ThinkOn Expands VMware Architecture to Support Sovereign Private AI Infrastructure

Cloud infrastructure provider ThinkOn Inc. is expanding its VMware-based operational architecture to deliver private artificial intelligence capabilities, targeting organizations seeking stronger control over proprietary data, operational spending, and regulatory compliance. The Canadian company is structuring its sovereign cloud model to ensure sensitive workloads remain subject to domestic legal oversight and engineering management.

ThinkOn's sovereign cloud framework serves as the underlying technical foundation for Canada's Sovereign AI Platform, which supports thousands of federal employees across 23 government departments. Speaking in an interview first reported by SiliconANGLE during the VMware Explore conference, ThinkOn founder Craig McLellan noted that public sector agency and enterprise interest in dedicated AI systems has climbed substantially as organizations re-examine public cloud expense variability and data protection rules.

McLellan stated that inbound customer inquiries reflect a clear market pivot toward localized infrastructure. Despite operating as a modestly sized firm, ThinkOn receives between three and four operational inquiries per week, with roughly half focused specifically on private AI deployments.

To manage these AI workloads, ThinkOn utilizes VMware Cloud Foundation as a unified control plane across both its established private cloud offerings and its emerging artificial intelligence services. The firm pairs this unified software platform with Canadian data center facilities, domestically based engineering staff, and audited compliance workflows to meet strict jurisdictional and supply-chain criteria.

Addressing the technical definition of sovereign infrastructure, McLellan emphasized that physical and legal access rights remain paramount. He noted that true sovereignty depends on whether an organization maintains complete, unrestricted access to the underlying computing hardware hosting its applications.

Financial predictability is also encouraging organizations to move away from variable, consumption-based public cloud pricing models in favor of private AI infrastructure. ThinkOn's deployment strategy centers on aligning graphics processing unit resources directly to individual workload requirements rather than default provisioning.

McLellan highlighted that many enterprise workloads perform effectively on prior-generation accelerators, such as Nvidia's A-series processors, without requiring costlier H-series or B-series silicon. He added that deploying appropriate hardware tiers offers direct financial benefits to clients while maintaining operational consistency through a single management plane.