The utilization of renewable energy in Indonesia continues to grow to reduce dependence on fossil fuels. Sidrap Wind Power Plant as the first large-scale wind power project in Indonesia plays an important role in supporting electricity demand, but wind speed variability causes fluctuations in power output that can disrupt system stability. This study aims to predict the power output of Sidrap Wind Power Plant based on wind speed data by comparing two methods, namely Autoregressive Integrated Moving Average (ARIMA) and Extreme Gradient Boosting (XGBoost). The data used consists of wind speed data and the average power output of the 10 largest and smallest turbines over a three-month period. The ARIMA model was built using Minitab, while XGBoost was built using Google Colab with Python. Accuracy was evaluated using Mean Absolute Percentage Error (MAPE). The results showed that for the lowest power output data, the ARIMA model produced better results with a MAPE of 4.1%, while XGBoost had a MAPE of 4.42%. For the highest power output data, the ARIMA model produced a MAPE of 7.11%, while XGBoost provided better results with a MAPE of 0.73%. This comparison demonstrates that XGBoost provides more adaptive results for non-linear patterns, while ARIMA is effective for linear data patterns. This comparison provides recommendations for methods that are more suitable for the characteristics of the data for power management and improving the stability of the electrical system.