This paper considers the problem of energy efficient beamforming allocation in a cellular wireless network. Unlike most of the existing work that considers the problem at the physical layer, we include in our analysis the impact of the bursty traffic patterns of the users. We formulate a stochastic optimization problem that minimizes the average transmit power of the network such that the queues of the users are stable. We then develop a solution that combines Lyapunov optimization and mixed integer quadratic problem (MIQP). We provided a fundamental complexity analysis of the resulting optimization problem and show that it is NP-hard. We then provide a Semi-Definite Relaxation (SDR) of the problem and develop an algorithm with low complexity to get as close as possible to the desired optimal solution. Our claims are corroborated by simulation results.