rather than jobs or sites. On paper, the families are the natural constituency for reform. In practice they have been quiet for years: the lesson of Dieselgate and the botched Porsche listing was that public confrontation with IG Metall and Hannover costs the share price more than any structural victory could return. The one player who could press the issue has concluded that pressing it is the worst trade available. The veto architecture does not even need to outvote them. At least, that has been true for a decade — hold the thought; this week put it in play. And so three proposals land in front of one board. Each party optimizes its own subsystem — jobs, sites, control — and each values what it might give up more highly than anything the collective might gain. This is loss aversion at institutional scale, which is simply what a veto architecture produces. The final feature is elegant: with three competing proposals, whatever happens, no one will have decided it. The structure manufactures outcomes without authors. Responsibility does not disappear; it is diffused so evenly that it can never again be located. Even the personnel are outcomes without authors: CEO and board chair hold their offices less because anyone trusts them to solve the problem than because no faction retained the strength to install alternatives. Made in Germany, priced in Wolfsburg Abstractions about overhead have a way of sounding deniable, so here it is priced in showroom euros. Kia’s PV5, an electric van, starts at €38,390 in Germany; the seven-seater at €40,190. Volkswagen’s ID. Buzz — the electric Bulli, the most emotionally loaded product in the portfolio, the one that is supposed to embody “we were once affordable and beloved” — starts at €52,271, runs to €62,903 with the long wheelbase, and €73,239 as the GTX. The Buzz carries more power and bigger battery options, which the PV5 buyer does not miss — and the gap survives any fair configuration: €15,000 to €25,000, spec for spec. That difference does not buy a better vehicle. It buys Wolfsburg: the coordination layers, priced and shipped with every van. The sharper comparison is not price but margin. A BYD Seal crosses the planet on a company-owned ship, pays EU punitive tariffs, and lands in Germany at €47,990 — at a profit. Volkswagen builds the Golf at home, pays neither tariff nor ocean freight, and clears €1,300. Not all of that distance is Wolfsburg’s making: part of it is BYD’s vertical integration, part is Chinese industrial policy and what it does to steel and energy prices. But the part Volkswagen itself has quantified — the 30 percent against comparable manufacturers —
