Policies promoting AI adoption seem to be becoming a burden for companies, especially Big Tech firms. Even though token prices may fall in the future, AI-related expenses are causing production costs to rise sharply, exceeding the cost of employing human workers.
Leading global technology companies, commonly known as Big Tech, have implemented policies encouraging employees to use AI to boost work efficiency. However, this policy has begun to impact their financial structures.
The Verge news agency revealed that Microsoft is gradually withdrawing employee access to Claude Code and directing engineers to use GitHub Copilot CLI instead. Although this policy was only piloted six months ago, rapid growth in usage at large scale has forced Microsoft to reverse course to control expenses.
Certainly, Microsoft is not alone in scaling back AI usage internally. Uber exhausted its entire 2026 budget for AI coding tools within just the first four months. These developments reflect that AI-related costs are increasingly problematic and could hinder company growth.
According to Goldman Sachs data, by 2030, the use of automated or Agentic AI will cause token consumption to surge 24-fold, reaching 120 trillion tokens per month. Although Gartner reports that large model processing costs will drop by 90 percent by 2030, the lower token prices do not translate into actual budget savings. This is because Agentic models require many more tokens per instruction than standard models. Therefore, the reduced basic token price should not be confused with the true cost of advanced processing.
This reality may disrupt the grand plans of many companies, especially the vision of Jensen Huang, CEO of Nvidia, who envisioned that in the future each employee would have up to 100 AI assistants working alongside them. However, given the rapid rise in token consumption outpacing the basic token price decline, the future of AI-driven work could come with prohibitively expensive AI bills that organizations may struggle to afford.
: Source:Fortune