On the road to ROI, six percent have arrived
McKinsey's 2026 State of AI survey of 1,719 respondents in 97 countries finds AI use is now nearly universal, yet only 6 percent of organisations qualify as AI 'high performers,' unchanged from 2025 even as adoption climbed.
TL;DR
- McKinsey's 2026 Global Survey on the state of AI, fielded May 4 to June 8 among 1,719 respondents in 97 countries, finds nearly nine in ten organisations now regularly use AI, but only 37 percent attribute any EBIT impact to it, about the same share as 2025.
- Just 6 percent of organisations qualify as "AI high performers," attributing at least 5 percent of EBIT to AI and describing its impact as significant, a figure that has not moved since last year despite the adoption surge.
- A growing share of respondents, 39 percent versus 32 percent in 2025, now expect their employer to cut jobs because of AI in the coming year, and one in five organisations say rising AI operating costs are already limiting further use.
McKinsey's State of AI in 2026 report, titled "On the Road to ROI," surveyed 1,719 professionals and business leaders across industries and 97 countries. The headline is optimistic: adoption is essentially solved, with nearly nine in ten organisations now using AI regularly and agentic AI proliferating fast. But the report's own subtitle undersells its central finding. Three years into the gen AI era, the share of organisations converting that adoption into measurable enterprise value has been stuck in place for a full year.
Adoption climbed. The value line didn't.
| McKinsey State of AI 2026 (n = 1,719, 97 countries) | Figure |
|---|---|
| Organisations that say they regularly use AI | ~90% |
| Respondents attributing at least some EBIT impact to AI | 37% (flat vs. 2025) |
| "AI high performers" (≥5% EBIT impact, described as significant) | 6% (flat vs. 2025) |
| Report AI has improved individual productivity | 80% |
| Report AI helps them make better decisions | 50% |
| Organisations limiting AI use because of operating costs | ~20% (1 in 5) |
| Expect their employer to cut jobs because of AI this year | 39% (up from 32% in 2025) |
The two numbers that matter most sit next to each other and don't move together. Adoption is nearly universal. The share of organisations that can point to a measurable EBIT effect, 37 percent, is unchanged. And the share achieving real scale, the 6 percent of high performers, hasn't budged either. Individual-level gains are real and widely felt, 80 percent report a personal productivity lift, but that has not translated into enterprise-level financial impact for the vast majority of the organisations reporting it.
Where the value shows up, when it shows up at all
McKinsey's function-level data explains part of the gap. Cost reductions concentrate in supply chain management, service operations, and manufacturing, functions with well-defined, repeatable processes AI can be pointed at directly. Revenue gains cluster in marketing and sales, followed by product and service development and software engineering, functions where AI augments judgment calls rather than replacing a fixed process. High performers aren't distinguished by using more AI; McKinsey finds they are distinguished by using AI to transform how the organisation works, deploying a broader set of best practices and technologies, and actively managing AI-related risk, while the other 94 percent mostly chase efficiency inside their existing workflows.
The operating-cost finding adds a second constraint most 2025 commentary missed: it isn't only measurement discipline holding value back now. One in five organisations say they are actively limiting AI use because running it costs too much, a ceiling that scales with usage rather than shrinking as models improve. Layer that against 39 percent of respondents now expecting AI-driven job cuts this year, up from 32 percent, and the workforce story around this survey is getting harder, not easier, even as the ROI story stays flat.
What this means for enterprise leaders
- Stop treating adoption metrics as a proxy for value; 90 percent usage sitting next to a flat 6 percent high-performer rate shows they have decoupled.
- Study the high-performer playbook, not the average deployment: broader best practices, wider technology coverage, and active risk management, not simply more AI, is what separates the 6 percent.
- Target functions where the process is well-defined for cost wins (supply chain, service ops, manufacturing) and functions where AI augments judgment for revenue wins (marketing, sales, product, engineering).
- Put AI operating costs on the same governance dashboard as ROI; a fifth of organisations are already hitting a cost ceiling that adoption-stage plans didn't budget for.
Related reading
- McKinsey says enterprise AI is finally 'on the road to ROI' (The Register)
- Production is up, ROI is flat: enterprise AI's discipline gap
Source: The State of AI in 2026: On the Road to ROI, McKinsey & Company.