Eurostat’s 2025 business digitalisation data show a wide spread in the share of enterprises using at least three artificial intelligence technologies. Among 31 reporting countries with comparable observations, Denmark records the highest share at 22.96%, followed by Finland at 21.81%. They are the only two observations above 20%. Sweden stands at 16.97%, Belgium at 15.66% and Norway at 15.17%. At the other end, Montenegro is at 2.21%, Poland at 2.40%, Romania at 3.31%, Türkiye at 3.63% and Serbia at 3.64%.
The unweighted mean of the 31 country rates is 8.75%, while the median is 6.87%. The mean is about 1.88 percentage points higher than the median, indicating that the largest observations pull the average upward. The first quartile is 5.00% and the third quartile is 11.52%, so the middle half of the observations lies roughly between those two values. The gap from the maximum to the minimum is 20.75 percentage points. Reading the median, quartiles and bands alongside the ranking gives a clearer picture than relying on the average alone.
Table of Contents
This is a narrower measure than general AI use
E_AI_TGE3 does not measure whether an enterprise uses any AI technology at all. An enterprise must use at least three of the artificial intelligence technologies listed in the Eurostat survey to enter the numerator. A business that uses one or two AI technologies can therefore be an AI user while remaining outside this indicator. Denmark’s 22.96% should not be read as saying that only 22.96% of enterprises use AI. It says that 22.96% of enterprises in the statistical universe meet the higher threshold of using at least three listed technologies.
The enterprise-size condition matters as well. GE10 covers enterprises with ten or more persons employed, and PC_ENT expresses the result as a percentage of enterprises. A very large company still counts as one enterprise, just like a company slightly above the ten-person threshold. Micro-enterprises below that threshold are outside this size class. The indicator also does not measure AI expenditure, the number of workers using AI, model calls, project counts, productivity gains or financial returns. It is best treated as a measure of the breadth of AI-technology adoption within the defined business population.
Denmark and Finland exceed 20%, with several Nordic countries near the top
Denmark at 22.96% and Finland at 21.81% lead the comparison. Sweden follows at 16.97%, Belgium at 15.66% and Norway at 15.17%. The Netherlands reaches 14.83%, Luxembourg 13.71%, Austria 12.85% and Germany 10.19%; these nine are the only observations at or above 10%. The simple average of Denmark, Finland, Norway and Sweden is 19.23%, which is 10.48 percentage points above the 31-country mean. This is an unweighted comparison of four national percentages, not a regional estimate weighted by the number of enterprises.
The three Benelux observations also form a relatively high cluster: Belgium 15.66%, the Netherlands 14.83% and Luxembourg 13.71%, for a simple average of 14.73%. The Baltic states average 7.26%, with Lithuania at 8.84%, Estonia at 7.27% and Latvia at 5.66%. Geographic clusters can be useful descriptive patterns, but this cross-section alone cannot establish that policy, sector mix, labour supply or digital infrastructure caused the differences. The evidence supports describing where the rates are higher or lower, not assigning a causal explanation.
The mean is above the median because the upper tail is comparatively strong
The mean is 8.75% and the median is 6.87%. Croatia, at 6.87%, is the sixteenth observation once all 31 values are ordered and therefore matches the median. The 1.88-point difference between mean and median reflects the influence of observations such as Denmark and Finland above 20%. The first quartile at 5.00% and third quartile at 11.52% show that the central half of reporting countries is concentrated between about five and eleven and a half percent, even though the full range extends much farther.
By broad bands, 2 countries are at 20% or more, 3 are from 15% to below 20%, 4 are from 10% to below 15%, 14 are from 5% to below 10%, and 8 are below 5%. Only 9 of the 31 observations reach 10% and only 5 reach 15%. The top eight countries average 16.74%, while the bottom eight average 3.58%, a ratio of about 4.7 to one. These bands make it clear that the 8.75% mean does not describe a uniformly distributed set of national rates.
Many countries in the middle are separated by only small percentage-point gaps
Spain records 9.02%, Lithuania 8.84%, Slovakia 7.98%, Malta 7.69%, Czechia 7.66%, Estonia 7.27%, Croatia 6.87%, Slovenia 6.78%, France 6.54%, Ireland 6.47% and Italy 6.46%. Ireland and Italy differ by only 0.01 percentage point, while Malta and Czechia differ by 0.03 point. A movement of one or two ranking positions in this part of the table can therefore correspond to a very small underlying difference. Percentage-point distance is more informative than rank alone when observations are tightly clustered.
Germany, France, Italy and Spain have a simple average of 8.05%, about 0.70 percentage points below the overall mean. That figure should not be treated as a combined rate for all enterprises in those four economies because each national percentage receives equal weight. To estimate an enterprise-weighted regional share, the number of eligible enterprises in each country would also be required. The value of the present comparison is that it holds the indicator definition, year, size class, sector aggregate and unit constant across the reporting countries.
Low values do not mean that enterprises make no use of AI
Montenegro at 2.21%, Poland at 2.40%, Romania at 3.31%, Türkiye at 3.63%, Serbia at 3.64%, Bulgaria at 3.96%, Hungary at 4.69% and Cyprus at 4.79% are below 5%. These percentages should not be expanded into a claim that AI use is nearly absent. Enterprises using only one or two listed AI technologies can still be active AI users while falling outside E_AI_TGE3. A 3% rate in this indicator does not imply that the remaining 97% use no artificial intelligence.
The threshold of at least three technologies provides information about adoption breadth, but it does not identify the exact technology combinations. Two enterprises may both qualify while using completely different sets of tools, and identical combinations can also differ greatly in frequency, organisational importance and scale. The single country-level percentage does not isolate differences by industry or enterprise size beyond the GE10 cut-off. Explaining why countries differ would require additional breakdowns, time series and evidence on investment, skills, sector composition and other conditions.
All 31 comparable observations for 2025
The table orders every observation from highest to lowest while keeping the year, enterprise-size class, sector aggregate, indicator and unit constant. The rank is a navigation aid, not a composite league table of AI competitiveness, innovation, productivity or economic performance. Each percentage is the share of enterprises with ten or more persons employed that use at least three of the AI technologies covered by the Eurostat indicator.
| Rank | Country | Share of enterprises |
|---|---|---|
| 1 | Denmark | 22.96% |
| 2 | Finland | 21.81% |
| 3 | Sweden | 16.97% |
| 4 | Belgium | 15.66% |
| 5 | Norway | 15.17% |
| 6 | Netherlands | 14.83% |
| 7 | Luxembourg | 13.71% |
| 8 | Austria | 12.85% |
| 9 | Germany | 10.19% |
| 10 | Spain | 9.02% |
| 11 | Lithuania | 8.84% |
| 12 | Slovakia | 7.98% |
| 13 | Malta | 7.69% |
| 14 | Czechia | 7.66% |
| 15 | Estonia | 7.27% |
| 16 | Croatia | 6.87% |
| 17 | Slovenia | 6.78% |
| 18 | France | 6.54% |
| 19 | Ireland | 6.47% |
| 20 | Italy | 6.46% |
| 21 | Bosnia and Herzegovina | 5.98% |
| 22 | Latvia | 5.66% |
| 23 | Portugal | 5.20% |
| 24 | Cyprus | 4.79% |
| 25 | Hungary | 4.69% |
| 26 | Bulgaria | 3.96% |
| 27 | Serbia | 3.64% |
| 28 | Türkiye | 3.63% |
| 29 | Romania | 3.31% |
| 30 | Poland | 2.40% |
| 31 | Montenegro | 2.21% |
What this comparison can and cannot establish
The data can directly describe the national level and distribution of the 2025 share of enterprises using at least three AI technologies. Denmark and Finland exceed 20%, the median is 6.87%, and the middle half of observations lies between roughly 5.00% and 11.52%. The data also show a substantial distance between the upper and lower groups and very small gaps among several middle-ranked countries. These are descriptive findings that can be reproduced from the 31 values.
The same evidence cannot establish that a higher rate causes better productivity, profitability or AI quality. Counting technologies is not the same as measuring how deeply they are embedded in operations or whether their use produces successful outcomes. A single year cannot reveal growth rates, and the country-level total cannot show which industries or business sizes drive the differences. For that reason, the interpretation here remains deliberately narrow: it compares the prevalence of meeting the ‘at least three AI technologies’ threshold under one consistent Eurostat definition.
Source and indicator definition
The source is Eurostat isoc_eb_ai annual data for 2025. The indicator is E_AI_TGE3, the enterprise-size class is GE10, the sector aggregate is C10-S951_X_K and the unit is PC_ENT. The official statistical definition covers the percentage of enterprises with ten or more persons employed using at least three listed artificial intelligence technologies. The ranking, mean, median, quartiles and band counts in this article are calculated directly from the 31 observations under that common definition.
Frequently Asked Questions
Which country had the highest share of enterprises using at least three AI technologies in 2025?
Denmark had the highest share among the 31 reporting countries at 22.96%, followed by Finland at 21.81%. They were the only two observations above 20%.
Is E_AI_TGE3 the same as the overall share of enterprises using AI?
No. It covers enterprises with at least ten persons employed that use at least three listed AI technologies. Enterprises using only one or two can still be AI users while remaining outside this indicator.
What are the mean and median across the 31 countries?
The unweighted mean is 8.75% and the median is 6.87%. The higher mean reflects the influence of the largest observations.
Is the 8.75% mean an enterprise-weighted average for Europe?
No. It is the simple average of 31 national percentages. An enterprise-weighted European rate would require the number of eligible enterprises in each country.
Related Articles
Green Map creates custom-edited map images using open geographic data sources such as geoBoundaries, Natural Earth, OpenStreetMap, and government open data. These maps are edited visual materials, not raw data files, and are provided for education, documents, presentations, and graphic reference.
