AI has become more expensive than real employees—Uber’s CTO allocated the entire annual budget to neural networks.

IT budgets of companies are breaking apart because of expensive AI computing

Many organizations already feel the pressure: costs for cloud and hardware resources for artificial intelligence exceed even their employees’ salaries. Uber’s chief technology officer, for example, has completely exhausted the planned AI budget through 2026 due to rising token prices.

Global figures

Forecasts predict that global IT spending in 2026 will reach $6.31 trillion, which is 13.5 % higher than in 2025. The main contributors to this growth will be:

- AI infrastructure;
- Software;
- Cloud services.

Who’s already feeling the blow

* Nvidia – Brian Catanza (VP Applied Deep Learning) said in an Axios interview that his team’s computing costs are “much higher” than personnel expenses.
* Swan AI – CEO Amos Bar‑Yosef noted on LinkedIn that the company is building a “first autonomous business,” scalable through intelligence rather than staff, and praised the bill from Anthropic.

Assessing effectiveness

Even large players with substantial IT budgets will have to prove ROI for AI investments. This can manifest as increased productivity or specific return‑on‑investment metrics. Brad Owens (VP Digital Workforce Strategy, Asymbl) emphasized that “the tone is changing: more often the question is about the real value of a worker – human or digital.”

Impact on the startup ecosystem

High token prices also change dynamics among AI startups themselves. One OpenAI investor told Axios that price increases could benefit the company: according to him, Codex uses tokens more efficiently than Claude Code, reducing costs. Anthropic has revised its pricing after a sharp rise in demand for its services.

Thus, rising AI computing costs are forcing companies to rethink business models and prove the value of digital solutions to shareholders and investors.