When The Frontier Moves: Outsider Innovation Across AI Paradigm Shifts

Discrete technological shocks can radically reshape technology landscapes, yet their impact on innovation from the margins remains poorly identified. Building on theories of outsider and peripheral innovation, this paper conceptualises frontier-moving technology shocks as external shocks that change how permeable field boundaries are, and which actors and technology combinations fall inside or outside the dominant paradigm. Focusing on artificial intelligence (AI), we examine two paradigm-shifting shocks: the 2012 deep learning breakthrough and the 2017 introduction of transformer architecture. Using OECD-classified AI patents, we construct measures of technological centrality and track how the centrality changes, induced by the frontier-moving shock, relate to subsequent innovative significance. We find that patents whose centrality increases receive significantly more forward citations, particularly those combining core and peripheral technologies. After the shocks, the top performing breakthrough patents increasingly come from patent applicants who were new entrants, further confirming the frontier repositioning recognition effect with outsiders. Taken together, these results provide evidence consistent with the view that technology shocks reshape which innovations are recognised and rewarded, grounding recognition penalties in field-level knowledge structure rather than relational position alone.