European governments and technology firms are intensifying efforts to build domestic artificial intelligence capability, as reliance on US-based platforms increasingly intersects with regulation, security and trade concerns. New funding commitments, regulatory frameworks and research initiatives aim to strengthen Europe’s ability to develop and deploy advanced AI systems within its own legal and infrastructure boundaries. The push reflects a broader reassessment of technology sovereignty, as artificial intelligence becomes embedded in public services, industry and national security.
The European Union’s AI Act, which entered into force in 2024, provides binding rules for the development and use of artificial intelligence, including large general-purpose models. Alongside regulation, public investment and access to state-backed supercomputing resources are being positioned as tools to support European labs and companies seeking to reduce dependence on US technology providers.
Structural gap with US AI leaders
Across much of the AI supply chain, US firms currently hold a clear lead. American companies dominate areas including advanced processor design, data centre capacity, large-scale model training and commercial deployment. Investment patterns reflect that dominance, with a significant share of global AI funding flowing to US-based firms and contributing to their rapid scale and market reach.
Some policymakers and security officials have warned that Europe risks replicating earlier dependence patterns seen in cloud computing, where US providers became deeply embedded across government and enterprise systems. Concerns have been raised that critical digital infrastructure could become a point of leverage in future trade or regulatory disputes, particularly as AI services expand into sensitive domains.
Public funding and onshore capacity
In response, European governments have committed hundreds of millions of euros to support domestic AI research, infrastructure and commercial development. Funding programmes focus on areas such as model training, language coverage for European markets and partnerships between universities, startups and public institutions.
State-backed supercomputing facilities are being opened to researchers and industry, providing access to computing power that would otherwise be prohibitively expensive for smaller labs. These resources are intended to reduce barriers to entry for European projects and allow experimentation with alternative approaches to model design and optimisation.
Some initiatives focus specifically on developing large language models for European languages and use cases. Projects such as GPT-NL in the Netherlands and Apertus in multilingual research contexts aim to address gaps left by predominantly English-language models developed elsewhere.
Open development as a strategic approach
One area where European labs differ from many US counterparts is a greater emphasis on open development. Several research groups and startups publish model architectures, weights or training methods for external use and modification. Supporters argue that this approach allows improvements to compound more quickly through collaboration, rather than remaining locked within proprietary systems.
Advocates of open development also point to its alignment with European regulatory and academic traditions, where transparency and peer review are long-established norms. By contrast, many leading US AI firms operate closed systems, limiting external scrutiny of training data and model behaviour.
While open development does not remove the need for substantial computing resources, proponents argue that recent advances show efficiency and design choices can narrow performance gaps without relying solely on the largest hardware clusters.
Regulation and transatlantic tensions
Europe’s AI strategy is unfolding alongside wider regulatory and diplomatic tensions with the United States. EU authorities have taken enforcement action against US technology platforms under digital competition and content rules, prompting criticism from US officials. Disagreements over regulation of social media, data protection and platform accountability have become increasingly public.
These disputes have sharpened focus on the risks of technological dependence. While experts generally view extreme scenarios, such as the withdrawal of AI services, as unlikely, the possibility of commercial or political pressure has reinforced calls for domestic alternatives and greater resilience.
At the same time, European policymakers have stressed that sovereignty does not necessarily mean isolation. The debate centres on how much self-sufficiency is required to preserve choice, security and regulatory autonomy, rather than replacing one dependency with another.
Competing visions of AI sovereignty
There is no single definition of “digital sovereignty” across Europe. Some stakeholders argue for policies that actively favour domestic suppliers through procurement incentives or market access rules, similar to approaches used in other regions. Others caution that restricting access to global providers could raise costs or limit innovation for European businesses.
Industry groups have highlighted the need for clarity on whether sovereignty implies full control across the AI supply chain or simply the availability of viable European alternatives. This distinction affects how governments design funding schemes, procurement rules and regulatory guidance.
Despite these differences, there is broad agreement that Europe’s current position leaves it exposed in negotiations over trade, regulation and technology standards, particularly as AI becomes more deeply integrated into economic and public systems.
Performance gap and long-term prospects
US-developed models continue to set benchmarks for general-purpose AI performance, reinforcing network effects that draw users and developers towards established platforms. Some analysts warn that AI markets may follow winner-takes-most dynamics, making late entry difficult without distinctive advantages.
However, recent developments in model efficiency and training techniques have challenged assumptions that only the largest computing clusters can deliver competitive results. European researchers point to examples where smaller teams have achieved meaningful performance gains through optimisation and design choices rather than scale alone.
Several European projects have set targets for releasing large-scale language models over the next year, supported by public infrastructure and collaborative development frameworks. While outcomes remain uncertain, these efforts illustrate a strategic shift towards sustained domestic capability rather than short-term competition.
What this means
Europe’s renewed focus on sovereign AI reflects a convergence of regulation, geopolitics and technological maturity. Binding AI rules, public funding and access to computing resources are being used together to shape how artificial intelligence is built and governed within Europe.
For technology firms and public institutions, the shift signals a landscape where compliance, infrastructure access and jurisdictional control matter as much as raw performance. For policymakers, the challenge lies in balancing openness and competition with resilience and autonomy.
The success of Europe’s AI push will depend less on matching US firms feature for feature, and more on whether domestic ecosystems can deliver reliable, competitive systems aligned with European legal and social frameworks.
When and where
This development was reported in January 2026, drawing on coverage of European AI policy, funding initiatives and industry activity published by WIRED.
