<p>Artificial intelligence and Internet of Things are transforming traditional infrastructures, including the energy sector, buildings, smart grids, etc. These innovations have shifted the traditional building paradigm into smart buildings. Smart buildings are energy-efficient, make automatic decisions and maintain user comfort. This survey paper presents a comprehensive review of the integration of IoT infrastructures and various energy prediction models in smart buildings. Initially, it details core components of the IoT landscape and their deployment challenges in smart buildings. Next, the existing literature is classified based on machine learning, deep learning, hybrid models, and spatio-temporal models for energy prediction models. Further, it highlights the emerging role of explainable artificial intelligence to enhance prediction accuracy and trust and its integration with IoT-enabled buildings. Subsequently, various benchmark datasets of electricity consumption are reviewed, including the performance metrics associated with electricity demand prediction. This paper elaborates current research trends, limitations, and open issues, and provides a foundational base for future work on smart building systems that are intelligent, understandable, trustworthy, and energy efficient.</p>