This study examines AI-based modeling for multi-physics simulations of rotary mechanisms, focusing on the complex interactions between mechanical, thermal, and fluid domains. While conventional simulation methods are often limited by high computational costs and long processing times, this study evaluates data-driven and physics-aware alternatives—specifically Artificial Neural Networks (ANN), Physics-Informed Neural Networks (PINN), and Graph Neural Networks (GNN). Due to hardware and time constraints, rather than large-scale data generation, the research establishes a conceptual and methodological framework for integrating simulation-aware AI into rotary systems. To assess practical implementation challenges, a Monte Carlo-based feasibility analysis was developed in Python; this analysis estimated a success probability of approximately 0.77% for training these models under current resource limitations. Ultimately, this study contributes to the literature by providing a structured roadmap for AI-supported multi-physics modeling and offering practical guidance for engineering applications operating under significant computational constraints.