In this paper, we investigate the system performance of a multi-cell multi-user (MU) hybrid millimeter wave (mmWave) communications in a multiple-input multiple-output (MIMO) network. Due to the reuse of pilot symbols among different cells, the performance of channel estimation is expected to be degraded by pilot contamination, which is considered as a fundamental performance bottleneck of conventional multicell MU massive MIMO networks. To analyze the impact of pilot contamination to the system performance, we first derive the closed-form approximation of the normalized mean squared error (MSE) of the channel estimation algorithm proposed in [2] over Rician fading channels. Our analytical and simulation results show that the channel estimation error incurred by the impact of pilot contamination and noise vanishes asymptotically with an increasing number of antennas equipped at each radio frequency (RF) chain at the desired BS. Furthermore, by adopting zero-forcing (ZF) precoding in each cell for downlink transmission, we derive a tight closed-form approximation of the average achievable rate per user. Our results unveil that the intra-cell interference and inter-cell interference caused by pilot contamination over Rician fading channels can be mitigated effectively by simply increasing the number of antennas equipped at the desired BS.