On the contrary, Europe’s reliance on American-made AI is starting to look like a liability. In a worst-case scenario, though experts consider the possibility remote, the US could choose to withhold access to AI services and critical digital infrastructure. Clearly, the Trump administration could use Europe’s dependence as a springboard as the two sides continue to hammer out a trade deal. “That dependence is mandatory in any negotiations—and we will be negotiating more with the US,” Taddeo said.
The European Commission, the White House, and the UK’s Department of Science, Innovation and Technology did not respond to requests for comment.
To curb those risks, European countries have tried to bring AI production to shore, through funding programs, deregulation, and collaboration with academic institutions. Other efforts have focused on building large-scale competitive language models for European vernaculars, such as Apertus and GPT-NL.
However, as long as ChatGPT or Claude continues to improve the European-made chatbot, America’s lead in AI will only grow. “These fields usually take everything. If you have a really good field, everybody goes there,” Nejdl said. “Not being able to produce modern technology in this industry means you will not be able to participate. You will always be feeding the big players with your input, they will be much better and you will be far behind.”
Note the Gap
It is not yet clear how far the UK or the EU intend to take “digital sovereignty,” campaigners say. Does sovereignty require complete autonomy across the entire AI supply chain, or only enhanced capabilities in a small set of disciplines? Does it need to be issued by US-based suppliers, or only available in domestic alternatives? “It’s not clear,” said Boniface de Champris, senior policy manager at the Computer & Communications Industry Association, a membership organization for technology companies. “It seems to be the talk of the town at this stage.”
And there is no broad agreement on what policies should be drawn upon to create the conditions for Europe to become self-sufficient. Some European suppliers are advocating a strategy where European businesses would be required, or at least encouraged, to buy from domestic AI firms—similar to China’s reported approach to its domestic processing market. Unlike grants and subsidies, such an approach could help with seeding, said Ying Cao, CTO at Magics Technologies, a Belgian outfit developing AI-specific processors for use in space. “That’s more important than just getting money,” Cao said. “The most important thing is that you can sell your products.” But those who advocate for open markets and deregulation say that trying to cut off US-based AI companies puts domestic businesses at risk of global peers, left to choose which AI products are best for them. “From our point of view, sovereignty means choice,” said de Champris.
But for all the disagreements over the minutiae of policy, there is widespread belief that closing the performance gap in America’s leaders is still clearly possible even in labs with limited budgets and resources, as DeepSeek has shown. “If I just thought we wouldn’t fight, we won’t [try],” said Nejdl. SOOFI, an open-source modeling project in which Nejdl is involved, aims to put out a general-purpose language model to compete with nearly 100 billion parameters in the next year.
“Progress in this field will no longer depend on large GPU clusters,” Nejdl said. “We will be the European DeepSeek.”
