Uber questions AI returns after blowing through Claude Code budget

Uber is moving from AI enthusiasm to AI accountability. After the company said it had already blown through its 2026 Claude Code budget by April, president and chief operating officer Andrew Macdonald said Uber still could not clearly connect higher token consumption with more useful consumer features.

The comments land at an inflection point for the corporate AI boom. Praveen Neppalli Naga, Uber’s chief technology officer, had already flagged the budget problem in April, underscoring that the issue is not whether Uber is using AI, but whether the spending is translating into measurable improvements. Naga leads engineering and science strategy, while Macdonald oversees Mobility, Delivery, advertising, customer support and the company’s autonomous-vehicles strategy.

Uber has been willing to spend heavily on technology. In its February 4, 2026 full-year results, the company said 2025 research and development expenses reached $3.4 billion, up 9% from a year earlier, even as quarterly trips and gross bookings both rose 22% year over year. It also reported record quarterly operating cash flow of $2.9 billion and free cash flow of $2.8 billion. By March 31, Uber said it operated in 70 countries and had $215 billion in annualized run-rate gross bookings.

The scale helps explain why Uber has been pushing deeper into AI. The company has said AI agents generate about 10% of the code it produces, and it has also been slowing hiring as it invests more in AI. But the latest remarks suggest a tougher internal test is emerging: not how quickly AI can be deployed, but whether the output justifies the bill.

That question matters because Claude Code is priced by tokens, a model that can make enterprise costs difficult to forecast. Anthropic’s published Opus-class pricing is $5 per million input tokens and $25 per million output tokens, with prompt caching and batch processing available to reduce expenses. For companies running large engineering operations, that means usage can climb fast even when the business case is still fuzzy.

Uber’s shift is a warning sign for the rest of corporate America. If a platform with billions in cash flow, operations across 70 countries and a $215 billion gross bookings run rate is asking harder questions about AI returns, smaller companies may soon face the same pressure: show the productivity gains, or rein in the spending.