In the hushed corridors of corporate America, executives have been whispering the same question for months: when does all this AI spending actually pay off? Andrew Macdonald, Uber's Chief Operating Officer, just became the first major executive to say it out loud.
In comments that sent ripples through the investment community, Macdonald acknowledged that it's getting "harder to justify the money spent on AI token spending." Translation from corporate-speak: we're burning cash on AI and the ROI isn't there yet.
This matters because Uber isn't some struggling startup trying to appear relevant. They're a $150 billion company with actual revenue and a real business model. When their COO questions AI economics, investors should pay attention.
Here's what he's really saying. Companies have been buying AI tokens and compute capacity at a staggering rate, convinced that being "AI-first" is essential for competitiveness. The problem? Most of these implementations are expensive solutions looking for problems. The productivity gains are often marginal, the infrastructure costs are substantial, and the measurable business impact remains elusive.
For investors, this is a potential inflection point. The AI trade has been built on the assumption that enterprise spending will continue accelerating indefinitely. If companies start scrutinizing ROI more carefully, who gets hurt first?
The obvious exposure is in the AI infrastructure stocks that have run up on the narrative of unlimited enterprise demand. Companies selling AI services, cloud compute, and specialized chips have been priced for perfect execution and endless growth. If customer acquisition costs are rising while customers are getting more selective about spending, those margins compress quickly.
This doesn't mean AI is over or that the technology isn't transformative. It means we might be entering the "show me the money" phase, where executives need to demonstrate actual business value rather than just checking the AI box for their board.
Macdonald isn't alone in these concerns; he's just the first to say it publicly. Behind closed doors, CFOs have been asking the same questions about their AI budgets. The difference between 2026 and the early internet era is that companies learned from the dot-com crash. They won't burn capital indefinitely on unproven business cases.
