Uber capped Claude Code and Cursor after exhausting its AI budget in four months

Create a watch on this topic Receive upcoming developments with an AI watch configured from this article.

Create my free watch

Uber institutes a monthly cap of $1,500 per employee and per agent coding tool, whether it be Claude Code from Anthropic or Cursor, reports TechCrunch based on a Bloomberg dispatch. The move follows a revelation from the company's technical director in April 2026: Uber had exhausted its entire annual AI budget in four months, after encouraging its teams to use these tools as much as possible and implementing internal rankings that valued higher consumption.

The gesture comes as a Bain & Company survey, relayed by Bloomberg (June 1, 2026), notes that 40% of companies that tracked their AI spending have not met their initial cost-cutting goals, and that 83% of CFOs still plan to increase their AI budgets by more than 15% over two years. According to Bain, the gap between spending and productivity gains "should make leaders uneasy." Uber exemplifies this moment where the bill catches up with the promise.

To manage spending, Uber provides each employee with an internal dashboard tracking their consumption, and allows case-by-case exceptions beyond the cap. The signal matters because Uber is not a laggard: according to Pragmatic Engineer, 84% of its developers are already users of agent-based coding tools, and the company itself commercializes an offering Agentic AI Solutions to third-party clients. Its COO, Andrew Macdonald, actually acknowledged on a podcast cited by Fortune that «it is very difficult to draw a line» between AI usage and the creation of new consumer features, a rare admission about the measurability of returns. For IT departments that are industrializing Claude Code or Cursor, the operational lesson is clear: as usage becomes agent-based and billed per token, the initial mantra (use to the fullest) gives way to governance through headquarters, dashboard, and traced exceptions.