Summary
The EU is pursuing two tracks at once in cloud markets. On one, it is promoting competition with an unusually broad arsenal: the Data Act’s binding switching rules, enforcement under Articles 101 and 102 TFEU and, since 25 June 2026, the preliminary designation of AWS and Azure as gatekeepers under the Digital Markets Act (DMA). On the other, it is assembling an industrial-policy architecture, from the Cloud Sovereignty Framework to the proposed Cloud and AI Development Act, that conditions access to public demand on sovereignty requirements. Given that cloud adoption is a precondition for European competitiveness and for artificial intelligence, how these two tracks interact deserves closer scrutiny than the policy debate has so far given it. This Policy Paper examines whether they are in contradiction, and answers: yes, but only under certain configurations, and those configurations are a choice.
On the competition side, the Paper organises the obstacles documented by authorities across jurisdictions into four families of barriers (contractual, technical, strategic and structural) and matches them against the three instruments. The exercise exposes real gaps within the track itself: restrictive software licensing, the most active lock-in mechanism, is a usage problem that no ex ante instrument reaches and that enforcement has addressed only at the tying margin; and the DMA’s encounter with cloud carries serious procedural frictions, from designating while the obligations catalogue is still under review to fitting a gateway concept designed for two-sided platforms onto a vertically integrated B2B market.
On the industrial-policy side, the Paper traces how establishment and control requirements have migrated from administrative practice into proposed binding law, and what the exclusionary configurations cost: price premiums of 20% to 30% in comparable regimes, fewer effective bidders in procurement markets that already show competition deficits and legal exposure under Article 63 TFEU where restrictions reach EU-established subsidiaries of foreign parents. On the older question of whether public procurement should serve any objective beyond competitive purchasing, the position taken here is neither purist nor permissive: strategic use of public demand is admissible, but the evidence shows the exclusionary variant vindicating the purists’ warnings, while only Made with Europe configurations survive the strategic school’s own test of disciplined design.
The central argument is that the contradiction between the two tracks is real but contingent, not inherent. Contestable markets and reduced strategic dependency are both legitimate objectives; what makes them collide is instrument design and the absence of joint authorship or evaluation. The Paper closes with five coherence conditions: (a) each instrument confined to its own job; (b) a corrected sequence within the competition track; (c) Made with Europe as the default configuration with the burden of justification on structural exclusion; (d) coherence obligations across instruments; and (e) joint measurement of the two tracks’ interaction. The conditions do not settle the political choice between openness and strategic autonomy; they demand that whichever balance is struck be done so consciously, against shared standards, rather than emerge as the residue of instruments drafted without regard to one another.
Servers in a data center in Lithuania. Photo: BalticServers.com (CC BY-SA 3.0).
