Uber blew through its entire 2026 AI budget in just four months, and the culprits were Claude Code and Cursor. According to Hacker News, the ride-hailing giant’s CTO revealed engineers couldn’t stop reaching for these tools despite individual API costs running between $500 and $2,000 per person each month. By April, the annual allocation was gone.
This is a striking signal about where developer tooling economics are heading. When a company with $3.4 billion in annual R&D spending can’t budget accurately for AI coding tools rolled out four months earlier, every engineering org needs to pay attention.
How Claude Code Took Over
Uber rolled out Claude Code to its engineering team in December 2025. Usage doubled by February as developers got comfortable with its multi-step agentic capabilities. Two months later, the bill had eaten the full year’s AI budget.
A few numbers from the Hacker News report stand out:
- 95% of Uber engineers now use AI tools monthly
- Per-engineer API costs hit $500 to $2,000 per month
- Usage doubled in roughly 60 days post-rollout
- Cursor adoption plateaued while Claude Code kept climbing
The CTO described the company as “back to the drawing board” on AI budgeting. Translation: leadership is figuring out whether productivity gains at this price are sustainable at Uber’s scale.
Why Claude Code Won the Internal Race
Cursor and Claude Code entered Uber on roughly equal footing. They didn’t stay there. Hacker News reports Cursor’s adoption flattened while Claude Code became the dominant engineering workflow tool.
The likely reason: Claude Code’s multi-step execution model fits how engineers actually work. It runs longer chains, touches more files, and handles bigger tasks autonomously. That makes it more useful, but it also makes it more expensive to run, since each session burns far more tokens than a typical autocomplete-style assistant.
What stands out here is that engineers chose the costly tool over the cheaper one because it actually moved work forward. Productivity beat price sensitivity.
What This Means for Everyone Else
Uber is a leading indicator, not an outlier. If 95% adoption and $1,000+ per-engineer monthly bills are showing up at one of the largest engineering orgs in tech, similar patterns are coming for everyone running serious software teams.
Three things to watch:
- Budget cycles are broken. Annual AI allocations made in late 2025 assumed adoption curves that no longer hold. Expect quarterly resets across the industry.
- Pricing pressure on Anthropic. Enterprise customers will push hard for volume discounts, predictable pricing tiers, or cost caps. The current per-token model doesn’t scale gracefully for orgs with thousands of developers.
- Build vs. buy conversations. Some companies will start running open-source coding agents on their own infrastructure to control costs, even if quality is lower.
The Real Story
The issue isn’t that Claude Code is expensive. The issue is that the budget was set before anyone knew how transformative the tool would become. Uber’s engineers didn’t burn the budget out of carelessness. They burned it because the tool was worth using.
That’s a healthier problem than the alternative, which is paying for software nobody touches. But it’s still a problem, and CFOs across tech are about to have the same conversation Uber’s leadership is having now.
More details at the original source.