The literature suggests that firms invest substantially in further developing their past research efforts. However, less is known about how such cumulative invention efforts are influenced by the multi-purpose potential of the original research effort. Taking a question-driven approach, I examined firms’ cumulative invention efforts around past research on deep learning. Difference-in-difference analysis suggests that firms radically increased cumulative invention efforts following a shock elevating the multi-purpose potential of past deep-learning research. Furthermore, firms publicly disclosed their cumulative inventions to attract application-sector innovation efforts while learning from the attracted efforts to innovate further. Grounded in these findings, I suggest that the multi-purpose potential of a research effort, by invoking interdependence between the inventing firm and application sectors, drives cumulative invention efforts alongside disclosure and learning.
