The transition to renewable energy sources has intensified interest in marine energy due to its predictability and availability. Accurate wave height prediction is critical for optimizing wave energy converters, designing offshore structures, and ensuring maritime safety. Traditional forecasting methods, based on physical and statistical models, often lack adaptability to dynamic ocean conditions. This study explores the use of transfer learning for predicting maximum wave height in the Bay of Biscay, leveraging data from two offshore buoys. Various neural network architectures—MLP, CNN, RNN, LSTM, and GRU—were evaluated, comparing their performance with and without transfer learning. The results demonstrate that transfer learning enhances generalization capabilities, particularly in recurrent architectures, reducing prediction errors while maintaining model stability. However, its effectiveness depends on dataset characteristics and model fine-tuning. These findings highlight the potential of artificial intelligence in improving wave height forecasting, contributing to the development of more efficient marine energy systems and offshore operations.