Token fatigue
Also known as: token pricing fatigue, token opacity problem
Tokens are a natural unit for AI providers: cost scales directly with tokens processed, so charging per token aligns vendor pricing with underlying cost. But for most buyers, especially non-technical ones, tokens are meaningless. 'How many tokens do you need per month?' is not a question most procurement teams can answer, and surprise bills from unexpected usage make finance teams nervous.
Token fatigue is the accumulated frustration from this mismatch. Buyers want to know: what does this cost, what do I get, and will my bill be predictable? When the answer involves estimating tokens, context windows, and input-output ratios, the conversation breaks down. Some enterprise deals have been lost not on product quality but on pricing complexity.
The industry response has been to layer abstractions over tokens: credits, outcomes, seats with allowances. These don't eliminate the underlying cost reality, they just translate it into terms buyers can reason about. Builders setting pricing strategy today often start with 'what does success look like for my customer?' and work backward to a pricing unit that reflects that, rather than starting from token math and working forward.