Discontinuous innovations reshape technological trajectories, but identifying them in advance remains a significant challenge. In this project, we argue that the origin and nature of knowledge flows leading to an innovation can signal its potential for discontinuity. Specifically, we find that knowledge derived from basic research increases the likelihood of an innovation being discontinuous, while knowledge from applied research tends to reduce that likelihood. Using patent data, we empirically test these predictions and find strong support for our hypotheses. Our study advances the literature on discontinuous innovations and deep tech, offering new insights into how the sources of knowledge influence the disruptive potential of technological breakthroughs.