Energy exchange occurs between commercial buildings and external environment as well as internal activities. External disturbance factors such as solar radiation and outdoor temperature and internal disturbance factors such as personnel activities and equipment operation will affect the thermal performance of buildings, which together act on the air conditioning system and increase the difficulty of prediction. Therefore, a method for predicting air-conditioning load of commercial buildings under dual-carbon background is proposed. The importance of each load factor to the prediction was calculated by Morris sensitivity analysis method, and external and internal disturbance loads of air conditioning loads in commercial buildings were analyzed and calculated, which were used as the training data set of least squares support vector machine (LSSVM) model and extreme gradient boosting (XGBoost) model to fully consider internal and external disturbance factors and combine two different models to complete load prediction under the dual-carbon background. And the accuracy of air conditioning load forecasting is improved. The experimental results show that the proposed method can accurately predict the air conditioning load, and the highest mean absolute percentage error (MAPE) is only 0.026.