AI Companies Should Stop Charging for Tokens and Start Charging for Outcomes
We trust math. And the math shows that a model that uses five times as many tokens but achieves ten times better outcomes is economically superior, despite higher COGS.
So, while charging per token seems to be the default strategy, it is a mistake for most of them.
A foundation lab prices per token because token count is its actual cost – the compute it consumes to provide an answer. An Application built on top of that model sells something else: a resolved ticket or a shipped feature. Passing the lab’s cost structure straight through to the customer prices the wrong thing.
Uber found this out directly. Fortune reported in May 2026 that an internal leaderboard rewarding teams for total AI usage led engineers to pile on, and the company burned through its entire 2026 AI budget in four months – Goodhart’s Lawstrikes again. The magazine quoted Uber’s own...
Copyright of this story solely belongs to hackernoon.com. To see the full text click HERE
