Uber exhausted its annual AI budget just four months into 2026 and now questions whether the spending delivers measurable returns. President and COO Andrew Macdonald said the company sees no connection between rising token consumption for Claude Code and useful features shipped to consumers.
Uber spent $3.4 billion on research and development in 2025, 9 percent more than the previous year. The company uses Anthropic's Claude Code extensively across its engineering organization, driving token consumption in what Macdonald described as an astronomical upward trajectory.
“That link is not there yet, right? I think maybe implicitly there is more that is getting shipped, but it's very hard to draw a line between one of those stats and, 'Okay, now we're actually producing 25 percent more useful consumer features.'”— Andrew Macdonald, Uber president and chief operating officer
Macdonald acknowledged that more features may be shipping implicitly but said drawing a direct line between AI usage and a 25 percent increase in useful consumer functionality remains impossible. The metrics gap creates a justification problem for a company that has publicly committed to offsetting AI costs with reduced headcount.
Key facts
- 01Uber exhausted its annual AI budget just four months into 2026 with no clear productivity gains.
- 02The company spent $3.4 billion on research and development in 2025, 9 percent more than 2024.
- 03Rising token consumption for Claude Code shows no measurable link to consumer features delivered.
- 04CEO Dara Khosrowshahi said Uber is offsetting AI costs by hiring fewer human employees.
CEO Dara Khosrowshahi said earlier this month that Uber is hiring fewer human employees to cover its growing AI investments. Macdonald confirmed the company now frames its AI budget decisions as a tradeoff between token consumption costs and employee salaries.
The admission comes as enterprises across sectors face similar measurement problems. Token usage as a unit cost is easy to track; productivity gains from AI-augmented workflows remain stubbornly difficult to quantify, especially when the output is software features rather than discrete tasks like customer-service tickets resolved or documents reviewed.
“We're going to have to start talking about token consumption and the associated cost versus headcount.”— Andrew Macdonald, Uber president and chief operating officer
Macdonald said the link between AI spend and deliverables may become clearer over the coming quarters and years. For now, Uber joins a lengthening list of companies that ramped AI budgets in 2025 and are struggling to prove the investment paid off in 2026.
The transparency is unusual. Most AI-heavy companies cite productivity improvements in earnings calls without offering the kind of granular measurement Macdonald says Uber lacks. Whether Uber pulls back on AI spending or continues betting on deferred returns will signal how much patience the market has for unproven ROI.
Uber's AI budget crisis is a canary for the broader enterprise AI market. If a company spending $3.4 billion annually on R&D cannot measure whether its AI tools are working, smaller companies with thinner margins and less room for experimental budgets face an even steeper justification problem. The tradeoff Macdonald names — tokens versus headcount — will define 2026 enterprise IT conversations, and Uber just gave every CFO the language to start asking harder questions.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
