This study examines how the openness of ecosystem architecture shapes complementor innovation. More specifically, we focus on the openness of core technologies central to the development of ecosystem interdependencies that influence complementor’s search for external knowledge in subsequent innovation. Our key arguments draw on the literature on technology ecosystems and knowledge sourcing strategies, as well as the discussions on the role of openness in product design and highlight that open architecture of core ecosystem technologies could reduce complementor’s reliance on external knowledge in innovation. The systematic key knowledge provided by the open architecture of the core is a more comprehensive substitution that complementors can refer to in understanding the interdependent dynamics within the ecosystem, while the public accessibility of such knowledge accentuates the importance of internal knowledge in identifying novel and unique complementary opportunities to gain advantages through innovation. Our empirical examination adopts a difference-in-differences design that leverages the context of the deep learning artificial intelligence ecosystem, in which the open source of the TensorFlow framework by Google shifted the ecosystem of core architecture from closed to open. Our hypotheses are confirmed based on complementor innovation using data on complementary code repositories of deep learning on GitHub. Moreover, we also found that the decrease in external knowledge sourcing is stronger for complementors who rely on extended external knowledge sources, while it is weaker for those who maintain frequent communication with external knowledge contributors.
