Uber has set a $1,500 monthly cap on employee spending for specific AI coding tools after the company exhausted its entire 2026 budget for those tools in the first four months of the year.
The limits apply to agentic coding platforms such as Anthropic’s Claude Code and Cursor, according to Stocktwits. Prior to the caps, some engineers were generating monthly bills between $500 and $2,000 in token consumption as adoption of the tools surged following their rollout in late 2025.
In April, Uber CTO Praveen Neppalli Naga went on the record saying the company had burned through its entire 2026 Claude Code budget in four months. Roughly 5,000 engineers, monthly usage rates between 84 and 95%, per-engineer bills ranging from $150 to $2,000. Neppalli reportedly torched $1,200 in tokens during a two-hour internal demo. Macdonald would later describe his reaction to learning the number as a "head-exploding moment."
President and COO Andrew Macdonald has been open about the challenge - noting that while AI tools are seeing strong adoption, the direct connection between higher token spend and the delivery of useful new features for customers remains unclear.
A company spokesperson said the new framework is intended to “responsibly encourage agentic AI adoption and experimentation at scale” while keeping costs under control. The restrictions do not apply to all AI tools used at Uber - only the higher-cost agentic coding software.
Uber’s move is part of a broader shift among large companies as they confront the real costs of scaling artificial intelligence. Walmart has similarly capped employee access to its internal AI assistant, Code Puppy, after usage far exceeded expectations. The retailer shifted from unlimited tokens to fixed per-employee allocations and is now emphasizing training to ensure AI is used for high-value tasks.
Microsoft has also scaled back internal access to Claude Code for engineers in one of its major divisions, directing staff toward its own GitHub Copilot tools amid rising costs. But then, GitHub started charging for actual use vs. a flat monthly fee.
Across industries, companies are discovering that agentic AI tools can deliver meaningful productivity gains but often consume tokens at rates that quickly outpace traditional budgeting models. Many organizations initially rolled out these tools with minimal guardrails. The result has been budget overruns and increased scrutiny from finance teams and boards over return on investment.
Rather than abandoning AI, most companies are responding by adding governance: usage dashboards, spending caps, approval workflows, and clearer prioritization of high-value use cases. The goal is to make AI adoption sustainable rather than uncontrolled.
As model efficiency improves and costs per token continue to decline, the economics of AI are expected to become more favorable. For now, the most disciplined organizations are treating AI spending with the same rigor applied to other major technology investments - balancing rapid experimentation with financial accountability.