McKinsey AI enterprise ROI gap earnings The McKinsey AI enterprise ROI gap on earnings is widening rather than closing, according to the consulting firm’s annual State of AI survey released this week. Despite record spending on generative and agentic systems across industries, only 37% of the 1,719 executives surveyed said they could attribute any earnings impact to artificial intelligence, a share that has barely moved year over year. The headline finding, that 94% of organizations cannot book any EBIT impact from AI at all, has become the defining statistic of the current spending cycle and underscores how little of the billions deployed at the corporate level is reaching the income statement.
McKinsey & Company’s QuantumBlack division published the survey on Monday, based on responses collected between May 4 and June 8 from senior leaders in 97 countries. The figures paint a picture of an enterprise technology market in which adoption is near-universal but measurable financial returns remain rare. Bob Sternfels, McKinsey’s global managing partner, disclosed that his own firm operates roughly 25,000 AI agents running alongside its 40,000 human consultants, and that back-office functions have shrunk by 25% as agents took over research synthesis and document preparation.
McKinsey AI enterprise ROI gap earnings: The High Performer Gap
McKinsey defines “AI high performers” as organizations whose respondents attribute at least 5% of EBIT to AI and describe that impact as significant. Just 6% of respondents in the 2026 survey qualify under that definition, unchanged from the prior year and flat since 2025. Nearly three-quarters of those high performers, 73%, reported fundamentally redesigning workflows around their AI deployments, up from 55% a year earlier. By contrast, only about a quarter of all other respondents said the same.
Michael Chui, a senior fellow at McKinsey’s QuantumBlack unit, framed the gap in historical terms. “High performers are already seeing real ROI,” Chui said. “It should not be surprising that it has taken time, because it is a reflection of trends we’ve seen with other technologies. History doesn’t repeat itself, but it rhymes.” The comment reflects a long-running argument inside the firm that the current diffusion curve resembles prior enterprise technology waves, in which early broad adoption produced diffuse productivity gains that only later consolidated into earnings.
Where the Money Goes
The survey points to a structural split between where cost savings appear and where revenue gains appear. Respondents reporting cost reductions named supply chain management, service operations and manufacturing as the leading sources of measurable savings. Those reporting revenue gains pointed to marketing and sales and product development. McKinsey’s analysts describe this pattern as the “Gen AI paradox,” in which horizontal tools such as chatbots and coding assistants scale quickly across an organization but produce diffuse productivity gains that never aggregate into the bottom line.
Vertical deployments, meaning AI built into a specific revenue or cost process with a measurable outcome, remain rare. The survey found that 90% of more specific function-level AI use cases remain stuck in pilot mode rather than scaled into production. Nearly a third of respondents said their organizations had opted to build AI tools in-house rather than purchase a commercial software product, often relying on agentic coding tools to do so. One in five respondents said AI-related operating costs, including token costs for inference, had actively constrained their use.
Agentic AI and the Cost of Inference
Agentic systems, which can plan and execute multi-step tasks, are scaling faster inside larger organizations. Among respondents at companies with annual revenues above $1 billion, 40% reported scaling agents in at least one function, up from 27% a year earlier. Adoption at smaller organizations remained flat at 22%. The shift carries a meaningful cost implication. Gartner’s March 2026 analysis found that agentic AI workflows can generate 5 to 30 times more inference tokens per task than a standard chatbot, raising the operational cost of moving from a working pilot to a production deployment.
KPMG’s first-quarter 2026 Global AI Pulse survey put the average enterprise AI budget at $186 million, with only 8% of respondents reporting tangible ROI. A February 2026 study from the National Bureau of Economic Research, which surveyed nearly 6,000 C-suite executives, found that close to 90% of firms reported AI had produced no measurable impact on employment or productivity over a three-year window. Within McKinsey’s own data, 60% of respondents said they expect their organizations to increase AI investments over the next year, while only 14% said AI had contributed to an overall workforce decline in the past 12 months. Two-thirds reported little or no AI-related change in total employment.
The Workforce Question
The workforce data point to a widening expectation gap. Thirty-nine percent of respondents said they expect net job reductions attributable to AI in the near term, up from 32% in 2025, but only 14% reported any actual workforce decline over the prior year. McKinsey’s high performers were 3.3 times more likely than peers to say they intend to use AI to fundamentally transform their business within three years, suggesting that the divide between the 6% and the rest may widen before it narrows. Inside McKinsey itself, Sternfels said client-facing roles grew by 25% as back-office output rose 10% on fewer people, with agents saving 1.5 million hours of search and synthesis work and producing 2.5 million charts in the past six months, illustrating the kind of internal restructuring that the 94% of enterprises without an EBIT impact have yet to execute as the McKinsey AI enterprise ROI gap on earnings continues to define the market. McKinsey AI enterprise ROI gap earnings.
Source: https://www.techtimes.com/articles/325590/20260826/record-ai-spending-cant-move-earnings-needle-94-enterprises-mckinsey-finds.htm