Massive multi-input multi-output (MIMO) can support high spectral efficiency (SE) with simple linear transceivers, and is expected to provide high energy efficiency (EE). In this paper, we analyze the EE of downlink multi-cell massive MIMO systems under spatially correlated channel model, where both transmit and circuit power consumptions, training overhead, channel estimation and pilot contamination (PC) are taken into account. We obtain the maximal EE for the systems with maximum-ratio transmission (MRT) and zero-forcing beamforming (ZFBF) for given number of antennas and users by optimizing the transmit power. The closed-form expressions of approximated optimal transmit power and maximal EE, and their scaling laws with the number of antennas M are derived for the systems with MRT and ZFBF. Our analysis shows that the maximal EE decreases with M for both systems with and without PC, but with different descending speeds. For the system without PC, the optimal transmit power should be configured to increases with M, while for the system with PC, the optimal transmit power should be configured as a constant independent from M. The analytical results are validated by simulations under a more realistic three-dimensional channel model.