With energy demand increasing year by year, high-yield crop production requires the use of Controlled Environment Agriculture (CEA) facilities, such as modern greenhouses and precision technologies. However, the energy consumption for maintaining such an outstanding indoor climate, including temperature, humidity, and concentration, is exceptionally high. This study primarily focuses on the engineering challenge of improving energy efficiency through an Intelligent Hybrid Energy Management System (IHEMS) that leverages Artificial Intelligence (AI) to optimize in real-time solar photovoltaic (PV) and thermal energy storage systems in combination. The IHEMS employs a Model Predictive Control (MPC) algorithm integrated with a Deep Reinforcement Learning (DRL) agent. The MPC predicts future thermal and electrical loads based on crop-specific requirements, including external weather forecasts and historical, operational, and statistical data. The DRL agent then learns the most cost-effective and energy-efficient control plans for the hybrid system's components (e.g., maximizing PV self-consumption, coordinating the charging/discharging of the thermal storage tank, and modulating HVAC unit operation). The main objective is dual-target optimization. Firstly, minimize the net energy cost, and secondly, maintain the microclimate within a narrow, ideal band for maximum crop growth. The experimental validation using a test greenhouse facility shows that the IHEMS achieves up to a 25% reduction in peak power demand and an overall 18% reduction in energy consumption compared to traditional and conventional rule-based control systems, while significantly enhancing the potential for crop productivity. This research provides a robust engineering solution for sustainable, high-precision energy management in the future of the agriculture sector.