McKinsey's Chart of the Week series highlighted new research on AI spending discipline, published July 30 alongside the firm's broader analysis on managing AI costs at scale. As organizations move from isolated AI pilots to enterprise wide deployment, McKinsey found that 62 percent have moved beyond experimentation into active deployment. The catch shows up on the financial side, where 93 percent of respondents report exceeding their AI budgets.

The overrun data breaks down unevenly. Thirty nine percent of respondents say their organization went over budget by less than 10 percent, and 46 percent went over by 10 to 30 percent. A smaller group, 7 percent, exceeded budget by 31 to 50 percent, and 1 percent went over by 51 to 70 percent. Only 5 percent of organizations stayed under budget entirely.

McKinsey consultants Pankaj Sachdeva and Wasim Lala, who coauthored the underlying research, found that businesses which build the capability to optimize spend, improve accountability and redirect savings toward the highest value AI opportunities could cut their AI costs by 20 to 30 percent.

That gap between budget planning and actual spend points to a broader pattern across enterprise AI programs. Forecasting models built for a smaller pilot scale are struggling to hold once AI use expands company wide, and cost governance is becoming as important to AI programs as the underlying technology choices themselves.

Source: McKinsey & Company - https://www.mckinsey.com/featured-insights/charts/burning-through-the-ai-budget