The value of radiation energy is considered to be the main component of the design of clean energy devices. Due to the fact that this amount is not measured or in some cases the radiation station may not be available in Iran, so it is necessary to estimate this component. In this study, Kerman and Yazd were selected as the regions with the maximum potential of solar energy. Then Artificial Neural Network (ANN) and Support Vector Machine (SVM) capabilities were evaluated in solar energy estimation. For this purpose, the daily data of 25 years (1992-2017) including maximum temperature, mean temperature, relative humidity, sunshine hours and solar radiation were collected at the synoptic stations of these regions. To evaluate the performance of models, common evaluation statistics were used. The results showed that at the Yazd station, in the ANN method, the lowest values of RMSE, MAE and the highest values of IA and R2 were 2. 381, 1. 760, 0. 869, and 0. 962, respectively. These values at Kerman Station are equal to 2. 708, 2. 050, 0. 945 and 0. 810, respectively. In the SVM method, the lowest values of RMSE, MAE and the highest values of IA and R2 at the Yazd station were 2. 028, 1. 540, 0. 901 and 0. 973, respectively, and at Kerman station were 2. 407, 1. 896, 0. 956 and 0. 846, respectively. Overall, the efficiency of the SVM method in both regions was more accurate than the ANN method.