Knowledge-base Modularity and Boundary Choice under Technological Discontinuity

How do firms choose between internal development and external sourcing when adopting AI-related technologies under technological discontinuity if a firm's knowledge base can be modular? We argue that knowledge-base modularity has two opposing effects on outsourcing: it initially increases outsourcing by reducing coordination and interface burdens, but at even higher levels it decreases outsourcing by raising the option value of internally retained, redeployable, and recombinable knowledge. We further argue that supplier capability distinctiveness under concentrated supply increases the level of modularity at which outsourcing is most attractive. Using patent data and outsourcing-event data on global automotive manufacturers in Advanced Driver Assistance Systems (ADAS) and autonomous driving from 2000 to 2021, we find support for both predictions. We contribute by showing that modularity does not uniformly expand market use under technological discontinuity.