SMEs and Mid-sized Companies: AI, Between Ambitions and Complex Realities

AI: A Strategic Challenge for SMEs and Mid-sized Enterprises

Artificial intelligence has quickly established itself as a central topic in the strategic discussions of companies, particularly within SMEs and mid-sized enterprises (ETIs). The promises of AI are numerous: task automation, process optimization, and the development of conversational assistants and intelligent agents. These technologies generate significant enthusiasm, but a crucial question arises: are companies truly ready to leverage data and AI?

For several years, hundreds of organizations have been analyzed to assess their maturity regarding data and AI. The consensus is clear: this maturity is not improvised; it is built patiently over time. Companies must develop a deep understanding of the technologies and processes necessary to integrate AI effectively and sustainably.

Data Maturity of SMEs: A Persistent Challenge

The results from the Data & AI Maturity Observatory for Enterprises reveal a significant gap between the aspirations of SME/ETI leaders and the reality of their implementation. In 2025, a survey showed that 88% of leaders consider data to be an essential lever for performance, but only 19% believe they are capable of fully exploiting it.

Paradoxically, nearly half of the companies continue to manage their operations with tools like Excel or Google Sheets. The identified obstacles are recurring: lack of organization, skills deficits, absence of in-depth knowledge, and a strategic vision. Leaders recognize the potential of data and AI to improve their performance but struggle to master the prerequisites necessary for sustainable exploitation.

Leaders at a Crossroads

Leaders of SMEs and ETIs today face structural choices in a constantly evolving environment. The rapid pace of change forces them to frequently reevaluate their decisions. They must juggle imperatives such as the acceleration of AI, the fear of missing out, and immediate competitiveness issues. Additionally, concerns related to digital sovereignty, cybersecurity, regulatory uncertainties, as well as environmental and societal challenges, come into play.

This situation creates constant pressure to act, often in urgency, without the necessary benchmarks to effectively structure their approach. Companies are advancing in a context where reference frameworks are still under construction, making strategic decisions particularly complex.

2025: A Year of Disordered Experimentation

In 2025, many companies chose to react quickly to the pressure. This year was marked by a wave of experimentation. Nearly one-third of companies had already integrated an AI solution, and half planned to do so by 2026.

However, in practice, this translated into the purchase and deployment of licenses for tools like ChatGPT, Copilot, Mistral, and Gemini. Usage often remains individual, without a clear overall strategy. It is not yet an organizational transformation but rather a phase of experimentation aimed at meeting the expectations of employees and leaders.

The Classic Mistake: Starting with the Tool

A common mistake is to begin with the acquisition of tools before clarifying business issues and the company's level of maturity. This leads to scattered usage, teams advancing in a heterogeneous manner, and human barriers emerging. Scaling up then becomes a major obstacle.

Companies often find themselves blocked by initiatives that are not aligned with a global strategy, limiting their ability to leverage technologies coherently and effectively.

The Limits of AI Without Solid Data

Initially, companies notice time savings in tasks such as content writing, meeting minutes, and document analysis. However, when it comes to industrializing these processes or creating sustainable value, limitations quickly become apparent.

The real issue is not AI itself, but the quality and exploitation of data. Companies face difficulties related to data quality and the inability to scale. Without a solid database, AI projects remain limited.

Three Levels of AI Usage

On the ground, three levels of AI usage are distinguished: individual, business/project, and organizational. Most SMEs and ETIs still find themselves between the first two levels. Few companies have truly initiated a comprehensive transformation. This phase of experimentation is necessary, but it cannot last indefinitely. Companies must structure their approach to reassure their often-skeptical employees.

Human Barriers Outweigh Technological Obstacles

Contrary to popular belief, the main obstacles to AI adoption are not technical but organizational, cultural, and human. Concrete questions arise: who leads the initiative internally? What skills are necessary? How to mobilize teams?

Broader issues also emerge, such as environmental impact, social responsibility, and the role of humans in the organization. Companies must navigate a complex landscape where human and ethical considerations play a crucial role.

2026: Towards Necessary Structuring

After a year of experimentation in 2025, 2026 could very well be the year of structuring. Companies will need to focus on creating sustainable value rather than hastily adopting AI. The companies that succeed will be those that take a pragmatic and methodical approach.

An Essential Question for the Future

The crucial question is no longer whether to adopt AI, but rather whether one is building a genuine transformation or merely accumulating tools. Before accelerating, it is essential to take stock, clarify use cases, business issues, and priorities. It is crucial to lay solid foundations, train teams, and then accelerate in a controlled manner.