European companies using artificial intelligence intensively have a 4% higher probability of hiring compared to others. This data, from a European Central Bank survey of 5,000 companies in 2025, contradicts the catastrophic predictions that have dominated the debate. The impact of AI on employment depends less on the technology itself than on the strategy of company leaders.
AI-Intensive Companies Are Hiring More
The ECB survey reveals an apparent paradox: companies that deploy artificial intelligence on a large scale create more jobs than those that ignore it. This positive correlation is observed across all sectors studied, from finance to services to manufacturing.
The 5,000 companies analyzed represent a representative sample of the European economy. Among those classified as “intensive users” of AI, 34% carried out net hiring in 2025, compared to 30% for companies with low technology use. This 4 percentage point difference is maintained even after adjusting for company size, sector, and financial situation.
Job creation is concentrated primarily in three areas: technical positions related to the deployment and maintenance of AI systems, quality control and human supervision functions, and new roles in analyzing generated data. The companies surveyed also report increased training needs, creating temporary but recurring jobs.
Job Destruction Remains Limited and Targeted
Contrary to alarmist projections, positions directly attributable to AI losses remain marginal in Europe. Only 12% of AI-using companies report having reduced their workforce due to automation, according to the ECB study.
The sectors most affected by job losses are basic accounting, data entry, and certain repetitive customer service tasks. But even in these areas, companies often prefer internal retraining rather than layoffs. 68% of employers who automated administrative positions offered training to redirect their employees toward other functions.
This restraint is explained by the cost of recruitment and the importance of human capital in the European economy. Training an existing employee costs an average of 3,000 euros according to Eurostat data, compared to 8,000 to 15,000 euros for recruiting and integrating a new colleague. The economic tradeoff therefore favors retraining rather than replacement.
Two Opposing Strategies Shape Social Impact
The effect of AI on employment fundamentally depends on the approach adopted by company leaders. The ECB survey identifies two distinct models: companies focused on innovation and those centered on cost reduction.
The first group, representing 62% of intensive AI users, deploys technology to develop new products or services. These companies show an average revenue growth of 8.3% in 2025, compared to 4.1% for the European average. This expansion mechanically generates a need for additional workforce, explaining the positive correlation between AI and hiring.
The second group, 38% of intensive AI users, uses AI primarily to automate existing tasks and reduce operating costs. These companies record average productivity gains of 15%, but their net hiring remains stable or slightly negative. They partially offset job losses by recruiting for new technological functions.
This strategic duality explains why Europe imposes its AI rules with caution, recognizing that social impact depends more on management intent than on the technology itself.
European SMEs Resist Massive Adoption
Intensive use of AI remains concentrated in large European companies. Only 23% of SMEs with fewer than 250 employees are classified as intensive users, compared to 67% of companies with more than 1,000 employees according to the ECB.
This disparity is explained by three main factors. Implementation cost represents a major obstacle: integrating a complete AI system requires an initial investment of 50,000 to 200,000 euros depending on complexity, an amount prohibitive for many SMEs. The technical skills required constitute the second barrier: 78% of SMEs surveyed report lacking internal expertise to evaluate and deploy AI solutions.
The third obstacle concerns regulation. New European AI rules impose transparency and compliance requirements that are particularly burdensome for small structures. 34% of SMEs consider these requirements a barrier to adoption, preferring to postpone their technology investments.
This situation creates a paradox: companies most likely to engage in mass layoffs (large corporations) are also those creating the most AI-related jobs. Conversely, SMEs that employ 60% of European workers adopt the technology more slowly, limiting the overall impact on employment in the short term.
Job Qualification Progressively Increases
Artificial intelligence transforms the qualitative structure of European employment more than it reduces its quantity. 76% of new positions created in AI-intensive companies require higher education or specialized training, compared to 45% across the entire economy.
This rise in qualification levels concerns three main categories. Direct technical professions (AI engineers, data scientists, machine learning specialists) represent 28% of job creation. Supervision and quality control functions for automated systems account for 31%. Positions at the interface between humans and machines (training, user support, AI ethics) constitute the remaining 41%.
The qualitative evolution poses challenges for low-skilled workers. While AI does not massively destroy employment in volume, it renders certain skills obsolete. European companies are investing 1.2 billion euros in retraining programs according to the ECB study, but this sum remains insufficient given identified needs.
Continuing education becomes a strategic issue. Nordic countries, which dedicate 2.8% of their GDP to professional training, show higher rates of adaptation to AI than the European average. This correlation suggests that public investment in skills largely determines an economy’s ability to benefit from AI without social exclusion.
Europe Faces the Challenge of Regulatory Uniformity
The nuanced results of the ECB survey question Europe’s regulatory approach. The AI Act, which entered into force in 2024, applies uniform rules to all uses of artificial intelligence without distinguishing business strategies or differentiated social impacts.
This uniformity is problematic given the diversity of business models observed. Innovative companies, which massively create jobs through AI, face the same regulatory constraints as those that automate to reduce costs. The European legal framework does not explicitly encourage job-creating AI uses.
Several member states are experimenting with differentiated approaches. Germany is testing tax incentives for companies that maintain their workforce when deploying AI. France is developing a “responsible AI” label that values job-creating practices. These national initiatives reveal the limitations of purely technical regulation.
The European Commission is studying a revision of the AI Act to integrate social impact criteria. This evolution could distinguish “augmentative” AI from “substitutive” AI, applying looser rules to the former. But this legal complexity risks further burdening the regulatory framework, already criticized by European companies.
The issue goes beyond pure technology. Europe must arbitrate between social protection and economic competitiveness, in a context where the United States and China deploy AI without equivalent constraints. ECB data suggests that intelligent regulation could reconcile these objectives by encouraging job-creating uses while regulating destructive practices. But this reconciliation requires analytical refinement that current legal instruments do not yet permit.
Sources: