Uber spent its entire 2026 AI budget in just four months after widely adopting Anthropic’s Claude Code. The move highlights the challenges of balancing tech innovation with financial control as AI becomes more central to daily work.
How Claude Code became a major expense
Claude Code, Anthropic’s AI-powered coding assistant, was rolled out to Uber engineers in late 2025. By March 2026, most engineers were using the tool to handle tasks like code refactoring and test generation. By April, 95% of engineers were using AI tools monthly — with 70% of code being influenced by these systems.
The result? A sharp rise in API costs. Monthly expenses per engineer ranged from $150 to $2,000 — with some power users reaching six-figure bills in a single month. You might be wondering how this happened — the answer lies in how Uber structured its internal incentives.
Internal incentives drove overuse
Engineers were ranked on leaderboards based on Claude Code usage, creating a culture of overconsumption. Teams focused on adoption weren’t the same as those managing budgets — and that gap became a problem.
Uber’s finance teams weren’t prepared for the cost structure. The company had planned for a steady AI spend over 12 months — but the rapid shift to AI-driven workflows left them scrambling. You can see how quickly things can spiral out of control when tech adoption isn’t matched by financial planning.
The ROI question remains unclear
Despite the costs, Claude Code worked — engineers used it for exactly what it was built to do. Productivity gains were real, but so were the expenses. You might be thinking — is this worth it?
Uber’s president and COO said, “We can’t yet draw a clear link between increased token usage and new consumer features.” That’s the crux of the issue — ROI is still unclear. Companies need to rethink how they budget for AI tools — you can’t just plug in an assistant and expect the same cost structure as before.
What this means for tech companies
The token-based pricing model — which charges based on usage — is faster and more flexible than traditional software licensing. But it’s also harder to predict. Practitioners in tech are taking note — this isn’t just about Uber.
“It’s a warning shot for any company that assumes AI tools will be cost-effective without rethinking their financial models,” said one software engineer. The challenge now is balancing innovation with fiscal responsibility — a tightrope walk that many tech leaders are only just beginning to understand.
Uber’s wake-up call
For Uber, it’s a lesson in the power of AI — and its potential to spiral out of control. They’ve seen the benefits, but also the risks. You need to be prepared if you’re planning to adopt similar tools.
The future of AI in tech depends on how companies handle these challenges. You can’t ignore the financial side — it’s just as important as the tech itself. The question is: how many other companies are in the same boat?
