From principle to control

UNCTAD’s Digital Economy Report 2021 treats sovereignty as a broader question of legitimacy, authority, and power over data and its use. It examines localization and cross-border-flow rules as particular policy mechanisms. It also recognizes multiple legitimate reasons for governments to regulate flows, including privacy, security, and economic-development objectives. [3]

For an institution, the distinction quickly becomes practical. A critical dataset may be stored domestically. That does not answer who holds administrator privileges, whether offshore personnel can support the environment, or which subcontractors can reach it. Nor does it say where analytics or AI inference occurs, or what information a provider retains. Nor does location explain what happens when the technology relationship needs to change. A useful sovereignty model therefore has to operate across the digital stack.

The Sovereignty Control Plane

Five controls translate sovereignty into decisions.

Classify. What type of data or workload is involved? Public, internal, personal, sensitive, strategic, and critical information should not automatically carry identical requirements; classification allows controls to be proportionate to consequence.

Place. Where may the data or workload be stored, processed, backed up, and recovered? This is where residency and processing-location requirements become architectural and contractual decisions.

Access. Who can technically or operationally reach the data and systems? This includes government users, technology providers, cloud administrators, support teams, subcontractors, applications, and automated processes.

Move. Under what circumstances may information cross an institutional or national boundary? The answer should reflect classification, purpose, authority, receiving environment, and security requirements.

Exit. Can the institution change technology direction while preserving data, continuity, and operational knowledge? Exit includes portability of data and configurations, workload migration, transition support, and the practical ability to continue a critical service.

Classification, placement, access, movement, and exit apply across data, cloud and compute, AI models, and the digital service itself. Sovereignty therefore does not reside in one database or one data center.

Control does not require owning every layer

The five controls are not independent of each other, and recognizing that is what turns the plane from a checklist into a design tool.

Where a residency requirement has been set, the practical question is which combination of the remaining controls actually delivers what the requirement was intended to protect. Four things contribute to the same protection that placement was reaching for. Encryption keys held under institutional control. Administrator access restricted to named domestic personnel. Processing locations constrained by contract rather than by assurance, and an exit that has been tested rather than merely written.

The substitution does not run both ways with equal force. Domestic placement does not compensate for weak access and exit controls. A dataset held inside the country, administered by personnel an institution cannot identify, on a platform it cannot leave, is located correctly and controlled poorly.

This is a statement about control design under a given policy, not about what the policy should be. Where a jurisdiction has set a residency rule, the rule is the requirement. The useful executive discussion is about which controls make the requirement real, and which have been assumed to follow from location and do not.

This distinction matters as African governments combine public infrastructure, domestic providers, regional platforms, and global technology ecosystems. A government may intentionally use an external cloud provider for scale, security capability, or access to advanced technology. It may use a global AI model because building an equivalent capability independently would be uneconomic. Those choices do not automatically weaken sovereignty. The executive question is what control the architecture and the contract preserve: administrative access, encryption arrangements, permitted processing locations, audit rights, subcontractor transparency, portability, and continuity. This moves the sovereignty discussion away from a binary choice between building locally and relying externally, and focuses leadership on the dependencies it is prepared to accept.

Exit is a strategic control

Provider choice receives significant attention during procurement; the ability to make another choice later often receives less. The European Union’s Data Act provides a useful international comparator. Its cloud-switching requirements seek to remove obstacles to moving between data-processing providers, and include measures on contract transparency, export of data, and switching functionality. [4] The relevance for Africa is the underlying control principle, not adoption of the European regulatory model. A public institution that can choose a provider today but cannot practically leave later has transferred strategic discretion. That transfer was rarely visible in the original procurement decision. It matters more each year, as data, compute, and models become harder to separate from one another.

Retaining a control has a price. For some workloads it will mean higher cost, slower delivery, or a narrower set of technology options. An institution not prepared to pay any of those for a given dataset has not yet decided that the control matters.

Make sovereignty reviewable

For a critical dataset, workload or digital service, five questions can make the policy operational.

1. How is it classified? Use consequence to determine the strength of the remaining controls.
2. Where may it reside and be processed? Translate policy into architecture and contract requirements.
3. Who can access or administer it? Include indirect provider and subcontractor access.
4. Under what conditions may it move? Make institutional and cross-border movement explicit.
5. How do we change direction? Define portability, continuity, and exit before dependency becomes difficult to reverse.

This creates a more useful executive discussion than asking whether a solution is “sovereign” in the abstract. It asks which decisions remain under institutional authority and which dependencies have been consciously accepted.

Sovereignty becomes operational when leaders can explain what they control, what they deliberately delegate, what they permit to move, and how they retain the ability to change direction.