A national average leaves a lot out

More than half of surveyed businesses in Denmark's Midtjylland used at least one AI technology in 2025. Eurostat's 2026 regional yearbook puts the share at 50.8%. Across the EU, it was 20.0%, or about one business in five.

Midtjylland was the only available regional observation above 50%. Another Danish region, Nordjylland, recorded 32.1%. The gap is about 18.7 percentage points within the same country. A national adoption figure is useful, but it is not a reliable description of every local economy.

Which businesses, and what counts as AI?

These are enterprise survey results, not counts of individual chatbot users. The population covers businesses with at least ten employed people in the non-financial business economy. The smallest firms and the financial sector are outside this measure, so it cannot describe every employer or self-employed person.

AI is also broader here than ChatGPT. The survey includes technologies such as text analysis, speech recognition, generated text or code, machine learning and systems that automate decisions or movement. A business counts if it uses at least one covered technology. That does not tell us how often it is used, how many workers use it or whether it has improved the business.

The map is not a level playing field

Most observations are for NUTS 2 regions, a standard European statistical geography. Belgium is shown at the broader NUTS 1 level. Several countries, including Germany, France, Italy and Finland, are represented by national figures instead of regional ones.

That matters when comparing places. A countrywide observation can hide high- and low-adoption regions, while a smaller regional observation exposes them. Midtjylland's position among the available regional figures is a clear finding. It is not proof that every other local area in Europe has a lower rate.

Adoption is moving, but it is not an outcome

The EU share rose from 8.1% in 2023 to 20.0% in 2025. That is an increase of 11.9 percentage points on the rounded published figures. It describes a wider spread of use, not a measured gain in productivity, revenue or wages.

The geography raises useful questions about skills, infrastructure and the types of businesses located in each place. This dataset does not settle which of those factors caused the gaps. And the dates matter: the chapter belongs to the 2026 yearbook, but the AI observations are from 2025. It is a baseline for understanding where adoption has spread, not a live reading of the market in September 2026.

Sources

  1. Eurostat: Digital society statistics at regional levelPrimary regional yearbook chapter, AI findings, survey coverage, reference year and geographic comparability notes.
  2. Eurostat: EU enterprise AI dataset, 2025Direct primary data cross-check: 19.95%, which rounds to 20.0%; equivalent 2023 query returns 8.06%, rounding to 8.1%.
  3. Eurostat: Midtjylland regional AI dataset, 2025Direct primary data cross-check: 50.82%, rounding to 50.8%; changing geo to DK05 returns Nordjylland at 32.13%, rounding to 32.1%.