The OECD’s AI Adoption Numbers Show a Second Gap
Large firms are more likely to use AI. A new OECD report shows how far that lead extends when cloud computing and data analytics are counted alongside it.
Editorial visualizationAbout 12% of enterprises in the OECD’s comparison used AI, cloud computing and data analytics together. Among large firms, the share was 44%. The OECD’s Digital Economy and Society Outlook, released November 5, adds this second measure to the familiar headline about AI adoption. It shows how much more common it is for large businesses to have all three tools in use at once. Read the report’s analysis of combined adoption.
The combined measure covers businesses with at least 10 employees and uses 2025 figures where available. Several countries contributed older observations. It counts a business that reports using all three technologies, so the figures cannot tell whether those tools share data or even sit in the same department. See the notes for Figure 3.4.
What the size breakdown shows
The basic AI gap was already visible in January’s OECD release. The November outlook’s annex puts reported AI use at 52.1% for large firms, compared with 17.5% for small firms. The 20.3% headline rate includes medium and large firms alongside small ones. OECD statistical annex, Table 1
| Enterprise group | Share reporting AI use |
|---|---|
| Small | 17.5% |
| Medium | 29.5% |
| Large | 52.1% |
| All covered sizes | 20.3% |
The EU figures show a similar divide within their own coverage. In 2025, 55% of large EU businesses used AI, compared with 19% of small and medium businesses combined. Eurostat’s grouping differs from the small-firm row above, and the OECD draws in part on European survey data. Eurostat’s 2026 digitalisation release
Editorial visualizationWhere tools meet the work
Take a wholesaler that wants to flag stock shortages. Cloud software could hold the order records, analytics could show which items repeatedly run low, and an AI tool could help turn the findings into a draft for staff to review. That is one possible workflow. The survey’s adoption figures say little about whether a particular firm has connected those steps or improved its stock decisions.
Editorial visualizationThe OECD identifies skills, privacy concerns, implementation costs and difficulties with existing systems as barriers. Its chapter also notes that a lack of in-house expertise was the most frequently cited obstacle among EU businesses considering AI. These are reported barriers, so the survey leaves open how much each one contributes to the gap between large and small firms. OECD discussion of adoption barriers
A small firm judging an AI pilot can track a narrower outcome than adoption. For the stockroom example, that means checking how often a draft alert identifies a real shortage and how long staff spend correcting it. Those results tell the firm whether the tool earns a place in that workflow.
Publication history
— First published.
No published corrections or earlier versions.
How this was made
UpThat uses AI-assisted research and writing. Source checking and publication follow our editorial review workflow. See the original documents above to assess the reporting.
How we use AI ↗
