Imagine your company buys every employee a gym membership, installs a leaderboard tracking who logs the most hours on the treadmill, and then — four months later — the HR director quietly admits that nobody has actually lost any weight.
That, in essence, is what just happened at Uber.
Uber's Chief Technology Officer recently disclosed something that most companies would bury in the fine print: the company burned through its entire 2026 budget for two premium AI coding tools — Claude Code and Cursor — in just four months. Not because the tools stopped working. Because engineers used them so enthusiastically that they blew past the spending ceiling before the year was half over.
On the surface, that sounds like a success story. Adoption through the roof. Engineers embracing new technology. Leadership investing boldly in the future. Except Uber's Chief Operating Officer, Andrew Macdonald, then said the quiet part out loud in a public forum: the connection between all that AI tool usage and actual improvements that riders or drivers would notice — better routes, faster support, smoother experiences — is simply "not there yet."
Seventy percent of Uber's committed code is now written by AI. Ninety-five percent of its engineers use AI tools every month. And by the company's own admission, none of that has translated into a meaningfully better product.
This is not a story about AI failing. It's a story about what happens when companies reward the act of using a tool instead of the results it produces — and why Uber's unusual candour about the gap makes this one of the most important business admissions of the year.
