TL;DR
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
Companies are adopting AI, but recent surveys show that measurable financial returns remain limited. McKinsey found 37% of respondents reported some effect on EBIT, while 6% attributed at least 5% of EBIT to AI; BCG reported value at scale at 5% of surveyed firms. The surveys use different samples and definitions, and do not establish that AI caused every reported change.
Recent company surveys show a widening gap between AI adoption and reported financial returns: while use is spreading, only a small share of respondents say AI has produced a large effect on profit. In McKinsey’s 2026 survey, 37% reported some effect on EBIT, but just 6% attributed at least 5% of EBIT to AI, according to coverage by The Register and TechTimes. The findings suggest that deploying AI alone does not guarantee a measurable profit gain.
McKinsey’s survey included 1,719 respondents. The 37% who said AI had some effect on company EBIT was about the same share as a year earlier. The 6% classified as “high performers” said AI accounted for at least 5% of EBIT. These are respondents’ reports, not audited measures establishing that AI caused the results.
A separate BCG survey of 1,250 companies found a similar concentration of reported value: 5% said they were getting value at scale, while 60% reported no material value. BCG said 35% were scaling AI with some returns. The survey categories and sample differ from McKinsey’s, so the percentages should not be treated as direct comparisons.
Adoption is also uneven by company size and location. Eurostat reported that 20.0% of EU enterprises with at least 10 employees used AI in 2025, up from 13.5% in 2024. Usage stood at 55% among large enterprises and 17% among small ones. Eurostat measures reported use, not business value.
The gap between AI use and profit
Companies are adopting AI, but surveys report that substantial financial returns remain concentrated among a small share of firms. Reported results do not prove AI caused the gains.
AI use is spreading
Eurostat tracks adoption, while business surveys report value using different measures.
More enterprises use AI
20.0% used AI in 2025, up from 13.5% in 2024. Use does not show whether firms profit.
Adoption varies
AI use reached 55% among large enterprises and 17% among small ones.
Reported value remains concentrated
BCG’s separate survey found different levels of progress across companies.
What may separate leaders
The surveys point to implementation and business fit, but do not establish what causes stronger returns.
Tie each use case to a clear business result, such as cost, revenue or output.
BCG’s leading firms had deployed 62% of their AI initiatives, compared with 12% among laggards.
Expertise gaps, costs, unclear value and weak risk controls are cited as barriers.
Turning AI Use Into Returns
The survey gap matters because companies may spend on software, staff and process changes without seeing a clear financial return. The figures point to a distinction between having AI in use and making it work across operations in ways that affect costs, revenue or output. They do not, by themselves, explain why particular firms succeed.
BCG’s findings suggest that implementation capacity may be part of the divide: its leading firms had deployed 62% of their AI initiatives, compared with 12% among laggards. That is an association within BCG’s survey, not proof that deployment rates alone cause better results. For executives, the practical question is whether a use case improves a defined business measure, rather than whether a pilot can produce a convincing demonstration.
Employees also have a stake in how companies measure AI’s effects. In McKinsey’s survey, 80% of respondents who use AI said it made them more productive. At the same time, 39% expected job cuts, up from 32% the prior year. The source reports that previous expected cuts had exceeded the cuts that later occurred, underscoring that expectations are not the same as realized employment changes.
Products Worth Considering
Logitech MX Keys Combo for Business Keyboard and Mouse – Graphite
Kindle Paperwhite Signature Edition 32GB (2024) – Auto-adjusting front light, wireless charging, weeks of battery life – Metallic Black
Bose QuietComfort True Wireless Earbuds with Active Noise Cancellation, Black
As an affiliate, we earn on qualifying purchases.
From Digital Tools to Agents
Earlier digital transformation often moved paper processes into software while leaving people to interpret cases and make decisions at each step. AI systems can handle some unstructured inputs, such as emails, invoices and support tickets, and may take actions under set limits. That changes the role software can play in a workflow, though the surveys do not show that every firm has moved from pilots to dependable, integrated systems.
Agent adoption is one measure of this shift. BCG respondents said agents already delivered about 17% of their AI value and expected that share to reach 29% by 2028. McKinsey reporting said 40% of respondents at companies with more than $1 billion in revenue were scaling agents, compared with 27% a year earlier. These figures are survey responses and expectations, not independently verified outcomes.
Barriers remain. Among European firms that had considered AI but did not use it, 71% cited a lack of expertise, according to Eurostat figures in the source material. Gartner forecasts that more than 40% of agentic AI projects will be cancelled by the end of 2027 because of costs, unclear value or weak risk controls. It also forecasts that organizations will abandon 60% of AI projects that lack AI-ready data. These are forecasts, not counts of cancellations that have already happened.
“80% of respondents who use AI said it made them more productive.”
— McKinsey survey respondents, as reported by The Register and TechTimes
What the Surveys Cannot Show
The surveys rely on different samples, definitions and measures. McKinsey’s EBIT figures are self-reported, BCG uses its own categories for value and scaling, and Eurostat tracks whether firms use AI rather than whether they profit from it. Their percentages cannot be combined into a single rate of AI success or compared as if they measured the same outcome.
The material also does not identify which specific applications generated reported gains, how returns varied by industry, or how much was due to AI rather than other business changes. It remains unclear how many agent projects will reach sustained use, how Gartner’s forecast will compare with actual cancellations, and whether respondents’ productivity gains will translate into lasting financial results or changes in employment.
Tracking Projects Through 2027
Gartner’s forecast puts the end of 2027 in focus for agentic AI projects, but the available material does not name a confirmed next survey or company reporting milestone. The practical test for businesses will be whether projects move beyond trials, work with their data and controls, and produce measurable results over time. Future survey results and company disclosures may clarify whether reported gains have broadened beyond the small group identified as high performers.
Key Questions
How many companies reported a significant AI effect on profit?
In McKinsey’s 2026 survey, 6% of 1,719 respondents qualified as high performers by attributing at least 5% of EBIT to AI. This is a self-reported survey measure.
Does the 37% figure mean AI raised profits by 37%?
No. It means 37% of McKinsey respondents said AI had some effect on EBIT. It is not the size of a profit increase, and the survey does not establish causation.
How widespread was AI use in the EU?
Eurostat said 20.0% of EU enterprises with at least 10 employees used AI in 2025, up from 13.5% in 2024. Use was reported by 55% of large firms and 17% of small firms.
Why are some AI projects expected to be cancelled?
Gartner forecasts cancellations tied to cost, unclear value and weak risk controls. It also forecasts abandonment of projects that lack AI-ready data. These are forecasts; the source does not provide a final count of cancellations.
Source: Thorsten Meyer AI
