Share of enterprises using at least one AI technology. Focus: EU average, survey year 2025.
Europe.
Up to speed. AI adoption across Europe
Discover where you stand.
Explore the data and compare your country.
CountryEU average
Year2025
THE OBSERVATORY
Eurostat · Epoch AI · EuroHPCFour perspectives. One Europe.
AT A GLANCE / 2025
Three things to know
- In 2025, 20% of enterprises with 10 or more employees in the EU used at least one AI technology. In the 2024 survey, the share was 13%; because the question changed, the comparison is only partly valid.
- Member States range from 5% (Romania) to 42% (Denmark).
- Large enterprises (250+ employees) use AI 3.2 times as often as small ones (10–49): 55% compared with 17%.
Source: Eurostat, isoc_eb_ai, last updated June 15, 2026 · Why other figures differ
ANALYSIS / 2025
EU average · 2025AI statistics. With context.
Key figures for your selected country and year, with the relevant comparison group. This analysis follows your selection on the chronograph. Compare enterprise sizes →
Published country figures in 2025 range from Romania to Denmark. This does not establish a ranking of overall AI capabilities.
The surveys are only partly comparable. We therefore do not calculate a pace for this period.
OVER TIME
Each survey is a snapshot.
The bars show published shares for each survey year. Years without a survey are left blank.
EU average
2021
EU average 7.7 %
2023
EU average 8.1 %
2024
EU average 13.5 %
2025*
EU average 20.0 %
* Break relative to an earlier survey. From 2025 the survey includes additional AI technologies. Exact values and flags are in the data table.
Source: Eurostat, isoc_eb_ai, last updated June 15, 2026
Enterprises with 10 or more persons employed, all activities except agriculture, mining and the financial sector
Why AI statistics can differ.
Who was counted?
The data cover enterprises with 10 or more persons employed. Surveys of microbusinesses or individual workers cover different groups. Comparisons also need to account for the economic activities covered.
What was asked?
Use of at least one AI technology, generative AI and AI use at work are different indicators. Internet and cloud adoption remain separate series here.
Survey or forecast?
An AI adoption survey is not a forecast of the AI market. The publication date is not necessarily the survey year. Comparisons require the same measurement basis.
Detailed figures and sources
| Year | EU average | Comparability notes |
|---|---|---|
| 2021 | 7.65 % | Published observation |
| 2023 | 8.06 % | Published observation |
| 2024 | 13.48 % | Published observation |
| 2025 | 19.95 % | Break in series |
Source: Eurostat, isoc_eb_ai, last updated June 15, 2026
Enterprises with 10 or more persons employed, all activities except agriculture, mining and the financial sector
Data retrieved: November 7, 2026. Missing values are shown as a dash and are not estimated.
Cite the figure and its source
Eurostat: EU average, 2025: 19.95% of the enterprises covered used at least one AI technology. Dataset isoc_eb_ai, last updated June 15, 2026, retrieved November 7, 2026. Presentation: AIpace.eu.
Enterprises with 10 or more persons employed, all activities except agriculture, mining and the financial sector · When reusing a figure, include the year, population and primary source.
Open the Eurostat datasetBY ENTERPRISE SIZE / 2025
EU average · 2025Three sizes. Three perspectives.
One country figure does not tell the whole story. Compare AI adoption in small, medium and large enterprises — for the same year and using the same definition.
EU average
0117.0 %
Small enterprises
10–49 persons employed
0230.4 %
Medium-sized enterprises
50–249 persons employed
0355.0 %
Large enterprises
250+ persons employed
From 2025, the survey also covers AI that generates images, video or audio. Results are therefore only partly comparable with earlier years.
Flags for each size class are shown in the table. Missing values are not estimated.
All size classes as a table
| Size class | EU average | Data & methodology |
|---|---|---|
| Small enterprises | 17.0 % | Source: Eurostat, isoc_eb_ai, last updated June 15, 2026 Enterprises with 10–49 persons employed, covered economic activities excluding agriculture, mining and finance size_emp=10-49 |
| Medium-sized enterprises | 30.4 % | Source: Eurostat, isoc_eb_ai, last updated June 15, 2026 Enterprises with 50–249 persons employed, covered economic activities excluding agriculture, mining and finance size_emp=50-249 |
| Large enterprises | 55.0 % | Source: Eurostat, isoc_eb_ai, last updated June 15, 2026 Enterprises with 250+ persons employed, covered economic activities excluding agriculture, mining and finance size_emp=GE250 |
BY SECTOR / 2025
AI use by sector
Share of enterprises with 10 or more employees using at least one AI technology, by economic activity, survey 2025. Ordered by activity, not by value.
Manufacturing
EU27 17 %
Energy, water and waste
EU27 18 %
Construction
EU27 11 %
Trade and motor vehicle repair
EU27 19 %
Transport and storage
EU27 11 %
Accommodation and food services
EU27 12 %
Information and communication
EU27 63 %
Real estate
EU27 25 %
Professional, scientific and technical services
EU27 40 %
Administrative and support services
EU27 20 %
Eurostat does not publish some country values, for example for confidentiality; they appear as “no data”. Each sector is its own population.
Source: Eurostat, isoc_eb_ain2, last updated June 15, 2026
Enterprises with 10 or more employees in each economic activity (NACE Rev. 2)
USE IN DETAIL / 2025
What with, what for – and what holds back
Which AI technologies enterprises use, what they use them for and why others do not, survey 2025. Each column has its own population.
What with: AI technologies
% of all enterprises with 10 or more employees
Analysing text (text mining)
EU27 12 %
Generating images, video or sound
EU27 10 %
Generating text, speech or code
EU27 9 %
Speech recognition
EU27 7 %
Show 4 more
Automating workflows, supporting decisions
EU27 5 %
Machine learning for data analysis
EU27 5 %
Image recognition
EU27 4 %
Autonomous movement of machines
EU27 1 %
What for: areas of use
% of enterprises using AI
Marketing or sales
EU27 35 %
Administration and management
EU27 31 %
Accounting, controlling, finance
EU27 23 %
Production
EU27 21 %
Show 3 more
ICT security
EU27 20 %
Research, development, innovation
EU27 19 %
Logistics
EU27 6 %
What holds back: reasons against AI
% of enterprises not using AI
Of the enterprises not using AI, 14 % had ever considered using it (EU27).
Lack of expertise
EU27 10 %
Unclear legal consequences
EU27 8 %
Data protection and privacy concerns
EU27 7 %
Required data missing or of low quality
EU27 6 %
Show 4 more
Incompatible with existing systems
EU27 6 %
Costs seem too high
EU27 5 %
Ethical considerations
EU27 3 %
Not useful for the enterprise
EU27 3 %
Multiple answers possible, so a column does not add up to 100%. Values unchanged from the Eurostat survey.
DIGITAL DECADE 2030
The EU's targets for 2030
The EU has set measurable digital targets for 2030. Each card shows the latest published value next to the target. The targets apply to the EU as a whole, not to individual countries.
AI, cloud or data analytics
Share of enterprises using AI, sophisticated or intermediate cloud services, or data analytics
- EU 2025
- 63 %
- EU trajectory for 2025
- 63 %
- 2030 target
- 75 %
Basic digital skills
Share of people with at least basic digital skills
- EU 2025
- 60 %
- EU trajectory for 2025
- 66 %
- 2030 target
- 80 %
ICT specialists
Employed ICT specialists
- EU 2025
- 10.45 million
- EU trajectory for 2025
- 13.36 million
- 2030 target
- 20.0 million
Eurostat flag: estimated
Digitalisation of SMEs
Share of small and medium-sized enterprises using at least 4 of 12 digital technologies
- EU 2025
- 71 %
- EU trajectory for 2025
- 70 %
- 2030 target
- 90 %
Solid: published values. Dashed: the European Commission's official trajectory to 2030. No projection of our own.
Sources and populations
Source: Eurostat, isoc_eb_ai, last updated June 15, 2026
Enterprises with 10 or more persons employed, all activities except agriculture, mining and the financial sector
Source: Eurostat, isoc_sk_dskl_i21, last updated April 17, 2026
People aged 16 to 74
Source: Eurostat, isoc_sks_itspt, last updated April 17, 2026
Persons in employment according to the Labour Force Survey
Source: European Commission, State of the Digital Decade 2026, trajectories (EUPL)
Targets: Decision (EU) 2022/2481 establishing the Digital Decade Policy Programme 2030
WORLDWIDE
Europe and the world
How widely is AI used by enterprises outside the EU? Each country is shown with the latest figure published by the OECD and its survey year. The data cover enterprises with 10 or more persons employed.
Comparability is limited: questionnaires, reference periods, economic activities covered and survey years may differ.
European survey using Eurostat methodology
- EU average202520.0 %
- Norway202528.9 %
- Türkiye20257.4 %
National surveys
- Australia20223.4 %
- Brazil202412.9 %
- Canada20238.3 %
- Japan2023 · definition differs9.2 %
- New Zealand20229.0 %
- South Korea202432.9 %
- Switzerland2023 · definition differs10.0 %
- United States2021 · break in series5.7 %
Bars: 0 to 40%. Alphabetical within each group.
All values as a table
| COUNTRY | 2021 | 2022 | 2023 | 2024 | 2025 |
|---|---|---|---|---|---|
| EU average | 7.7 % (B) | — | 8.1 % | 13.5 % | 20.0 % |
| Norway | 10.8 % (B) | — | 9.2 % | 20.8 % | 28.9 % |
| Türkiye | 2.7 % (B) | — | 5.5 % | 4.4 % | 7.4 % |
| Australia | — | 3.4 % | — | — | — |
| Brazil | 12.9 % | — | 12.9 % | 12.9 % | — |
| Canada | 4.7 % | — | 8.3 % | — | — |
| Japan | 3.6 % (D) | — | 9.2 % (D) | — | — |
| New Zealand | — | 9.0 % | — | — | — |
| South Korea | 2.7 % | 28.0 % (B) | 30.3 % | 32.9 % | — |
| Switzerland | 9.6 % (D) | — | 10.0 % (D) | — | — |
| United States | 5.7 % (B) | — | — | — | — |
B = break in series, D = definition differs, E = estimated (according to the OECD). Values from 2021.
IN CONTEXT
Why AI figures differ
Depending on who is asked and what counts as using AI, the results differ. None of these official figures is wrong – they measure different things and cannot be combined.
- 20.0 %
Eurostat · enterprises in the EU
Survey 2025
- Who
- Enterprises with 10 or more employees, excluding agriculture, mining and the financial sector
- What counts
- At least one AI technology in use, e.g. text or image generation, speech recognition, machine-learning data analysis
- 32.7 %
Eurostat · people in the EU
Survey 2025
- Who
- People aged 16 to 74
- What counts
- Used a generative AI tool in the last 3 months, for private, work or education purposes
- 15.1 %
Eurostat · people, for work
Survey 2025
- Who
- People aged 16 to 74
- What counts
- Used a generative AI tool for work in the last 3 months
- 19.8 %
US Census Bureau · businesses in the United States
Period ending 3 May 2026
- Who
- All employer businesses except farms, including very small firms
- What counts
- Used AI in a business function in the last two weeks
Official statistics only, values unchanged from each source. Last checked November 8, 2026.
AI adoption across Europe
AI in enterprises · 2025: 20 % · 2:24
Eurostat · isoc_eb_ai · June 15, 2026
Enterprises with 10 or more persons employed, all activities except agriculture, mining and the financial sector
From 2025, the survey also covers AI that generates images, video or audio. Results are therefore only partly comparable with earlier years.
The clocks show saved annual Eurostat data, not live measurements. Missing values are never estimated. Percentages are mapped to a 12-hour dial.
The share in percent is shown on a 12-hour dial: 0% is 0:00, 100% is 12:00.
12:00 represents 100% adoption, not a policy target.
All countries at a glance
| COUNTRY | AI ADOPTION | AS A TIME OF DAY |
|---|---|---|
| Austria | 30 % | 3:36 |
| Belgium | 35 % | 4:09 |
| Bulgaria | 9 % | 1:02 |
| Croatia | 15 % | 1:49 |
| Cyprus | 9 % | 1:07 |
| Czechia | 18 % | 2:07 |
| Denmark | 42 % | 5:03 |
| Estonia | 23 % | 2:48 |
| Finland | 38 % | 4:32 |
| France | 18 % | 2:11 |
| Germany | 26 % | 3:07 |
| Greece | 9 % | 1:04 |
| Hungary | 10 % | 1:15 |
| Ireland | 20 % | 2:21 |
| Italy | 16 % | 1:58 |
| Latvia | 12 % | 1:28 |
| Lithuania | 21 % | 2:33 |
| Luxembourg | 34 % | 4:02 |
| Malta | 22 % | 2:35 |
| Netherlands | 33 % | 3:59 |
| Poland | 8 % | 1:00 |
| Portugal | 12 % | 1:23 |
| Romania | 5 % | 0:38 |
| Slovakia | 18 % | 2:10 |
| Slovenia | 22 % | 2:36 |
| Spain | 20 % | 2:26 |
| Sweden | 35 % | 4:12 |
Eurostat · isoc_eb_ai · June 15, 2026
Enterprises with 10 or more persons employed, all activities except agriculture, mining and the financial sector
WHAT’S NEW
Recently updated
- European Commission: Digital Decade trajectories
- EuroHPC: AI Factories and antennas
- OECD: AI in businesses worldwide
- LMArena: language model leaderboard
- Epoch AI: AI models
- Eurostat: AI by sector (isoc_eb_ain2)
- Eurostat: AI in enterprises (isoc_eb_ai)
- Eurostat: ICT specialists (isoc_sks_itspt)
Subscribe via RSS · Data date per source; new entries appear as soon as a source publishes new data.
FURTHER SOURCES
More perspectives on AI
These freely accessible resources cover further aspects of AI and complement the data on AIpace. Each uses its own methods.
- Stanford AI Index 2026Annual report on research, performance, investment, policy and public opinion, with open data.
- Stanford AI Economic IndicatorsMonthly indicators on employment, productivity and AI use, mainly for the United States.
- Artificial AnalysisIndependent tests of models by performance, speed and price.
- LMArenaModel ranking from head-to-head votes. Excerpt on the Models page.
- OECD.AIPolicies, strategies and indicators on AI across OECD countries.
Read the numbers.
What does the chronograph show?
The chronograph shows the share of enterprises using at least one AI technology. 0% corresponds to 0:00 and 100% to 12:00. The bold hand shows hours; the pale hand shows minutes. The thin orange seconds hand is decorative and does not indicate changes in the data. The clock measures AI adoption, not a country’s overall AI capabilities.
How is the pace calculated?
The change between two surveys is annualised: 1 percentage point equals 7.2 minutes on the AI clock. No pace is calculated across a break in the time series. In 2025 the list of surveyed AI technologies was expanded.
What can be compared across countries?
The table shows published AI adoption figures for all countries in the selected year. Countries are listed alphabetically by default, with an option to sort by value. Differences from the EU average are shown in percentage points. Internet and cloud use are separate indicators covering their own populations.