AI Adoption Outruns Financial Returns

Nearly a decade after McKinsey began tracking corporate use of artificial intelligence, the question for many companies is no longer whether to use AI but how to scale it and turn adoption into measurable business value. The latest report, “The state of AI in 2026: On the road to ROI,” shows that deployment has entered a more operational phase. AI is being used in more business functions, agentic systems are moving beyond pilots, and larger organizations are starting to reconsider what they need to buy from vendors and what they can build themselves. Yet enterprise-level financial impact continues to lag behind the pace of adoption.

Almost nine in ten respondents say their organizations regularly use AI in at least one business function. More importantly, 44 percent of organizations using AI have reached the stage of scaling it across the enterprise, up from 38 percent a year earlier. The share using AI in three or more functions has also increased, from 51 percent to 56 percent. Experimentation remains widespread, but the center of gravity is clearly shifting toward operational deployment.

The gap between large and smaller organizations is becoming more pronounced. Among companies with at least $1 billion in annual revenue, 54 percent of respondents say AI is being scaled across the enterprise, compared with roughly one-third at smaller organizations. The difference is even sharper for AI agents. At large companies, the share scaling agents in one or more functions rose from 27 percent in 2025 to 40 percent this year. Among smaller organizations, the figure remained essentially unchanged at 22 percent.

Chatbots are still the most widely scaled AI tool. Across all organizations, 47 percent of respondents say their companies are scaling them, while the figure reaches 64 percent among companies with more than $1 billion in revenue. Software coding agents are also moving quickly into production: 31 percent of respondents at large enterprises say their organizations are already scaling them. That matters not only for software development itself, but for a more strategic question – whether a company should buy a capability at all.

Nearly one-third of respondents, 32 percent, say their organization decided not to purchase at least one software product or feature because the functionality could be built internally with agentic coding tools. Such decisions are most common in technology, followed by healthcare, professional services, and energy and materials. The finding does not suggest that external software vendors are becoming unnecessary. It does show that the build-versus-buy boundary is shifting as AI lowers the cost and effort required to create certain capabilities in-house.

The most striking tension in the survey appears when individual benefits are compared with enterprise results. Eighty percent of respondents say AI has improved their personal productivity, while about half say it helps them make better decisions and develop new skills. Those positive effects are broadly consistent across seniority levels. At the same time, midlevel managers and individual contributors are more likely than executives to report downsides such as greater stress, mental fatigue, an overwhelming volume of AI output, or pressure to take on more work.

At company level, the picture is more restrained. Only 37 percent of respondents say AI has made a positive contribution to their organization’s EBIT, essentially unchanged from last year. That is notable because the share of organizations scaling AI increased over the same period. Wider access to the technology and broader deployment have therefore not yet translated automatically into stronger enterprise-wide financial performance.

There are, however, measurable gains inside individual business functions. Respondents most often report cost reductions from AI in supply chain management, service operations and manufacturing. Revenue gains are most commonly associated with AI use in marketing and sales, product and service development, and software engineering. In other words, value is appearing in specific parts of the organization, but those gains are not yet consistently large enough to move overall EBIT.

Costs are also becoming a more visible part of the equation. One in five respondents says their organization has limited AI use because of operating costs, including tokens, computing resources and storage. Even so, investment remains strong. Twenty-eight percent say their organizations already spend more than 10 percent of their total enterprise-wide ICT budget on AI technologies, while 60 percent expect AI investment to increase over the next year. Respondents in pharmaceuticals and medical products, insurance, and financial institutions are among the most likely to expect higher spending.

McKinsey pays particular attention to a small group it calls AI high performers. These are organizations that attribute at least 5 percent of EBIT to AI and also describe the value created by AI as significant. They account for only about 6 percent of respondents, unchanged from 2025. Their advantage is not simply a larger budget. They are more likely to use AI for growth and innovation as well as efficiency, and they are far more willing to redesign entire workflows around the technology.

Nearly three-quarters of high performers say they have fundamentally redesigned workflows because of AI, compared with about one-quarter of other respondents. They are also 3.3 times more likely to expect AI to fundamentally transform their business within the next three years. They more frequently build human validation into AI processes, actively manage token, compute, and storage costs, conduct strategic workforce planning, define processes for measuring AI impact, and report stronger ownership from senior leadership.

High performers also commit a larger share of technology spending to AI. They are more than twice as likely as other organizations to spend over 15 percent of their ICT budget on AI technologies, and more than half expect to increase AI investment by at least 10 percent over the coming year. At the same time, they are more aware of operating constraints. In software coding agents, cost limitations are reported roughly three times as often among high performers as among other respondents, which is consistent with heavier use of those tools rather than weaker confidence in them.

The survey also provides a useful correction to last year’s expectations about employment. In 2025, 32 percent of respondents expected AI to reduce their organization’s workforce over the following 12 months. In the 2026 survey, only 14 percent say AI actually contributed to an overall workforce decline during the past year, while about two-thirds report little or no change. Expectations have nevertheless risen again: 39 percent now anticipate an AI-related reduction in total headcount over the next year, while 43 percent expect little or no change.

Concern about personal careers is much lower than the broader expectation of workforce reductions. Just 13 percent of respondents say AI makes them anxious about their own career prospects. That contrast captures much of the report’s central message. Employees are already seeing tangible benefits in their own work, companies are embedding AI more deeply into operations, and investment continues to rise, but the organizational and financial impact has not yet caught up.

The survey was conducted online from May 4 to June 8, 2026, and received responses from 1,719 participants in 97 countries, covering a wide range of industries, company sizes, functions, and levels of seniority. The data were weighted by each respondent’s country contribution to global GDP, and 36 percent of participants work for organizations with more than $1 billion in annual revenue. The core issue emerging from the research is therefore not whether companies are benefiting from AI, but whether they can close the gap between individual productivity gains and measurable enterprise-wide returns.