Against the background of increasing urbanization rate and intensifying global warming, conflicts between humans and the natural environment continue to arise, and regionalized forms of spatial organization have become an important research direction. This paper constructs a green innovation city network. It empirically tests the evolution process of the green innovation city network and its carbon emission effect by combining the social network approach and the spatial Durbin model. The conclusions are as follows: (1) the strong ties among green innovation city networks are mainly distributed in and around the provincial capital cities and the middle and lower reaches of the Yellow River Basin; the density of green innovation city networks has been strengthened, and the degree centrality and closeness centrality have been improved. (2) The carbon emissions of cities in the Yellow River Basin are generally increasing. Still, the rate of increase is slowing down. The carbon emissions from liquefied petroleum gas show a decreasing trend yearly, and the energy structure tends to improve. (3) The impact of the green innovation city network on carbon emissions mainly comes from its externality's direct and indirect effects; the increase of degree centrality will reduce the total carbon emissions in the region and the network-associated regions.