Economic growth is slow. Political instability is high, with the transatlantic alliance in tatters. The continent, fearing dependence on US software, Chinese rare earths, and Taiwanese chips, has rallied around a new cry: digital sovereignty.
But “decoupling” from US and Asian technology would cost Europe between €3 and €5 trillion, a sum it cannot afford and a move that would ultimately stifle innovation. Europe can neither afford nor achieve this fantasy. The continent instead must embrace Pragmatic Digital Sovereignty. This is not about building walls, but about carving out a position of indispensability within the global tech stack.
Despite the gloom and doom, Europe should not overlook its considerable strengths. It holds chokepoints that are key to tech success, starting with lithography and chemicals, and boasts world-leading industrial data. It is not just Europe that depends on American companies; a large part of the US public sector depends on European companies, too. Software giant SAP, Germany’s most valuable company, is the third-largest software vendor to the US government. Unlike Europe, the US isn’t raising sovereignty alarm bells.
No one country or continent, not even the US or China, can achieve digital sovereignty defined as total control and complete tech independence. Pragmatic digital sovereignty means combining offensive and defensive tactics designed to ensure that Europe remains indispensable to other partners while protecting its own security.
Offensive tactics aim to secure European leverage over its trading partners by making the continent indispensable in the tech value chain. It must boost its poor self-esteem: Europe boasts a strong industrial base, with a world-class tech workforce. The continent needs to focus on its competitive strengths in the various layers of the AI stack — chemicals, optics, and semiconductor lithography — and build new ones in areas such as applied AI and fusion energy, rather than expending energy trying to duplicate US and Asian strengths.
Defensive tactics promote European interests by avoiding lock-in and dependence on a single supplier or a single nation. Security of supply, never absolute, can be insured only by maintaining multiple suppliers, interoperable data and operational security guarantees.
The quest for digital sovereignty should not and does not require a divorce from non-European vendors. Europe needs to ensure competition in key tech markets like AI chips, AI data centers, and AI software choices. Tough antitrust action could help.
Start with chips. Europe possesses world-leading strategic assets that other nations rely on. It holds a virtual monopoly on the lithography machines required to make advanced chips via ASML and leads in essential chemicals and R&D. Rather than trying to manufacture the “shiniest” new chips, Europe should double down on these niche strengths and secure the supply of basic chips vital to its automotive and industrial sectors.
Continue with cloud. The “big three” US hyperscalers — Google, Microsoft, and Amazon Web Services (AWS) — have captured 70% of Europe’s cloud computing market, and roughly 80% of spending on cloud and software services in Europe goes to American companies. Europe’s own position in its cloud market is dire. In 2017, European providers like SAP and Deutsche Telekom still held a 26% market share on the continent. Since then, it has dropped to 15%, despite a growth in revenue for European providers. Rather than failing at “sovereign clouds” like Gaia-X, Europe should focus on interoperability to prevent vendor lock-in and accelerate its lagging tech adoption. AWS, Microsoft Azure, and Google Cloud represent hundreds of billions of euros in capital expenditure for data centers, networking, and software. In 2024 alone, the big three invested close to $150 billion in AI infrastructure.
In order to catch up, Europe needs to ease the construction of cloud infrastructure. European energy costs run up to three times more than in the US. Permitting to build data centers in Europe is often slow and unwieldy. Counterproductive data localization measures proliferate. At present, companies are building data centers in Luxembourg, which suffers from high land and energy costs. But Luxembourg requires all financial data held in its large banking system to be located inside the country.
The bottom line is clear: reducing vendor-lock in should be a priority, giving customers control over their data and choice of cloud provider. Beyond this, Europe should boost cloud adoption and European companies ought to have freedom of vendor choice. Public procurement should prioritize operational control over geographic sovereignty requirements. European vendors should be strategically promoted only in sensitive sectors.
Onto the crown jewel of the AI stack: end-user software. American AI models such as ChatGPT, Llama, and Claude have captured the public’s imagination. Today, 70% of foundation models are being developed in the US, with models like OpenAI’s ChatGPT, Google’s Gemini, and Meta’s Llama among the most developed in the world. An additional 15% goes to China, particularly after it burst on the scene in January 2025 with its DeepSeek, providing a more affordable, open-source alternative to the American models.
Since the race for massive Large Language Models (LLMs) is already crowded, Europe’s future lies in Applied AI — specialized, smaller models used in healthcare, robotics, and B2B sectors where European industrial data is a competitive advantage.
Europe’s current strategy takes the goal of AI sovereignty to be independence, and focuses on economic sovereignty as its primary goal. The cost-benefit analysis for such a project doesn’t add up. Even if Europe invests billions in its digital independence, it (like every other nation) is unlikely to succeed. As long as Europe is always going to be dependent on others, duplicating other countries’ capacities won’t give Europe leverage. Leverage is created by knowing one’s strengths, building on them, and putting in place insurance policies for when things go wrong.
Europe suffers from a crisis of self-confidence. It’s sitting at the poker table, and doesn’t know the cards it holds. It is too easy to buy into the picture of Europe as a hanger-on to American success. But the picture is more complicated. Europe holds strategic strengths. It should pivot from the “do everything” strategy to deepening in areas where it already leads globally.
For those who define sovereignty as independence, these recommendations will be insufficient. But let’s be practical. The continent can achieve interdependence, not independence. Through smart policy choices, it can still secure a critical role in the AI value chain.
William Echikson is a Brussels-based non-resident Senior Fellow at CEPA and editor of the Bandwidth blog.
Bandwidth is CEPA’s online journal dedicated to advancing transatlantic cooperation on tech policy. All opinions expressed on Bandwidth are those of the author alone and may not represent those of the institutions they represent or the Center for European Policy Analysis. CEPA maintains a strict intellectual independence policy across all its projects and publications.
Tech 2030
A Roadmap for Europe-US Tech Cooperation
