McKinsey: Eight in ten faster with AI, EBIT unchanged

05.09.2026

5 min read

Eight in ten respondents work faster with AI. Yet its profit contribution has remained unchanged for a year. That is the finding of McKinsey’s latest survey of 1,719 professionals and executives. Three days later, the same consultancy presents a playbook promising 20 percent more EBITDA.

Key takeaways

37 percent of organizations report a positive EBIT impact from AI, the same share as in 2025. Its profit contribution stagnates despite far more companies deploying AI enterprise-wide.
80 percent of respondents report higher personal productivity. Leaders achieve this by redesigning workflows; most companies add tools to existing processes.
14 percent report that AI reduced headcount last year. More than twice as many had expected this.
McKinsey’s playbook draws on 20 selected companies rather than the full survey. The 20 percent EBITDA increase comes from these cases rather than the survey sample.

RelatedDefine the decision thesis before the org chart / Why AI fails at master data

Employees work faster, profits stand still

For its State of AI 2026 survey, McKinsey questioned around 1,700 professionals and executives in 97 countries in May and June 2026. Eight in ten respondents say AI has accelerated their own work.

McKinsey’s analysis identifies the resulting gap: people work faster, but their companies earn no more. 37 percent of organizations report a positive contribution to EBIT, or earnings before interest and taxes. That figure was unchanged from a year earlier, despite substantially more organizations deploying AI enterprise-wide since then. Even the leaders have stalled: six percent attribute at least five percent of their EBIT to AI. McKinsey calls them high performers.

Survey metric	2026	Previous year
Positive EBIT contribution from AI	37 percent	37 percent
High performers (at least 5 percent of EBIT from AI)	6 percent	6 percent
Higher personal productivity	80 percent	Not reported
High performers fundamentally redesigning workflows	nearly 75 percent	55 percent
Other respondents redesigning workflows	around 25 percent	Not reported
AI reduced headcount (reported)	14 percent	32 percent (previous year’s expectation)

Source: McKinsey, The State of AI 2026, online survey of 1,719 participants in 97 countries, conducted May to June 2026, weighted by share of gross domestic product (GDP).

The burden is uneven. 47 percent of middle managers and administrative staff report at least one negative consequence of AI use, compared with 31 percent of board members and senior executives. Those working with the tools daily feel the friction. Those approving the investment mainly see the acceleration.

The 20 percent EBITDA increase comes from 20 selected companies

Three days after the survey, McKinsey Quarterly published “The new management playbook for AI”. Its opening sentence says 94 percent of companies have yet to generate meaningful value. The footnote points back to the same survey: this is the other side of the six percent of high performers. There is no second study behind it.

Next come 20 selected companies that McKinsey itself advised. After three years, their EBITDA, or earnings before interest, taxes, depreciation and amortization, is up a fifth on average. Behind this are investments in the tens to hundreds of millions, involving those same 20 client cases rather than a global sample.

Both figures can be correct because they describe different things. The survey is a weighted global sample. The playbook is a selection of successful cases from the same consultancy that sells these transformations. Across the broader survey, AI’s EBIT contribution remains unchanged.

Job cuts remain an expectation

When profits fail to materialize, cutting staff seems a likely next step. The survey provides no support for it. 14 percent of organizations using AI report that it helped reduce headcount last year. A year earlier, 32 percent had expected precisely that, more than twice the share that experienced it.

Organizations expect staff reductions, but respondents rarely see their own jobs at risk. 39 percent expect headcount reductions next year, while only 13 percent feel their own careers are threatened.

32 percent of organizations declined to buy at least one software product because AI programming tools, known as coding agents, built the functionality internally. This shifts costs from software budgets to in-house development but does not explain the flat EBIT contribution.

High performers redesign workflows, the rest buy tools

Redesign sets high performers apart. For McKinsey, this means rebuilding processes around agents rather than adding a tool to an employee’s existing workflow. Nearly three-quarters of high performers now do this, up from 55 percent a year earlier; among other respondents, the share remains one-quarter. They pursue growth or innovation alongside efficiency and measure the impact. The report identifies organizations’ capacity to absorb change as the limiting factor.

Kate Smaje, McKinsey’s global technology and AI leader, places responsibility where the workflow resides: quality and judgment remain with the business unit owning the process. This does not make the chief technology officer the manager of a thousand agents; responsibility stays with the business unit.

One in five already limits AI use because of operating costs, as billing per token, or unit of processed text, begins to bite. Still, 60 percent plan to invest more next year. While workflows stay unchanged, faster individual work saves calendar time without appearing in the income statement. Only the next survey will show whether 2027 spending translates into profits.

What is the State of AI survey? McKinsey annually surveys professionals and executives worldwide on AI adoption and impact. It calls the six percent of organizations attributing at least five percent of operating profit to AI high performers.

Frequently asked questions

What exactly did McKinsey measure?

An online survey covered 1,719 participants in 97 countries from early May to early June 2026. Data are weighted by countries’ GDP shares. 36 percent of respondents work in organizations with revenue exceeding one billion.

Why does the share reporting an EBIT contribution remain at 37 percent?

Individuals work faster, but company profits follow only when workflows are redesigned and effects measured. High performers do precisely this far more often than the rest.

Does the 20 percent EBITDA increase contradict the survey?

No. The survey covers the broader picture. The playbook describes 20 selected cases from the same consultancy. The 94 percent without meaningful value are the other side of the six percent of high performers.

Are companies now building their own software?

32 percent declined to buy at least one product or feature because coding agents built it internally. Among high performers, nearly half did so. This changes the build-or-buy decision without explaining the EBIT impact.

Does the survey point to job cuts?

No. It shows an expectation. More than twice as many had expected headcount reductions as subsequently experienced them.