Contemporary research and development (R&D) increasingly faces a paradoxical situation: despite exponential growth in scientific publications and technological capabilities, innovation outcomes remain strongly constrained by rising costs and institutional bottlenecks. This paper addresses the question of how R&D efficiency can be substantially increased without proportional growth in investment, particularly under conditions of budgetary limitations and uneven access to advanced technologies. The study introduces a sociologically grounded framework for understanding artificial intelligence (AI) as an awakening mechanism for dormant scientific knowledge, conceptualized as "sleeping beauties"-research outputs that remain unnoticed for extended periods before gaining relevance. Rather than focusing on AI as a generator of entirely new knowledge, the paper emphasizes its role in systematically reinterpreting, recombining, and recontextualizing existing scientific results. Two complementary institutional strategies are analyzed. The first is a circular economy of knowledge, in which organizations reuse and reinterpret their own previously produced research outputs, transforming sunk costs into renewed sources of scientific and economic value. The second is the economy of co-optation, where research organizations appropriate external knowledge assets-particularly under Open Access regimes-without formal ownership or cooperation. While legally permissible and economically efficient, this strategy redistributes intellectual rents toward actors possessing superior analytical and organizational capacities. The paper argues that AI-enabled reinterpretation decouples innovation performance from proportional increases in experimental investment, but simultaneously introduces new forms of stratification within the scientific system. Inequality increasingly reflects disparities in analytical capability, institutional infrastructure, and access to AI, rather than differences in experimental resources alone. The findings contribute to the sociology of knowledge and innovation policy by demonstrating how AI reshapes institutional practices of R&D, intensifies competition over intellectual rents, and challenges existing policy assumptions regarding openness, collaboration, and equitable access to scientific progress.