A number of authors (Aytac 2022, Oldenbourg 2022) have recently made the case against private ownership of the social media infrastructure of democratic societies. While these arguments often refer to the Habermasian idea of a democratic public sphere as a precondition of legitimate government, their normative work is mostly done by a neo-republican conceptual framework, highlighting the risk of domination that individual users are subjected to when their freedom to participate in political debate is conditioned by uncontrollable corporate power. On the one hand, this focus on forms domination that always primarily affect individuals, and only secondarily the public sphere itself, runs the risk of obscuring, however, a normative concern that is central for Habermasian theory: A concern with preserving the specific form of communicative freedom and rationality that is a distinctive potential of a self-reflexive public sphere. On the other hand, Habermas has recently (2022) drawn attention to the dangers that digitally mediated communication poses to a functioning public sphere if it produces, as an outcome, isolated and increasingly irrational “semi-publics”. While this intervention succeeds connecting his theoretical framework to widely discussed worries, its concern with empirically contingent effects makes it less radical than the neo-republican accounts that are concerned with unconstrained power (and that consequently do not depend on any empirical claims about undesirable outcomes). In my paper, I argue that there is a more radical, non-individualist argument available to Habermasians: Allowing private parties to control those means of communication that are hegemonic in digital public spheres not only carries the risk of exposing individuals to uncontrolled power and to undesirable forms of interference, but democratic publics in their entirety (including citizens who refrain from participating in social media activities) to the risk of being dominated, because it introduces systematic obstacles to social reflexivity into democratic publics. Such forms of structural or constitutive domination form a distinct danger to democratic legitimacy. This argument has practical implications: While accounts that focus on the danger to individual users often suggest that democratic social media consumer cooperatives and similar arrangements could effectively safeguard against digital domination (Muldoon 2022), such particularist solutions are not promising if the democratic public as a whole is the subject for which non-domination is to be ensured. Because Habermas’s theory clearly also entails that a democratic public sphere cannot be directly organized through interventions by the state bureaucracy, I will argue that the digital infrastructure of democratic public spheres needs to be provided by institutions following the model of European public broadcasting services, independent of the state but subject to citizens’ contestatory control and to constitutional requirements.