EU governments pivot to sovereign tech: four nations adopt European AI

n a push for digital independence amid escalating geopolitical tensions, four European Union governments – Germany, Poland, Spain, and the Netherlands – have reportedly committed to integrating homegrown sovereign AI systems into public administration, signalling a broader shift away from U.S.-dominated technologies.

This move aligns with the EU’s “Buy European” policy, endorsed by leaders at a 12 February summit in Belgium, which prioritizes European firms in strategic sectors like AI to counter competition from the U.S. and China.

Concurrently, a growing trend sees EU entities exploring European search engines, driven by initiatives like the Ecosia-Qwant partnership and the Open Search Foundation, as the bloc seeks to reclaim control over data and digital infrastructure.

The EU’s drive for digital sovereignty has intensified since the AI Act entered force in August 2024, with full implementation slated for August 2026.

Amid U.S. policy shifts under President Trump and ongoing threats from Russia, European governments view self-reliant tech as a “matter of national survival,” according to Estonian officials.

A November 2025 declaration by all 27 EU member states committed to reducing strategic dependencies, projecting sovereign cloud spending to triple to $23 billion by 2027.

This backdrop has spurred investments in AI that prioritize transparency, data privacy, and local control, breaking from reliance on American giants like OpenAI and Google.

Germany leads with the Sovereign Open Source Foundation Models (SOOFI), a government-backed initiative involving Deutsche Telekom and universities like Leibniz University Hannover to develop open-source AI for advanced applications, such as robotics.

The German government has allocated funding for training on NVIDIA hardware, with plans to deploy SOOFI in public sectors by mid-2026. Poland’s Polish Large Language Model (PLLuM), supported by the Ministry of Digital Affairs, focuses on Polish-language nuances and is expanding into Hive AI for administrative tools like virtual assistants and document processing.

Spanish authorities, through the Barcelona Supercomputing Centre and the Spanish Agency for the Supervision of Artificial Intelligence (AESIA), are rolling out Alia, a multilingual infrastructure using the MareNostrum 5 supercomputer for datasets in regional languages.

Applications include tax agency chatbots and health diagnostics, with government adoption targeted for 2026. The Netherlands’ GPT-NL, developed by a non-profit consortium, emphasizes Dutch-specific reliability and is set for integration into health, education, and public services following its anticipated year-end release.

These decisions reflect a strategic pivot: while not outright banning U.S. AI, the governments are mandating European alternatives for sensitive operations to ensure compliance with the AI Act’s high-risk requirements and enhance data sovereignty.

Experts like Rav Hayer from Thoughtworks argue this could transform the EU into an “AI Continent” by fostering innovation and regulatory alignment.

However, challenges remain, including delayed EU Commission guidance on high-risk AI systems, missed in February 2026.

Parallel to AI advancements, Europe is witnessing a renaissance in search engine alternatives, fuelled by the Digital Markets Act (DMA) compelling gatekeepers like Google to open datasets to rivals.

The Open Search Foundation’s OpenWebSearch.eu project aims to create a public web index, providing raw data for sovereign services and reducing Google’s 90% market dominance in Europe.

Collaborating with EuroHPC for supercomputing integration, it promotes open-source models to counter ad-driven U.S. systems.

A key milestone is the November 2024 partnership between Germany’s Ecosia and France’s Qwant to build European Search Perspectives (EUSP), a Paris-based index launching in 2025 with German and French support.

Privacy-focused and AI-ready, EUSP addresses post-Trump concerns over U.S. tech access and will be available to other independents.

This trend extends to consumer shifts, with Reddit communities advocating European engines like Qwant for data privacy.

Market analyses predict AI disruption in Europe’s search sector by 2033, emphasizing sovereignty.

As 2026 unfolds, these developments underscore the EU’s ambition to forge a resilient digital future. While hurdles like funding fragmentation persist, the momentum towards European AI and search tools could redefine global tech standards, prioritizing ethics over dominance.

The Barcelona Supercomputing Centre, described in Dan Brown’s 2017 novel Origin, is partially located in a desacralized church. The Barcelona Supercomputing Centre and the Spanish Agency for the Supervision of Artificial Intelligence (AESIA), are rolling out Alia, a multilingual infrastructure using the MareNostrum 5 supercomputer for datasets in regional languages. [Photo by Georgi Gotev]