France doesn't want to be a digital colony

On the two walls of the corridor loom two enormous diagrams, each twelve metres long. In white on a black background they contain a bewildering profusion of terms, numbers, pictograms, schematic representations of people and technical systems, curves and historical quotations. There are inventions, developments and events from the past five centuries.

These range from Leonardo da Vinci’s Vitruvian Man and the wampum beads of North American Indigenous peoples to the founding of the United Nations. Also included are fascist bureaucracies, as well as psychoanalysis and robotics, global refugee movements and high-tech communication systems.

In 2025 it all culminates in AI – as if this represented a kind of provisional final stage of evolution in terms of technologies, infrastructures and systems of power. The installation, entitled Calculating Empires by the artist duo Kate Crawford and Vladan Joler, formed part of the exhibition Le monde selon l’IA (The World According to AI), which was on show at the Jeu de Paume in Paris from April to September last year.

The title and timing of the exhibition were apt. The United States and China have a clear lead in the global AI race, but France wants to play catch-up.

“French data centres crammed full of American technology”

Shortly before the start of the AI Action Summit, the major AI conference held in Paris in February 2025, President Emmanuel Macron announced a programme providing for €109 billion in private-sector investment in artificial intelligence in France. This came hot on the heels of US President Trump's announcement, a month earlier, of the launch of the “Stargate” project: a total investment of $500 billion to strengthen AI in the United States. The French project, which involves both domestic and foreign investors, aims to build new data and computing centres for AI. At the Élysée Palace there is hope that these “factories of tomorrow” will give rise to start-ups and contribute more broadly to the modernisation of companies, triggering a domino effect across the economy.

Since then, a number of companies have made binding investment commitments totalling around €26 billion. If there is one name that epitomises France's AI ambitions, it is Mistral. Founded in 2023, the start-up specialises in large open-source language models, including for chatbots, and is regarded as one of Europe’s most promising AI companies.

Its growth to date has been rapid: at €11.7 billion, its valuation has almost doubled since its last funding round in June 2024. Other French companies, such as PhotoRoom – one of the world’s leading providers of AI-based image editing – or LightOn, which develops and markets AI tools in the form of large language models for corporate applications, have also managed to establish themselves in a fiercely competitive market.

But are these encouraging developments due to the policies adopted by the French president? Macron has indeed pursued “an investor-friendly policy and a policy to promote start-up financing”, says Julien Pillot, a lecturer in economics at INSEEC Business School in Paris. But, he adds, “above all, the country has decisive structural advantages that make it the most attractive in Europe in terms of AI.”

For the economist, the foremost of these is the ability to train talented specialists in the field. Once they enter working life, however, it is still “difficult to keep them in the country; they go abroad, to the United States, because they want to work in labs with the best resources and in the best teams”, he says.

Another structural advantage for investors is that, in France, AI is expected to be powered largely by nuclear energy – “low-cost, decarbonised energy,” according to Pillot. It remains controversial to what extent nuclear power is truly sustainable, given the CO₂-intensive extraction of uranium, radiation risks, the issue of waste storage, and the construction and dismantling of plants.

For the moment, however, it enjoys the support of a majority of the French population. A further locational advantage lies in measures such as the tax credit for competitiveness and employment (CICE), in force since 2013, which offers incentives for business investment. These are helpful in opening up new markets and in promoting research and innovation.

Building on these strengths, France aims to spearhead the European AI sector, enabling it to keep pace with the United States and China. For this reason it is working closely with Germany. At the initiative of both countries, a summit on European digital sovereignty was held in Berlin on 18 November, at which they called on Europe to assume a leadership role in AI. Among other announcements, a strategic partnership with Mistral and SAP for use in the public sector was unveiled.

Such cooperation is vital for France and for Europe as a whole. In September 2024, Mario Draghi, former president of the European Central Bank, warned in a widely noted report of an existential threat to Europe: “The first – and most important – task for Europe is to fundamentally refocus its collective efforts on closing the innovation gap with the United States and China, particularly in advanced technologies.”

Julien Pillot stresses that, in the field of AI, there is already a technological dependence on foreign companies, since “we do not fully control the value chains of the digital industry.”

In France, the infrastructure required for these technologies – above all data centres and hardware – is largely provided by non-European suppliers. The investments announced by Macron will, says Pillot, “create data centres on French soil. But they will be crammed full of US technology and operated by American companies.”

In this sense, “regulation is the only lever at our disposal to pursue a genuine industrial policy aimed at strategic autonomy”, the economist says. In effect, this means resisting the pressure from American tech giants to loosen regulation. This applies to calls such as Apple’s in September 2025 to withdraw the Digital Markets Act (DMA), the EU regulation designed to make the digital economy fairer and more competitive.

There are indeed laws such as the DMA, the EU’s General Data Protection Regulation (GDPR), or the AI Act in force since 2024. On their basis, Google could be fined very heavily for abusing its dominant market position, or platform X could be banned from the European market. Yet, according to Pillot, those responsible lack the “political will and technical capabilities” to enforce these measures.

“It’s hard to ban X in Europe when you need Starlink in Ukraine”

It is difficult, he says, “to tell Elon Musk that X is finished on European territory when you need Starlink in Ukraine and are rolling out the red carpet for him to build gigafactories in Europe”. All the more so as European companies themselves fear over-regulation. They complain about distortions of competition vis-à-vis US firms, which are able to circumvent European data-protection rules more easily.

“The question,” Pillot concludes, “is this: will we become a digital colony of the United States, with data centres using American technology and operated by American companies? Or will the European Union wake up, regard the US as a rival, and raise its voice to defend its interests?”

As for the promise to promote sustainable and inclusive AI, voiced by various politicians at the AI Action Summit, it failed to convince many. “It was a flashy summit without a genuine ethical, regulatory or sovereign vision for a European AI model serving the common good and the public interest,” commented David Cormand, an MEP from the French Greens.

Frédéric Bordage, a member of the Green IT collective, has also levelled a long list of criticisms. Among other things, he objects that the declaration on sustainable and inclusive AI published at the end of the Paris summit – which was not signed by the United States or the United Kingdom – contains no call for regulation on transparency. Bordage would like to see regulation that would make it possible to estimate the carbon footprint of AI applications with a minimum degree of accuracy, so that customers could compare competing models on that basis.

But France’s position in this field has been deeply undermined by its lack of political stability. In the three years since the last presidential election in May 2022 and the start of Emmanuel Macron’s second term, the nation has seen five prime ministers. There are repeated calls for Macron to resign and, by dissolving the National Assembly, to allow new elections.

According to polls, the Rassemblement National, Marine Le Pen’s party, could win the 2027 presidential election – a prospect that would represent a political turning point and could make some investors think twice. Another problem for the development of the AI sector is the strained state of public finances: France’s debt now stands at over 115.5 per cent of GDP (as of June 2025). The interest rates at which the state borrows have risen in recent months.

Against this backdrop, “France may appear riskier for investors,” says Julien Pillot. As a result, in both artificial intelligence as in politics, the nation finds itself at a crossroads.