Uber’s AI spending productivity link problem is becoming impossible to ignore. The ride-sharing giant burned through its entire 2026 budget for Claude Code in just four months, forcing leadership to pump the brakes on hiring while still increasing overall AI investment. The crisis exposes a growing tension in tech: companies are spending aggressively on AI tools without clarity on whether that spending actually converts into better products or just higher token consumption.
Key Takeaways
- Uber exhausted its full 2026 Claude Code budget by April after rapid adoption across engineering teams
- CEO Dara Khosrowshahi says AI adoption is creating “employees with superpowers” but admits no clear link exists between usage and shipped products
- The company is offsetting AI investment increases by slowing hiring rather than cutting AI spending entirely
- Claude Code adoption among Uber engineers jumped from 32% in December to 63% by February, driving budget burndown
- Leadership is now reassessing whether productivity gains justify the spending velocity
How Uber’s AI budget went sideways
Uber’s CFO Balaji Krishnamurthy admitted the company had “underestimated the amount of impact the AI tools could have” when it set its 2026 budget in late 2025. That miscalculation became obvious fast. Engineers adopted Claude Code at a pace that caught management off guard. The tool rolled out to 32% of the engineering workforce in December, then jumped to 63% by February. By April, the annual budget was exhausted.
This is not a story about Uber abandoning AI. CEO Dara Khosrowshahi made clear the company is still increasing AI investment overall. The difference is how that investment gets funded. Instead of requesting a bigger budget, Uber is offsetting AI spending by hiring less. That trade-off signals confidence in AI’s value but also reveals uncertainty about what that value actually is.
The AI spending productivity link problem nobody wants to admit
Here is where Uber’s situation gets uncomfortable. Khosrowshahi acknowledged that the company has seen AI adoption across legal, marketing, and developer teams, and he described AI as creating “employees with superpowers”. But when pressed on whether this translates into measurable business outcomes, the answer becomes vague. Khosrowshahi said if employees can raise throughput by “20%, 30%, 50%, or 100%,” then the investment would be worth it. Notice the conditional language: “if.” Not “we have measured,” but “if it happens.”
This gap between enthusiasm and evidence is the real story. Uber is a data-driven company that obsesses over metrics for driver matching and trip pricing. Yet on AI productivity, leadership is essentially saying: we believe it works, we are betting on it, but we cannot yet prove the AI spending productivity link exists. That is a stunning admission from a company spending millions on coding tools.
What this means for the broader AI arms race
Uber is not alone in this bind. Every major tech company is racing to integrate AI into workflows without clear evidence that the spending converts into shipped products or competitive advantage. The difference is that most companies are still in the hype phase, announcing AI initiatives and hiring AI teams. Uber has moved past the announcement stage and hit reality: you can buy a lot of Claude Code tokens, but tokens do not ship features.
The company’s decision to slow hiring while maintaining AI investment is a pragmatic hedge. If AI tools genuinely improve productivity, you need fewer headcount to do the same work. If they do not, you have wasted less on hiring and can redirect that money to other priorities. It is a way to test the hypothesis without committing the entire budget to the bet.
CTO Praveen Neppalli Naga reportedly said he was “back to the drawing board” after the budget exhaustion, suggesting leadership is rethinking not just spending levels but the entire approach to measuring AI impact. That is overdue. The industry has been obsessed with adoption rates and token consumption while ignoring the harder question: does this actually make us better at shipping products?
Is Uber cutting AI spending?
No. Uber is increasing overall AI investment but offsetting it by hiring less. The company is not retreating from AI; it is restructuring how it pays for AI adoption while it figures out the real return on investment.
Why did Uber exhaust its Claude Code budget so quickly?
Adoption was faster and more widespread than leadership anticipated. Claude Code rollout jumped from 32% of engineers in December to 63% by February, and the tool was being used across legal, marketing, and development teams. Management underestimated both the velocity of adoption and the cost per user.
What is Uber’s next move on AI spending?
Leadership is reassessing the link between AI tool usage and actual business outcomes before committing to higher budgets. The company will continue using AI but is slowing hiring to fund that usage rather than requesting larger budgets. This suggests a period of consolidation and measurement rather than aggressive expansion.
Uber’s budget crisis is a reality check for the entire industry. Enthusiasm for AI tools is not the same as proof that those tools drive better products. Until companies can measure the AI spending productivity link with the same rigor they apply to pricing algorithms and driver matching, they are essentially guessing at ROI while competitors do the same. Uber is at least honest about the uncertainty. That honesty might be more valuable than the hype.
Edited by the All Things Geek team.
Source: Tom's Hardware
