Uber Caps Employee AI Tool Access After Exhausting 2026 Budget

Uber has implemented strict usage caps on employee access to AI coding tools after burning through its entire 2026 artificial intelligence budget in just four months, a striking example of how quickly enterprise AI costs can spiral.

The ride-hailing company now limits employees to $1,500 per month in token spending for each AI-powered coding tool they use, according to Bloomberg, which first reported the policy change. The restrictions apply specifically to agentic coding software—AI systems like Cursor and Claude Code that can modify code with minimal human oversight.

The productivity paradox

The budget overrun comes despite measurable productivity benefits. CEO Dara Khosrowshahi disclosed last month that approximately 10 percent of Uber's codebase is now AI-generated. Legal and marketing departments have also increased their reliance on AI tools, Khosrowshahi told The Information.

Yet the financial reality has forced Uber to moderate its 2026 hiring plans relative to initial projections, even as it realizes internal efficiency gains from AI adoption.

How the caps work

Uber has created individual AI usage dashboards for every employee, allowing them to monitor their token consumption across different tools in real time. The company also established a request system for workers who need to exceed their monthly allocation.

A company spokesperson characterized the approach as "a pretty straightforward way to responsibly encourage agentic AI adoption and experimentation at scale across the company," according to Bloomberg.

The caps target the most resource-intensive category of AI tools—those that autonomously generate and modify code—rather than simpler chatbot interfaces or copilot assistants that require more human direction.

Why it matters

Uber's budget crisis illustrates a broader challenge facing enterprises racing to deploy AI: the technology delivers genuine productivity improvements, but at costs that can quickly become unsustainable without careful management. Companies that assumed AI would immediately reduce expenses are discovering that the compute-intensive nature of modern AI models, particularly agentic systems, creates significant ongoing operational costs. As more organizations scale AI adoption beyond pilot programs, Uber's experience suggests that usage governance and cost controls will become standard practice, even for tools that demonstrably improve output.

The details were first reported by Bloomberg.