Dependence is rarely a comfortable word for a continent that prides itself on regulatory power, yet it is the word the European Commission keeps returning to when it describes the bloc’s digital infrastructure. The EU relies on suppliers outside its borders for more than 80 percent of key digital products, services and infrastructure, and on 3 June the Commission put forward its most ambitious attempt yet to change that arithmetic: the Cloud and AI Development Act, the centrepiece of a wider technological sovereignty package.
The headline target is physical and unusually concrete for a piece of EU legislation. The Act aims to at least triple the bloc’s data centre capacity within five to seven years, an expansion the Commission expects to be financed largely by private capital, with estimates of the investment required running toward 200 billion euros. The reasoning is that artificial intelligence, cloud services and the public administrations and businesses that depend on them cannot run on infrastructure the EU neither owns nor controls, and that without a step change in domestic capacity the gap between European demand and European supply will keep widening through the decade.
To get the steel and silicon built, the Act reaches for the lever the EU has used repeatedly in recent years against slow permitting. Each member state would be required to designate at least one data centre acceleration zone, an area where qualifying projects benefit from streamlined authorisation, aggregated baseline permits and a hard ceiling of twelve months for the permit-granting process. Data centres are notoriously contentious at local level, drawing objections over their electricity demand, water use and land footprint, and the acceleration-zone model is an explicit bet that designated sites with pre-cleared environmental ground rules can move faster than the case-by-case permitting that has stalled projects elsewhere.
The more politically charged half of the proposal concerns who gets to run the resulting infrastructure. The Act introduces a four-tier cloud sovereignty framework and new requirements governing how public institutions and member states procure cloud and AI services, an attempt to translate the abstract goal of sovereignty into procurement rules that favour providers meeting defined standards of European control. That is where the hardest fights are likely to land. Hyperscale cloud is dominated by a small number of American companies whose services are deeply embedded across European governments and firms, and any framework that nudges public buyers toward sovereignty criteria runs straight into questions about cost, lock-in and whether genuinely European alternatives exist at the scale required.
The Cloud and AI Development Act does not stand alone. It was unveiled alongside a revived Chips Act, sometimes branded Chips Act 2.0, and an updated open source strategy, the three together forming a package the Commission frames as Europe’s bid to reduce its strategic dependencies across the digital stack from semiconductors up to applications. Skeptics note that the bloc has announced sovereignty drives before, and that previous cloud initiatives produced more memoranda than market share; the difference this time, supporters argue, is binding permitting deadlines and procurement leverage rather than voluntary coordination.
As a proposal, the Act now enters the long passage through the European Parliament and Council, where the permitting provisions, the procurement tiers and the sheer cost of the build-out will all be reopened. The target dates, tripled capacity within seven years, EU needs fully met by 2035, are the kind of numbers that are easy to legislate and hard to pour concrete behind. Whether Europe’s appetite for digital sovereignty survives contact with the price tag attached to it is the question the next two years will answer.
