An AWS outage disrupted thousands of services in 2025. Earlier that year, US sanctions reportedly affected the ICC prosecutor’s email access. Different causes, same lesson: critical European services rely on infrastructure beyond Europe’s control. Geopatriation can reduce that dependency.

Digital sovereignty: why Luxembourg should care

DORA highlighted Luxembourg’s dependence on non-European technology providers and the resulting sovereignty and concentration risks across cloud, payments, financial data, software and AI. Exit strategies remain largely theoretical, making rapid geopatriation difficult.

These risks go beyond data location and remain even when data is stored in Europe. AWS, Microsoft and Google account for more than 60% of the global public-cloud market. Under the CLOUD Act, US authorities may access data controlled by providers subject to US jurisdiction, regardless of where it is stored. European safeguards can mitigate this exposure but not remove the dependency on foreign legal frameworks.

To reduce this risk, Europe already offers alternatives through several European cloud providers (OVHcloud, Scaleway, StackIT) and sovereign cloud providers (Bleu, S3NS, Clarence). Initiatives such as Gaia-X, IPCEI Cloud and the EU Digital Identity Wallet are further expanding this ecosystem.

The challenge is no longer a lack of European sovereignty initiatives, but their coordination and execution at scale.
Hector Dumas

Hector Dumas, Business Consultant, BearingPoint

Geopatriation: reallocate, don't retreat

Geopatriation (the selective relocation of workloads to European sovereign clouds, regional providers or private infrastructure) can strengthen digital sovereignty and resilience. Gartner ranks it among its top technology trends for 2026 and expects adoption among European and Middle Eastern enterprises to rise from less than 5% in 2025 to more than 75% by 2030.

Geopatriation is not technological isolation or an anti-American stance. It uses a risk-based approach to identify which workloads require greater sovereignty and resilience and which can continue to benefit from public-cloud innovation.

Sovereignty by design: turning strategy into action

Cloud strategies are now common and largely driven by Client Identifying Data (CID) exposure. What is often missing is a workload-level approach translating sovereignty requirements into architectural decisions. BearingPoint observed three operating models in the market

·       Fully on-premises (rare): Maximises control, but limits access to cloud and AI innovation at scale.

·       Opportunistic adoption (most common): Cloud and AI decisions are made on a case-by-case basis in function of the immediate need, without a consistent workload-level framework.

·       Risk-based architecture (emerging): Applies a coordinated approach to data classification, cloud tiering and AI tool selection, matching each workload with the right environment and technology according to the institution’s risk appetite.

This risk-based model provides the foundation for embedding sovereignty at the workload level, through five principles:

·       Classify workloads, not providers: Assess each workload (e.g. KYC, TA register, reporting, AI, archiving or payments) based on criticality, data sensitivity, legal exposure and business continuity requirements. Use the service portfolio, sourcing readiness assessment and Make-or-Buy / In-Out-Hybrid analysis to determine the appropriate sourcing model.

·       Set differentiated sovereignty targets: Based on workload criticality, use public cloud for lower-sensitivity workloads and private, sovereign or on-premises environments where loss of control or availability could cause material harm. Benchmarking, business cases and market screening can identify viable options without making sovereignty the sole criterion.

·       Look through the supply chain: Assess subcontractors, shared providers and concentration up to the fourth-party level, not only direct providers. Confirm data residency, operational and legal control, and substitutability. Translate these requirements into RFPs, vendor evaluations and contracts, and assess gaps in existing agreements.

·       Make exit capability real: Keep data and applications portable, maintain alternatives and internal skills, and regularly test exit and recovery plans. Reversibility should be designed upfront, supported by a capable Retained IT Organisation and SIAM model.

·       Use European rails, stay open: Use European cloud, payment, identity and settlement infrastructures where they strengthen sovereignty, while retaining trusted global technologies where risks are controlled. Transition progressively, prioritising workloads by sensitivity, exposure and existing safeguards rather than pursuing a wholesale move from global providers.

The objective is not to maximise sovereignty across the entire IT estate, but to determine where technological independence is justified. BearingPoint's sovereign infrastructure offering illustrates how this approach can translate into technology choices for critical workloads. The appropriate model depends on risk exposure, workload criticality, regulatory requirements and economics.

The objective is not sovereignty everywhere, but sovereignty where dependency matters most.
Nicolas Mansuy

Nicolas Mansuy, Senior Business Consultant, BearingPoint