One figure stopped me cold while building my thesis. In a Bpifrance Le Lab survey published in 2025, run with more than 1,200 leaders of small and mid sized companies, 58% see artificial intelligence as a matter of survival within three to five years. Survival. The word is heavy, and it is sincere.
The same survey gives a second figure, far less quoted: only 43% have formalised an adoption strategy, and 26% actually use a generative tool. Between the fear and the act, there is a chasm. That chasm is what interests me, because it is there, and nowhere else, that the real decision is made.
Fear is not enough to make you act
Feeling you must move and knowing how to move are two different things. Most leaders I meet do not lack conviction. They lack an entry point. AI reaches them as one indistinct block, both urgent and vague, and a vague block does not get handled, it gets postponed. We will come back to it at the next board meeting, then the one after.
Add to the vagueness the surrounding noise, the spectacular demonstrations, the suppliers who promise the moon, the peers who claim to have transformed everything. You get the polite paralysis that marks so many leadership teams on this subject. Everyone nods about its importance, no one knows where to start, and fear, instead of pushing to act, ends up freezing them.
It is a cruel paradox: the more vital the stake seems, the more it intimidates, and the more you put it off. Perceived survival produces wait and see.
The detail that makes the SME fascinating
Here is what many forget in these statistics. In the same survey, in nearly three quarters of cases, it is the leader themselves who decides to commit the company, with no technical intermediary. Not a committee, not an IT department. One person.
That concentration of the decision is the signature of the small company, and it is precisely what makes it a unique field of study, the one I chose for my research. The strategic choice is not diluted in an organisation there. It forms in one head, with its intuitions, its limits, its relationship to risk. You watch the decision being made almost in the raw.
This has a direct practical consequence. In a small company, the quality of AI adoption depends less on a budget or a technology than on the clarity of a single person. It is a fragility, but also a strength: a decided leader can move the whole house fast, without the heaviness that paralyses large groups.
Where to start, concretely
When a leader asks me where to begin, I never answer with the name of a tool. I answer with a place.
Look for the task that comes back every day, that takes time, and whose error shows quickly. That is where a first brick of AI can be tested without risk, with a benefit measurable in a few weeks. Not the most impressive task, the most tameable. One person, one week, one single use. You measure, you adjust, you extend if it works.
Why so small, when the stake is judged vital? Precisely because it is judged vital. Faced with a matter of survival, the classic mistake is to want a response that matches the fear, a grand transformation plan. And grand plans never get going. A modest but real first success is worth a thousand times more, because it turns paralysing fear into confidence that moves forward. An adoption strategy is not decreed in a meeting, it is built through small gains that make you want the next one.
The real risk, for a small company today, is not picking the wrong tool. A bad tool choice is fixed in a month. The real risk is staying among the 57% who watch the train go by, telling themselves they will board the next one, until the day the platforms are empty. Survival does not play out in the fear you feel. It plays out in the first step you dare to take.