Earlier this month, a pervasive fear hung heavy in the air at Vivatech, a major tech conference in Paris: the unsettling prospect of Europe becoming perpetually reliant on American AI, implicitly trained on American values. As the US and China continue their intense AI arms race, France and Germany, despite their deep confidence in their own engineering talent, feel increasingly marginalized. They are not only demanding a seat at the table but are actively promoting ambitious plans to carve out their own path in artificial intelligence.
The term “sovereignty” was uttered so frequently that it could have fueled a drinking game for conference attendees. Throughout my decades covering the tech industry, I’ve witnessed numerous attempts by countries to replicate Silicon Valley’s unique ecosystem. While individual successes abound, no nation has ever truly matched the dynamic environment that fostered giants like Google, OpenAI, and Anthropic.
A stark disparity in investment further highlights Europe’s challenge. While American AI companies attract massive funding, European ventures receive comparatively modest sums. A statistic frequently cited at Vivatech underscored this imbalance: Anthropic’s recent $65 billion fund-raise surpassed the entire sum invested in European and UK AI startups last year. Data from the EU seems to corroborate this concerning trend.
Europe’s Ambitious Vision for AI Sovereignty
Despite the daunting statistics, discussions around AI sovereignty at Vivatech were tinged with optimism. Proponents pointed to significant new funding initiatives, burgeoning collaborative efforts, and the emergence of next-generation AI technologies that might be less resource-intensive than today’s leading large language models. Intriguingly, many attendees also cited a wild card that could prove to be the biggest boon for European tech in decades: the unpredictable influence of Donald Trump.
The Vivatech conference coincided with the G7 meeting in Evian-les-Bains, France, where French President Emmanuel Macron directly addressed AI executives on the critical issue of sovereignty. Macron firmly stated that if the US continued its trajectory of nationalistic AI, France would take decisive steps to develop its own independent capabilities. Aiden Gomez, CEO of Toronto-based Cohere, echoed this urgency, emphasizing the need for a democratic player to occupy the number two position in AI, a role he believes is currently vacant.
The notion of Europe building the world’s second-best AI might seem audacious, almost delusional, to some. It would require unprecedented collaboration among more than 20 nations, a radical shift from continental impulses that often stifle innovation with bureaucracy, and the attraction of immense investment. Most crucially, Europe must shed its risk-averse mindset and embrace a “moonshot” mentality.
Building a Collaborative AI Future
President Macron has already made tangible progress through his “Choose France” initiative, which has secured pledges of over 100 billion euros for AI infrastructure. This effort is anchored by Softbank’s staggering 75 billion-euro commitment to establish massive data centers in France, pending regulatory approvals. These investments signal a serious intent to build robust AI capabilities within Europe.
On the collaboration front, Aiden Gomez revealed Cohere’s efforts to forge a multinational network of partnerships. This includes a key alliance with German AI firm Aleph Alpha, aimed at pooling engineering and infrastructure resources for a “sovereign-first” approach. Gomez also mentioned signing a Memorandum of Understanding with Indra, Spain’s largest tech company, during a recent visit with the King of Spain.
Adding another layer to this collaborative vision is Yann LeCun, the renowned AI pioneer who recently stepped down as Meta’s chief AI scientist. He is championing Project Tapestry, a monumental undertaking that seeks to unite governments and private industry in developing a state-of-the-art frontier foundation model. LeCun believes that the universal desire for AI sovereignty can only be achieved through an open, free foundation model, upon which any entity can build specialized assistants tailored to their specific language, culture, values, and political leanings.
The Trump Effect: A Wake-Up Call for Europe
While these plans appear sound, they bear a resemblance to previous attempts to elevate European tech that ultimately fell short. However, a critical new element emerged in 2026: the disruptive policies of the Trump administration, which have rendered the status quo intolerable. For generations, the US has attracted Europe’s brightest scientific minds; now, many feel increasingly unwelcome, leading to a noticeable decline in European enrollment at US universities.
Jakob Uszkoreit, CEO of the AI-based biotech firm Inceptive, observed a discernible shift in talent away from the US even during Trump’s first term, a trend that has gained considerable momentum under the current administration. Uszkoreit suggests that it wouldn’t take much to significantly draw this talent away from American shores, provided two crucial conditions are met: reasonable personal incentives and the opportunity for these researchers to perform their best work.
It’s worth noting that both Uszkoreit and Gomez themselves exemplify the talent the US risks losing. They were among the eight co-authors of the seminal Transformers paper that ignited the generative AI revolution, with seven of those eight being foreign-born. This highlights the immense contributions made by international researchers to American technological advancement.
Export Controls: A Catalyst for Independence
Perhaps the most significant incentive for Europe to supercharge its own AI ecosystem came earlier this month, when the Trump administration attempted to place Anthropic’s powerful Claude Fable model under strict export control regulations, effectively denying foreigners access. While the justification for this clampdown remains debatable, from a European perspective, it served as a stark wake-up call: AI sovereignty now equates to survival.
A foreign company simply cannot build a reliable business around Claude, or any other American model, if the US government can arbitrarily deny access at any moment. Even more alarming, the designation initially banned access even for foreign nationals working at Anthropic, including those who helped develop Fable! Though Anthropic quickly pulled the model off the market, the message was unmistakable: Europeans cannot depend on American AI companies.
This episode has already provided a significant boost to European startups. Michael Förtsch, CEO of the chip startup Qant, noted that the current attention his company receives in Europe would not have been possible without Trump’s policies. He emphasized that the short-lived export controls on Fable “triggered a complete new discussion on sovereignty in Europe.” As Uszkoreit eloquently put it, “Europe had gotten pretty complacent in light of a very reliable, well-aligned, near-hegemonic [situation]. The US just made it clear that in the new world order, that’s over.” Europe now perceives no alternative but to aggressively pursue AI sovereignty.
Source: Wired – AI
