AI 'Tokenomics' Is a Trap. CIOs Should Avoid It at All Costs.
CIO, Friday, October 23rd, 2026
A columnist argues token-based AI cost metrics repeat the mistakes of function point analysis and don't tie spend to business outcomes.
Responding to Accenture authors who propose tokenomics as the way to account for and control enterprise AI costs, the columnist argues the idea only pretends to link spend with outcomes.
He compares tokens with Function Point Analysis, which IBM introduced in 1979 to size software but which gave business leaders cost metrics they could not understand.
Tallying tokens, like counting bricks in a wall, says little about the value delivered.
The piece questions whether boards care about tokens at all and urges CIOs to measure AI against business results rather than consumption units.