Uber has imposed a limit of 'up to 240,000 yen per person per tool per month' because its employees were overusing AI tools.
Uber Caps Employee Spending on AI Tools Like Claude Code to Manage Costs - Bloomberg
https://www.bloomberg.com/news/articles/2026-06-02/uber-caps-usage-of-ai-tools-like-claude-code-to-cut-costs
In a podcast appearance in February 2026, Uber Technologies CEO Dara Khosrowshahi revealed that Uber Technologies' software engineers use AI in approximately 90% of their work, and that about 30% of employees are 'power users' of AI. Some team members are even building 'Dara AI (AI Khosrowshahi CEO)' to prepare presentations for Khosrowshahi himself. Khosrowshahi said of AI, 'It's really changing employee productivity in ways I've never seen before,' indicating that AI is contributing to increased efficiency in operations.
Uber reveals that 90% of engineers use AI in their work, and one employee has even created an AI-powered CEO - GIGAZINE
Amidst this, Praveen Neppari Naga, Chief Technology Officer (CTO) of Uber Technologies, revealed in an interview with The Information in April 2026 that Uber Technologies would have exhausted its annual AI budget by then. According to Naga, approximately 11% of real-time updates to the code in Uber Technologies' backend systems are created by AI agents built primarily with Claude Code , a significant increase from just a few percent three months prior. Naga did not disclose the exact figure for the annual AI budget or how much is spent on AI coding tools, but it is known that Uber's research and development spending reached $3.4 billion (approximately 540 billion yen) in 2025, a 9% increase year-on-year, and Uber Technologies stated in a filing with the Securities and Exchange Commission that 'we expect these costs to continue to rise.'
Following the depletion of its AI budget, Uber Technologies has imposed restrictions on the use of certain AI tools by its employees for cost-cutting purposes, Bloomberg has revealed. According to an Uber Technologies spokesperson who spoke to Bloomberg, the company is limiting token spending per employee per AI coding tool to $1,500 (approximately 240,000 yen) per month.
AI services are often billed in units called 'tokens' based on the amount of text processed, and this cap applies to the amount of tokens used. Note that the price of tokens is higher for enterprise plans than for general users using the same amount of data. Each tool is calculated independently, and excessive use of one tool will not affect the budget for other tools.
All employees are provided with a dashboard that allows them to track their usage of various tools. Additionally, a process is in place to request permission if they wish to use AI coding tools beyond their normal usage limits.
A spokesperson for Uber Technologies said, 'We see this as a very clear way to responsibly accelerate the adoption and experimentation of agent-based AI at scale across our entire company.'
Simon Willison, the engineer who devised a unique benchmark to have AI draw a ' pelican riding a bicycle ,' explains the '$1,500 per month' cap. According to Levels.fyi , which collects engineer salaries, the median annual salary for software engineers at Uber Technologies in the United States is approximately $330,000 (approximately 52 million yen). Since the $1,500 per month cap is per tool, if one engineer uses two different tools, the annual amount would be '1,500 x 2 x 12 = $36,000 (approximately 5.7 million yen),' which is about 11% of the engineer's annual salary. Willison points out that 'the cap information is interesting because it suggests the actual amount that Uber Technologies is making from these AI tools.'
Related Posts:
in AI, Posted by log1e_dh