Recently, literature on urban network research from the perspective of firm networks has been increasing. Focusing on firms' multi-location distribution, existing studies use mainly the internal organization structural information of firms to portray one-type industry or all-industry networks, and there are deficiencies in reflecting the authenticity of networks and comparing the differences across industries. In contrast to studies applying intra-firm networks, this study collects real investment data of all listed firms in Jiangsu Province and studies urban networks through quantifying inter-firm linkages. Based on inter-firm investment linkages, this study provides fresh insights on inter-city connections in China. The results show that Nanjing, Suzhou, Shanghai, Beijing and Shenzhen are the constant network centers across different industries. Particularly, Shenzhen has been favored by more investment inflows far beyond its GDP. In China's unique institutional context, state power, geographical location, market and other factors jointly affect the flow of investment, and the role of state power deserves special attention. Since real economic connections are used to describe urban networks, the findings might contribute to ongoing debates regarding economic centers in China and bring further implications for policy making in investment environment improvement. It enriches the research of urban networks based on real inter-firm connections, and provides ideas for the wider regional study and the combination of econometric techniques and social network analysis.Sustainability 2020, 12, 89 2 of 19 questions about its accuracy and universality remain. Neal [12] and Liu [5] critique the model on the basis that the "rich" inter-city interactions generated by the interlocking network model would lead to a "flat" network which hides the real clusters. Another question about the usefulness of this model is that the producer service firms in developing countries tend to be concentrated in a few cities, and the network status of most cities with underdeveloped economies embedded in the producer services network is difficult to evaluate. Compared with the interlocking network model, the headquarter-subsidiary ownership method is a more straightforward and precise approach. Built on concrete data rather than assumptions, this method focuses on the control from the city where the headquarters are located on the location of branches [13,14]. The merit of the headquarter-subsidiary ownership method lies in converting 'tangible' intra-firm linkages into inter-city connections, which does not go "beyond what is strictly supported by the available data" [14]. However, this method also has some defects. For instance, in the myth of global management of multinational firms, the relations between headquarters and branches are quite complex rather than following a pattern of asymmetric control [15]. Besides, since large cities are more likely to be the location site for large firms establishing multi-locational branches [16], the he...