Business resilience is becoming increasingly vital in an uncertain world where technology offers both a route to growth and risks that must be managed.
Enterprises undergoing digital transformation can be seriously disrupted by cyber attacks or geopolitical tensions. As companies transition from experimentation to operational deployments, both data and AI sovereignty are now firmly on the boardroom agenda.
In a new CIO webcast, AMD Senior Vice President of Global Public Sector, Thomas Zacharia, explains what the concept actually means, and how IT leaders can harness the benefits of AI sovereignty without blocking innovation.
Understanding sovereignty
In the IT world, the concept of sovereignty is often associated with data. But when it comes to AI, there’s a much wider scope. Organisations must consider the models themselves, their governance, the infrastructure layer, the software and operating environment, and even the people managing it. Above all, it’s about ensuring enterprises have sufficient control over the AI-related capabilities on which they increasingly depend.
“We are early in the phase of a truly disruptive technology,” explains Zacharia. “We should keep our eyes wide open and recognise that there are known unknowns and unknown unknowns that we need to grapple with.”
One way to help future-ready AI sovereignty is through open standards and open-source technology. They can help to improve portability, reduce vendor lock-in risk, and enable organisations to benefit from broad research and developer communities.
However, an open environment doesn’t automatically guarantee success. There’s plenty more for CIOs to consider, such as where to run their workloads. A hybrid cloud approach may offer flexibility, but context is important. Workload placement must be intentional, and based on factors such as sensitivity, data, residency, latency, scale, and cost.
Taking the first steps
Building true AI sovereignty will require CIOs to navigate multiple regulatory mandates without slowing innovation. Yet it is possible to addressdifferent compliance requirements without redesigning the entire system each time.
The key is to generate evidence automatically and continuously, and treat compliance as an architectural input, not a late-stage approval gate.
When it comes to innovation, it’s important to apply governance appropriate to the level of risk exposure. CIOs should first create a risk register to review the potential impact of applications. And provide safe, sandboxed places for people to experiment.
Additionally, by using pre-approved deployment patterns, teams can move quickly without recreating security and compliance processes for every project.
Zacharia continues: “The objective is disciplined speed and adoption of these innovative technologies—enabling people to experiment rapidly, while giving them a predictable path to turn good ideas into trusted operational systems.”
As for first steps, CIOs could begin by focusing on four steps: “understand, design, test, and institutionalise.”
That means first understanding where the best opportunities are for AI to make a meaningful impact. Designing the target architecture; preferably one that’s open and hybrid, with clear provisions for identity, security, observability and portability. Then testing everything, because sovereignty cannot exist only in a policy document. And finally institutionalising the capability.
Most importantly, CIOs must stay focused.
“Sovereignty is not about building walls or owning every layer,” says Zacharia. “It’s about understanding what is critical, retaining control where it matters, and preserving the freedom to act when circumstances change.”
Watch the full webcast and develop your understanding of AI sovereignty.
