What boards really mean by tech sovereignty

European tech sovereignty has moved quickly from conference language to procurement reality. In Brussels, the argument is framed as economic security: reduce dependence on a small number of foreign providers, keep greater control over critical digital infrastructure, and build more domestic capacity in cloud, compute, and AI. Business groups are now warning that a harder push to end reliance on US technology could hit profits and competitiveness. That warning matters because it captures the central tension in the debate. Sovereignty is politically persuasive. Procurement is not political in the same way. It is governed by uptime, integration, contract risk, skills, and cost.

This week, governments tried to give sovereignty more industrial substance. In Germany, Polarise said it plans a 30-megawatt AI data centre in Bavaria, with scope to expand to 120MW, in a move that would materially increase domestically run AI computing capacity. In France, Emmanuel Macron argued that the country’s nuclear power base gives it the ability to open data centres and build more computing capacity for artificial intelligence. These are not symbolic gestures. They point to the physical foundations of digital independence: land, power, financing, chips, cooling, and operating expertise. Once sovereignty is discussed at that level, it stops being a slogan and becomes a capital allocation question.

The case for greater control is not difficult to understand. Maria Luís Albuquerque, the EU’s financial services commissioner, has said Europe must retain control over the key technologies that underpin its economies. That concern is especially visible in finance, where regulators are already grappling with concentration risk. Under DORA, the European Supervisory Authorities designated critical ICT third-party providers for direct oversight, describing the move as a crucial step in managing the systemic risks that come from heavy reliance on a limited number of technology providers. In other words, sovereignty is not only about industrial pride. It is also about resilience, substitutability, and the question of what happens when too much of the system sits on too few external platforms.

Yet the commercial objections are real, and they are not confined to lobbying. Most companies do not buy technology as a declaration of values. They buy it to reduce downtime, shorten deployment cycles, improve security, and avoid unnecessary complexity. Replacing hyperscalers or core software providers is expensive. Migrating workloads is slow. Retraining teams takes time. Boards may have a strategic interest in avoiding lock-in, but very few want to swap one dependency for another if the substitute is less mature, more expensive, or harder to integrate. That is why the language of “independence” often gives way, in boardrooms, to a more sober word: optionality.

The likelier outcome is not a clean break from US technology, but a more selective model of sovereignty. Public-sector systems, regulated workloads, critical infrastructure, and AI compute may attract stronger domestic requirements and more local investment. Private companies are more likely to pursue layered strategies: sovereign hosting where the risk is highest, foreign providers where scale and capability remain decisive, and architectures designed to reduce deep lock-in over time. That approach is less dramatic than the rhetoric, but it is also more credible. Sovereignty, in practice, may prove to be less about replacing everything than about knowing which layers of the stack truly matter.

The difficulty for Brussels is that technology policy is being asked to do too many jobs at once. It must improve resilience, answer political anxieties, support local industry, and preserve competitiveness. Some of those goals align. Others collide. When executives say they support sovereignty, many mean they want more bargaining power, better local infrastructure, and fewer single points of failure. That is already a significant shift. It is also a more realistic one. The harder work begins when that sentiment reaches procurement committees, budget holders, and live enterprise systems. That is where sovereignty stops being a principle and starts becoming an operating model.