As governments accelerate AI adoption, the debate around sovereign AI has shifted from where technology is built to who ultimately controls it. The recent launch of the Sovereign AI Power Index reflects this growing focus on national AI capability, resilience and governance. While the Index measures compute, models and data centres, these metrics do not tell the whole story. Sovereignty is a property of governance, not simply an inventory of infrastructure.
True sovereign AI is about control, not simply ownership. Governments must understand the systems they deploy, govern how they operate and ensure AI-driven decisions can be explained, audited and challenged when necessary. Regardless of who supplies the technology, the defining question remains: are governments, not vendors, in control?
Sovereignty requires more than ownership
The debate around technological sovereignty often centers on ownership and location. While these factors matter, true sovereignty depends on governance, visibility and oversight.
Domestic hosting offers value, but only if governments retain the ability to understand, oversee and adapt systems over time. Equally, public bodies can work with global technology providers while maintaining sovereignty by establishing strong governance, transparency and accountability. For example, Microsoft’s “Bleu” and “Delos” are marketed as sovereign clouds while Microsoft retains the codebase, update cycles and key mechanisms. Servers moved; control did not.
As AI becomes embedded in public-sector operations, sovereign cloud does not equal sovereign decisions. Governments need to understand how decisions are made, identify risks and ensure AI continues to support public priorities.
Trust must be built into public-sector AI
Public trust will determine the long-term success of AI in government. Citizens expect technology used in essential services to operate responsibly, transparently and fairly. That trust depends on accountability. People need confidence that AI-assisted decisions can be understood and reviewed, particularly when they affect access to public services or resources—such as when a system flags a traveller at a border or selects a taxpayer for enquiry.
Explainability and auditability are therefore essential. Public bodies must be able to understand why a system produced a recommendation, trace decisions back to trusted information and apply human oversight whenever required. These capabilities help ensure AI supports better decision-making without diminishing accountability.
Furthermore, AI systems are only as reliable as the data that underpins them. Because public-sector information is often fragmented across departments, building strong data foundations is critical to producing dependable outcomes.
Competition strengthens digital sovereignty
A resilient technology environment depends on choice. Competition, interoperability and open ecosystems enable governments to benefit from innovation while maintaining long-term flexibility.
Interoperable systems reduce the risk of critical public services becoming locked into a single platform or vendor. They also allow governments to combine global innovation with domestic priorities, maintaining control over how technology is implemented, governed and improved over time.
Resilience is equally critical. Recent drone and missile strikes affecting data-centre infrastructure in the Middle East highlight how quickly concentrated infrastructure dependencies become operational vulnerabilities. Recovery can extend over months, reinforcing the need for diversified infrastructure, interoperable systems and the ability to move workloads when circumstances change.
The strongest sovereign AI strategies will combine access to world-leading technology with clear principles around transparency, accountability and choice.
Responsibility remains with governments
As AI becomes more influential in public decision-making, governments must retain oversight of how systems operate and affect citizens. That means establishing clear governance, setting expectations for transparency and ensuring decision-makers have the information needed to evaluate AI-driven outcomes.
Public institutions should always be able to answer three fundamental questions:
- How was this decision reached? Â
- What information informed it? Â
- Can the system be reviewed, challenged or adjusted as circumstances change?
Maintaining the ability to answer these questions is essential to preserving public confidence. Ultimately, sovereign AI is as much a governance challenge as a technical one. It is not defined by where technology is built or who owns it, but by whether governments retain the authority to understand it, govern it and remain accountable for the outcomes it enables. In the age of AI, sovereignty is measured not by possession, but by control.