The Commodification of AI Sovereignty: Lessons from the Fight for Sovereign Oil

TL;DR

The paper investigates how sovereignty is being embedded and commodified within the AI ecosystem, arguing that private vendors now offer sovereign AI infrastructures and models that can redefine national autonomy on their terms. By contrasting the historical concept of sovereignty with parallels from oil production, it dissects how sovereignty is negotiated across the AI stack—hardware, cloud, and model development—through corporate discourse and state policy. It highlights risks such as vendor-driven control, data localization dynamics, and labor precarity in data annotation, while suggesting governance-focused approaches like measuring technology transfer and tracing data as a resource. The work emphasizes the need for a historically informed lens to prevent sovereignty from becoming a market value that constrains democratic control and national self-determination, and it points to careful scrutiny as AI sovereignty expands into new domains, including space and beyond.

Abstract

"Sovereignty" is increasingly a part of national AI policies and strategies. At the same time that "sovereignty" is invoked as a priority for global AI policy, it is also being commodified along the AI stack. Companies now sell "sovereign" AI factories, clouds, and language models to governments, enterprises, and communities -- turning a contested value into a commercial commodity. This shift risks allowing private technology providers to define sovereignty on their own terms. By analyzing the history of sovereignty and parallels in global oil production, this paper aims to open avenues to interrogate the implications of this value's commercialization. The contributions of this paper lie in a disentangling of the facets of sovereignty being appealed to through the AI stack and a case for how analogizing oil and AI can be generative in thinking through what is achieved and what can be achieved through the commodification of AI sovereignty.