Amazon and Uber Cap Employee AI Tool Usage as Token Costs Surge

Amazon and Uber have placed limits on how much employees can use AI tools after rising token costs began straining technology budgets. TechStartups reported in June 2026 that the caps reflect a broader tension between the productivity gains AI tools offer and the significant expense of running them at scale.

Token costs refer to the fees charged by AI providers for each unit of text processed by a large language model. As employees at large companies have adopted AI tools for tasks ranging from writing code to drafting emails, the cumulative cost of those interactions has grown substantially.

Amazon and Uber are among the first major companies to publicly acknowledge that AI tool costs have become significant enough to require management. Both companies have reportedly set monthly or per-user limits on AI tool consumption, requiring employees to prioritize which tasks they use AI for.

The development highlights a challenge that many enterprises are beginning to face. AI tools have been widely adopted in part because they were initially offered at low or subsidized prices to attract users. As providers move toward sustainable pricing, the economics of widespread AI use are becoming clearer.

Some technology analysts say the caps are a sign of maturity in enterprise AI adoption, not a retreat. Companies are learning to treat AI compute as a managed resource, similar to cloud storage or software licenses, rather than an unlimited utility.

Others argue that the caps could slow productivity gains if employees are forced to ration their use of tools that have become central to their workflows. The tension between cost control and productivity is likely to shape how companies structure AI budgets over the next several years.

The issue is particularly acute for companies that have built internal tools and workflows on top of third-party AI APIs, where costs scale directly with usage.