This paper investigates the negative impact of spatial fading correlation in multiple-input multiple-output relaying systems on the performance of energy beamforming. Namely, a source and destination nodes equipped with multiple antennas which have a general correlation structures and arbitrary eigenvalue multiplicities, exchanging information through a dual-hop amplify-and-forward single antenna energy-constrained relay. To facilitate longer-distance wireless power transfer, the overall scavenged energy needs to be maximized. Hence, the energy-constrained relay harvest energy from the source radio-frequency signal through energy beamforming, the harvested energy is then used to forward the source information symbol to the destination. The time switching-based receiver along with the power splitting-based receiver protocols is examined in order to perform wireless information and power transfer at the relay. To this end, tight closed-form lower and upper bounds for the outage probability and ergodic capacity are derived, and used to examine the throughput of the delay-constrained and delay-tolerant transmission modes, respectively. Numerical results supported by simulations manifest the tightness of the presented analytical formulas. The effect of several parameters, like, energy harvesting ratio, source transmit power, number of antennas, and spatial fading correlation on the overall throughput is investigated. It is shown that increasing the number of antennas could be used to improve the system throughput or facilitate longer-distance wireless power transfer. On the other hand, the ramification of spatial correlation on the system throughput is also studied for arbitrary correlation structure. Moreover, it is revealed that the performance of power-splitting receiver outperforms the time-switching receiver at high signal-to-noise ratio. Finally, simulation results for the case of statistical channel state information is also included for comparison purposes.