Europe’s AI Ambition: Sovereignty, Innovation, and Ethical Tech

The European Union is making a strategic push to carve its niche in the global AI race with the launch of OpenEuroLLM, an ambitious open-source language model initiative designed to reflect Europe’s linguistic diversity and ethical values.

By Matthaios Tsimitakis
February 04, 2025

Open Euro LLM has received €52 million in funding, including €20.6 million from the EU’s Digital Europe Programme. As a recognition of its strategic importance, it has earned the European Commission’s STEP Seal.

The collaboration unites 20+ research institutions, startups, and super computing centres across Europe, including Germany’s Aleph Alpha, Finland’s CSC, and France’s Lights On. Led by computational linguist Jan Hajič and Silo AI co-founder Peter Sarlin, the initiative aims to develop adaptable, transparent LLMs for all 24 EU languages, trained on European supercomputers.

The STEP Seal will help the project attract additional funding and increase its investment profile, potentially accelerating AI innovation across the European Union. Unlike competing projects from the United States and China, OpenEuroLLM emphasizes democratic principles, transparency, and cultural diversity.

"This isn't about creating another general-purpose chatbot,” Sarlin explained. “Our goal is to build digital infrastructure that enables European companies to innovate with AI tailored to their specific contexts." The models will comply with the EU’s AI Act and prioritize commercial, industrial, and public-sector applications.

The crackdown of DeepkSeek by different European Data Protection Authorities comes as DeepSeek’s rapid ascent highlights a paradox for Europe. The Chinese model, developed at a fraction of the cost of Western counterparts like OpenAI’s GPT-4, has been praised for its efficiency, opening the way for homegrown innovation. 

This poses a dilemma for Europe, which balances its regulatory leadership (via the AI Act) with the need to foster homegrown innovation. The stakes escalated with the U.S.’s 500 billion Stargate AI infrastructure project and China’s 140 billion state-backed AI investments. AI innovation no longer requires exorbitant budgets but it's urgent as it's underscored by rivals’ rapid advances. 

Reacting to these developments, European coalitions like the European AI Forum and EIT Digital are demanding bold action such as fast-tracking the EU AI Stack to boost energy efficiency, sovereign AI, and slashing bureaucracy to attract private investment. They are also prioritising the adoption of EU-developed AI in key industries.

The groups argue that rather than being passive, the EU should respond with a bold vision and strategy to build a powerful, energy-efficient, and strategically autonomous European AI ecosystem. This includes implementing initiatives like the "EU AI Stack" to support an open, sovereign, and innovative European AI landscape.

“It’s a bit like at the beginning of the internet when we thought that the big business model would be to make websites," says Thomas Wolf, chief science officer and co-founder of French-American AI company Hugging Face.

"The big takeaway is that the recipe to AI is easy, lots of teams can do it. There is no moat around the skillset required to create high-quality foundational models," he says. "The gigantic business model in AI won’t be building LLMs, but building companies that use LLMs."

The question is even deeper than that: "The challenge is one of digital sovereignty, to avoid becoming a mere digital colony," Corinne Narassiguin, a Socialist Party senator, member of the French Parliamentary Office for Scientific and Technological Assessment (OPECST), an independent body within the parliament. “The European Union is currently focusing on regulating AI, but this is still insufficient in the face of the size and progress of the American and Chinese powers,” she added.

OpenEuroLLM exemplifies Europe’s dual vision: leveraging collaboration and open-source principles to counterbalance tech giants while embedding ethics into AI’s foundation. The European Commission views the initiative as a potential landmark in the continent's technological development, supporting efforts to strengthen digital sovereignty. Detailed models are expected to be released in the coming months, with a focus on multilingual capabilities and adaptability across various sectors


Image: European AI Alliance - European Commission