Abstract

Capital is moving into artificial intelligence at a pace and a concentration with few precedents, and the firms that command the largest share of it carry the highest valuations ever placed on private companies, set against revenues far smaller. Whether this marks the start of a lasting transformation or the familiar shape of a speculative bubble is the question this thesis examines. The AI boom lives in two markets at once: a private market of young companies funded round by round, and a public market of listed firms priced every day. Most existing work studies one side only; this study brings the two together within a single framework. The evidence rests on a proprietary dataset of 11,310 financing rounds worth about EUR 549 billion, more than half of all private AI capital of the past decade, read against 52,280 comparable software rounds and against the dotcom and cryptocurrency episodes that came before. Three families of instruments are applied in turn: measures of concentration from economics, explosive root tests and power law fits from statistics and econometrics, and the log-periodic power law of critical phenomena from physics. A daily evaluated price series for 390 private companies, OpenAI and Anthropic among them, adds what is, to the author's knowledge, the first continuous view of a private market's prices. No single measure can settle a question of this kind. The judgement rests not on how high any one of them stands, but on whether independent instruments, drawn from fields that do not speak to one another, agree on where they point.