Techno-Feudalism vs Digital Sovereignty

A serious argument holds that platform economies have stopped being markets and started being something older. Assessing the claim, the objections to it, and what is genuinely new about algorithmic power.

Written by Ottavio Malatacca · Published 2026-07-15 · Updated 2026-08-30

Techno-feudalism is a deliberately provocative term, and the provocation is doing analytical work. The claim is that the largest platform companies have stopped behaving like firms competing in markets and started behaving like landlords extracting rent from territories they own, and that this constitutes a change of kind rather than of degree. The argument is contested and the label is arguably wrong. What it points at is real and badly served by the existing vocabulary.

The claim

Yanis Varoufakis's version, set out in Technofeudalism (2023), runs as follows. Capitalism is organised around profit, which is earned by producing goods and services and selling them in competitive markets. Feudalism was organised around rent, which is extracted by controlling access to land that others must use.

A dominant platform, on this account, resembles the second. It does not primarily produce the goods sold through it. It owns the space in which transactions occur, and it charges for access — commissions on sales, fees for visibility, a share of advertising revenue. Sellers on a large marketplace are not competitors of the platform in any ordinary sense; they are tenants, whose viability depends on terms the landlord sets and can change unilaterally.

Varoufakis adds a second element concerning labour. A substantial part of the value of these systems is produced by users who are not paid: reviews, posts, ratings, search behaviour and every other interaction that trains the ranking systems. He calls this unremunerated contribution cloud serfdom. The rhetorical stretch is obvious — nobody is bound to the land, and users can and do leave — but the underlying observation is accurate: users generate the asset and hold no claim on it.

The objections, which are strong

Several economists reject the framing, and their objections deserve more attention than the label usually receives.

• Feudal rent was backed by legally enforceable personal subordination. Serfs could not leave. Platform users can and regularly do, at a cost that is real but not comparable.
• Platforms do produce. Logistics networks, search infrastructure, payment systems and recommendation engines are genuine services with substantial capital and operating costs behind them.
• Economics already has a term for this. Rent extraction through control of a bottleneck is monopoly rent, thoroughly analysed since at least the nineteenth century. Invoking feudalism adds drama rather than explanatory power.
• The dominance may be less durable than it appears. The historical record of platform incumbency is mixed, and several apparently unassailable positions have eroded within a decade.

The counter to the third objection is the one worth taking seriously. Classical monopoly analysis concerns a firm that raises prices above the competitive level in a market whose rules it does not set. A platform sets the rules — who is visible, on what terms, under what ranking, subject to what policy — while simultaneously competing inside the space those rules govern. That combination of regulator and participant has no clean equivalent in standard market analysis, and it is what the feudal metaphor is reaching for even if it reaches imprecisely.

What is genuinely new

Setting the label aside, four features distinguish platform power from earlier concentrations.

Rule-setting combined with participation

A firm that owns a marketplace and also sells in it possesses information about its competitors that no ordinary rival could obtain, and controls the ranking that determines whether those rivals are seen. Several jurisdictions have concluded that this is not manageable through conduct rules and have moved toward structural remedies.

Adjudication without process

Platforms make decisions that determine whether a business survives — account suspension, demonetisation, ranking demotion — at enormous scale and largely by automated systems. These decisions frequently arrive without a stated reason, without prior notice, and without a meaningful route of appeal. The function is judicial; the procedure is not. This is the sharpest departure from anything liberal constitutional traditions have equipped themselves to handle, and it is the specific gap discussed in our essay on modern liberalism.

Behavioural surplus

Shoshana Zuboff's argument in The Age of Surveillance Capitalism (2019) is that the raw material of these businesses is behavioural data collected far in excess of what is needed to deliver the service, and that the resulting product is a prediction of future behaviour sold to third parties. Her stronger claim — that the object is behavioural modification rather than prediction — is more contested, and the empirical evidence on how far targeted persuasion actually shifts behaviour is more equivocal than the popular account suggests. The weaker claim is well established and sufficient.

Infrastructural dependency

Three cloud providers underlie a very large share of the world's digital services, including public administration. This is not a consumer market with substitutable options. It is infrastructure, and a state that runs its tax system, health records and identity services on foreign-owned infrastructure has a dependency that is invisible until it becomes acute.

Digital sovereignty: the counter-programme

Digital sovereignty is the umbrella term for state responses, and it covers projects with incompatible aims — which is why the term generates so much confusion.

The democratic version treats the goal as jurisdictional control: ensuring that data about citizens is governed by law they have a say in, that critical infrastructure has domestic alternatives, and that platform decisions are subject to due process and appeal. The European regulatory approach — data protection, digital markets and services legislation, and public cloud initiatives — is the fullest expression of this, and its results so far are mixed: real procedural gains, considerable compliance cost, and limited success at producing viable domestic competitors.

The authoritarian version uses identical vocabulary for the opposite purpose: control over the information environment, mandatory local data storage that facilitates surveillance, and the exclusion of foreign platforms that cannot be compelled to cooperate. Data localisation is the clearest example of a measure that is genuinely ambiguous — it can protect citizens from foreign surveillance or expose them to domestic surveillance, and the same technical requirement serves both.

The third version is corporate rather than governmental: firms reducing dependency on any single provider through multi-cloud architecture and open standards. This is the least discussed and possibly the most effective, since it is driven by commercial self-interest and does not require legislation.

Where this sits on the political compass

Platform power scrambles conventional alignment more thoroughly than almost any other contemporary issue, and the coalitions it produces are genuinely strange. Antitrust action against large technology firms draws support from the economic left, which objects to concentrated corporate power, and from the nationalist right, which objects to platform content moderation. Opposition draws together market liberals defending property rights and progressives worried that state control of speech is more dangerous than corporate control.

The vertical axis is more informative here than the horizontal one. The real question is not whether platforms should be constrained — a broad majority now agrees they should — but who should do the constraining and under what accountability. That is an authority question, not an economic one, which is precisely why a single left-right line cannot represent it.

Conclusion

Techno-feudalism is probably the wrong word. Feudalism involved legal bondage, and platform relationships, however asymmetric, remain voluntary in a sense that matters. But the term has been useful because it forced attention onto a question that market vocabulary was obscuring: what do you call an institution that sets the rules of a space, competes within it, adjudicates disputes in it, and answers to no electorate? Liberal democracies have well-developed answers for governments that do this and for firms that do this. They have no settled answer for something doing both at once, and building one is among the most consequential political tasks of the decade.