Information flow within urban networks profoundly affects production modes and the structure of urban agglomeration economic systems. Drawing on the panel smooth transition regression (PSTR) model and the mediating effect model, this study examined the effects of information flow on agglomeration externalities, with considerations on non-linear effects, proximity mechanisms, and industrial heterogeneity. The resulting findings revealed that the ability to control information at the city node level exhibited a significant non-linear relationship with the agglomeration externalities of the manufacturing, producer, and consumer services industries. Furthermore, the ability to exchange information promoted the agglomeration externalities of producer and public services, albeit only under specific intensity conditions. The proximity mechanisms demonstrated significant mediating effects at the city connection level. This study highlights the infrastructure-like nature of information flow networks and underscores the unique role of network externalities in reshaping agglomeration externalities. These findings offer insights for governments to improve regional planning and leverage virtual urban networks effectively.