Value Coverage: A Measurement Framework for Technology Investment Booms, with a Live Application to the AI Capital Build-Out

Is the AI technology boom a bubble? The question is being argued almost entirely by anecdote, one side pointing to the technology potential and the other to the spending. This paper offers a way to keep appropriate accounts rather than a direct answer to the question. It proposes that any technology investment boom can be described by three quantities, each measurable from public data. The first is private coverage (Cp): the earnings the capital throws off, set against the annual cost of owning it. The second is social coverage (Cs): the value the technology creates for the wider economy, net of its harms, against the same cost. The third is financing fragility (F): how much of the build is paid for with outside money, and where in the structure that borrowing sits. The paper defines the three, sets out two hypotheses, one for the probability of a correction and one for its cost, in a form that can be estimated, places past booms on a simple map, and reports a live, open implementation for the AI build-out. On data to mid-2026 the headline (capital-holder) readings are Cp approximately 0.31, Cs approximately 0.68 at a ten-year horizon and approximately 1.13 at thirty, and F approximately 0.72, in the speculative range; a second, ecosystem-wide coverage line reads approximately 0.54, a wedge of approximately 0.23 above the capital-holders. What is offered is a measurement programme with a working instrument, and a not-yet-tested theory of bubbles. Two steps would make the central claim testable: an expectations benchmark that says what coverage the market is pricing in, and a historical backtest against the price-based crash predictor of Greenwood, Shleifer and You. Both are specified. The paper is equally clear about what such measurement cannot do. It can deliver consistency rather than accuracy, and a characterised boom rather than a prophecy.