Socialize the Data Centres

In the AI Story

The essay's historical argument drew on the well-documented trajectory of previous infrastructure concentrations. Railroads, telephone networks, and electrical grids had all followed a recognizable pattern: initial private development, rapid concentration of ownership, the emergence of critical dependency relationships between infrastructure owners and the populations whose economic activity depended on the infrastructure, and eventually — after the consequences of unaccountable concentration became sufficiently visible — the imposition of democratic governance through regulation, public utility designation, or in some cases nationalization.

Morozov argued that digital infrastructure had reached the stage where comparable governance was both necessary and feasible. The concentration of ownership in a handful of platform corporations had produced dependencies that affected the economic, social, and increasingly cognitive lives of populations who had no voice in the governance of the systems they depended on. The rhetoric of market superiority that prevented democratic intervention was itself a political construct, not a natural law, and could be contested through the same political processes that had subjected earlier infrastructures to public accountability.

The argument has become sharper with the AI transition. AI infrastructure — the training data, the model weights, the computational resources, the deployment platforms — represents the most concentrated and the most consequential infrastructure humanity has built. It is governed by the least democratic mechanisms of any critical infrastructure in modern history. The dependency it creates is qualitatively different from previous infrastructure dependencies because it operates at the level of cognition rather than mere consumption — which means the stakes for democratic governance are correspondingly higher.

The essay did not prescribe a single institutional form. Morozov has subsequently elaborated various possibilities — public data trusts, cooperative ownership models, regulated-utility frameworks, and the institutional innovations that would make data sovereignty meaningful in practice. The common thread is the insistence that infrastructure this critical must be subject to governance mechanisms that reflect the interests of dependent populations rather than the strategic priorities of private owners.

Origin

The essay emerged from Morozov's engagement with European debates about digital sovereignty and the political economy of platform capitalism, synthesizing his critique of solutionism with a constructive institutional alternative drawn from the long history of infrastructure regulation.

Key Ideas

Infrastructure analogy. Data centers and AI platforms are infrastructure in the same sense as railroads, grids, and telephone networks — critical systems whose governance is a political question.

Historical trajectory. Previous infrastructures followed a predictable pattern: private concentration, dependency emergence, eventual democratic accountability. Digital infrastructure is traversing the same arc.

Cognitive dependency. AI infrastructure creates a form of dependency that operates at the level of cognition itself, raising the stakes of governance beyond anything previous infrastructures produced.

Plural institutional forms. The argument does not prescribe a single model; public trusts, cooperatives, and regulated utilities all represent viable alternatives to current private ownership.