Artificial Neural Network (ANN) technique has developed rapidly in the field of estimation. ANN can predict based on data on events and related factors that existed in the past. ANN has advantages in parallel computing in classifying patterns. ANN is also capable of self-regulating the data to be processed without requiring an explicit function specification. The advantage of using ANN is the elimination of complex analytical and numerical iterative computations. The ANN method that is often used in prediction case studies is the Backpropagation Algorithm. This algorithm has the ability to solve problems in the real world by building trained methods that show good performance on large data scales and are able to overcome complex pattern recognition. This study aims to predict the demand for salt optimally using the ANN Method with the Backprogation Algorithm at PT. Kurnia Garam Prosperous Padang City. This forecasting is needed because of the high cost of production with the large number of requests that occur to be more effective. Proper forecasting will be able to optimize production so that it can reduce the required production costs. The data processed is salt production data from 2016 to 2018 at PT. Kurnia Garam Prosperous. The momentum results obtained are 3-9-1 for dividing the data into 2, namely 24 training data and 12 test data. The optimal prediction result is 0.98946, so this research is very helpful in forecasting optimal and efficient production costs.