Accenture says AI token use could rise 78%, outpacing falling prices

Accenture Research expects AI token prices to fall while businesses use far more of them, a combination that could leave total bills rising rather than shrinking. Its figures also point to a basic accounting problem: many organisations do not know what their AI use costs per business outcome.

Accenture AI Watch analysis

What happened

A syndicated ANI report in The Tribune says Accenture Research projects average AI token prices will drop by about 19% over the next two years, while token consumption rises by 78%. The report warns that lower prices per token do not guarantee lower overall spending.

In Accenture’s survey, only 35% of organisations could precisely calculate the cost per business outcome for their highest-volume AI use case. If prices fall by at least 25%, just 5% of respondents said they would retain the savings; 42% would expand AI use to more employees, while 35% would introduce applications previously considered too expensive.

Why it matters

Cheaper AI can make each task less expensive and still make the overall bill larger if people and businesses use it much more. That is the familiar bargain of efficiency, with a less familiar invoice attached.

For organisations deciding whether to expand AI, the useful measure is not just the price of a token. It is what the extra usage costs, what work it enables and whether that work is worth the spend. The survey suggests many companies still lack a precise way to make that comparison.

Our read

Accenture’s projections are a prompt to measure AI against outcomes, not a forecast every business should treat as its own budget. The striking finding is that only 35% of surveyed organisations could calculate the cost of their highest-volume AI use case per business outcome. That is a rather large blind spot to bring to a spending review.

What to watch

• Whether AI use grows enough to offset falling prices in actual company budgets.
• Whether more organisations start tracking cost per business outcome.
• How businesses use any savings: more users, new applications or money kept in reserve.

Discussion spark: If AI gets cheaper per task but companies use much more of it, should success be measured by lower bills or by the extra work it enables?

not affiliated with or endorsed by Accenture AI