Carbon neutrality hinges on effectively harnessing renewable energy sources, a critical factor as countries worldwide enact low-carbon legislation to mitigate global warming. In this context, this study introduces an innovative soft-computing framework based on the adaptive neuro-fuzzy inference system (ANFIS) for predicting wave height and wave period, which are crucial parameters for energy harvesting. This model incorporates inputs such as temperature, wind speed, and wind direction to predict wave energy. The model was trained using data from the Lianyungang (LYG) and Dachen (DCN) stations in the East China Sea, and its performance was contrasted with that of artificial neural networks (ANNs) encompassing 10–50 neurons and utilizing subtractive clustering (SC) and grid partition (GP) techniques. The ANFIS SC method, with a 0.1 cluster radius (C.R), outperformed both the ANNs and ANFIS GP in wave energy prediction. The statistical analysis confirmed that this model yielded root-mean-squared error, R, and R2 values of 0.0017, 0.987, and 0.974 for the LYG station and 0.0152, 0.93, and 0.867 for the DCN station. These metrics imply that the ANFIS SC algorithm excels in wave energy prediction, thus rendering it a potent instrument for wave-powered ship navigation. This study underscores the value of soft-computing techniques in pursuing renewable energy forecasting, contributing to sustainable and efficient marine transportation.