Uber’s operations chief Andrew Macdonald told Hacker News in a rapid‑response interview that the company’s AI spend is no longer justifiable. He cited CTO Praveen Neppalli Naga’s earlier claim that Uber had already exhausted its Claude Code budget for 2026, a revelation he called a “head‑exploding moment.” The episode sparked internal debate over token consumption and its impact on headcount.

Macdonald explained that higher token usage has not produced a proportional rise in consumer‑facing features. “That link is not there yet,” he said, noting it’s difficult to tie a 25 % bump in useful features to raw token counts. The metric gap fuels internal skepticism. Uber CEO Dara Khosrowshahi has already warned that hiring slows as AI investments swell, underscoring the cost pressure.

The discussion mirrors a broader industry shift. While some firms double down on “tokenmaxxing” and even tie employee reviews to AI usage, others like Duolingo have reversed that policy after staff complained it forced irrelevant work. Macdonald concluded that without a clear ROI, Uber must treat AI spend like any other line item—subject to scrutiny and budget caps. Uber’s finance team now tracks token spend weekly.