The rideshare giant is implementing stricter controls on its employees’ use of generative AI coding tools like Anthropic’s Claude Code and Cursor. Internal data revealed Uber exceeded its initial 2026 AI budget within just four months, prompting the change.
The new policy limits each employee to $1,500 per month in token spending for any single AI tool. This means developers can distribute their allowance across multiple platforms rather than concentrating it on one.
According to analysts, this move reflects a broader trend among enterprise clients grappling with the unexpectedly high costs of generative AI adoption as initial subsidized pricing tiers expire and usage scales up. For Uber, which employs thousands of software engineers globally, these cumulative expenses can quickly reach tens or even hundreds of millions of dollars annually.
While the caps may constrain some developers’ access to cutting-edge tools, they also provide a framework for more sustainable AI investment—allowing Uber to explore use cases that deliver clear business value rather than pure experimentation. The company is reportedly focusing on applications where AI can automate repetitive tasks and improve operational efficiency across its engineering organization.