Experts debate the path to sovereignty at Ardian's AI x AI conference

Can Europe still compete?

After exploring the geopolitics of the internet routing and the challenges of building a sovereign future for data centers in Europe, the morning concluded with a, if not the, major question: can Europe still compete in the AI race? 

Vincent Luciani, Co-founder and Executive Chairman at Artefact opened the discussion with a pragmatic assessment. He argued that generative AI is rapidly becoming a commodity, with prices plummeting and open-source models quickly closing the performance gap with their proprietary counterparts. He contended that Europe cannot win a direct "race to scale" against the US and China, given the difference in investment firepower, noting that US investments in AI have been 20 times higher than Europe’s. Instead, Europe's winning strategy lies in the "battle of adoption," emphasizing the use of AI technology to build competitive advantages in key asset sectors. Viewing AI as a general-purpose technology similar to electricity, its true value comes from its widespread application across the economy.  

Audrey Herblin-Stoop, VP and Head of Global Public Affairs and Communication at Mistral AI, stated that, indeed, can Europe win – but more importantly it needs to. The picture she painted was stark: beyond the economics, the soft power impact of AI will be huge.

Losing this race could mean losing the ability to preserve European languages and values. 

Herblin-Stoop championed open source as a key accelerator for adoption and technological progress, arguing that value is created through specifying, fine-tuning, and co-training models with partners. She also highlighted the necessity of increased talent, attracting European researchers back from Silicon Valley, and favoring a global playing field where European companies can compete effectively.  

Hugo Le Picard, Associate Research Fellow at the international relations think-tank Ifri broadened the perspective, identifying two distinct AI races: an "upstream" race for cloud infrastructure, data centers, and GPUs requiring massive, continuous investment, and a "downstream" race focused on integrating AI into the real economy. For Europe, he argued, accelerating the downstream race is paramount. Given Europe's long-term economic slowdown, widespread AI adoption is essential to create value and foster growth. Le Picard noted a significant bottleneck: traditional companies often lack the knowledge to effectively implement AI, while AI companies may not deeply understand specific industry needs. He advocated for bridging these worlds, possibly through "forward-deployed engineers" to facilitate integration. Ultimately, he said a realistic short-term goal for Europe is to lead in integrating AI across its economies. 

A lot of ifs

Jean-Louis Missika, Senior Advisor at Ardian, offered a more conditional optimism, asserting that Europe can make it, but there are a lot of ifs. His "ifs" included the need for dramatically increased investment in research and higher education. Pointing to Europe’s higher collective GDP than the US or China, he asked: “Why are we investing less in research than the US and China?”. He pointed out the brain drain of European researchers to American companies and questioned Europe's historical choice not to invest in defense, which often drives research in other global powers. Missika prescribed a "civilian DARPA" model – a structure that enables quick decisions led by experts and strong public-private partnerships. For Europe to compete, he suggested, the continent needs to adapt its investment governance. However, as he reminded the audience, a good start has been made with the Draghi report, which recommended €800 billion of investment per year to enhance European competitiveness. 

A note of cautious optimism

Throughout the panel, several cross-cutting themes emerged. On public policy, Luciani urged policymakers to lead by example, integrating AI into their administrations, and to craft subtle regulations that allow for hybrid cloud solutions while protecting sensitive data. Herblin-Stoop emphasized that for startups, revenue-generating public procurements are preferable to mere subsidies. The challenge of talent attraction and retention was a recurring point, with calls for competitive compensation, harmonized equity schemes, and a vibrant ecosystem that prioritizes researchers' freedom and ability to publish. Luciani also highlighted that European venture capital often exhibits a risk-averse attitude, reluctant to fund zero-revenue projects, a critical barrier for disruptive innovation.

The panel concluded with a unanimous, albeit cautious optimism: all panelists answered "Yes" to the question put by moderator Sophie Pedder of The Economist of whether Europe can compete. Although no single path was prescribed, the consensus was for a distinct European strategy – one that prioritizes rapid AI adoption, leverages open-source technology, increases investment in talent and research, streamlines governance, and fosters a more ambitious, risk-taking mindset. 

While the future is complex, all agreed that Europe possesses the capability and will to forge its own impactful AI future.