AI capabilities and entrepreneurial outcomes in high-tech ventures: exploring the (in)congruent role of innovation ambidexterity
Abstract
This study examines how artificial intelligence (AI) capabilities influence innovation ambidexterity, comprising exploratory and exploitative innovation, and how these innovation strategies contribute to business model transformation and key entrepreneurial outcomes, including new venture performance and entrepreneurial resilience. A three-wave longitudinal survey of 218 high-tech entrepreneurial ventures in Vietnam was conducted. Polynomial regression and response surface analysis were used to test a temporally ordered process model grounded in the dynamic capabilities framework. The results indicate that AI capabilities significantly enhance both exploratory and exploitative innovation. These innovation types independently and jointly foster business model transformation. Bootstrapping results are consistent with indirect relationships from AI capabilities to entrepreneurial outcomes through exploratory innovation, exploitative innovation, and business model transformation. The response surface results further show that when exploratory and exploitative innovation increase together, business model transformation improves in a positive and largely additive manner. By contrast, incongruence between the two innovation modes does not exhibit a strong detrimental effect, although a small directional effect is observed. The findings provide practical guidance for entrepreneurs and managers on integrating AI into organizational routines and reinforcing both exploratory and exploitative innovation, and remaining agile in adapting business models to seize new opportunities and withstand external shocks. This study contributes to research on AI and entrepreneurship by showing how AI capabilities activate dual innovation pathways and by clarifying that the configurational effect of innovation ambidexterity on business model transformation is mainly additive, while the effect of incongruence is weak and directional rather than strongly penalizing.