We present a first-of-its-kind empirical study of technological sovereignty in artificial intelligence, adopting a competence-based perspective. We use patents and publication data to map competencies across AI techniques, functions and applications, and develop a novel measure of integration based on relative specializations and complementarities. We argue that our measure approximates technological sovereignty by capturing local capabilities to innovate in AI. We use our novel measure to explain AI innovation, and unpack integration determinants. Our focus is on the European Union, given its lagging position yet key role in a global landscape increasingly characterized by growing rivalries and fragmentation.

Read the document