Ninety-three percent of enterprises now report exceeding their AI budgets, according to a McKinsey survey of 120 enterprise participants published in the firm's July 2026 report on enterprise AI cost management. The finding points to a widening gap between how companies plan for AI spending and how much they actually consume once agentic workflows and generative tools move into daily use.

McKinsey found that AI spend increases nearly fourfold as organizations move from isolated pilot projects into enterprise-wide adoption, and 62 percent of surveyed organizations have already made that jump from experimentation into active deployment. A majority of respondents expect their AI spending to climb by at least 25 percent over the next 12 months, even as many report having little visibility into where the money is going. The report notes that 20 to 30 percent of AI spend often goes unaccounted for because investments are scattered across cloud providers, software platforms, and individual business units.

Despite the overruns, McKinsey found that companies who actively manage consumption, through steps such as routing tasks to lower-cost models, shortening prompts, and reusing cached context, can cut their AI costs by 20 to 30 percent. Only 20 to 25 percent of companies currently have mature financial operations practices in place for AI spending, leaving most enterprises still building the governance needed to track usage against outcomes.

The report frames AI cost management as a discipline distinct from traditional cloud budgeting, given how unpredictably token consumption can swing between tasks that look identical on the surface.

Source: McKinsey and Company - https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-cost-of-intelligence-how-cios-can-manage-ai-demand-at-scale