Cloud Control: A Comparative Analysis of National Strategies for Data Localization and AI Development

Abstract

Over the course of the internet’s development from its techno-libertarian origins to the global communications network seen today, governments have increasingly made attempts to establish their sovereignty in the digital space and control how the internet works within their borders. This thesis examines these attempts and seeks to understand the future of digital sovereignty through two policy domains: data localization and national AI development. Data localization policies constrain the movement of data across borders by requiring it to be stored and/or processed physically within a country, and national AI development policies seek to create homegrown foundation models, which are general-purpose systems that can be used for a variety of tasks. Using a desk research approach, six countries (Canada, Turkey, and Vietnam for data localization, and France, Japan, and India for national AI development) are evaluated by the ways in which laws, social norms, markets, and infrastructure affect policy and each other.

The research reveals that infrastructure and markets are two of the most important constraints for the success of digital sovereignty policies. Well-established data infrastructure is necessary for private companies to be able to easily comply with localization, otherwise governments risk them leaving the market entirely, even as they are the impetus for many localization policies. Supercomputing capabilities are similarly necessary for the development of national AI models, and countries have seen great success by working closely with private-sector AI startups. Taken together, the two analyses of this paper suggest that implementations of sovereign digital policy will lead to alignment, not fragmentation, where the version of the internet that citizens can access is closely tied with a government’s political and strategic goals.