McKinsey Reports: AI Scaling in Large Enterprises Reaches 40% in 2026
A groundbreaking survey from McKinsey & Company discloses that agentic AI has surpassed a crucial milestone in corporate America, fundamentally transforming the software market. The firm’s report, “State of AI in 2026: On the Road to ROI”, is based on feedback from 1,719 professionals across 97 countries.
The study reveals that 40% of respondents from large organizations (those with over $1 billion in annual revenue) are actively scaling AI agents. This is a significant increase from the previous year’s 27%. However, the percentage of smaller enterprises scaling AI agents remained unchanged at 22%.
One of the most disruptive findings is the shift from purchasing to building software in-house. 32% of all surveyed organizations have opted not to buy at least one software product or feature because agentic coding tools enabled them to develop it internally. This trend is particularly noticeable among AI “high performers” — the 6% of respondents attributing at least 5% of their EBIT to AI — with nearly half of this group forgoing vendor software entirely.
Despite the increase in adoption, the ROI gap continues. While 80% of respondents report individual productivity gains, only 37% report any measurable enterprise-level EBIT impact — a figure that remains relatively unchanged from the previous year’s survey. McKinsey points out that cost reductions are most commonly observed in supply chain management and service operations, while revenue gains are typically associated with marketing, sales, and software engineering.
Source: McKinsey & Company