A trilogy that explores what a Chinese model hosted by a leading French AI company reveals about the new geopolitics of intelligence.
By Dr. Tawhid CHTIOUI, Founding Presidentof aivancity School of AI & Data for Business & Society; selected by Keyrus as one of the 25 most influential global figures in the field of AI and data (January 2025).
What if we had gotten the sovereignty wrong?
Sometimes a national ambition is put to the test by what, at first glance, appears to be a simple business decision… On August 11, 2026, Mistral AI announced that it would expand its so-called “sovereign” infrastructure to include models developed by other companies. The first of these, GLM-5.2, was designed by Z.ai, one of the major players in Chinese artificial intelligence. Mistral makes no secret of either its origin or its status: it is indeed a third-party model offered within its European infrastructure (Mistral AI, 2026a).
A few days later, the South China Morning Post summarized—with irony that was hard to ignore—how this decision might appear from an Asian perspective: Europe was supposedly “plugging a Chinese AI brain into what would, in practice, be a European shell ” (South China Morning Post, 2026, author’s translation).
The statement is deliberately provocative. It simplifies a much more complex technical and industrial reality, but it highlights a contradiction we can no longer ignore: What does sovereignty even mean anymore when we control the infrastructure, the data, and the operating conditions, but the intelligence itself was designed elsewhere?
Taken on its own, Mistral’s decision is perfectly justifiable. An AI company can hardly claim to meet all needs with its models alone, especially at a time when architectures are proliferating and customers are seeking more choice, performance, and control. However, in Mistral’s case, this decision has implications that go far beyond its commercial strategy alone.
Because Mistral is not just another French technology company. In just a few years, the company has become one of the most visible symbols of France’s—and, more broadly, Europe’s—ambition to not entirely cede the development of large-scale artificial intelligence models to American platforms.
That is precisely why the inclusion of a Chinese model in its sovereign offering is not merely a minor catalog detail. It serves as a telling indicator.
It forces us to ask a much more fundamental question: What if we were wrong—not about the need for sovereignty, but about what that word really means in the age of artificial intelligence?
A few months earlier, in December 2025, Emmanuel Macron was still urging the French to choose Le Chat over ChatGPT precisely because it was French. He explicitly linked Mistral to the reclaiming of “cognitive sovereignty,” alongside computing capabilities, data centers, and other technological building blocks that France and Europe are seeking to rebuild (Office of the President of the Republic, 2025). This is where the paradox arises.
Mistral has been held up as a symbol of the possibility—for France and, more broadly, for Europe—of not ceding artificial intelligence to major American platforms. Three years after its creation, this same champion has chosen to incorporate Chinese-designed AI into its sovereign offering.
It would be too easy to see this as a contradiction or a retreat. Above all, that would be to miss the most interesting question: What if this episode revealed that we still don’t quite know what we mean when we talk about sovereignty in artificial intelligence?
For several years now, digital sovereignty has been understood primarily in terms of control over data, infrastructure, the cloud, computing capacity, or applicable law. While these dimensions remain essential, academic literature already highlights the ambiguity of a concept that encompasses very different—and at times even contradictory—objectives. Daniel Mügge thus shows that sovereignty in AI can refer to technological capability, economic autonomy, political control, or the protection of citizens; Francesca Musiani, for her part, points out that digital sovereignty is also built into the concrete infrastructure on which our capacity for action depends (Musiani, 2022; Mügge, 2024).
Artificial intelligence now adds a more unsettling dimension to this discussion. A model is not simply software hosted on a server. It is the result of corpora, languages, architectural choices, training methods, alignment rules, and decisions about what to respond to, reject, prioritize, or consider sensitive. In other words, artificial intelligence has no nationality in the legal sense of the term, but it is never without a history, a cultural context, or human choices.
Given that, can we consider an AI system to be sovereign simply because it operates in a data center located in France or Europe?
This issue becomes particularly sensitive when the models come from China. It is all the more so given that China no longer occupies the position of a mere follower in AI, as it was still considered to be until recently. The Stanford AI Index 2026 notes that the performance gap between the top U.S. and Chinese models has virtually closed: since early 2025, models from both countries have alternated at the top of several evaluations, while China now holds a leading position in AI research, publications, and patents (Stanford HAI, 2026).
This rise of China presents us with an uncomfortable question. If we ask ourselves which values, norms, or worldviews can be incorporated into a model developed in China, why did we ask this question far less often when virtually all the models we used came from the United States? American AI is no more culturally neutral than Chinese AI. The difference may lie in the fact that several decades of American technological dominance have gradually led us to confuse what is globally dominant with what is universal.
The Mistral case thus raises a question that goes far beyond the nationality of a model. It compels us to reflect on what might be called our “cognitive sovereignty”—that is, our ability to understand, choose, and, when necessary, challenge the systems to which we delegate an increasing share of our access to knowledge, reasoning, and decision-making.
But it also raises another issue, one that may be even more difficult for France.
Our country boasts an exceptional tradition in mathematics, internationally recognized researchers, prestigious higher education institutions, universities, and laboratories capable of training talent that is sought after by the world’s leading companies. However, the experience of recent years shows that scientific excellence, however remarkable it may be, is not enough to create sustainable technological strength. Between the researcher and the global leader lie capital, computational power, energy, infrastructure, market access, government procurement, industrialization capacity, and speed of execution.
France knows how to generate intelligence. The question is whether it still knows how to turn that intelligence into power.
This series of three opinion pieces invites readers to reflect on this paradox.
The first will examine what is really at stake behind Mistral’s integration of Chinese models, as well as the cultural, political, and cognitive dimensions of sovereignty in AI.
The second will seek to understand why France and Europe, despite the quality of their research and talent, continue to struggle to translate this excellence into industrial power in the face of competition from the United States and, now, China.
The third will finally attempt to reconstruct the very concept of sovereignty. For in a world where no single country is likely to control the entire technology chain on its own, autonomy can no longer mean producing everything ourselves. It may instead have to mean something more demanding: remaining capable of choosing the artificial intelligence systems we use, understanding what they entail, replacing them when we wish, and, above all, never giving up on producing our own.
Mistral may not be the subject. Mistral may be the indicator…