A recent publication by 8ra puts a precise question at the centre of Europe’s AI strategy: how do you scale artificial intelligence when the underlying data cannot move?
In sectors such as healthcare, industrial automation, and critical infrastructure, data residency requirements, cross-border compliance regimes, and IP protection constraints make centralised data aggregation either legally impermissible or operationally unviable. The conventional approach of pooling training data in hyperscale cloud environments is, in these contexts, simply not an option.
E-Group’s response, developed within the 8ra Initiative under its IPCEI-CIS project FedEU.ai, is to invert the paradigm entirely. As Antal Kuthy, Founder and CEO of E-Group, and Ákos Tényi, who leads the company’s Federated AI product line, describe it: rather than routing data to centralised compute, federated learning brings the model to the data. Private datasets, AI processing services, and insight generation are architecturally decoupled; each can be deployed independently across cloud, edge, or on-premises environments, with privacy-preserving techniques applied throughout the training and inference pipeline.
Antal Kuthy, Founder and CEO, E-Group and Ákos Tényi, Head of Federated AI Photo: © 8ra
This architecture is enabled by the Multi-Provider Cloud-Edge Continuum: a distributed infrastructure layer that allows services to be deployed across different national environments, cloud providers, and regulatory jurisdictions without requiring data to leave its point of origin. Within this context a project led by E-Group, Ambiti8n pilot, advances a complementary concept: Component Readiness Level. Where traditional Technology Readiness Level metrics assess whether a component is built, Component Readiness Level assesses whether it can be orchestrated, federated, and activated horizontally across a multi-provider ecosystem.
The strategic argument, as articulated in the 8ra article, is deliberately pragmatic. European competitiveness in AI will not be achieved through centralisation alone. It will depend on the ability to make federated intelligence deployable at scale; allowing organisations to retain data sovereignty while participating in shared, cross-border learning environments. Federated AI, in this framing, is not a constraint to work around. It is the architecture that makes data sovereignty and AI capability mutually reinforcing.
Source: https://www.8ra.com/blog/when-data-is-distributed-ai-should-work-that-way-too/
