
Tech • AI • Robotics
A French tech commentator used a live Monday program to argue that France should lead a sovereign artificial intelligence strategy built on public procurement rather than subsidies, positioning it as a third path between the United States and China.
The central argument was that AI developed under the current US economic and geopolitical model does not match French or European interests. The issue was framed less as model quality than as control over infrastructure, data, energy use and strategic dependence. The proposed answer was a clear French-led alternative, with European partners joining voluntarily rather than through a slow, top-down EU design.
The proposed doctrine rejects a binary choice between the American platform model and the Chinese state-centered model. The idea is to create a sovereign ecosystem able to operate in French and European public services, companies and institutions, while remaining exportable. That “third way” was presented as politically urgent as France approaches the next presidential cycle.
A key policy proposal was to replace broad public aid with state orders and guaranteed demand. The argument is that subsidies often create fragile companies dependent on public money, while procurement forces delivery on clear timelines. In that view, the state should specify what it needs, set a date, and let capable firms compete to build it.
The strongest practical example concerned education. If an AI system could save every teacher one hour per day on grading, paperwork, schedules and lesson preparation, the productivity gain would already be substantial without inventing a new market. The same logic was extended to administration more broadly, where automation of repetitive tasks could free time for direct human work.
The discussion stressed the speed of technical progress in smaller models. Reference was made to local Nemotron models, described as already outperforming frontier systems from about 18 months ago in perceived usefulness. That rapid pace was used to support the idea that the main obstacle is no longer purely technical, but strategic and organizational.
The argument held that large-scale European coordination is too slow to carry such a project from the start. France was presented as better placed to move first, while keeping the door open to later cooperation with neighbors on infrastructure such as data centers. The broader point was blunt: France knows how to function as France, Germany as Germany, but Europe often struggles to act as a single operational power.
The case for sovereignty extended beyond software to chips, robotics, data centers, ecology and space. The wider warning was that AI is reshaping labor markets, training systems and business models at the same time that the US-China tech confrontation intensifies. In that context, national leadership over computing capacity and industrial tools was portrayed as a matter of long-term autonomy.
Rather than presenting a finished doctrine, the session called for criticism, examples and counterarguments from the wider community to refine the proposal. The goal is to turn the concept into a structured, actionable plan that can circulate publicly and influence decision-making. The emphasis was on moving beyond complaint toward a concrete roadmap for the technologies shaping the next decade.
The debate highlighted a growing French concern that AI sovereignty cannot be reduced to regulation or slogans. The main stake is whether France can turn strategic intent into working systems, contracts and institutions before dependence on foreign platforms becomes harder to reverse.
Explain this