Rising AI spend is outpacing payroll for some firms. Nvidia’s vice‑president of applied deep learning, Bryan Catanzaro, told Axios that compute costs now dwarf employee salaries on his team. Uber’s CTO already burned through the entire 2026 AI budget on token fees, while Swan AI CEO Amos Bar‑Joseph posted his Anthropic bill as proof that his “autonomous business” scales with intelligence, not headcount.

Gartner projects global IT outlays to hit $6.31 trillion in 2026, a 13.5 percent rise from the prior year, driven largely by AI infrastructure, software and cloud services. Companies pouring money into models must now justify returns, whether through measurable productivity gains or clear cost‑benefit metrics, as shareholders demand evidence on earnings calls.

When AI providers raise prices, heavy spenders risk turning a strategic advantage into a liability. An OpenAI investor noted that more token‑efficient models like Codex could help curb costs, while Anthropic recently adjusted its pricing amid demand spikes. Enterprises will need to balance the lure of cutting‑edge models against the hard arithmetic of budget constraints.

The emerging cost debate forces CIOs to treat AI not as a free add‑on but as a line‑item comparable to hiring. As token usage bills climb, firms may revisit staffing models, potentially reinstating human expertise where AI proves uneconomical.