<p>Thermal management remains a major challenge for CubeSats in Low Earth Orbit (LEO), where rapid temperature fluctuations, limited radiating surfaces, and strict power budgets limit the effectiveness of traditional control strategies. This work presents a MATLAB-based hybrid thermal model that combines passive buffering with phase change material (PCM) and an AI-enhanced proportional–integral–derivative (PID) controller for active regulation. The AI tuner adaptively adjusts PID gains using machine learning, allowing the controller to respond more effectively to dynamic orbital heat loads. Four operating modes passive-only, PCM-only, PID-only, and hybrid PCM + AI-PID were simulated under representative orbital conditions. Results show that the hybrid approach reduced peak-to-peak temperature variation by approximately 25% compared with PID-only control, while maintaining heater energy use at the same level (2.86 Wh per orbit). This demonstrates that improved thermal stability can be achieved without additional power cost. The model’s thermal response is consistent with reported CubeSat PCM experiments, supporting its physical credibility. Although this study is simulation-based, the framework provides a scalable and energy-aware foundation for future hardware-in-the-loop testing and eventual in-orbit deployment on resource-limited CubeSat platforms.</p> Graphical abstract <p></p>