Uber Technologies Inc. has enacted firm controls on employee spending for certain artificial intelligence tools after exceeding its AI budget for the year. The company now caps per-employee token expenditures at $1,500 per month for each qualifying agentic coding tool, a measure intended to rein in costs while maintaining scope for experimentation.
The limit applies independently to each tool, meaning usage on one platform does not consume the allowance for another. These restrictions, rolled out in recent months, target agentic coding software - examples cited by the company include Cursor and Anthropic PBC's Claude Code - and do not encompass all AI products broadly.
Employees can track their consumption through a centralized dashboard that aggregates activity across the various tools. The company has also implemented a formal process through which staff may request approval to exceed the standard monthly ceiling when business needs justify it.
According to the company, the move is intended to strike a balance between cost discipline and continued internal experimentation with agentic AI. The company's spokesperson said that the policy is "a pretty straightforward way to responsibly encourage agentic AI adoption and experimentation at scale across the company," and framed the limits as part of that responsible approach.
Executives have acknowledged a rapid uptick in internal AI use. Chief Technology Officer Praveen Neppalli Naga told the Information in April that Uber had already exhausted its full-year AI budget. Chief Executive Officer Dara Khosrowshahi said last month that roughly 10% of the company's code had been submitted and built by AI agents, and that the legal and marketing teams in particular had increased their use of these tools.
Operational impacts have followed. Last month Uber announced it would slow the pace of hiring relative to its initial plans for the year, attributing that change to productivity gains stemming from internal AI use. At the same time, Chief Operating Officer Andrew Macdonald noted on the Rapid Response podcast that it remains difficult to draw a straight causal link between higher AI-tool usage and the production of new, consumer-facing features. "It's very hard to draw a line between one of those stats and 'OK, now we're actually producing like 25% more useful consumer features,'" he said.
Key points
- Uber set a $1,500 monthly token spending cap per employee for each agentic coding tool after surpassing its annual AI budget.
- The caps are tool-specific; separate dashboards allow employees to monitor usage and submit requests for exceptions.
- Executives report substantial internal AI adoption - including an estimate that about 10% of code has been submitted and built by AI agents - and cite AI-driven productivity changes that influenced hiring plans.
Risks and uncertainties
- Spending caps could constrain experimentation and development speed for teams relying heavily on agentic coding tools - this may affect engineering and product development functions.
- It remains unclear whether increased AI use is directly translating into more or faster consumer features - placing uncertainty on product roadmaps and feature delivery expectations in technology and consumer services sectors.
- Budgetary overruns and subsequent controls highlight potential volatility in AI-related operating expenses for firms adopting agentic tools - relevant to finance, operations, and corporate planning.
This report presents the company's stated measures and executive comments about internal AI adoption and budget effects. The policies described address a subset of tools defined as agentic coding platforms, include employee monitoring and an exceptions workflow, and follow the company's disclosure that its full-year AI budget had been consumed earlier in the year.