Microsoft Adds Support for Fully Disconnected Operations & Large AI Models to Sovereign Cloud

Microsoft has expanded its Sovereign Cloud portfolio with new capabilities that allow organisations to run infrastructure, productivity workloads and large AI models in fully disconnected environments, the company said on February 24.

The update introduces Azure Local disconnected operations, Microsoft 365 Local disconnected, and expanded Foundry Local support for large multimodal AI models. 

The offerings are intended for governments, regulated industries and enterprises that require operations to continue without connectivity to public cloud services.

Douglas Phillips, President and Chief Technology Officer of Microsoft Specialised Clouds, said organisations are reassessing how they deploy infrastructure and AI under tighter regulatory conditions.

“As digital sovereignty becomes a strategic requirement, organisations are rethinking how they deploy critical infrastructure and AI capabilities under tighter regulatory expectations and higher risk conditions,” Phillips said. According to him, Microsoft’s approach enables customers to operate “securely, independently and on their own terms.”

Azure Local’s disconnected operations, now available, allow customers to run mission-critical infrastructure with governance and policy controls without cloud connectivity. Management, policy enforcement and workload execution remain within customer-operated environments. Microsoft said organisations can deploy and govern workloads locally using Azure-consistent tools and policies without depending on continuous connection to public cloud services.

Microsoft 365 Local disconnected enables core server workloads, including Exchange Server, SharePoint Server and Skype for Business Server to run inside a customer’s sovereign boundary on Azure Local infrastructure.

Microsoft said these services will be supported through at least 2035, allowing teams to communicate and collaborate within the same controlled environment as infrastructure and AI workloads.

Foundry Local now supports larger multimodal AI models running directly on customer-owned hardware inside sovereign environments. The system integrates with Azure Local and supports GPU infrastructure from partners, including NVIDIA. 

Microsoft said customers will be able to run local AI inferencing and APIs within their own data boundaries without external connectivity, with deployment and operational support provided by the company.

Microsoft said the three offerings together form a Sovereign Private Cloud model that supports connected, intermittently connected and fully disconnected deployments, enabling customers to standardise governance across environments while choosing where workloads run based on mission and regulatory requirements.