Across Europe, artificial intelligence is spreading through workplaces at a pace that has caught many economists off guard. Yet the productivity dividends that policymakers have been banking on remain stubbornly difficult to measure, creating a paradox that is now at the center of a growing debate among researchers, business leaders, and government officials. A major new analysis from the Centre for Economic Policy Research sheds fresh light on this tension, offering granular data on how AI is reshaping European labor markets—and where the technology’s promise has yet to materialize.
The research, published by VoxEU at the Centre for Economic Policy Research, draws on firm-level and worker-level data across multiple European economies to assess AI’s real-world impact on productivity and employment. The findings paint a nuanced picture: AI adoption is accelerating, particularly among larger firms and in knowledge-intensive sectors, but aggregate productivity statistics have not yet reflected the kind of transformational gains that Silicon Valley evangelists have long predicted.
Adoption Is Surging, But Unevenly
According to the CEPR analysis, AI adoption rates among European firms have risen sharply in recent years, with the technology now embedded in operations ranging from customer service and logistics to research and development. However, the distribution of adoption is far from uniform. Large firms with more than 250 employees are significantly more likely to have integrated AI tools than small and medium-sized enterprises, which still form the backbone of most European economies. This gap raises serious questions about whether AI will widen existing disparities between firms rather than lift productivity across the board.
The sectoral divide is equally pronounced. Industries such as information and communications technology, finance, and professional services have moved aggressively to deploy AI-powered systems. Meanwhile, sectors like construction, hospitality, and traditional manufacturing lag considerably behind. The researchers note that this pattern mirrors earlier waves of digital technology adoption, where gains concentrated in already-productive sectors while others struggled to keep pace. The implication is that without targeted policy intervention, AI could reinforce rather than reduce structural inequalities within European economies.
The Productivity Paradox Persists
Perhaps the most striking finding from the CEPR research is the persistent gap between firm-level AI adoption and measurable productivity improvements at the macroeconomic level. While individual firms that adopt AI report efficiency gains—particularly in automating routine tasks and improving decision-making processes—these micro-level improvements have not yet translated into the kind of broad-based productivity acceleration that Europe desperately needs. The continent’s productivity growth has been sluggish for over a decade, and many had hoped AI would provide the jolt required to close the gap with the United States.
This echoes the famous Solow Paradox of the 1980s, when the economist Robert Solow quipped that computers could be seen everywhere except in the productivity statistics. The CEPR researchers suggest several explanations for why the same dynamic may be playing out with AI. First, adoption is still in relatively early stages for many firms, and the organizational changes required to fully capitalize on AI—restructuring workflows, retraining workers, redesigning management practices—take years to implement. Second, measurement itself is a challenge: traditional productivity metrics may not fully capture the quality improvements and new capabilities that AI enables, such as better customer experiences or faster product iteration cycles.
Labor Market Effects: Displacement Fears Meet Complementarity
On the employment front, the CEPR analysis offers a more reassuring picture than many had feared. While AI is clearly automating certain tasks—particularly routine cognitive work such as data entry, basic analysis, and standardized reporting—there is limited evidence so far of large-scale job displacement across European labor markets. Instead, the data suggests that AI is primarily changing the composition of tasks within existing jobs rather than eliminating positions outright. Workers in AI-adopting firms report spending less time on repetitive activities and more time on tasks requiring judgment, creativity, and interpersonal skills.
That said, the distributional effects are real and significant. Workers with higher levels of education and digital skills are far better positioned to benefit from AI adoption, while those in lower-skilled roles face greater risk of task displacement. The research highlights that this dynamic is already contributing to wage polarization in several European countries, with AI-complementary workers seeing wage gains while those whose tasks are most susceptible to automation experience stagnation or decline. This finding aligns with broader concerns raised by organizations including the Organisation for Economic Co-operation and Development, which has warned that AI could exacerbate income inequality if left unaddressed by policy.
Europe’s Regulatory Approach: Asset or Liability?
The European Union’s approach to AI regulation—most notably the AI Act, which entered into force in 2024—adds another layer of complexity to the adoption picture. Proponents argue that clear regulatory frameworks give European firms and consumers confidence to engage with AI, ultimately supporting sustainable adoption. Critics counter that compliance costs and regulatory uncertainty may slow deployment, particularly among smaller firms that lack the resources to manage complex legal requirements. The CEPR research does not directly assess the AI Act’s impact, but the data on the adoption gap between large and small firms lends some credibility to concerns about regulatory burden falling disproportionately on SMEs.
Recent reporting from Reuters and the Financial Times has documented growing frustration among European tech executives who argue that the regulatory environment is putting the continent at a competitive disadvantage relative to the United States and China, where AI deployment is proceeding with fewer constraints. At the same time, European policymakers point to the risks of unregulated AI—from algorithmic bias to labor market disruption—as justification for a more cautious approach. The tension between innovation speed and regulatory prudence remains one of the defining policy debates on the continent.
What the Firm-Level Data Reveals About Organizational Readiness
One of the most valuable contributions of the CEPR analysis is its attention to organizational factors that mediate AI’s impact. The research finds that firms with higher levels of pre-existing digital maturity, stronger management practices, and greater investment in worker training see significantly larger productivity returns from AI adoption. This suggests that AI is not a plug-and-play solution; rather, its effectiveness depends heavily on complementary investments in human capital and organizational design.
This finding has direct implications for European industrial policy. Governments that focus solely on subsidizing AI technology purchases without addressing the broader organizational context may see disappointing returns. The CEPR researchers argue that training programs, management development initiatives, and support for organizational restructuring should be central components of any AI strategy. Several European countries, including Germany, France, and the Netherlands, have launched national AI strategies that include workforce development components, but the scale of investment remains modest relative to the challenge.
The Skills Gap Threatens to Widen
The workforce dimension of AI adoption may ultimately prove to be the most consequential. The CEPR data shows that demand for AI-related skills—including data science, machine learning engineering, and AI-literate management—is growing rapidly across European labor markets. Yet the supply of workers with these skills remains constrained, creating bottlenecks that slow adoption and drive up wages for in-demand talent. This dynamic is particularly acute in smaller European economies, where the talent pool is limited and competition with larger markets for skilled workers is intense.
European universities and vocational training systems are beginning to respond, with new programs in AI and data science expanding across the continent. However, the pace of educational reform has historically lagged behind technological change, and there are legitimate concerns that the current generation of mid-career workers may be left behind. The CEPR researchers emphasize the need for lifelong learning systems that can rapidly upskill existing workers, rather than relying solely on new graduates to fill the skills gap. Without such systems, the productivity benefits of AI may remain concentrated among a relatively narrow segment of the workforce and firm population.
Where Europe Goes From Here
The picture that emerges from the CEPR research and related analyses is one of cautious optimism tempered by significant structural challenges. AI is clearly gaining traction across European workplaces, and the technology holds genuine potential to boost productivity and create new forms of economic value. But realizing that potential will require more than technology deployment alone. It will demand sustained investment in skills, organizational change, and policy frameworks that balance innovation with inclusion.
For European policymakers, the stakes are high. The continent’s aging population and slowing productivity growth make the successful integration of AI into the economy not just desirable but essential. The CEPR analysis provides a valuable empirical foundation for the debates ahead, grounding discussions in data rather than hype. As the researchers conclude, the question is no longer whether AI will affect European productivity and jobs—it already is. The question is whether Europe can shape that transformation in a way that delivers broad-based prosperity rather than concentrated gains.