We take a resource-based view to investigate how the adoption of Artificial Intelligence (AI) affects competitive capabilities and performance. Following prior work on using chess as a controlled setting for studying competitive interactions, we compare the same players’ capabilities and performance across conventional, centaur, and machine chess games. We find that, in the presence of AI-based chess engines, human chess-playing capabilities are substituted, with a new source of persistent heterogeneity emerging as humans interact with chess engines. Our analysis further shows that these complementary human-machine capabilities are unrelated, or even negatively related, to human chess-playing capabilities. We contribute to the strategy literature by describing how AI-based technologies’ adoption changes the sources of competitive advantage by driving interrelated substitution and complementarity dynamics between humans and machines.