The Gap Between AI Adopters and Everyone Else

Nearly every large survey of businesses now says the same first thing: AI is in use almost everywhere. In McKinsey’s 2025 global survey, 88 percent of respondents said their organisation uses AI regularly in at least one business function, up from 78 percent a year earlier. The second thing the same survey says gets less attention. Only 39 percent of respondents reported any effect on enterprise-level earnings from AI, and most of those put it below 5 percent of EBIT. McKinsey’s August 2026 edition found that share essentially unchanged, at 37 percent.

Both findings are true at once. Using AI and benefiting from it are different conditions, and the distance between them is where the interesting evidence sits. This piece looks at what the 2025 and 2026 research says about the firms on the far side of that distance, how large the gap has become, and why it is more likely to widen than to close.

Boston Consulting Group put numbers on the split in its September 2025 report, The Widening AI Value Gap, based on a survey of 1,250 senior executives across nine industries. It sorted companies into three bands. Five percent qualified as “future-built” for AI, 35 percent were “scalers” beginning to generate value, and the remaining 60 percent were “laggards” reporting minimal revenue and cost gains.

The gap between the first and last band is large. BCG reports that future-built companies achieve 1.7 times the revenue growth, 3.6 times the three-year total shareholder return and 1.6 times the EBIT margin of laggards. Looking ahead, they expect twice the revenue increase and 40 percent greater cost reductions than laggards in the areas where they apply AI.

Two cautions apply. These are survey responses from executives, and BCG built the categories itself. The direction is still consistent with what other researchers find, and the 60 percent figure is a reminder that “we use AI” covers a very wide range of practice.

McKinsey’s survey, which drew 1,993 participants in 105 countries between June and July 2025, isolates a similar group. It defines “high performers” as respondents who attribute 5 percent or more of EBIT to AI and report significant value from it. They make up about 6 percent of the sample.

What separates them is mostly operational. High performers are nearly three times as likely as others to say they have fundamentally redesigned individual workflows. McKinsey ran a relative-weights analysis on 31 variables and found that this redesign has one of the strongest contributions to meaningful business impact of all the factors it tested. About three-quarters of high performers say their organisations are scaling or have scaled AI, against about one-third of the rest, and they report deploying twice as many use cases.

A chatbot subscription does not appear on that list. The pattern is a process that used to need a person at each step, rebuilt so that software handles the routine steps and a person handles the exceptions. The tool matters less than the redesign around it.

PwC’s 2026 Global AI Jobs Barometer, which analysed more than one billion job advertisements in 27 countries and territories alongside company financial data, looks at the question from the outside. It compares firms by how exposed their sector is to AI, which is a measure of whether AI can do their work, and not of whether they have adopted it.

Companies in the most AI-exposed sectors recorded 34 percent productivity growth in 2025 relative to 2018. The least exposed recorded 24 percent. Within the most exposed group, PwC finds what it calls a “super-star” effect: the top 20 percent achieved average labour productivity growth of 163 percent relative to 2018, nearly five times the group average.

That last figure is the useful one for a business owner. Inside a single AI-exposed sector, a minority is far ahead of the average, and the average is itself ahead of less exposed sectors. Exposure creates the opportunity, and what a company does with it decides where it lands. The comparison is by sector exposure, so it shows an association and does not prove that AI caused every point of the difference.

Eurostat’s 2025 survey of enterprises with ten or more employees, collected in the first months of 2025, found that 19.95 percent of EU enterprises used at least one AI technology, 6.47 percentage points more than in 2024. The average conceals a steep gradient by size: 17 percent of small enterprises, 30.36 percent of medium-sized ones and 55.03 percent of large ones.

Intent shows the same slope. Among firms not using AI, 36.54 percent of large enterprises had considered it, against 12.65 percent of small ones. Eurostat suggests the complexity of implementation, economies of scale and the cost of investment as likely reasons large firms move first.

Bulgaria sits near the bottom. At 8.55 percent, it recorded the third-lowest share in the EU in 2025, after Romania at 5.21 percent and Poland at 8.36 percent. Denmark led at 42.03 percent. According to the Bulgarian Chamber of Commerce and Industry’s AI Council, working from the same Eurostat data, the gap between Bulgaria and the EU average has grown from 4.36 percentage points in 2021 to 11.40 in 2025.

These are adoption rates, which count any use of AI, and they say nothing about depth. Given McKinsey’s finding that most users have not scaled, the share of firms with AI wired into daily operations is smaller than any of these figures.

The data above are snapshots, so what follows is interpretation, though it rests on the survey evidence. Leaders invest more. BCG reports that future-built firms planned to spend more than twice as much on AI as laggards in 2025, and more than one-third of McKinsey’s high performers commit over 20 percent of their digital budgets to it.

A redesigned process also leaves something behind. Each time a firm rebuilds a workflow, it accumulates cleaner data, written-down rules, staff who know where the software helps and a record of what went wrong. The next project starts from that base. A firm that has only handed its staff a chatbot starts every project from zero.

Scale matters too. McKinsey found that nearly half of respondents at companies with more than $5 billion in revenue had reached the scaling phase, against 29 percent at companies under $100 million. Smaller firms start further back, which is also what Eurostat’s size gradient shows.

The evidence does not say that every business needs an ambitious AI programme. It says that the benefit goes to firms that change how a specific piece of work is done, and that the first firms to do so in a given market gain a lead that is expensive to close later.

A practical starting point is modest. List the recurring tasks that consume the most hours, such as answering the same enquiries, booking and rescheduling, chasing leads and posting content. Pick the one with the clearest rules, rebuild it end to end, and measure it for a quarter. Looking at where your hours go is the first step in deciding which of those tasks to rebuild.