McKinsey Warns of Rising Costs for Agentic AI

McKinsey & Company warns that corporate AI spending is poised for a significant increase as organizations transition from basic text-based tools to AI agents capable of performing multistep tasks. According to the firm's 2026 State of AI survey, 60% of organizations plan to increase their AI spending next year, with one-third already allocating more than 10% of their technology and communications budgets to the technology.

The shift toward agentic AI introduces substantial cost volatility. Lari Hämäläinen, a senior partner at the firm, noted that the cost of completing tasks can vary by as much as 30 times depending on the process used. While these agents can reduce human labor on transformation-office tasks by 35% to 70%, approximately 20% of surveyed organizations report that AI costs are already creating operating constraints, especially for token-heavy software development teams.

In response to rising bills, several companies have implemented cost-control measures. Coinbase and Salesforce, Inc. have introduced AI usage limits, while Amazon.com shut down an internal employee-created leaderboard that tracked AI token consumption. McKinsey advises companies to strictly measure whether the value produced by these tools justifies the escalating expenses.