While EU AI adoption rose from 8% to 20% by 2025, a new JRC report reveals deep integration lags, hindered by persistent skill gaps and regulatory uncertainty.
AI adoption among EU firms is growing, but uptake remains uneven across sectors, while skills shortages and regulatory uncertainty continue to pose barriers. A recent European Commission Joint Research Centre (EC JRC) report examines these patterns across the strategic sectors covered by the Apply AI Strategy.
Alongside the AI Act and the AI Continent Action Plan, the Apply AI Strategy – adopted in October 2025 – is crucial to the EU’s approach to AI policy-making. It aims to increase competitiveness and innovation across strategic sectors, as well as strengthen the EU’s technological sovereignty. Accurately measuring AI adoption is therefore essential for assessing whether these policy instruments are effectively translating into real-world economic impact.
Between 2021 and 2025, the share of firms using at least one AI technology increased from 8% to 20%. However, this growth remains unevenly distributed: adoption in media, energy, and electronic communications has reached at least 30%, compared to a lagging 13% in mobility, transport and automotive, tourism, and agri-food.
Beyond adoption rates, the report highlights a distinction between the use of readily accessible AI tools and deeper integration into core operational processes. Much of the uptake centres on off-the-shelf applications, rather than uses such as workflow automation or advanced data analytics. The most widely adopted technology across sectors is the analysis of written language, followed by the generation of pictures, videos, and audio, and the production of written or spoken language and programming code.
And while the number of firms relying on commercial AI systems and on open-source software has risen in most sectors, defence and energy stand out as exceptions. Firms operating in the defence sector have increasingly developed AI internally, with reliance on commercial software having declined. Similarly, the energy sector has seen a sharp drop in the use of systems created or modified by external providers, alongside growth in the adoption of commercial systems and in the development of in-house AI solutions. This trend reflects the heightened security sensitivity of these critical infrastructure sectors.
The report also examines the barriers preventing firms from adopting AI. A lack of relevant expertise is the most frequently reported hurdle across strategic sectors, except in energy, where data protection and privacy concerns rank highest. Overall, specialised skills shortages and regulatory uncertainty emerge as central bottlenecks. Concerns about data protection, privacy and the legal consequences of AI deployment intensified between 2023 and 2025, particularly in energy, pharmaceuticals and defence.
Digital maturity is another key factor: more digitally mature sectors show both higher uptake and faster growth, with highly digitalised firms recording the steepest increases in AI adoption. For sectors such as agri-food and tourism, translating AI adoption into productivity gains may therefore require broader organisational changes alongside investment in the technology itself.
Ultimately, the findings highlight that wider access to AI tools alone is insufficient to ensure their effective deployment. The report calls for measures to address persistent skills shortages, clarify regulatory frameworks and strengthen firms’ digital foundations. For the European research and innovation community, these insights underline the pressing need for specialised expertise and enhanced capacity to translate AI capabilities into practical applications. Together, these enabling conditions could support deeper, production-oriented AI integration and contribute to the EU’s central objectives, such as the twin transition, technological autonomy and sovereignty.