Tokenomics - why AI revenue is out of synch with costs, says Bain & Co – for users and vendors alike
AI vendors' unit economics are broken—and Bain just connected the dots between their margin collapse and enterprise customers' runaway token bills.
Why it matters
A Bain & Co analysis reveals structural misalignment in AI vendor pricing and cost models: as inference costs rise and token consumption accelerates unpredictably, neither vendors nor customers can forecast spend or margin. This directly affects enterprise budget planning and vendor selection—the economics of scale that were supposed to save money are working in reverse.
The key facts
5 to knowBain & Co analysis: AI vendor costs and customer expenditure are out of sync
Token-based pricing creates unpredictable consumption patterns for enterprises
Vendor inference cost inflation vs. revenue realization mismatch identified
Economics-of-scale assumptions failing in real deployments
Published October 1, 2026
Go to the source
diginomicadiginomica.com
Publisher excerpt: AI vendors’ costs are out of control, and so is users’ token expenditure. Turns out those problems are intimately related.