Britain’s defence nuclear expansion is normally read through the money that moves: appropriations voted, programme budgets allocated, contracts signed, capital projects begun. That reading captures what the state buys. It does not capture what the state keeps. Nuclear propulsion, submarine construction and through-life support expose contractors to categories of loss that are very large, very long-lived, difficult to price and, in some configurations, poorly matched either to commercial insurance or to the balance sheet of the company doing the work. Where that mismatch becomes acute, the government does not simply pay more. It absorbs part of the exposure itself, through indemnities, through limits on what it can recover from its supplier, through decisions not to insure, and through promises about what happens if it changes its own mind. None of this appears as expenditure. All of it changes the terms on which industrial work can be contracted at all. The question this raises is narrow but consequential: what, precisely, has the British state undertaken to bear alongside the prices it pays, and how much of that undertaking is visible in the public record?