Preferential attachment mechanism in innovation networks: from overall and individual perspectives of NEV industry

Purpose Innovation networks provide the dominant lens for understanding contemporary innovation processes and shape adaptability and performance. A key mechanism explaining their evolution is preferential attachment (PA), whereby central organizations attract disproportionate collaborations, influencing resource distribution and cumulative advantages. While prior studies have focused on the structure and treated PA as uniform, they have overlooked heterogeneity and contextual moderation. Therefore, this study proposes a novel framework to examine PA in regulated innovation networks. Design/methodology/approach The framework integrates network structure analysis and mechanism testing into an analytical process, examining PA at both the overall and individual levels. Drawing on the virtual breeding environment (VBE) perspective, we analyze the global new energy vehicle industry patent cooperation data (2001–2024). Findings The network exhibits a hub-dominated structure consistent with PA, and three nodes' evolutionary patterns emerge: PA, partial PA and non-PA. Practical implications Firms and policymakers should strategically manage node positions and evolutionary trajectories to optimize collaboration portfolios, and balance centralization, diversity and ecosystem stability. Originality/value By integrating the VBE perspective, the study reframes PA as a selectively enabled process rather than a uniform structural law, demonstrating how heterogeneous evolutionary paths coexist within PA-driven networks and extending understanding of differentiated actor trajectories.