New AI Sovereignty Model maps how nations compete for power

Timothy Clancy and Asmeret Naugle have published a seminal paper on AI sovereignty, defined as the extent to which a nation independently controls its artificial intelligence technologies. Their first-of-its-kind qualitative model addresses a critical gap: the lack of definitions and conceptual frameworks to navigate the escalating costs and strategic importance of frontier AI. The model operates at three levels—micro, meso, and macro—and generates qualitative forecasts of competitive dynamics and AI-driven national power. It identifies five key leverage points: accelerator hardware, electricity grids, water resources (for cooling data centers), high-quality datasets, and a skilled AI workforce. These levers can be activated through both direct kinetic actions—such as Iran’s drone strikes on data centers—and indirect non-kinetic effects including cyber operations, space warfare, information campaigns, economic coercion, and diplomacy.

The paper, submitted to arXiv on June 5, 2026, spans 19 pages and includes 10 figures, with 19 pages of supplementary material. It is slated for presentation at the 2026 International System Dynamics Conference in Delft, Netherlands. The authors hypothesize that strategic competition over AI sovereignty may define how nations improve their economic situations, competitive advantage, and overall power in the 21st century. While the model is qualitative, it offers a robust starting point for policymakers and strategists to understand the multi-faceted nature of AI as an instrument of national power, moving beyond purely technical metrics to include geopolitical, resource, and infrastructure considerations.