Altman Questions Whether AI Token Counts Measure Business Value

OpenAI CEO Sam Altman said token counts show how much AI models process, but not the value businesses receive, a distinction relevant to companies using tokens to gauge AI productivity.

Altman made the remarks during an Oct. 31 appearance on the Cheeky Pint podcast with John Collison. He called tokens a “kind of fake unit” for assessing business value while acknowledging they remain a valid way to measure model use.

An AI token is a small unit of text or data that a model reads or produces. Providers use token totals to meter activity and calculate customer bills. Altman said many customers prefer the stability and predictability of token-based pricing, even when that means accepting inflated token counts. Outcome-based pricing ties charges to results but is harder to forecast.

Some companies have treated token consumption as a signal of productivity and AI adoption. Amazon and Uber criticized token-consumption measures as performance indicators. Both companies were setting usage caps after exhausting their budgets earlier in 2026, with the caps reported in October.

The shift reflects concern that token totals track computing costs more directly than productive output. Firms are increasingly treating AI token spending as a cost center rather than a growth driver.