Most Companies Use AI. Only About 6 Percent Profit From It

By Thorsten Meyer | November 6, 2026 | ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on tech for your team

• Business-only prices and quantity discounts
• Tax-exempt purchasing
• Multiple users, one account, clear invoices

Create a free Business account →Free to register

As an affiliate, we earn on qualifying purchases.Hide

For twenty years, digital transformation meant one thing: move the paper into software. Forms became web forms, filing cabinets became databases, and the work still happened the same way. A person opened a screen, read a case, made a decision and clicked a button.

AI changes the second half of that sentence. The new promise is not “digitise the process” but “let software process the work”: read the invoice, check it against the contract, flag the exception, draft the reply, and only call a human when it matters. That is a much bigger change than a new ERP system, and the numbers show it. Most large companies now use AI somewhere. Only a small minority can show that it moves their profit.

I went through the primary surveys and forecasts behind the headlines: Eurostat’s official data on AI use in European companies, BCG’s study of 1,250 firms, Gartner’s published forecasts, McKinsey’s 2026 State of AI survey as reported by The Register (and, in a second report, TechTimes), and MIT’s much-quoted “GenAI Divide” report as reported by Fortune. Here is what really changes, where it pays off, where it fails, and what it means for companies and for the people who work in them.

Key takeaways

• Use is broad, profit is narrow. In McKinsey’s 2026 survey of 1,719 respondents, 37 percent said AI has some effect on their company’s EBIT, about the same as a year earlier. Only 6 percent qualify as “high performers” with at least 5 percent of EBIT attributed to AI, as reported by The Register and TechTimes. BCG finds a similar split: 5 percent of 1,250 companies get value at scale, 60 percent report no material value.
• The job of a transformation programme is changing. BCG’s respondents say AI agents already deliver about 17 percent of their AI value and expect 29 percent by 2028. Agents are AI systems that take steps on their own, not just answer questions. Among companies with more than $1 billion in revenue, 40 percent now say they scale agents, up from 27 percent, according to McKinsey as reported by The Register and TechTimes.
• Europe is earlier in the curve than the hype suggests. Eurostat: 20.0 percent of EU enterprises with ten or more employees used AI in 2025, up from 13.5 percent. Among large enterprises it was 55 percent, among small ones 17 percent.
• Pilots die for boring reasons. Gartner forecasts that over 40 percent of agentic AI projects will be cancelled by the end of 2027 because of cost, unclear value or weak risk controls, and that organisations will abandon 60 percent of AI projects that lack AI-ready data. European firms that considered AI but did not use it most often name missing expertise (71 percent).
• For business: the winners redesign whole workflows, pick few use cases, track value and govern what agents may do. BCG’s leaders have deployed 62 percent of their AI initiatives; laggards 12 percent.
• For employees: 80 percent of McKinsey respondents who use AI say it made them more productive. More of them now expect job cuts (39 percent, up from 32), but last year’s cuts fell well short of what respondents had expected, as reported by The Register and TechTimes.

Products Worth Considering

Logitech MX Keys Combo for Business Keyboard and Mouse – Graphite

AmazonView Latest Price

Apple AirPods Pro 3 Wireless Earbuds with Active Noise Cancellation

AmazonView Latest Price

Samsung Galaxy Tab A11 8.7-Inch Wi-Fi Tablet, 64GB, 4GB RAM, Silver

AmazonView Latest Price

As an affiliate, we earn on qualifying purchases.

From “digitise the process” to “let software process the work”

Classic digital transformation programmes had a clear shape. You mapped a process, picked a system, migrated the data, trained people on new screens. The human stayed in the middle of every step. The system stored and moved information; people interpreted it.

AI, and in particular AI agents, moves the line. A system can now read unstructured input (an email, a scanned invoice, a support ticket), decide what kind of case it is, look things up in other systems, and take an action within limits someone has set. Gartner describes the split in practical terms: use AI agents “when decisions are needed, automation for routine workflows and assistants for simple retrieval.”

Three things change in a programme when that happens:

1. The unit of change is the workflow, not the tool. Gartner warns that bolting agents onto legacy systems “can be technically complex, often disrupting workflows” and says that “in many cases, rethinking workflows with agentic AI from the ground up is the ideal path.” BCG makes the same point from its survey: big value comes “not from AI pilots or isolated use cases, but from reshaping and reinventing core business workflows end-to-end.”
2. People move from doing to supervising. BCG describes agents as “digital workers” that operate alongside or under the supervision of people. Someone has to set their limits, review their exceptions and own the result.
3. Governance becomes part of the design. An agent that can send an email or book a payment needs permissions, logs and a stop button. I wrote earlier about how to delegate to AI agents without handing over the vault. In a transformation programme, that question moves from the security team to the programme plan.

There is also a quieter change. McKinsey’s 2026 survey found that nearly a third of respondents’ organisations decided not to buy at least one software product or feature because they could build it in-house with agentic coding tools, The Register reported. A part of the classic “buy a system” step is turning into “build a small one.”

Who is getting value, and how many

The headline numbers depend on who is asked and what counts as “using AI.” So it helps to look at three different kinds of sources side by side.

Official statistics (Europe). Eurostat’s survey of enterprises with ten or more employees is the most representative source here. It found that 20.0 percent used at least one AI technology in 2025, up from 13.5 percent in 2024 and 8.1 percent in 2023. Size matters enormously: 55.03 percent of large enterprises used AI, 30.36 percent of medium-sized ones and 17 percent of small ones. Denmark led with 42.0 percent, Romania was lowest with 5.2 percent. The most common use was analysing written language (11.8 percent of all enterprises).

Executive surveys (global). McKinsey’s “The state of AI in 2026: On the road to ROI” surveyed 1,719 professionals and business leaders around the world. According to The Register’s report on it (TechTimes reported the same figures a day later), 37 percent “attribute at least some EBIT impact to AI use,” roughly unchanged from 2025, and only 6 percent count as high performers, those who attribute at least 5 percent of EBIT to AI and call its impact significant. That share was also flat. McKinsey’s own summary: “Organizations’ conviction in AI is growing faster than the immediate financial returns they can attribute to it.”

Maturity studies. BCG’s “The Widening AI Value Gap” (September 2025) scored 1,250 companies in 68 countries on 41 capabilities. Five percent are “future-built” and achieve value at scale. Thirty-five percent are scaling and see some returns. Sixty percent “are not achieving material value at all.” BCG reports that the future-built group achieves 1.7 times the revenue growth and 3.6 times the three-year shareholder return of stagnating and emerging companies. Those are correlations from a consultancy survey, and BCG itself notes that responses “may be subject to perception bias.”

The most-quoted number is the bleakest: MIT’s NANDA initiative reported in August 2025 that “about 5% of AI pilot programs achieve rapid revenue acceleration” while most stall, as Fortune reported. That study rests on 150 interviews, a survey of 350 employees and 300 public deployments, and it defines success through what leaders said. Treat the “95 percent fail” framing as a strong signal, not a measurement.

What the sources agree on is the shape: wide adoption, a thin layer of companies with real financial results, and a large middle that is stuck between pilot and production. In an earlier piece, I looked at an economist’s argument that AI pays off almost nowhere outside tech yet. The enterprise surveys point in the same direction.

Why so many pilots never reach production

Almost none of the published reasons are about the AI model being too weak. They are about money, data, people and control.

• Cost. Gartner names “escalating costs” first among the reasons it expects more than 40 percent of agentic AI projects to be cancelled by the end of 2027. In McKinsey’s 2026 survey, 20 percent of respondents said AI operating costs had constrained their use, according to The Register.
• Unclear value. Gartner’s analyst Anushree Verma said most agentic projects “are early stage experiments or proof of concepts that are mostly driven by hype and are often misapplied.” Gartner also estimates that only about 130 of the thousands of vendors selling “agentic AI” are real; the rest rebrand chatbots or older automation tools, which Gartner calls “agent washing.”
• Data. Gartner predicts that through 2026, organisations will abandon 60 percent of AI projects that are not supported by AI-ready data. In its survey of 248 data management leaders, 63 percent said they did not have, or were not sure they had, the right data management practices for AI.
• Skills and legal uncertainty. Among EU enterprises that had considered AI but not used it, Eurostat found the most common reasons were a lack of relevant expertise (70.89 percent), unclear legal consequences (52.52 percent) and data protection concerns (48.83 percent). Only 20.68 percent said AI was not useful for them.
• Integration. MIT’s report blames a “learning gap”: generic tools work well for individuals but stall in companies because they do not learn from or adapt to a company’s workflows, lead author Aditya Challapally explained to Fortune.
• Security. BCG says 72 percent of companies “already report unmanaged AI-security risks.”
• Misplaced budgets. According to MIT, more than half of generative AI budgets go to sales and marketing tools, while the biggest returns showed up in back-office automation.

What is solid, and what is contested

A few cautions before anyone builds a strategy on these numbers:

• Most figures are self-reported. McKinsey, BCG and MIT ask executives what they see. Eurostat counts use, not value. Nobody in this set measures profit from AI directly.
• Forecasts are not results. Gartner’s 40 percent cancellation figure and its 60 percent data figure are predictions. The same Gartner release also forecasts that by 2028, 15 percent of day-to-day work decisions will be made autonomously by agentic AI and 33 percent of enterprise software will include it.
• Buy or build is unresolved. MIT found that buying from specialised vendors and partnering succeeded about 67 percent of the time, while internal builds succeeded “only one-third as often,” per Fortune. McKinsey’s 2026 survey points the other way for some software: nearly a third of respondents chose to build in-house with coding agents, as reported by The Register and TechTimes. Both can be true, depending on how big and how core the thing is.
• Big companies are not average companies. Executive surveys hear mostly from larger organisations. Eurostat shows that small businesses are only about a third as likely as large ones to use AI at all.

What companies should do differently now

Read the surveys together and the playbook of the companies that get value is not mysterious. It is mostly discipline:

1. Start from a workflow with a number attached. Pick a process where you can measure time, cost or error rates before and after. BCG’s future-built and scaling companies report a 76 percent higher match between where AI is deployed and where it delivers impact, and typically reach impact in 9 to 12 months instead of 12 to 18.
2. Fewer pilots, more production. BCG’s future-built companies have already deployed 62 percent of their AI initiatives; laggards only 12 percent. A pilot without a path to production is a cost, not an asset.
3. Fix the data you need, not all data. Gartner’s advice is to align data to a specific use case, define governance requirements and assure quality continuously.
4. Give agents a job description. Write down what an agent may read, what it may change, what it must escalate, and who signs off. Log every action. Review grows in importance as agents produce more: on GitHub, AI agents now start four in ten pull requests and the bottleneck has moved to review.
5. Budget for people, not just licences. BCG’s rule of thumb puts 70 percent of the effort on people and processes, 20 percent on technology and 10 percent on algorithms. Its leaders plan to upskill more than half of their workforce.
6. Track value from day one. More than 60 percent of BCG’s future-built firms “rigorously track AI value,” compared with 17 percent of stagnating companies.
7. Watch the running costs. Agent workflows can make many model calls per task. Put a cost per completed case next to the time saved.

What changes for employees

For people inside these companies, AI-driven transformation feels different from the last wave. The last wave changed which screens you used. This one changes which parts of the job are yours.

• Productivity is the most-reported effect. In McKinsey’s 2026 survey, 80 percent of respondents who use AI in their role said it improved their individual productivity, The Register reported.
• Expectations about jobs are shifting, results less so. 39 percent of respondents expect their employer to cut jobs because of AI in the coming year, up from 32 percent; 43 percent still expect little or no change, The Register reported. McKinsey noted that workforce reductions in 2025 “fell well short” of what respondents had anticipated a year earlier.
• The quieter pattern is not backfilling. MIT’s researchers found that companies are often not replacing people when roles become vacant, especially in customer support and administrative work, rather than running mass layoffs.
• Supervision becomes a skill. When an agent drafts, someone has to check. When an agent escalates, someone has to decide. The new work is in setting limits, handling exceptions, owning the result and improving the process.
• Skills are the bottleneck. The most common reason European firms give for not using AI is missing expertise. That is also an opportunity for anyone who learns to work with these systems early.

The bottom line

AI does not make digital transformation obsolete. It raises the bar. The old programme ended when the paper was gone. The new one ends when software does a defined share of the work, under rules people set, with results someone measures.

Right now most companies are at the start of that road. The surveys put the group that gets real financial value at around 5 to 6 percent. What separates them is not a better model. It is a redesigned workflow, clean enough data, clear limits for agents, and the patience to count what actually changed.

Sources

• Eurostat, “20% of EU enterprises use AI technologies,” news release, 11 December 2025; and Statistics Explained, “Use of artificial intelligence in enterprises” (data extracted December 2025). EU survey on ICT usage and e-commerce in enterprises, enterprises with 10 or more employees.
• BCG, “The Widening AI Value Gap: Build for the Future 2025,” September 2025. Survey of 1,250 CxOs and senior executives in 68 countries; 41 capabilities.
• Gartner, “Gartner Predicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027,” press release, 25 June 2025 (incl. January 2025 poll of 3,412 webinar attendees).
• Gartner, “Lack of AI-Ready Data Puts AI Projects at Risk,” press release, 26 February 2025 (survey of 248 data management leaders, Q3 2024).
• McKinsey & Company, “The state of AI in 2026: On the road to ROI,” August 2026, as reported by The Register, “McKinsey says enterprise AI is finally ‘on the road to ROI’,” 25 August 2026. Survey of 1,719 respondents. Also reported by TechTimes, “Record AI Spending Can’t Move Earnings Needle for 94% of Enterprises, McKinsey Finds,” 26 August 2026.
• MIT NANDA, “The GenAI Divide: State of AI in Business 2025,” as reported by Fortune, “MIT report: 95% of generative AI pilots at companies are failing,” 18 August 2025.

This article explains public surveys and forecasts. It is not business, legal or investment advice.