European defence budgets have risen faster in the last three years than in the preceding three decades, and the public argument about that money has settled into a contest between two accounts. In the first, established prime contractors absorb the increase, because large platforms are what defence ministries know how to buy. In the second, a generation of cheaper and faster manufacturers is displacing them, because attritable systems have proved themselves in a war that the incumbents did not anticipate. Both accounts take the firm as the unit of analysis, and both are therefore answering a question that procurement documents do not pose. Programmes are not organised around firms. They are organised around a small number of allocations that a public authority makes when it writes a contract: who holds integration authority over the system as a whole, who qualifies it and against which standard, who holds the design data and the mission data, who carries support responsibility across the service life, and how frequently the authority obliges itself to reopen the market. Those allocations determine which companies can enter a programme at all, how much of the money they retain once inside it, and — the question a customer asks only when something has already gone wrong — whether a supplier can be replaced, or a capability sustained, without reopening the entire system architecture. The useful question is therefore not who receives the money, but how the architecture that distributes it is constructed, and what that construction permits the buyer to do afterwards.