The Jurisdictional Stack: AI Sovereignty and the New Geopolitics of Compute

Recent reporting indicates a significant international trend towards establishing sovereign AI capabilities, a phenomenon described as a multi-hundred-billion-dollar 'gold rush'. This imperative is driven by a state-level desire to control national data, infrastructure, and the core logic of decision-making systems. Yet, the concept of AI sovereignty is more complex than simple data localisation. It involves asserting control over a complete 'jurisdictional stack'—from the physical cloud infrastructure to the application layer where AI-driven operations are executed.

This re-architecture of national technical capacity is creating new forms of geopolitical alignment and dependency. The immense capital and research required to develop and maintain cutting-edge AI, particularly foundational models, means that few nations can achieve true autarky. The result is often a strategic choice of technological patronage, where a state aligns with a hyper-scale cloud and AI provider. Reporting this week of a planned US$10 billion investment by Microsoft in its Middle East cloud and AI footprint exemplifies this dynamic. Such investments are not merely commercial; they establish technological spheres of influence, embedding a provider's architecture, standards, and legal jurisdiction deep within a client nation's critical infrastructure.

The Layers of Control

The pursuit of sovereignty manifests at distinct layers of the technology stack, each presenting unique challenges and intelligence surfaces.

• —The Physical and Virtualisation Layer: Control begins with the physical location of data centres and the cloud environment. A nation may build its own infrastructure, a costly and complex undertaking, or partner with a provider that offers a region-specific or private cloud solution. This choice dictates fundamental dependencies concerning hardware supply chains, security protocols, and legal susceptibility to foreign government requests.

• —The Model Layer: Reliance on foundational models trained predominantly in one cultural and linguistic context, such as the United States, poses a strategic risk. States now seek to develop or procure models that reflect national language, culture, and values. The discourse around 'confidential AI' speaks to this need, focusing on building trusted and verifiable computational foundations for models that handle sensitive state or commercial data.

• —The Application Layer: This is where sovereign control has its most tangible effect. It involves the deployment of AI in defence, intelligence, critical infrastructure management, and public services. The ability to audit, direct, and trust these applications without foreign interference is the ultimate objective of a sovereign AI strategy. As political discourse in states like Brazil shows, the term 'digital sovereignty' itself is malleable, interpreted differently by various political actors to mean anything from strict protectionism to regulated market access.

For intelligence and security organisations, this evolving landscape demands a new analytical framework. Assessing a state's strategic posture requires mapping its dependencies across this entire jurisdictional stack. Who provides the compute? What are the origins and architecture of its core models? What contractual and legal frameworks govern these relationships? The answers to these questions provide a more accurate and predictive map of national capability and intent than traditional metrics alone.

Hankevahti Watch

Strategic ambition is most reliably observed at the point where intent becomes expenditure. The development of a national AI capability, whether through domestic build-out or foreign partnership, generates a clear and persistent trail of procurement signals. Tenders for data centre construction, cloud service contracts, AI software licensing, and specialist consultancy services are leading indicators of a national strategy in motion. Monitoring this signal plane is therefore a foundational discipline in modern strategic intelligence.

Reporting from the US Defense Logistics Agency (DLA) on its use of AI to accelerate procurement timelines highlights a concurrent trend: AI is not only the object of procurement but increasingly the tool of it. This creates a feedback loop where the efficiency and scale of procurement itself become a reflection of a state's technological sophistication.

CARIO's Hankevahti service is engineered to operate on this principle. By systematically monitoring and structuring public procurement and project intelligence, it provides an empirical basis for mapping strategic investments and supply chain dependencies. The underlying methods—large-scale data aggregation, entity resolution, and AI-driven analysis—are central to CARIO's all-source approach. This same core capability, which powers our national security analysis, is also productised within our commercial ecosystem, such as in the TajuLeads platform, which applies procurement intelligence tradecraft to private-sector lead generation in the Nordics.

This illustrates a core doctrine at CARIO: an operational portfolio of commercial services like Taju and Hankevahti provides a persistent, real-world aperture for observing systemic economic and strategic activity. The intelligence discipline of tracking procurement signals is not merely theoretical; it is an active, operationalised capability.

Conclusion

The global drive for sovereign AI is redrawing the map of international power. It creates a new substrate for competition and cooperation, built on the exchange of compute capacity, model access, and architectural control. For states, the challenge lies in balancing the need for sovereign control with the reality of technological dependency. For the intelligence community, the imperative is to develop the analytical tools and methodologies required to understand this new, multi-layered jurisdictional stack. Success will belong to those who can see the system behind the signal, fusing technical analysis with the observable, foundational evidence of procurement intelligence.