Table of Contents

Every modern state runs on infrastructure. Roads, electricity grids, postal systems, courts, registers of births and deaths — the physical and procedural rails on which ordinary life runs. For most of the twentieth century these were either publicly owned or privately owned and tightly regulated. The twenty-first century has added a new layer of infrastructure on top of this — identity, communication, payments, maps, information retrieval — and that new layer is overwhelmingly private and largely unregulated.

This is not a conspiracy, and framing it that way misses what is interesting. The shift is an equilibrium, arrived at by obvious local decisions that made sense at every step. It is also a structural change in how civic life is organized, and the consequences are becoming impossible to ignore.

What got privatized, quietly

Start with a list that should be uncontroversial. How most people in a developed country in 2026 actually access government services: through a browser owned by one company, running search paid for by ad impressions on a different company’s property, sending email through a third, identifying themselves via an OAuth flow owned by a fourth, paying via card networks duopolized between two firms, navigating with mapping data owned by one of two providers, and communicating about what they are doing on messaging platforms owned by a small cluster of firms.

This is not a failure of the state. It is a success of the private sector at offering services that were too expensive, too slow, or too uncool for the state to build. In most cases the state did not lose anything; it simply never had the capacity to offer the service in the first place. In some cases — payments rails in countries with weak banking systems, identity in countries where the national ID system is dysfunctional — the private rails are strictly better than what came before.

But the cumulative effect is that the mundane infrastructure of civic participation now runs on private rails. A citizen in 2026 who cannot access Gmail, Apple Pay, and Google Search is meaningfully excluded from ordinary life. That is a civic fact, and it is new.

The accountability problem

What this does to accountability is the part the discourse has not caught up with.

When civic infrastructure is public, there is a clear — if imperfect — chain of accountability. Decisions are made by officials who are eventually answerable to elected bodies, who are eventually answerable to voters. The decisions are slow and often bad, but the chain exists.

When civic infrastructure is private, the chain is different. The platform is answerable to its shareholders, its regulators, and its user base, roughly in that order and with the first dominating. The voter is not in the loop. A platform can change the rules of civic participation — how you authenticate, what you are allowed to say, which information is retrievable — in ways that would be impossible for a public agency, because there is no equivalent legitimacy requirement.

This is fine in the limit where platform interests and civic interests are aligned. Most of the time they are. But the cases where they diverge — content moderation at election time, identity verification for services the platform has opinions about, algorithmic suppression of speech the platform disfavors — are exactly the cases where it matters most that the infrastructure be accountable, and it is exactly these cases that the private model handles worst.

Why the state cannot easily reclaim the ground

There is a standard response that goes: governments should simply build or regulate public alternatives. This is not wrong, but it is naive about the constraints.

The state’s problem is not budget. The state’s problem is speed. Platform infrastructure operates on software timescales — decisions made in days, deployments in weeks. Public infrastructure operates on procurement timescales — decisions in quarters, deployments in years. For any given capability, by the time the state has built a public equivalent, the private one has moved two generations ahead. This is not a failure of any particular administration. It is a structural mismatch between legislative process and software development velocity.

The other problem is that the state increasingly depends on the platforms for its own operations. Agencies run on the same cloud providers as the platforms they would regulate. Officials communicate on the same messaging tools. The notion of a state that stands above the platforms and regulates them from an independent vantage point is already more aspirational than descriptive in most jurisdictions.

What to watch for

I do not think the privatization goes all the way. I think the endgame is a set of functions that are eventually re-classified as critical civic infrastructure and subjected to some version of the utility-regulation model we applied to electricity and telecom in the twentieth century. Payments are already moving in this direction. Identity is next. Communications are further out but plausible.

The signal to watch is not grand policy announcements. It is boring procedural changes — interoperability requirements, mandated data portability, reserved-power provisions that let regulators intervene in platform operations during defined emergencies. Each of these nibbles a small margin of sovereignty back from the platforms. Over a decade they add up.

The alternative — a continuation of the current drift, where the state is a senior tenant on rails it does not control — is politically unstable. It does not end in a dramatic event. It ends in a series of crises, each one of which makes it fractionally less defensible that civic infrastructure is owned by people whose incentive structure is orthogonal to civic outcomes.

Either path ends with a civic infrastructure that looks different from today’s. The question is whether it gets there by design or by accident.