With the rapid accumulation of marine big data and the robust development of Artificial Intelligence (AI) technology, intelligent marine forecasting has shown greater precision and efficiency in this new era. Marine data can be categorized into point- and field-observation data based on the observation methods, providing foundational support for marine forecasting. Marine forecasting methods can be divided into three main types based on the characteristics of the dynamic marine processes and phenomena: point-to-point, field-to-point, and field-to-field forecasting. These forecasting approaches not only cover a variety of marine phenomena but also address different forecasting requirements. Through a case analysis, this study specifically introduces intelligent forecasting models and results for point-to-point internal solitary wave forecasting, field-to-point El Niño-Southern Oscillation (ENSO) forecasting, and field-to-field phenomena such as mesoscale eddies and sea ice. Finally, it explores the development directions for intelligent marine forecasting in the context of big data, suggesting that enhancing the integration of data-driven methods with physical mechanisms can improve forecast accuracy and real-time responsiveness, thereby providing technical support for marine environmental monitoring, disaster warning, and the sustainable use of marine resources.