Europe’s AI future as a meta-sovereign imaginary: negotiating global norms, sovereign machines, and infrastructural futures through AI simulations

The emerging AI Cold War exposes a paradox: the more states strive to territorialize AI infrastructures, the more they expose their dependence on globally entangled systems of chips, cloud, and compute that no actor fully commands. Sovereignty, once grounded in territorial control, has moved from the logic of possession to the symbolic and infrastructural processes through which authority is performed. It is enacted through symbolic gestures, strategic investments, and infrastructural imaginaries that project authority without securing it. We call this meta-sovereignty, a mode of rule sustained by infrastructures that do not yet exist, by fictions that persuade before they materialize, and by costly signals that substitute for control. Using the EU’s 2025 “AI Continent” plan alongside U.S. and Chinese initiatives, we examine the transition from multilateral coordination toward signaling and the political labor of imagination. We argue that AI Cold War strategies function simultaneously as costly signals of intent and as political fictions that project agency amid structural interdependence. Drawing on Fearon’s signaling theory and Ezrahi’s account of political fiction, we conceptualize these as meta-sovereign imaginaries in which legitimacy is negotiated through narrative projection and partial infrastructural realization. Building on this framework, we introduce an AI Cold War governance simulation in which policy-grounded LLM agents stage sovereignty under asymmetric constraints, revealing how such imaginaries adapt, fracture, or consolidate.