In this paper, simultaneous wireless information and power transfer (SWIPT) concept is introduced for multi group multicast beamforming. Each user has a single antenna and a power splitter which divides the radio frequency (RF) signal into two for both information decoding and energy harvesting. The aim is to minimize the total transmission power at the base station while satisfying both signal-to-interference-plus- noise-ratio (SINR) and harvested power constraints at each user. Unlike unicast and certain broadcast scenarios, semidefinite relaxation (SDR) is not tight and global optimum solution cannot be found for this problem. We propose an iterative algorithm where a convex optimization problem is solved at each iteration. Both perfect and imperfect channel state information (CSI) at the base station are considered. Simulation results show that the proposed solution is very close to the SDR lower bound and a few number of iterations are enough for the algorithm convergence.