This article was originally published on ainews.it on March 9, 2026 (Italian only).
AI can become a lever for competitiveness for SMEs, but cultural and regulatory barriers are hindering its adoption. The challenge lies in transforming it into concrete innovation.
For years, artificial intelligence has been portrayed as a revolution led by major global players, confined to Big Tech laboratories and multinational strategies. However, the most profound economic transformation is unfolding elsewhere: within the widespread fabric of micro, small, and medium-sized enterprises (SMEs). It is here that AI can cease to be a technological promise and become a concrete lever for competitiveness, capable of optimizing processes, reducing costs, generating new business models, and improving customer relations.
Yet, precisely where its impact could be most systemic, adoption remains limited. Data published by the Milan Polytechnic’s Observatory for 2025 shows a clear divide: only 7% of small businesses use AI consciously, and about 15% of medium-sized enterprises adopt it structurally, while its penetration in large companies is significantly higher. This is not a sign that the technology is useless, but rather evidence of a cultural, organizational, and regulatory lag that is holding back the transformation of the production system.
The SME Adoption Paradox
The low level of adoption highlights a systemic issue. The primary barriers are not technological but structural: there is a lack of widespread skills regarding data, risks, and opportunities; regulatory complexity generates compliance costs that are difficult to sustain; and interpretative uncertainty regarding rules slows down investment decisions. These factors make it harder to initiate innovation paths precisely where they are most needed, widening the competitive gap with large organizations.
Within the entrepreneurial landscape, a crucial distinction also emerges. On one hand, there are companies that directly develop AI solutions, equipped with technical skills and integrated into experimental ecosystems. On the other lies the vast majority of SMEs that do not produce technology but must necessarily adopt it to remain competitive, despite lacking adequate technical and regulatory knowledge. The overall competitiveness of the production system depends on this second group.
The “Regulatory Sea” Slowing Down Innovation
One of the main obstacles is represented by regulatory layering. Companies must simultaneously navigate the AI Act, GDPR, Data Act, and sector-specific regulations within a complex framework that creates uncertainty and leads to postponed investment choices. The problem concerns not only the quantity of regulations but their practical applicability and the interpretation of risk thresholds.
A clear paradox ensues: from a technical standpoint, many AI solutions are quick to implement and accessible even for small-scale entities; however, from a regulatory perspective, hidden costs emerge related to data management, risk assessments, liabilities, and necessary consultancy, which in some cases can exceed the cost of the technology itself.
A Concrete Case: Technological Simplicity vs. Regulatory Complexity
The gap becomes evident when observing operational situations. An SME wishing to adopt an AI system for document management may have access to ready-to-use, affordable, and fast-to-implement tools. Yet, alongside technological simplicity, regulatory questions immediately arise regarding the system’s risk classification, the presence of personal data, the need for impact assessments, the identification of liability, and transparency obligations.
The company thus finds itself forced to build unplanned compliance expertise, highlighting the divide between technological availability and the regulatory capacity to support its adoption.
A Lesson from GDPR for AI
The history of GDPR offers an encouraging perspective. Initially perceived as a cost and a source of complexity, over time it has helped spread a data culture, improve governance models, and even create competitive advantages for more mature companies. The same path could characterize new regulations linked to artificial intelligence, transforming a perceived constraint into a lever for trust and development.
For this to happen, specific conditions are required: continuous dialogue between institutions and businesses, clear indicators of AI diffusion and economic impact, simple guidelines proportionate to the size of SMEs, and, above all, a rapid reduction in interpretative uncertainty. Only when rules are understandable and applicable can businesses invest with confidence and accelerate their innovation journeys.
“Ai that Works” as an Industrial Paradigm
In this context, the paradigm of “AI that Works” takes shape: an artificial intelligence that is accessible, integrated into real processes, and capable of generating measurable impact. The real challenge is not building increasingly sophisticated algorithms, but making AI truly usable for millions of businesses, supporting its diffusion through skills, collaborative ecosystems, and technologies consistent with the European regulatory framework.
If adoption remains concentrated in large companies, the risk will be an increase in production inequalities. If, instead, it succeeds in spreading through SMEs, it can become the true engine of European economic growth.
