Abstract: With intensifying global technological competition and the emergence of a new wave of industrial revolution, breakthrough innovation has become a critical engine driving national development. Understanding the mechanisms behind breakthrough innovation is therefore an urgent research priority. Innovation entails novel combinations of production factors, and the introduction of new elements such as knowledge and data has enriched this notion of "new combinations." As a core production factor, scientific knowledge plays a pivotal role in breakthrough innovation. Investigating its connection to innovation value carries significant theoretical and policy implications. Drawing on knowledge recombination theory and using a Tobit model, this study analyzes biotechnology patents granted by the United States Patent and Trademark Office (USPTO) from 2010 to 2021, along with the scientific literature cited therein, to explore the influence of heterogeneous knowledge sources on the value of breakthrough innovation. The main findings are as follows: (1) From the perspective of organizational boundaries, the paper distinguishes the effects of internal and external sources of scientific knowledge. Internal scientific knowledge positively contributes to the value of breakthrough innovation. External scientific knowledge exhibits a U-shaped relationship with innovation value—while diversity in external sources can be beneficial, it also increases integration complexity, necessitating strong absorptive and transformative capacities within the organization. Once an organization’s internal knowledge accumulation surpasses a certain threshold, it is more capable of capitalizing on open innovation, thereby enhancing the value of its breakthrough innovations. This mechanism helps explain why knowledge-intensive organizations often emerge as leaders in open innovation. (2) From the structural perspective of knowledge sources, the study finds that the layered structure of knowledge within university-industry-government collaborations—ranging from basic research to applied development—is conducive to generating breakthrough innovations. This provides micro-level support for the "Triple Helix" model of innovation. (3) From the perspective of knowledge characteristics, the study reveals differentiated impacts of scientific and technological knowledge on innovation value. While greater cognitive distance in technological knowledge tends to enhance breakthrough innovation, an increase in cognitive distance in scientific knowledge may raise the risk and uncertainty of innovation. This diverges from prior literature that suggests broad search and distant recombination of technological knowledge enhance breakthrough innovation. By distinguishing between scientific and technological knowledge, the study deepens the theoretical framework of knowledge recombination. This research makes three primary contributions: (1) It refines the understanding of how internal versus external scientific knowledge affects breakthrough innovation, clarifying the distinct roles of internal innovation and inter-organizational collaboration; (2) it elucidates the role of scientific knowledge generated through university-industry-government collaborations, shedding light on the deeper value of such partnerships; (3) it reveals the differential mechanisms by which scientific versus technological knowledge contribute to innovation value, thereby enriching knowledge recombination theory. Based on the above discussion, this paper proposes the following two policy recommendations: First, innovation policies—especially in science-driven industries—should prioritize incentives for internal knowledge accumulation. Second, knowledge exchange and integration across academia, industry, and government should be strengthened by removing institutional barriers and encouraging deeper cross-domain knowledge construction in collaborative innovation.