The transition toward renewable energy sources is gaining significant attention to mitigate the environmental impact of fossil fuels. Wind energy systems are one of the most suitable solutions among renewable energy resources. For the reliable operation of these systems, predictive analysis plays a crucial role. In this chapter, a comparative analysis of deep learning models in wind speed forecasting is conducted. Seven different models were trained with real historical data: CNN, RNN, LSTM, GRU, CNN-RNN, CNN-LSTM, and CNN-GRU. The comparative analysis was conducted in terms of model performance and the required computational resources, measured by root mean square error (RMSE) and training time in seconds, respectively. The CNN-LSTM and CNN-GRU models significantly reduced the required training time by half compared to the LSTM or GRU models alone, also improving the accuracy. However, unlike these models, the CNN-RNN model did not show any performance improvement. In fact, the average error metrics increased along with a higher demand for computational resources.