VivaTech, the meeting dedicated to technological innovation which is held in Paris, was this year the great stage of the French project to accelerate the recovery of sovereignty in the field ofArtificial intelligence. The message for theEurope However, there are mixed feelings. On the one hand, it demonstrates that the presence of a strong national entity like Mistral and large computing hubs can serve as the "core" for a potential European platform, especially if combined with common financing instruments, as Macron proposes. On the other hand, if the French project results in institutional and technical solutions that are not very interoperable, or if governance is perceived by other member states as a nationalization, of the European agenda, an incentive is created for the proliferation of “mini national initiatives” in AI.
This could fragment technology development programs, duplicate investments in basic models, and ultimately reduce Europe's critical mass relative to the United States and China—exactly the opposite of what is needed.
The experience of the greatest example of technological coordination in Europe—the development of mobile telephony in the 1980s—suggests at least three pillars for a truly European initiative on AI. First, a strong European level of technical standardization, based on open-source technologies, common protocols, and shared reliability and security levels, complementing the AI Act with voluntary standards, but strongly incentivized for those wishing to operate in the single market.
Second, a commitment that binds Member States and major operators (both public and private) in terms of minimum investments in technology, training, and interoperability conditions, reducing the risk of incompatible national barriers. Finally, the integration of national initiatives (such as the French and German projects) into a framework of shared hardware, software, data, and programs, with closer coordination between national agencies and the Commission. In this framework, France can be a pillar, but not the political "owner" of European AI: the strength of the GSM model lay in the fact that no single state could claim ownership of the standard.
The real risk is not so much that France will proceed alone as that a patchwork of national strategies will consolidate, partially competing with each other, with cooperation limited to EU funds. In this scenario, Europe risks replicating the pre-GSM fragmentation in analog telecommunications: numerous solutions, each too small to benefit from significant economies of scale, forced to rely on non-European infrastructure. To avoid this outcome, it is crucial that European instruments (AI factories, funds, research programs) function as truly supranational "hubs," not simply as labels for already-decided national projects.
...
The published text is the second of three analyses dedicated to Artificial Intelligence and technological sovereignty by Franco Bernabe, President of TechVisory, which FIRSTonline hosts weekly courtesy of the author and the AI Magazine.
