Cloud radio networks employ coordinated transmission among base stations (BSs) to reduce the interference effects. The practical limitations in implementing coordination results in suboptimal systems with limited performance. In this paper, we analyze the performance of a cloud network with clustering, where geographically close BSs form a clustered cloud. Coverage probability and rate distributions are analyzed using stochastic geometric models. Specifically, we develop approximations for the coverage probability of a typical user in a clustered cloud network with zero-forcing dirty paper coding. The adverse effect of finite clusters on the achievable rate is quantified. © 2015 IEEE.