Developing biofuels from bioenergy crops is essential for replacing conventional fossil energy and addressing climate change. Timely and reliable water footprints prediction is imperative and prerequisite to mitigating climate risk and enhancing water-use efficiency. This research adopts an Artificial Intelligence-Seasonal ARIMA (AI-SARIMA) integrated model to estimate, model, and forecast cassava's green and blue footprints (WFg and WFb) in Nanning city of Guangxi, China. Three supervised learning algorithms: Artificial Neural Network (ANN), Support Vector Machine (SVM), and Random Forests (RF) and time series forecasting model: SARIMA model, are used in this study. The monthly meteorological data of minimum temperature (Tmin), maximum temperature (Tmax), precipitation (P), sunshine hours (SH), wind speed (WS), and relative humidity (H) data from 1994 to 2019 are collected for modeling WFs by applying different scenarios of climate variables. The results show that the ANN method with hidden layers (8, 6) combined with input variables Tmax, Tmin, P, Kc, SH, and H was the best model to estimate WFb, the ANN with hidden layers (7, 5) combined with Tmax, Tmin, P, WS, and SH was the optimal estimation model for WFg. These WF estimation models achieve satisfying accuracy and coefficients of determination close to 1. Additionally, SARIMA models are integrated with the developed Artificial intelligence (AI) estimation model for WFs prediction. The results show that the WFs forecasting values can effectively match the overall trend of WFs. Thus, the AI-SARIMA coupled prediction models developed in this study are recommended for WFs estimation and prediction of energy crops in similar regions. Furthermore, this research can guide decision-making for water, agricultural, and biofuels managers and planners.