Short － term grid load forecasting plays an important role in the operational decision for any power system． Because the real － time requirements for short － term load forecasting conflict with the high computing capability which is required to perform complex processes on large data sets，a short － term grid load forecasting algorithm based on support vector regression is proposed，which is implemented based on cloud computing platform． The proposed forecasting algorithm is compared between the cloud computing platform and the stand － alone computing platform，and the results show that the implementation based on cloud computing platform effectively improves the execution efficiency of the algorithm．