Ardian makes the case for European AI sovereignty

The geopolitical stakes of AI

In his opening remarks, Executive Vice-President of Ardian and CEO of Ardian France Mathias Burghardt laid out Europe’s current dependence on foreign infrastructure and tools, highlighting the consequences for the economy, geopolitics and Europe’s capacity to protect its culture and social values. He advocated for a more active approach to support the emergence of European champions capable of competing with American and Chinese companies. He believes the EU has relied too much on normative power rather than focusing on massive investments in infrastructure, technology, fundamental research and education as major levers. He pointed to Europe’s household savings, today mainly allocated to cash and deposits, that could be redirected to a more productive part of the economy, potentially starting to reduce the gap with the US and China. 

The conference’s keynote speaker was Europe’s foremost expert in the geopolitics of data and internet routes, Frédérick Douzet, Professor of Geopolitics at the French Institute of Geopolitics (University of Paris 8), Director of GEODE, and Principal Investigator of DATAROUTES, one of the most important projects on the topic in Europe, funded by the European Research Council. 

The first part of her keynote detailed the often-overlooked geopolitical dimensions of the internet and digital infrastructure, sharing her research on Iran’s isolation strategies as well as on Russia’s controlling internet routes after the 2014 Maidan Revolution. The second part focused on another trend: the extent of Europe’s digital dependencies and the increasing privatization of internet routes by major platforms, which also reinforces these dependencies. Professor Douzet stressed that the rise of generative AI could deepen existing dependencies, and advocated for an ambitious strategy involving major investment in fundamental research, higher education, start-ups, and critical infrastructure to protect sovereignty and European values.

The discussion continued over two panels, which explored the technological, political, commercial, cultural and infrastructure dimensions of digital sovereignty. 

Panel I: Sovereignty above all? Rethinking what matters most in data-center strategy 

The panel began by unpacking digital sovereignty. Michel Paulin, Board member Quandela (the Quantum Photonic Leader) and ex-CEO OVHcloud & SFR, defined it as Europe’s “freedom of choice” in technology. Henri Verdier, Ambassador for Digital Affairs, French Government, warned that outages or geopolitical tensions could quickly expose this vulnerability, urging companies to diversify their cloud dependencies. 

The discussion then turned to Europe’s chance to build its own AI-ready infrastructure. Gonzague Boutry, Managing Director – Head of Digital Infrastructure, Ardian, highlighted the continent’s strengths such as clean energy, strong connectivity, a lively startup scene, tempered by slow permitting and heavy regulation. Michel Paulin argued for a distinctly “European way” based on open source and specialized models, with multiple AI champions rather than one giant.

Michel Paulin and Henri Verdier warned that fragmented strategies are slowing Europe, which once led in global digital standards. Unified policies could restore that leadership. The debate also covered sustainability and talent: innovations like wafer-scale GPUs and liquid cooling are boosting data center efficiency, aided by Europe’s clean energy mix. However, talent shortages persist, requiring stronger research pathways and more female participation in tech.

The panel closed with cautious optimism: Europe has the ingredients for sovereign digital infrastructure arguing it is now the moment to act collectively. 

Panel II: Europe’s sovereignty in AI: Is there still a path ahead? 

The panel opened with a clear diagnosis: Europe may not catch up with the US or China in sheer AI scale, but it can lead in adoption. Vincent Luciani, Co-founder and Executive Chairman, Artefact, emphasized that real value lies in embedding AI across the economy, not in building the biggest models. Audrey Herblin-Stoop, VP, Head of Global Public Affairs and Communication, Mistral AI, added that staying competitive is also about safeguarding Europe’s cultural and linguistic influence, highlighting open source as a strategic advantage and calling for talent to return from Silicon Valley. 

The conversation then shifted to the two AI “races.” Hugo Le Picard, Associate Research Fellow, Center for Geopolitics of Technology, Ifri, explained that while the upstream race—cloud, GPUs, data centers—demands massive, continuous investment, the downstream race is about applying AI in industry, where Europe still lags. He argued that connecting AI firms with traditional sectors, even through “forward-deployed engineers,” is essential. 

Jean-Louis Missika, Senior Advisor, Ardian, underscored the scale of effort required to catch up, pointing to the need for far greater research funding, stronger universities, and DARPA (Defense Advanced Research Projects Agency) like innovation structures to stem the brain drain. The Draghi report’s €800 billion annual investment recommendation, he noted, reflects the magnitude of the challenge. 

The panel agreed Europe can compete in AI if it accelerates adoption, embraces open source, attracts top talent, and takes bolder risks. Public administrations should use AI, regulation must protect data without hindering innovation, and startups need revenue from procurement over subsidies.  

Is it too late for Europe?

In closing remarks, Pauline Thomson, Ardian’s Head of Data Science, underscored that Europe’s AI ambition must be pragmatic and coordinated. That means having clear priorities: investing in resilient digital infrastructure, building stronger public-private alignment through incentive programs to support R&D and strategic infrastructure, and developing an ecosystem that is efficient in retaining top talents. 

As the conference neared its end, Sophie Pedder, Chief of The Economist’s Paris Bureau, asked the panelists: “Is it not too late? Can Europe still compete in the AI race?”. 

Many agreed that there is a long process ahead and Europe will need to choose its battles. Some argued that Europe had to find its own way to compete. But given the stakes involved – geopolitical, technological, cultural – there is no other choice than competing. 

Audrey Herblin-Stoop, VP and Head of Global Public Affairs and Communication at Mistral AI, put it well: “Yes, we can win it – but more importantly we need to win it. Because the impact of this technology is such that if we do not, we will not just lose on the economic side – we will lose everything.”

Q&A with Gonzague Boutry, Ardian’s Head of Digital Infrastructure Europe

Q: Why is digital sovereignty important?
Gonzague Boutry: AI is no longer just a set of tools; it is infrastructure, power and geopolitics. Europe’s growing reliance on non-European AI and technology ecosystems, particularly U.S. and China-based providers, creates a “dependency path” that could constrain future digital autonomy, with implications for resilience, competitiveness, and regulatory enforcement. 

Q: What is the scale of the problem? 
As highlighted by our speaker Frederick Douzet, data flows are concentrated in a handful of private actors. Roughly 70% of Europe’s cloud services come from US providers, while Nvidia alone holds over 90% of global AI chip sales. On top of that, increasingly complex geopolitical tensions are bringing about new vulnerabilities we would not have envisioned even 10 years ago. That is the context in which investors like Ardian must assess risks and opportunities with a long-term view.

Q: What must we do about it? 
First, we need to recognize that digital infrastructure has become a strategic, essential asset – it is a matter of national security. Second, Europe’s AI ambition must be pragmatic and coordinated. The notion that we either replicate the US/China playbook or give up is a false choice. We can build on our advantages while tackling our dependencies. Third, regulation is a strategic asset if we use it to accelerate innovation rather than slow it. Europe’s approach should simply favor the so-called EU preference. European decision makers should promote resources to build European sovereign AI capabilities, train open language models, and foster innovation leveraging on our talents.