<p>Solid particle erosion (SPE) is a persistent challenge in engineering systems involving particle-laden flows and a complex, multiscale problem governed by coupled fluid dynamics, particle transport, and material response. From a computational perspective, SPE has emerged as a canonical benchmark for evaluating hybrid numerical and data-driven modeling strategies. This review provides a comprehensive synthesis of physics-based, computational, and artificial intelligence (AI) enabled approaches for SPE prediction, with a particular focus on their mathematical foundations, numerical implementation, and scalability. Classical mechanistic erosion models and computational fluid dynamics (CFD) based solvers are first examined to establish the physical kernels underlying erosion prediction. Recent advances in machine learning (ML), deep learning (DL), reduced-order modeling, and hybrid CFD-ML frameworks are then systematically analyzed, highlighting their ability to serve as computational surrogates that preserve physical consistency while reducing computational cost by several orders of magnitude. Particular attention is given to feature construction, kernel abstraction, uncertainty quantification consistent with ASME V&amp;V principles, and explainability techniques that enhance model interpretability. The review further integrates optimization strategies, digital twin concepts, and real-time predictive frameworks, thereby positioning SPE modeling within a broader computational hierarchy. Beyond erosion, the synthesized methodologies are transferable to erosion-adjacent degradation phenomena, including cavitation, corrosion-wear coupling, and particulate fouling. Overall, this work contributes a unified computational taxonomy that advances scalable, interpretable, and uncertainty-aware modeling of erosion phenomena.</p>