McKinsey Asked 1,993 Executives About AI. 88% Use It. Only 39% Can See It in Profit.

Via McKinsey & Company: The state of AI in 2025: Agents, innovation, and transformation (November 5, 2025)

The adoption number is huge. The profit number isn't.

McKinsey's annual State of AI survey is the one boards and consultants quote most. The November 2025 edition has a headline that sounds like victory: 88% of respondents say their organisations use AI regularly in at least one business function, up from 78% a year earlier.

Read further and the picture changes. Nearly two-thirds of respondents say their organisations have not yet begun scaling AI across the enterprise; about one-third have. And when McKinsey asks about the bottom line, only 39% report any impact on EBIT (earnings before interest and taxes) from AI at the enterprise level. Most of those say AI accounts for less than 5% of their organisation's EBIT.

So: nearly everyone is using it, a minority can see it in profit, and for most of that minority the effect is small. That lines up with the pattern in the MIT AI report, from a much bigger survey.

Who was asked, and why it matters

The online survey ran from June 25 to July 29, 2025, and drew 1,993 participants in 105 nations, across industries, company sizes and functions. Everything in it is self-reported. That doesn't make it wrong, but it means "39% see EBIT impact" is 39% of respondents saying so, not 39% of audited P&Ls. Treat the numbers as how executives perceive their AI programmes, which is useful in its own right.

Agents: lots of experimenting, little scale

The 2025 survey asked about AI agents for the first time in depth. 62% say their organisations are at least experimenting with agents. 23% say they are scaling an agentic system somewhere in the enterprise. In any single business function, no more than 10% report scaling agents.

For an owner being pitched an "AI agent" this quarter, that's the useful fact: even among larger firms, agents at scale are rare. Our AI agents for business guide covers when a plain automation is the better buy.

What the 6% do differently

McKinsey identifies a group of "AI high performers", about 6% of respondents, who report EBIT impact of 5% or more and significant value from AI. The interesting part isn't that they spend more. It's how they work:

• They redesign the work. High performers are nearly three times as likely as others to say their organisations have fundamentally redesigned individual workflows. Bolting a chatbot onto an unchanged process is the low-performer pattern.
• They aim beyond efficiency. 80% of respondents say efficiency is an objective of their AI work, but high performers more often also set growth and innovation goals. 64% of all respondents say AI is enabling their innovation.
• They check the output. High performers are more likely to have defined processes for when model outputs need human validation.

The small-firm gap in the data

McKinsey breaks scaling out by company size. Among companies with less than $100 million in annual revenue, 29% say they are scaling AI, compared with nearly half of those with more than $5 billion. If you run a smaller firm and feel behind, the survey suggests that's common, not a personal failing.

It also suggests where the gap comes from. Scaling needs the things large firms buy by default: someone who owns the workflow, clean inputs, and a habit of measuring. A 20-person firm can have all three; it just has to choose to. The AI readiness assessment shows which of those you already have.

The risk line most summaries skip

51% of respondents from organisations using AI say they've seen at least one negative consequence, and nearly one-third report consequences from AI inaccuracy. That's the cost side of the ledger, and it falls hardest on firms without a review step. A short hallucination check and a written governance checklist are cheap insurance.

What to take from it, if you run a small firm

1. Don't measure yourself on adoption. 88% use AI; that number tells you nothing about payback. Measure cost per finished task with the AI ROI calculator.
2. Pick one workflow and redesign it. That's the clearest high-performer trait. Use the AI use case scorer to choose which.
3. Write down who checks what. High performers define when a human validates output. You can do that in an afternoon.
4. Be patient with agents. If big firms aren't scaling them past 10% in any function, you don't need to be first.

For the wider context on what's real and what's sales talk, read AI hype vs reality and what the 95%, 80% and 40% failure numbers measure.

Frequently asked questions

What did McKinsey's State of AI 2025 find?

88% of respondents say their organisations use AI regularly in at least one function, but only 39% report any EBIT impact, and most of those say it's under 5% of EBIT. Nearly two-thirds have not begun scaling AI.

How many companies get profit from AI, according to McKinsey?

39% of respondents report some enterprise-level EBIT impact from AI. About 6% qualify as high performers, reporting EBIT impact of 5% or more.

Is the McKinsey AI survey reliable?

It's a large survey (1,993 participants in 105 nations, June–July 2025), but the results are self-reported. Read them as executive perceptions, not audited figures.

Are small companies behind on AI?

In McKinsey's data, 29% of companies under $100 million in revenue say they are scaling AI, versus nearly half of those over $5 billion.