58% of leaders call AI vital, 57% have no strategy for it

58% of French SMB and mid-sized company leaders see artificial intelligence as a survival issue for their business. At the same time, 57% of them have no formalized AI strategy. These two figures, published by Bpifrance Le Lab in January 2026 in its study "AI in French SMBs and mid-sized companies, a quiet revolution," describe almost exactly the same group of leaders. This is not an awareness problem. Awareness of the stakes is there. What is missing is the step from general conviction to a concrete action plan.

This finding matches what the actual depth of usage reveals. Roughly a third of French SMBs and mid-sized companies have adopted AI in a structured way. Of that third, half use free or generic solutions, with limited strategic impact. In other words, the adoption levels reported in perception surveys, often above 50% of companies surveyed, hide a more modest reality, a minority of SMBs have genuinely transformed a business process through AI, while the majority experiment occasionally without lasting integration.

Usage itself remains heavily concentrated on the existing. 94% of AI use in French SMBs and mid-sized companies goes toward optimizing activities already in place, drafting documents, analyzing data, researching information. Very few companies use AI to develop a new activity or product. This figure is not surprising, it is even reassuring on one point, AI is not yet being used as a risky bet on the unknown, it mainly helps companies do better what they already know how to do. But it also reveals considerable room for companies willing to go further.

The foundation most often missing is not technological, it is structural. 43% of French SMBs and mid-sized companies still do not use data analysis to steer their activity. A company that does not structure its sales, production, or customer data cannot build a reliable AI use case beyond assisted writing. This is the obstacle that comes first, well before the question of which model or subscription to choose, a company with no usable data cannot industrialize AI, whatever tool it picks.

The macroeconomic stakes give a sense of what is collectively at play. Bpifrance Le Lab estimates that AI-driven automation could add 1.3 percentage points to French GDP growth per year by 2034. That potential will not materialize if the gap between conviction and action persists across the fabric of SMBs and mid-sized companies, which make up most of the French economy. The responsibility does not rest solely on large, technologically advanced companies, it rests on the ability of the many to cross the line between occasional free-tool use and structured integration.

This gap between conviction and action is not unique to France, but it takes a particular shape here. The French economic fabric is made up of more than 99% micro, small, and mid-sized companies, often more cautious in their technology investments than large groups with dedicated budgets and in-house AI teams. That caution is not a flaw, it has long protected French SMBs from rushed investments in immature technology. But applied to AI in 2026, whose maturity and accessibility have improved sharply, that same caution is starting to become a competitive disadvantage rather than a protection.

Closing this gap does not start with choosing a tool. An SMB that decides to adopt AI because it rightly judges the issue vital, and rushes straight into a subscription or a general training session, reproduces exactly the pattern behind the 57%-with-no-strategy figure, a reaction with no method. The first question is not which tool to choose, but which AI uses already exist, even informally, within the company, and what data would be needed to develop them seriously.

This is precisely the purpose of the usage inventory, in week one: mapping the company's actual AI uses, official or spontaneous, before any decision on priorities or tools. The gap between 58% and 57% is not a statistical fatality, it is the direct consequence of this missing diagnostic in most companies. A five-business-day diagnostic turns this general conviction into a dated, prioritized action plan.