Addressing the issue of how breakthrough innovation alliances in emerging industries dynamically evolve, a system generation theory perspective was adopted to analyze the evolution mechanism of these alliances. A longitudinal single-case study of the biopharmaceutical industry was conducted. Based on the case analysis, utilizing a combined inductive and deductive reasoning approach, a two-stage evolutionary game model was further developed to explore the emergence patterns of breakthrough innovation alliances in emerging industries. The research indicates that leading enterprises, affiliated enterprises, external enterprises and academic research institutions constitute the core layer of these alliances. The evolution of the alliances can be divided into two stages including formation and stabilization. During the formation stage, the distribution of benefits positively influences the willingness to cooperate among alliance members. The cooperation willingness of leading enterprises follows a U-shaped trend with the increasing difficulty of knowledge matching, while tax rate changes have a minimal impact. In the stabilization stage, leading enterprises maintain a stable positive willingness to cooperate, affiliated enterprises are prone to “free-riding” behavior, and academic research institutions shift from an initial brief U-shaped fluctuation to a positive willingness to cooperate. Theoretical support and practical insights are provided for fostering breakthrough innovation alliances in emerging industries from an evolutionary perspective.