FORSEE D.5.1 Report-on-Digital-SMEs-perspectives-on-economic-viability
This report examines how European digital small and medium-sized enterprises (SMEs) understand and assess the economic viability of artificial intelligence (AI) adoption. It draws on accounts from 20 qualitative interviews with SME executives across 12 EU Member States, collected between April and July 2026. Thematic analysis, informed by the Social Construction of Technology and the Sociology of Expectations, explores how the anticipated benefits compare with implementation experiences. While these findings cannot be seen as representative of all European SMEs, they nevertheless offer important analytical insights into how some SMEs view AI systems and their adoption.
Productivity through the augmentation of human workers emerges as the main benefit associated with AI adoption. Participants describe existing teams undertaking more projects, performing a wider range of tasks and delivering improved outputs. Time savings allow effort to shift towards design, integration, review and client engagement. AI capability can also strengthen credibility and access to market share as SMEs signal their competitive advantage.
However, none of the participants establishes a clear link between AI adoption and increased profitability. Perceived productivity gains are often assessed through informal estimates, and their magnitude and relation to AI remain uncertain. Expanded capacity does not automatically reduce expenditure or generate additional revenue; GenAI adoption can increase costs and squeeze margins without corresponding revenue gains, echoing the GenAI divide between widespread adoption and limited economic returns.
Positive expectations and concern about falling behind nevertheless encourage continued investment, while practical experimentation leads firms to revise initial assumptions.
Changes regarding the workforce are similarly mixed: no participant reports AI-related job losses within their firm, yet several describe delayed or avoided recruitment. These findings suggest the operation of two related risks: the first is the risk of job stagnation and the second is that of workers’ lack of acceptance of new technologies as they fear for their own job.
Making AI economically sustainable requires organisational investment beyond purchasing tools. Costs include testing, data preparation, integration, training, security, human oversight and maintenance. Staff must verify outputs and assume responsibility, potentially creating additional review work. Dependence on large technology providers compounds these challenges through uncertain prices and costly switching, alongside concerns about data protection and compliance with the EU regulatory framework. European hosting and open-source systems offer greater control but require resources and expertise. Public funding can support development, although access difficulties, and delays can constrain its contribution to commercial viability.
The report concludes that reported productivity benefits coexist with uncertain profitability and substantial dependencies. Longer-term economic viability depends on whether SMEs can convert expanded capabilities into sustained value while managing costs and organisational demands. Longitudinal research incorporating financial evidence and employees’ perspectives is needed.
