This paper attempts to distinguish and estimate the direct and indirect effects of infrastructure on firm productivity. The latter arises from the infrastructure–agglomeration link and has been largely overlooked in the literature on infrastructure. An analytical framework is then developed to estimate both effects. Finally, empirical results are obtained using large-scale firm-level survey data from the People’s Republic of China (PRC). Major findings include: (1) all the three kinds of infrastructure—road, telecommunication servers, and cable—are found to directly promote firm productivity; (2) they also exert a positive indirect effect on firm productivity through the agglomeration channel; and (3) the empirical results are robust to different agglomeration indicators and different subsamples.