Maintaining the efficient and dependable operation of power systems requires load forecasting. Integrating Internet of Things (IoT) technologies provides a great opportunity to improve load forecasting accuracy by utilizing detailed data from smart meters, environmental sensors, and grid monitoring devices. This study examines the utilization of IoT-enabled load forecasting models, which encompass machine learning and deep learning techniques, as well as ensemble and hybrid models, to capture intricate patterns and nonlinear associations in energy usage. This paper focuses on the attributes of IoT-enabled load forecasting, including real-time data processing, scalability, and adaptability. These features allow for a wide range of applications in short-term, medium-term, and long-term forecasting, demand response, renewable energy integration, and energy trading strategies. Although there are notable advantages, there are also problems that need to be acknowledged and resolved, including those related to the accuracy of data, privacy concerns, the ability of different systems to work together, and the complexity of computing processes. Case studies and applications exemplify the effective utilization of IoT-enabled load forecasting in many areas such as short-term grid operations, demand response programs, renewable energy integration, and energy trading. The study explores future research directions, such as advanced data analytics methodologies, edge computing architectures, blockchain integration, and uncertainty quantification methods. Suggestions are given for promoting cooperation among individuals or groups involved in addressing difficulties and promoting the widespread acceptance of smart grids and sustainable energy practices.