By Andrea Falleni
As AI becomes business-critical, organizations must balance innovation with control by reducing dependencies that could undermine resilience and future growth.
As AI moves from pilots to enterprise-wide transformation, leaders face a critical question: can they scale value without sacrificing strategic control? The answer increasingly depends on how they design resilience, flexibility and choice into the technologies that will power AI at scale.
This is why AI has become central to the sovereignty agenda. The Capgemini Research Institute’s recent research finds that 75% of organizations prioritize AI in their pursuit of digital sovereignty, making it the top concern in the technology landscape. It is easier to design flexibility into AI environments up front now, than to retrofit it once systems, providers and processes are embedded.
Today’s AI choices will shape tomorrow’s resilience
The value case is becoming clearer: organizations are already achieving an average return of 1.7 times their AI investment. But scaling AI requires more than successful pilots; it requires resilient foundations from the start.
The risk is already visible in enterprise AI strategies. Forty-three percent of organizations identify AI as a technology layer that could affect operational continuity or resilience. Many also face supplier concentration, foreign dependency and protracted switching timelines. These are exactly the kinds of constraints that become harder to unwind over time.
Designing choice into AI
Digital sovereignty is broader than AI, but its implications are particularly urgent as AI moves into business-critical operations. The issue is no longer only where data or infrastructure sit, but who can access the data, how models are trained, and how easily organizations can move from one platform to another. Global providers will remain essential to innovation, scale and performance; the challenge is to retain the freedom to innovate without becoming locked into another organization’s technology choices.
That requires looking beyond the model itself. AI depends on data, cloud and compute infrastructure, cybersecurity, software, hardware, connectivity and energy. Taking a narrow focus on models alone risks missing the dependencies that could most affect continuity, scalability or control.
In aerospace and defense, for example, AI can support engineering, simulation, maintenance and secure operations. In that context, sovereignty depends on whether the organization retains sufficient control over the data, infrastructure, compute capacity and security layers that allow those systems to operate reliably.
The same principle applies more broadly. The Capgemini Research Institute report shows that 59% of organizations view complete digital sovereignty as unrealistic, while 67% define it as resilient interdependence, rising to 75% in Europe. For AI leaders, the priority is not to own every layer of the stack, but to choose, switch and adapt where AI underpins business-critical operations.
Knowing where control really matters
A proportionate approach is essential. The report recommends “minimum viable digital sovereignty” identifying the smallest set of capabilities, assets and controls that must remain under direct or assured control to safeguard critical operations.
As AI becomes embedded in core business processes, organizations should start by identifying the essential capabilities that must keep running in a major disruption, then map the AI systems, data, infrastructure and providers those capabilities now depend on.
Only then can leaders decide where stronger controls, alternative suppliers, contingency plans or portability are justified and where managed exposure remains acceptable.
The visibility gap makes early action critical. Only 14% of organizations report end-to-end visibility into dependencies across their broader technology ecosystem, underscoring the need to address AI dependencies before key design choices become costly to revisit.
From sovereignty to AI advantage
This is already a boardroom agenda item. Ninety-three percent of organizations worldwide have discussed digital sovereignty at board level, while 67% believe it can enable innovation. For AI, the opportunity is to treat sovereignty not as a constraint, but as a condition for trusted growth.
This is particularly relevant in Europe, where the debate on digital sovereignty is closely tied to industrial competitiveness, governance, trust and responsible technology deployment. These strengths can become differentiators if organizations use them to support AI adoption.
The organizations that harness AI to innovate without becoming dependent on the technologies that enable it, will be those that create competitive advantage and long-term growth.
