Six findings
- More than half of large companies use AI. In the European Union 55% of companies with 250 employees or more used at least one AI technology in 2025, up from 41% in 2024 and 30% in 2023.
- Small companies are far behind. The share is 17% among companies with 10 to 49 employees and 30% among those with 50 to 249.
- The north leads. Finland is first with 79% of large companies, followed by Belgium with 76% and Denmark with 75%. Romania is last with 21%, after Bulgaria with 26%. Large companies are above 50% in 18 of the 28 countries.
- France and Italy caught up in one year. France went from 33% to 58% and Italy from 33% to 53%.
- The gap between large and small differs by country. In Poland large companies use AI 7.9 times as often as small ones. In Denmark the ratio is 2.0.
- Skills and law hold large companies back more than cost. Among large companies, 10.3% do not use AI for lack of expertise, 9.3% over data protection and 8.1% because the legal consequences are unclear. Cost is named by 5.5%.
Large against small
Eurostat asks companies every year whether they use AI. The answer depends on size more than on anything else. In every one of the 28 countries large companies are ahead of small ones, and the European Union average is 55% against 17%.
Figure 1
Large companies use AI three times as often as small ones
Share of companies using at least one AI technology in 2025, by size. Countries are ordered by the share among large companies.
Large companies, 250 employees or more
Small companies, 10 to 49 employees
European Union average
Source: Solorai analysis of Eurostat data on the use of AI in enterprises, survey year 2025.
Data table: 28 countries
| Country | Small, 10 to 49 | Medium, 50 to 249 | Large, 250 or more | Large, 2024 | Large, 2023 |
|---|---|---|---|---|---|
| European Union | 17.0% | 30.4% | 55.0% | 41.2% | 30.5% |
| Finland | 33.5% | 51.4% | 79.4% | 70.4% | 53.3% |
| Belgium | 28.8% | 54.5% | 76.4% | 66.3% | 47.9% |
| Denmark | 37.5% | 58.3% | 74.5% | 63.4% | 51.4% |
| Sweden | 30.7% | 49.6% | 71.9% | 56.3% | 37.8% |
| Slovenia | 18.3% | 30.6% | 71.7% | 59.7% | 53.2% |
| Austria | 26.2% | 44.6% | 68.3% | 49.9% | 35.2% |
| Netherlands | 28.8% | 46.6% | 67.6% | 54.1% | 41.6% |
| Norway | 25.2% | 45.3% | 66.0% | 53.3% | 34.4% |
| France | 15.0% | 30.8% | 58.0% | 32.7% | 20.9% |
| Luxembourg | 30.5% | 43.7% | 58.0% | 45.6% | 41.8% |
| Spain | 17.2% | 30.1% | 57.5% | 44.0% | 39.7% |
| Ireland | 16.8% | 28.0% | 57.3% | 50.8% | 36.3% |
| Germany | 23.1% | 35.6% | 57.0% | 48.2% | 35.4% |
| Czechia | 13.4% | 28.8% | 54.1% | 40.5% | 28.3% |
| Italy | 14.2% | 27.6% | 53.1% | 32.5% | 24.1% |
| Estonia | 20.7% | 33.1% | 52.9% | 39.0% | 23.0% |
| Lithuania | 18.1% | 30.4% | 51.8% | 31.2% | 21.3% |
| Malta | 18.0% | 30.7% | 51.8% | 46.7% | 32.5% |
| Portugal | 9.4% | 18.2% | 49.1% | 41.9% | 35.4% |
| Latvia | 9.8% | 19.8% | 47.1% | 33.3% | 21.3% |
| Poland | 5.8% | 14.8% | 45.8% | 33.0% | 24.4% |
| Slovakia | 15.6% | 23.1% | 43.7% | 29.1% | 21.9% |
| Hungary | 8.7% | 15.3% | 40.0% | 23.5% | 17.4% |
| Croatia | 12.8% | 25.5% | 39.6% | 28.4% | 19.3% |
| Greece | 7.3% | 15.2% | 37.6% | 24.3% | 14.0% |
| Cyprus | 7.7% | 15.3% | 35.1% | 34.9% | 15.5% |
| Bulgaria | 7.2% | 13.3% | 26.2% | 20.2% | 13.8% |
| Romania | 4.1% | 7.8% | 20.8% | 11.3% | 8.1% |
One year of change
Adoption among large companies rose in all 28 countries between 2024 and 2025. The European Union average rose by 13.9 points. The largest gains are in countries that started low.
Figure 2
France, Italy and Lithuania gained more than 20 points in a year
Change from 2024 to 2025 in the share of large companies using AI, in percentage points.
| France | 25.3 |
|---|---|
| Italy | 20.6 |
| Lithuania | 20.5 |
| Austria | 18.3 |
| Hungary | 16.5 |
| Sweden | 15.5 |
| Slovakia | 14.6 |
| Estonia | 13.9 |
| Latvia | 13.8 |
| Czechia | 13.6 |
| Netherlands | 13.6 |
| Spain | 13.5 |
| Greece | 13.3 |
| Poland | 12.9 |
| Norway | 12.7 |
| Luxembourg | 12.4 |
| Slovenia | 12.0 |
| Croatia | 11.2 |
| Denmark | 11.1 |
| Belgium | 10.1 |
| Romania | 9.5 |
| Finland | 9.0 |
| Germany | 8.8 |
| Portugal | 7.3 |
| Ireland | 6.4 |
| Bulgaria | 6.0 |
| Malta | 5.0 |
| Cyprus | 0.2 |
Source: Solorai analysis of Eurostat data on the use of AI in enterprises, survey year 2025.
Sector by sector
By sector Eurostat publishes figures for companies of all sizes together. Information and communication is at 63%, professional and scientific activities at 40%. Construction and transport are at 11%.
Figure 3
Information and communication is far ahead of every other sector
Share of companies using AI in 2025, European Union, companies with 10 employees or more.
| Information and communication | 63% |
|---|---|
| Professional, scientific and technical activities | 40% |
| Energy supply | 34% |
| Real estate | 25% |
| Administrative and support services | 20% |
| Wholesale and retail trade | 19% |
| Manufacturing | 17% |
| Water and waste | 13% |
| Accommodation and food services | 12% |
| Transport and storage | 11% |
| Construction | 11% |
Source: Solorai analysis of Eurostat data on the use of AI in enterprises, survey year 2025.
What large companies use AI for
Among large companies the most common purpose is ICT security, followed by administration and by marketing and sales. Small companies use AI mostly for marketing and sales.
Figure 4
Large companies use AI first for security and administration
Share of companies using AI for each purpose in 2025, European Union.
Large companies, 250 employees or more
Small companies, 10 to 49 employees
Source: Solorai analysis of Eurostat data on the use of AI in enterprises, survey year 2025.
What holds the others back
45% of large companies in the European Union did not use AI in 2025. The reasons they give are about people and rules more than about money.
Figure 5
What holds large companies back is skills and law, not cost
Share of large companies in the European Union that do not use AI and give each reason, 2025. More than one reason can be given.
| Lack of relevant expertise | 10.3% |
|---|---|
| Data protection and privacy concerns | 9.3% |
| Unclear legal consequences | 8.1% |
| Availability or quality of data | 6.9% |
| Incompatibility with existing systems | 6.0% |
| Costs seem too high | 5.5% |
| Ethical considerations | 3.4% |
| AI not useful for the enterprise | 1.6% |
Source: Solorai analysis of Eurostat data on the use of AI in enterprises, survey year 2025.
What it means for a company
- Not using AI is now the minority position among large companies. In 18 countries more than half of them use it. A company of that size that has not started is behind its peers.
- Plan for skills first. Lack of expertise is the first barrier. Training and a clear owner matter more than the choice of tool. Use the EU AI Act checklist.
- Settle the legal questions early. Data protection and unclear legal consequences come second and third. Check the risk level of one AI use.
Method and limits
- Source: Eurostat, survey on ICT usage and e-commerce in enterprises, datasets isoc_eb_ai and isoc_eb_ain2, survey year 2025, as updated on 15 June 2026.
- Coverage: the 27 countries of the European Union and Norway. The United Kingdom, Switzerland and countries outside Europe are not in the survey.
- A company uses AI when it uses at least one of the AI technologies listed in the survey, such as text mining, speech recognition, language or image generation, machine learning or workflow automation.
- Size classes: small is 10 to 49 persons employed, medium 50 to 249, large 250 or more. Companies with fewer than 10 are not covered.
- The survey excludes the financial sector, agriculture and mining.
- Sector figures are for companies of all sizes together. Eurostat does not publish them for large companies alone.
- Barriers are shares of all large companies, not of the non-users only.
- For France and Sweden, Eurostat flags a break in the series in 2023, so we compare 2025 with 2024.
Sources
Cite this research
Solorai (2026). AI in large European companies: who uses it, country by country. https://solorai.com/insights/research/ai-adoption-large-companies-europe-2026/
Free to quote and to chart, with attribution and a link to this page.