AI Policy Wiki
Dashboard

The Commercial Landscape of AI Sovereignty Offerings (Stanford HAI, July 2026)

high confidence · updated 2026-07-26

Surveys commercial AI sovereignty offerings across the stack from Nvidia, Microsoft, Google, AWS, and OpenAI. Finds they provide real solutions for domestic control over computing infrastructure, data governance, and localized model deployment, but 'often reconfigure, rather than eliminate, dependence on these companies' — the basis of the 'sovereignty washing' criticism the brief examines.

An HAI Policy & Society issue brief by Caroline Meinhardt, Juan N. Pava, Caroline Yee, and James A. Landay.

The premise

"'AI sovereignty' has become one of the defining buzzwords of global AI development and governance. From Europe to Southeast Asia, from the Gulf to Latin America, governments are pouring resources into strategies" to "guard against overdependence on foreign AI providers — and a sprawling commercial market is emerging to meet that demand."

The brief's stated question is evaluative rather than descriptive: it surveys offerings "to assess how different actors are designing and marketing sovereignty solutions and to what extent these meaningfully increase operational control and reduce dependencies."

The finding

Offerings from Nvidia, Microsoft, Google, AWS, and OpenAI "provide solutions for increasing domestic control over computing infrastructure, data governance, and localized model deployment. However, they often reconfigure, rather than eliminate, dependence on these companies."

That formulation is the brief's central claim: the sovereignty being purchased is jurisdictional and operational rather than structural. A country may gain control over where data sits and which regulations apply while remaining dependent on the same vendor for the capability itself.

"Sovereignty washing." The brief reports that "many sovereignty initiatives have been criticized for 'sovereignty washing'," and states the two-sided assessment directly: the tools sold under this banner "may genuinely improve purchaser countries' control over certain aspects of AI development and deployment (e.g., telco-operated cloud infrastructure that is compliant with regulations)," while "they also ensure these countries will remain structurally dependent on them for the long" term.

Full-stack claims vary. Reviewing companies positioning themselves as end-to-end sovereign-stack providers "offering customers a single point of control across every AI layer," the brief finds that "in practice, however, the meaning of 'full stack' varies considerably depending on the company's core competency" — each vendor's definition of the full stack centres on the layer it already sells.

Relationships