smbai-adoptioncbseurostatai-strategy

AI adoption in Dutch SMBs: statistics by sector and size

November 9, 202610 min readPixel Management

This article is also available in Dutch

The Dutch cabinet wants at least 75 percent of the mkb, the Dutch term for small and medium-sized business, on advanced digital technology by 2030, but that target adds AI, cloud and data analytics together.

AI adoption among Dutch SMBs is 29.8 percent according to Statistics Netherlands (CBS, 10 to 249 people employed, 2025), and 46 percent in the 50 to 250 band. Within that band it runs from 29 percent in transport and storage to 78 percent in information and communication. The 70 and 78 percent you'll see from Exact and ABN AMRO measure something else: whether AI shows up in any form.

Which AI adoption figure for Dutch SMBs measures what?

Figures on AI adoption in Dutch SMBs range from 29.8 to 78 percent in 2026 because each source counts something different. These are the six numbers that turn up in board papers and press releases.

FigureSource and dateCoversWhat counts
29.8%CBS, 16 March 2026SMBs, 10 to 249 people employedAI technology used in 2025
33%CBS, 10 February 2026Companies with 10 or more people employedAt least one of eight AI technologies
46%CBS StatLine 86116NED, 12 December 2025, provisional50 to 250 people employedAI used in 2025
70%Exact SMB Barometer, 3 August 2026Over 1,700 SMB owners, size not statedOwner uses the technology
78%ABN AMRO, 23 June 2026, via Cybersecuritybeeld Nederland 2026SMBs in the studyAI "to some degree"
75%Cabinet vision on AI, 6 November 2026SMBs, target for 2030AI, cloud and data analytics together

Only the first three work as a benchmark. The last one isn't a measurement at all.

Why are CBS, ABN AMRO and Exact so far apart?

The gap between CBS and the ABN AMRO and Exact surveys comes down to the question, the respondent and the timing.

The question. CBS asked companies about eight specific AI technologies, such as text mining, speech recognition and machine learning. ABN AMRO reports use "to some degree", a far lower bar. Exact doesn't give its question wording in the press release.

The respondent. CBS measures the company. Exact polls the owner, and ABN AMRO asked 777 organisations. We couldn't open ABN AMRO's original report, so we don't know how the bank defines an SMB. Meanwhile CBS reported on 25 February 2026 that 43 percent of adults in paid work use AI in their job. Our reading, not a finding from the sources: at a company of a hundred people where a few colleagues open ChatGPT, a director answers yes to "to some degree", even if AI isn't part of any process.

The timing. The European figures for 2025 come, according to Eurostat, from surveys run in the first months of 2025. Exact did its fieldwork in the first quarter of 2026 and ABN AMRO published in June 2026. In the 50 to 249 band, Eurostat shows use rising from 31.19 to 46.62 percent in a single year. So a year's difference in fieldwork can be worth about 15 points.

Watch out for one trap. Exact reports a rise from 33 to 70 percent. That 33 isn't the CBS 33 percent: it's a different panel with a different question, and the match is a coincidence.

The working rule: use the official figures as your benchmark and the surveys as a signal. The signal is clear enough. ABN AMRO finds that 9 percent of SMBs have a formal policy on AI models, against 32 percent of large companies, while 78 percent use AI. How to close that gap is covered in our piece on a shadow AI policy and in this AI policy template.

AI adoption by sector for Dutch companies with 50 to 250 people

AI use among Dutch companies with 50 to 250 people differs by 49 percentage points between sectors, so the 46 percent average tells you little about your own market. An ICT company of 100 people without AI belongs to a minority of 22 percent. A transport company of that size without AI belongs with 71 percent of its sector.

AI use at companies with 50 to 250 people, by sector

Share of companies using AI in 2025. Provisional figures.

  • All sectors
  • Sector
  1. Information and communication78%
  2. Financial services71%
  3. Specialist business services66%
  4. Trade47%
  5. All sectors46%
  6. Manufacturing41%
  7. Construction41%
  8. Rental and other business services39%
  9. Health and social care39%
  10. Energy, water, waste38%
  11. Hospitality33%
  12. Transport and storage29%

Source: CBS StatLine, table 86116NED, as of 6 November 2026. CBS publishes no real estate figure for this band. The CBS news item on these figures gives 45 percent for all sectors, without explaining the difference: compare within one source at a time.

Size counts for more than sector in some industries. In construction, 16 percent of companies with 10 to 50 people use AI, 41 percent at 50 to 250 and 73 percent at 250 or more. The step from a small to a mid-sized builder is a factor of more than 2.5. In health and social care it's under 2: from 21 to 39 percent.

The group that considered AI but didn't adopt it also varies a lot. The average in this band is 14 percent. In energy, water and waste it's 23 percent, in health and social care 20 percent, in construction 18 percent. Interest isn't the issue there; getting it into use is. In ICT, where 78 percent already use AI, it's 10 percent.

For what this looks like by sector, see AI by industry.

How do Dutch SMBs compare with the EU?

Dutch companies sit above the EU average in every size class, and the lead is widest at 50 to 249 people: 46.62 against 30.36 percent in Eurostat dataset isoc_eb_ai, a gap of 16.26 points. At 10 to 49 it's 11.76 points, at 250 or more 12.61.

AI use by company size: the Netherlands against the EU, 2025

Share of companies using at least one AI technology.

  • Netherlands
  • EU average
  1. 10 to 49: Netherlands28.76%
  2. 10 to 49: EU17%
  3. 50 to 249: Netherlands46.62%
  4. 50 to 249: EU30.36%
  5. 250 or more: Netherlands67.64%
  6. 250 or more: EU55.03%

Source: Eurostat, isoc_eb_ai, doi.org/10.2908/ISOC_EB_AI (updated 15 June 2026), as of 6 November 2026. Excludes the financial sector, so it doesn't match the CBS figures above. Eurostat flags the Dutch value for 10 to 49 in 2025 as a break in the time series.

That lead looks smaller next to the neighbours. At 50 to 249, Denmark (58.27), Belgium (54.49), Finland (51.35) and Sweden (49.60) all score higher. Germany trails at 35.56 percent. The growth is real, though: in 2023 the Netherlands stood at 19.75 percent in this band.

What does the cabinet's 75 percent target measure?

The Dutch cabinet's 75 percent target doesn't measure AI use. It measures the use of three technologies together. The Cabinet vision of 6 November 2026 says, in our translation: "The cabinet wants to raise the use of advanced digital technologies (AI, cloud technology and data analytics) within SMBs to at least 75% in 2030." The letter gives no starting value, no indicator and no definition of an SMB.

The number resembles the EU target in Digital Decade Decision (EU) 2022/2481: 75 percent of EU enterprises taking up "one or more" of cloud, big data and AI. One of the three is enough there. Set that reading against the Eurostat series and Dutch SMBs have already passed it:

Yardstick, Netherlands 202510 to 24950 to 249
AI, advanced cloud or data analytics (at least one)79.02%91.58%
AI and advanced cloud and data analytics (all three)22.26% (series break)38.73%
AI only31.87%46.62%

The European Commission's report on the Netherlands (24 June 2026) uses the same "or" reading and arrives at 79.67 percent for all companies with 10 or more people. Eurostat also publishes stricter variants for data analytics, which put the 10 to 249 band at 71.35 to 74.43 percent, also close to the target.

This is our comparison, not a statement by the cabinet. The word "raise" implies a starting value below 75 percent, which would mean the cabinet has something else in mind: all three technologies, or a wider group that includes micro businesses. Until the letter says so, the target can't be tested. The Commission's scoreboard also lists a much tougher Dutch target for 2030: 85.1 percent for AI alone, measured at companies with 10 or more people, against a current 33.2 percent.

The cabinet itself says where the work is: "De volgende stap is het structureel toepassen van AI in bedrijfsprocessen", the next step being the structural use of AI in business processes. The 75 percent says nothing about that. The letter also leaves that step to business: "Het is in de eerste plaats aan bedrijven zelf om te innoveren", it's first of all up to companies themselves to innovate.

How deep does AI go at companies with 50 to 249 people?

AI at Dutch companies with 50 to 249 people sits mostly in text work on standard software, far less in processes. This is the benchmark we'd use, because it measures depth rather than presence. All percentages are of all companies in this band (Eurostat isoc_eb_ai, Netherlands, 2025).

IndicatorShare of companies
At least one AI technology46.62%
Text mining31.21%
Ready-made commercial software28.39%
Language or code generation24.63%
At least three AI technologies23.50%
Built or modified by an external provider15.06%
Workflow automation or decision support14.25%
AI for at least three purposes14.06%
AI in production processes10.28%
Built by own staff9.61%
AI in logistics3.42%

The ratio is the takeaway. Of the companies using AI, about three in ten use it for workflow automation (14.25 divided by 46.62 is 31 percent, our calculation). A company running an AI agent or automated order processing belongs to a much smaller group than the headline "nearly half use AI" suggests. We work through that gap between trying and building in AI automation for Dutch SMBs.

Why companies drop out is in the same dataset. As a reason not to adopt AI, 10.15 percent of companies in this band name a lack of expertise, 6.87 percent privacy, 5.54 percent unclear legal consequences and 2.61 percent cost. Knowledge weighs almost four times as much as money.

What should a director do with these figures?

A director uses AI adoption figures to establish a position, not to justify a purchase. CBS measures use, not return. That 78 percent of ICT companies use AI doesn't prove it pays there. That 29 percent do in transport doesn't prove it doesn't.

Three questions give you a position worth having:

  1. Is your company among the users in your sector and size class, and how large is that group? Take the CBS column, not the national average.
  2. How many processes run with AI, rather than how many staff hold a licence? Count workflows with an owner and a KPI. Measuring AI results explains how.
  3. Where does adoption stall: knowledge, privacy or legal uncertainty?

If you don't use AI and have no process with a provable bottleneck, the benchmark is no reason to buy anything. If you do use it, but with loose licences and no policy, the next step isn't another tool but implementing one process with a baseline. Hiring independent AI consulting mainly pays off on the third question, because a lack of expertise is the most cited reason for dropping out in this band. Within AI consulting and business automation, that measurement is where we start.

Service

Learn more about AI consulting?

View service →

Curious how much time you could save?

Request a free efficiency audit. We'll analyze your processes and show you where the gains are, no strings attached.