Wind power is playing an increasingly important role in present power grids due to innovative advancements in wind energy generation. Wind power and wind speed should always be accurately predicted in order to assess wind power for power system operation and planning properly. Due to the advancement of AI technologies, Deep learning, in particular, is increasingly being used in wind energy forecasting because of its outstanding capacity to handle complicated nonlinear challenges. The aim of this paper is to forecast the wind power, speed, and direction for a given historical wind data using two deep learning techniques; long short-term memory (LSTM) and gated recurrent unit (GRU). The experiment results show that GRU outperforms LSTM for a small number of epochs, but when epochs increase to a larger number, the behavior of both techniques is nearly equal, with a preference for LSTM with a mean square error of 0.03 %.