AI Sovereignty: A Qualitative Model of Strategic Competition as AI Becomes an Instrument of National Power

Abstract: AI sovereignty is the extent to which a nation independently controls its AI technologies. The race toward ever-more-sophisticated frontier AI models is of increasing strategic importance, with nations considering how AI might improve their economic situations, competitive advantage, and overall national power. However, the costs of AI sovereignty are enormous, and we lack definitions and conceptual models to navigate evolving AI sovereignty dynamics. We address this gap with definitions relevant to AI sovereignty, along with a first-of-its-kind qualitative model that incorporates micro, meso, and macro contributors. Model-based qualitative forecasts highlight competitive dynamics and evolving potential for AI-driven national power. The model identifies key leverage points that nations can use to enhance their own growth or degrade an adversary's, including consideration of accelerators, electricity, water, data sets and skilled workforce. These leverage points can be activated at strategic and operational levels through both direct kinetic actions, such as Iran's targeting of data centers with drones, and indirect non-kinetic effects including cyber, space, information, economic coercion and diplomacy. If our assumptions and hypotheses are valid, this strategic competition may come to define how nations improve their economic situations, competitive advantage, and overall national power in the 21st Century.

• The paper introduces a qualitative system dynamics model that formalizes AI sovereignty as a determinant of national power.
• It details micro, meso, and macro scales to capture resource limitations and strategic adversarial actions in national AI development.
• The study proposes quantitative metrics, such as zettaFLOPS and an AI Sovereignty Index, to forecast competitive advantages and security risks.

AI Sovereignty as a Systemic Driver of Strategic National Competition

Overview and Problem Framing

This work, "AI Sovereignty: A Qualitative Model of Strategic Competition as AI Becomes an Instrument of National Power" (2606.07245), rigorously formalizes the notion of "AI sovereignty" and provides a micro/meso/macro system dynamics qualitative model capturing its evolution and its consequences for strategic competition among nation-states. The study foregrounds how agentic AI capabilities are rapidly transitioning AI from a market-driven technology into a geostrategic instrument—where independent control over models, infrastructure, and requisite resources stratifies national power.

Clancy and Naugle present their model as a foundational scaffold for systematic inquiry, distilling key variables, feedbacks, and leverage points. They emphasize both operational constraints (hardware, energy, water, skilled workforce, and datasets) and national policy decisions, and explicitly map adversarial behaviors (e.g., targeting data centers kinetically, or supply chain degradation) as strategic levers. The qualitative forecasts and scenario reference modes probe regime shifts in the landscape of international competition as AI capabilities mature.

Definitions and Analytical Framework

AI sovereignty is precisely defined as the degree of independent national control over AI technologies. This encompasses not only the possession of frontier models and physical infrastructure but also control over critical resources, physical and human capital, and the ability to innovate new model generations without foreign dependencies.

The system dynamics model is architected across three interlocked scales:

• Micro (Cabinets/Accelerators): The minimal, physically-constrained units (AI server cabinets), defining hardware resource densities, cooling requirements, and energy/water draws.
• Meso (Data Centers): The aggregation mechanisms, provisioning compute clusters, and serving as both concentration points for national capability and targets for strategic action.
• Macro (National AI Power): The summation of all operational (delivered) zettaFLOPS, number and vintage of deployable agentic AI models, and the depth and distribution of skilled workforce.

The model's constructs echo with established measures of military power, especially airpower generational analysis, where the gap in sovereign generational capability directly determines competitive advantage. In analogy, the authors propose that mere access to foreign models or leased infrastructure is an unstable path; true power resides in sovereign innovation and operation.

Key Results and Model Insights

Resource-Limited Growth and Leverage Points

The causal loop diagrams codify both virtuous cycles (investment in accelerators, workforce, energy, data) and growth constraints (resource gaps, underinvestment, adversarial action). Three levels of resource growth dynamics are identified:

• Easy Growth: Periods where sovereign resources (hardware, energy, water, workforce, data) are abundant, resulting in rapid (S-shaped) increases in capability.
• Limits Activation: National bottlenecks in any resource activate, slowing or stalling AI capability growth; parties must choose between domestic expansion, seeking foreign supplementation, or accepting capability plateau.
• Foreign Dependence: Sourcing compute or AI models abroad mitigates limits but introduces profound security risks and strategic vulnerability to interdiction.

The model distinguishes kinetic actions (e.g., physical attacks on data centers, as seen in the 2026 Iran-UAE incidents), and non-kinetic measures (cyber operations, economic coercion, or information operations targeting AI labor supply, data pipelines, or societal support for infrastructure projects) as principal instruments of competition.

Competitive Dynamics and Arms Race Logic

Strategic competition is formalized by the introduction of adversarial system dynamics, where each actor can seek to augment their own growth or degrade adversary resources at the micro or meso level. The model supports scenario generation, including arms race behavior, dependency traps (where nations or alliances leverage but become reliant on foreign capabilities), and resource denial strategies.

The reference modes support the hypothesis that, as agentic AI matures, there will be stratified use-case regions: some nations will realize hoped-for economic and defense applications (sweeping automation, rapid scientific inference), while others, constrained by resources, will languish in suspense or feared regions, unable to operationalize next-gen capabilities.

Quantitative Metrics for Sovereignty and Power

The authors propose specific quantitative proxies for future simulation:

• National Power in Agentic AI: Total sovereign (delivered) zettaFLOPS, number of sovereign models by generation, breadth across multiple generations, and cumulative model depth.
• AI Sovereignty Index: Percentage of total national AI compute and model capacity under exclusive national (vs. foreign) ownership, expressed as ratios between 0 and 1.

These metrics are intended for scoring national positions and evaluating effects of resource shocks, adversary campaigns, or strategic investments in simulation-based analyses.

Theoretical and Practical Implications

The primary theoretical implication is the reframing of AI, specifically agentic AI, as a core, measurable input to national power, on par with nuclear, air, or cyber capabilities. The operationalization of sovereignty highlights that AI arms races will be governed not just by model innovation, but by the ability to allocate vast physical, energetic, and human capital, and to insulate critical infrastructure from foreign interdiction.

In practical terms, the model predicts the rise of targeted campaigns against data center infrastructure (as empirically evidenced in recent kinetic attacks in the Gulf), and information/cyber denial operations targeting workforce or resource supply chains. It also suggests that domestic policies (investment in workforce, energy/water infrastructure, supply chain security for accelerators, dataset curation and sovereignty) will become foremost levers in national AI strategy.

The forecasted arms race and competitive landscape may induce decoupling/regionalization of AI infrastructure, bifurcation of standards, and the emergence of alliance-based AI sovereignty coalitions. AI dependence relationships could become strategic liabilities, similar to energy or supply chain dependencies of previous eras.

Future Directions

The current model is qualitative and offers a system structure but not scenario-specific quantitative forecasts. Converting this model into a quantitative dynamic simulation is essential for policy analysis, sensitivity testing, and strategy optimization, especially as the parameters (resource costs, scaling curves, workforce development) are in rapid flux. Such extension would enable robust scenario gaming—evaluating strategic shocks, investments, or adversary campaigns—informing national policy, alliance formation, and infrastructure hardening.

The model should also be extended to explicitly incorporate model poisoning and advanced AI security threats as first-class limiting variables, as noted by the authors.

Conclusion

This paper provides a comprehensive, rigorous systems framework for understanding and forecasting AI sovereignty as a dynamic determinant of national power. By integrating physical constraints, human capital, geopolitical competition, and security vulnerabilities, it operationalizes sovereignty as neither binary nor parametric but as an emergent property of policy, investment, and international interaction. The work directly informs strategic planning in national security, industrial policy, and international relations, and lays the foundation for future quantitative modeling and scenario analysis of the evolving AI-powered global order.