Uber has become an unlikely case study in the hidden running costs of artificial intelligence. According to TechCrunch, the company capped per employee AI spending in early June after its chief technology officer disclosed that Uber's entire planned 2026 AI budget had been exhausted by the spring, just four months into the year. Employees are now limited to around 1,500 dollars a month per coding agent.
The cost no one priced in
The blowout has given fresh currency to the idea of AI tokenomics, the metered, usage based cost of running generative tools at scale. With 95% of Uber's engineers reported to use AI tools each month, the company found that adoption it had encouraged could send bills past forecasts at speed. The lesson lands far beyond Uber: agentic tools that act autonomously consume far more compute than a simple chatbot query, and that cost is not always visible until the invoice arrives.
The cuts that followed
Days later, on June 3, Uber eliminated nearly a quarter of its People and Places division, the unit covering human resources, recruiting, workplace facilities and company culture. The cuts amounted to 23% of that division but less than 1% of Uber's roughly 34,000 corporate employees. A newly appointed president insisted the restructuring had nothing to do with AI.
A tension companies keep hitting
That denial sits awkwardly against the timing and against Uber's own data on how deeply AI is now woven into its engineering. It echoes a wider pattern that even Sam Altman has acknowledged, in which some firms blame AI for cuts they would have made anyway, while others quietly absorb automation without naming it. Industry trackers put total 2026 tech layoffs near 150,000.
What it signals
For executives and workers alike, Uber's experience is a useful corrective to the idea that automation is simply cheaper. The tools carry real, variable costs that have to be budgeted, capped and governed, and the productivity case still has to clear that bar. The gap between AI's promise and its metered price is becoming one of the defining management questions of the year.
