A defensible total for annual US-linked external AI content licensing expenditure was not established in this research. Use transaction-level evidence to test your own opportunity; do not turn a broad market estimate into an archive valuation. The supplied $1–5 billion annual-spend and 20–40% annual-growth ranges are unvalidated planning hypotheses. They are excluded from the verified market dashboard below.
Define the perimeter before commissioning further market sizing: external payments for rights to existing content, split between training, adaptation, retrieval and evaluation. Keep internal data generation, annotation labour, general AI software, cloud infrastructure and corporate acquisitions separate. Otherwise, a large “AI training data” estimate can describe services that the content owner will pay for rather than royalties it might earn.
The most decision-useful denominator is smaller: buyers purchasing the owner's modality, subject area, freshness, rights scope and delivery format. No number of model developers establishes how many have an open budget for this archive.