The Joint Research Centre of the European Commission publishes a disturbing finding in 2026: administrations across the Union still poorly understand what their digital dependence truly implies. The report identifies four simultaneously failing levers—markets, infrastructures, skills, governance—and calls on administrations to identify, evaluate, and manage the dependencies and vulnerabilities that constrain their autonomy. Administrations have built dependencies through the accumulation of successive purchasing decisions.

The essentials

  • The capacity to reduce dependencies and diversify suppliers is one element of digital sovereignty, among others.
  • The Joint Research Centre of the European Commission (2026) identifies four simultaneous dependencies in EU administrations: markets, infrastructures, internal skills, and governance.
  • The report underscores the necessity of coordinated action across different areas of digital sovereignty.
  • The absence of reversibility clauses can increase dependencies with each contract renewal.
  • The direction of public procurement policies will largely determine whether supplier concentration or diversity increases.

Digital sovereignty in practice

The phrase has circulated through every European capital for roughly a decade. It has produced strategies, white papers, national plans. It has not always produced administrations capable of switching software.

The distinction matters. When an administration has the technical and contractual capacity to consider migration, its dependence remains manageable. Closed formats, the absence of internal capacity, and the absence of exit mechanisms can increase dependence, without on their own exhausting the evaluation of sovereignty.

The Joint Research Centre report makes this distinction with unusual clarity in institutional literature. Digital sovereignty comprises organizational capacities as well as issues of dependence on markets, infrastructures, and suppliers.

This capacity rests on four elements that the report examines separately but which form a single problem: digital public procurement, which creates or prevents dependence depending on how it is drafted; infrastructures, which lock in or liberate depending on whether they rest on open or proprietary standards; internal skills, without which an administration can neither evaluate its service providers nor consider migration; and governance—that is, decision-making procedures that allow, or do not allow, action when a dependence becomes problematic. Administrations must identify, evaluate, and manage the dependencies and vulnerabilities that constrain their autonomy. European administrations have built dependencies through the accumulation of purchasing decisions.

Four dependencies that reinforce each other

The problem of digital lock-in is not new. What makes it structural is the interaction among its components.

The absence of portability guarantees can increase the risk of dependence, but it does not alone prove the exclusivity of the service provider’s format control. After two years, internal staff have ceased to understand the underlying architecture, having failed to maintain externalized expertise. At renewal time, administrations often have reduced resources to evaluate other solutions. The service provider takes this into account in renewal conditions.

This cycle is mechanical. It involves no particular malice on the part of suppliers: they optimize their long-term revenues, which is their role. Public procurement decisions can contribute to reducing dependencies, notably through market diversification.

The Joint Research Centre report emphasizes that these four interconnected action areas require coordination and cannot be treated in isolation. Data portability and human capacity can be complementary: data that is theoretically portable requires trained staff to actually be moved. Staff training and public procurement reform are complementary: skills alone are insufficient if contracts do not provide room to exercise them. All four areas must be addressed in a coordinated manner to strengthen autonomy and resilience.

Gauchet’s analysis applied to this issue

Marcel Gauchet, in his analysis of neoliberalism, develops a thesis directly applicable to this matter: modern democracies lose their autonomy not under external pressure from a power that constrains them, but by delegating it from within, through the very logic of their free choices. The market, in promising efficiency, ends up occupying spaces that political deliberation had abandoned.

Applied to digital administrations, this reading is precise. Procurement decisions generally prioritized criteria other than contractual reversibility. Purchasing choices often favored short-term efficiency or cost gains. The accumulation of these decisions created structural dependencies. The loss of autonomy is linked to dependencies and vulnerabilities that administrations must identify and manage.

Liberal economists like Jean Tirole or Thomas Philippon defend a competing reading: the problem lies in the absence of competition, which should make any delegation reversible. In a sufficiently open market, lock-in would be a losing strategy for the supplier; customers would leave. If administrations remain captive, it is because the market for sovereign solutions is not sufficiently developed.

These two readings do not exclude each other. They identify different causes of the same dysfunction and call for partially different remedies. The first points to the necessity of regulation that protects public decision-making autonomy, by imposing standards and exit clauses. The second points to the necessity of developing a sufficiently dense European market for digital solutions so that competition can play its role. The Joint Research Centre report indicates the complementarity of both approaches: competition and regulation must work together to resolve market failures and supply shortcomings.

The debate remains open on the hierarchy among these levers.

Supplier nationality, an insufficient indicator

The European debate on digital sovereignty has long been dominated by the question of service provider nationality. The idea is intuitive: a European cloud guarantees European sovereignty. It is false in both directions.

The geographic origin of the supplier does not provide immunity against lock-in risks linked to proprietary formats. Supplier nationality alone is insufficient to evaluate dependence and lock-in risks. European origin can create false confidence and lead to neglecting essential contractual guarantees.

The inverse is also true. Some European administrations use services from non-European suppliers with a level of portability and internal control that genuinely allows them to consider migration. Their dependence is manageable.

The report organizes its analysis around four action areas: people, markets and products, infrastructures, and governance. Are the data in an open, exportable format? Do staff have the skills to manage migration? Does the contract provide an exit procedure with reasonable timelines and costs? The report calls for managing dependencies and vulnerabilities across several coordinated action areas.

This clarification has direct practical consequences for public procurement policy. It shifts the criterion for selecting service providers from their nationality to their architecture—open or closed—and to contractual exit conditions. This is a change of doctrine that implies reforming public procurement teams, who do not always have the technical skills to evaluate these criteria.

Brazil has undertaken comparable reflection on digital actor regulation, by holding platforms accountable for their algorithms with similar logic: what matters are concrete effects on institutional autonomy, not the origin of operators.

The two scenarios for 2040 and today’s decisions

The 2026-2040 horizon identified by the OECD in its Digital Government Outlook 2026 allows us to see what current decisions will produce in the long term. Two trajectories emerge, not as forecasts, but as logics at work if nothing changes or if something changes precisely.

A reinforcement scenario would include reversibility requirements in public procurement regulation. Public contracts could require defined portability criteria and exit procedures. Migrations remain costly and technically complex; they always will be. But they become possible, and their possibility alone is sufficient to shift the balance of power in renewal negotiations. Effective competition requires that customers have genuine alternatives and can consider changing suppliers.

A status quo scenario would assume the absence of significant changes in procurement practices. Contracts continue to be evaluated primarily on immediate cost and functionality. Absent changes in procurement practices, dependencies tend to increase with each contract renewal. In ten to fifteen years, a few actors dominate entire segments of the European digital state—not necessarily non-European actors, but actors who, whatever their nationality, set the terms of service. Services would function but administrations’ capacities to adapt them would diminish.

The difference between these two trajectories lies in decisions that appear technical but are profoundly political: the clauses of public tenders, staff training, audits of dependency. None of these decisions is spectacular. They are cumulative.

Two signals allow tracking which of the two trajectories is actually engaged: the share of new cloud contracts including a verifiable reversibility clause, and the number of actual migrations from one supplier to another in major European administrations over a three-year period. These figures do not exist today in consolidated form at the European level; their absence is itself an indicator of the degree of priority accorded to the issue.

Available measures without waiting for Brussels

European regulation advances slowly. The European Interoperability Framework has existed since 2017. The European Data Act of 2023 imposes portability requirements in the private sector. Their translation into concrete public procurement practices remains uneven across member states.

Several administrations are taking action without waiting. Some Nordic member states have begun auditing their critical dependencies and producing mappings of their systems, distinguishing functions where migration is feasible in the short term from those where lock-in is too deep to be resolved quickly. This is elementary administrative hygiene that, in many cases, had simply never been conducted. Generative AI is also disrupting administration practices, making the question of internal skills more urgent: without staff capable of understanding the tools, portability promises remain formal.

Staff training is the other lever that administrations can activate without waiting. An administration whose technical teams understand their systems’ architecture is better positioned to negotiate, to evaluate a competing offer, and to manage migration when it becomes necessary. This competence is not created in three months, but it develops progressively if deliberately prioritized.

There is a logic here close to what economist Philippe Aghion describes for businesses: the capacity to innovate and adapt depends less on available resources than on the quality of internal decisions and the competition that stimulates them. An administration that maintains its internal skills remains a credible public buyer, capable of exercising real pressure on its suppliers. An administration that has entirely outsourced them has abandoned its negotiating position of strength.

European public decision-makers have a straightforward question to ask, though difficult to answer: how many of their critical systems could migrate to another supplier in eighteen months, without major service interruption? This question calls for a numerical answer, by sector and by administrative level. It does not yet have one.


Sources

  1. Joint Research Centre of the European Commission, Advancing EU Digital Sovereignty in Public Administrations, 2026, https://publications.jrc.ec.europa.eu/repository/handle/JRC146997
  2. Marcel Gauchet, “Le néolibéralisme ne gagnera pas,” Sciences Humaines, https://www.cairn.info/revue-sciences-humaines.htm
  3. OECD, Digital Government Outlook 2026, Organisation for Economic Co-operation and Development (no guaranteed URL)
  4. European Commission, European Interoperability Framework, 2017, European Commission, Directorate-General for Informatics
  5. European Commission, European Data Act, 2023, Official Journal of the European Union