Designing for sovereign AI: A shared foundation from edge to cloud
Microsoft and Intel outline a shared architecture for sovereign AI that lets organizations run sensitive workloads across edge, cloud, and on-premises while maintaining consistent security and control.
AI processing is shifting closer to where data originates and decisions are made, expanding the concept of digital sovereignty beyond storage to include where workloads run, who operates them, and how they remain secure. This evolution requires alignment across silicon, infrastructure, cloud, security, data, and AI to ensure consistent controls as requirements change. Microsoft and Intel are collaborating to provide a flexible foundation that lets organizations place each workload in the right environment and apply the specific controls it demands.
Azure Local on Intel Xeon 6 processors integrates distributed Azure infrastructure with customer-controlled environments, enabling organizations to extend cloud and AI capabilities closer to their data while maintaining uniform security, management, and governance practices. For sensitive workloads, hardware-based protections such as Intel Trust Domain Extensions and hardware-backed attestation help isolate and verify environments, giving organizations verifiable evidence that systems meet their security requirements.
Microsoft Foundry and Foundry Local allow organizations to build and deploy AI models with options for where inference runs, including on-premises or in customer-controlled environments, while Intel Xeon 6 processors with Intel Advanced Matrix Extensions accelerate AI workloads directly on the CPU. Intel OpenVINO supports optimized model deployment across supported infrastructure, and Microsoft 365 Local and GitHub Enterprise Local extend sovereign controls to collaboration, source code, and development workflows.
A sovereign architecture must remain adaptable as regulations, technologies, and operating conditions evolve, requiring alignment across silicon, infrastructure, cloud, security, and AI software to avoid rigidity. Microsoft and Intel emphasize that workload placement and proportional protections—such as isolation, encryption, and attestation—enable organizations to apply AI within defined operational and governance boundaries while preserving a manageable operating model.