It’s time to reframe the data sovereignty conversation
Global Head of Public Sector & Workforce Intelligence
Originally appeared in ComputerWeekly
Once largely a technical concern, data sovereignty has become a question of public trust: who controls public data, who can access it, and under which jurisdiction. As a result, how much governments depend upon foreign platforms to operate critical infrastructure and power public sector departments is being heavily scrutinised.
Government services must now demonstrate appropriate data governance and confidentiality capabilities. In the UK public sector, attention has quickly turned to whether companies like Palantir pass that test. Whether it’s the NHS, Met Police, Armed Forces or intelligence agencies, citizens want reassurance that public institutions retain control of sensitive data and mitigate their potential exposure to extraterritorial laws, such as the US Cloud Act.
It is right that we continually evaluate whether our evolving sovereignty requirements are being met by public institutions. There are credible, genuine, open alternatives. But the conversation has so far focused on two questions: where our data sits geographically, and who owns the technology that holds it? Both matter, but neither alone is a true indication of data sovereignty.
We need to think instead about operational independence and control. A domestic platform that a public body cannot leave may offer data residency, but it does not offer sovereignty. The real test is whether an organisation can continue to access its data, operate and make decisions should it wish to change its relationship with a platform.
Local does not mean sovereign
Calls for greater control often become calls to prefer domestic technology providers. We’ve seen examples where the French government supports replacing the incumbent, Palantir, at the heart of its domestic intelligence services. While in Germany, Schleswig-Holstein replaced most of its Microsoft tools with open-source equivalents, and Bavaria has committed to the same path to build digital independence and avoid foreign control.
But history tells us that bringing systems and data back within national borders is not always enough. The presence of non-European platforms among the members of Gaia-X, the federated European data and cloud ecosystem, shows that the ecosystem isn’t defined by supplier nationality alone. It reinforces the broader point that sovereignty depends on governance, interoperability, and the practical ability to change providers.
Moreover, governments across the world, not just the UK, have deeply embedded dependencies that cannot be simply ripped out and replaced. Migrations require systems to be reconnected, data to be transferred safely, governance to be re-established, and teams to have the necessary skills. The past has shown us that unless those dependencies are understood in advance, switching suppliers can threaten service continuity.
The objective should not be a replacement for its own sake, but rather to ensure it remains possible. That starts with mapping critical dependencies and then deciding what to replace, open up, or diversify, and in what sequence. This is a more practical approach to sovereignty, in which our institutions retain control of their data, maintain its integrity, and connect their systems without compromising confidentiality.
In doing so, they can not only deliver better public services, but move faster, make better decisions, and take advantage of new technologies, like AI, without surrendering control to third-party platforms.
The three exit tests
There are three simple questions that highlight whether a public sector institution has operational independence and control, or if it has become reliant on a platform to store, connect and action insights, and function effectively.
- Can you remove a vendor or leave a deployment without the costs becoming prohibitive?
Data sovereignty begins with the ability to leave. This is not to say that switching platforms doesn’t incur costs or require a rebuild, redesign or migration. These are all reasonable expectations for a digital transformation project that seeks to address the complexities of public sector IT.
But crucially, leaving must remain a viable option. If the financial repercussions make switching practically impossible, then it indicates vendor dependency. This carries risk as it perpetuates lock-in, drives up costs, and ties customer data to a specific platform.
- Can you feasibly take your data or operating model to another vendor – or are you locked in?
The second test highlights who owns and controls the data and the capability built around it.
Public sector institutions should evaluate whether they could still function and provide services if their chosen vendor were sanctioned, went bankrupt or failed to meet their service agreements. Their data and systems should be portable and documented, using open standards where appropriate.
Highly proprietary or closed systems, which require specialised technical knowledge, can create an additional sovereignty challenge when a public body becomes dependent on a particular platform not simply to store its data, but to interpret relationships within it and turn that information into intelligence. The intelligence is outsourced, and so while the institution may technically own its data, it is dependent on a platform to make sense of it. This risk exists whether the supplier is based in the UK, the US or beyond.
- Can you bring in another vendor or third-party to work with your underlying data – or are the systems closed?
The third test is related to whether the technology is sufficiently open and interoperable.
This does not mean everything must be open-source, but it does mean that another vendor should be able to maintain, integrate, or build on the existing system. Data, interfaces, and underlying logic must be accessible during a supplier switch. If this is not possible, the institution lacks operational control and independence, failing the sovereignty test.
Sovereignty is a capability
Ultimately, data sovereignty is not about eliminating those relationships with certain technology providers or rebuilding from scratch. It means ensuring that working together does not cause institutional vulnerability.
This requires a shift in how public sector technology is evaluated. Procurement should not focus solely on deployment speed or capabilities, but on what should happen if circumstances change. Can we leave? Can someone else work with what we have built? Can services continue during the transition?
These questions become even more important as AI is embedded more deeply in public services and the technology that sits between data and decision-making grows more powerful. The institutions that retain control of their data, context and operating models will be best placed to adopt those technologies without surrendering their independence.
The sovereignty debate, therefore, needs to move beyond questions of where our data resides or where a supplier is headquartered. The real test is much simpler. When it matters most, who is in control?