Almost nine in ten big-company tech leaders say they have already cut back or reworked AI projects because the costs keep climbing. That is the main finding of a survey released on Oct. 26, 2026 by Cloudera, a data and AI software company.
Cloudera commissioned the survey, and Wakefield Research ran it in September 2026. It covered 1,500 IT leaders and decision-makers worldwide. In all, 89% said they limited or redesigned AI uses because of rising token consumption. Tokens are the small pieces of text an AI tool processes, and they are one way AI usage gets measured.
What else they said
A total of 84% said rules about where data must be kept, or which country's laws apply, shape where their AI runs. Asked what they would do differently if they started over, 49% said they would set stronger rules for managing data from the start. Another 46% would fix data quality sooner. Both 45% for upgrading technology systems and 45% for more employee training also made the list.
- 89% limited or redesigned AI uses because of rising usage costs
- 84% say data location or legal rules affect where AI runs
- 49% would build stronger data rules first
What is not known
The material does not say how much costs rose, which companies were surveyed, or how they changed their projects. Cloudera sells software for running AI where a company's data already sits, so it has a stake in the topic. For smaller businesses, the practical point is simple: a pilot that looks cheap can get expensive as more people use it.
The takeawayBefore expanding an AI tool, check how its usage-based pricing grows as more people use it.