The integration of physics-based principles with Artificial Intelligence (AI) offers a transformative solution to complex challenges in hydrogeology. While AI excels at capturing nonlinear relationships and processing large datasets, it often struggles with physical consistency and interpretability, particularly when data are sparse. This chapter explores the synergy between physics-based modeling and AI, focusing on their potential to enhance predictive capabilities in hydrogeology. We discuss the challenges of purely data-driven AI approaches, such as generalization and overfitting, and highlight how incorporating physical laws—such as mass conservation—into AI models improves robustness. A key example is Physics-Informed Neural Networks (PINNs), which directly integrate governing physical laws, expressed as partial differential equations, into the neural network training process. The advent of PINNs marked the beginning of a new paradigm—one where physics is not an afterthought but an intrinsic part of the learning process. This foundation has paved the way for a spectrum of hybrid approaches, from Theory-Guided Machine Learning (TGML), where domain knowledge is used to regularize or guide models, to Physics-Aware Machine Learning (PA-ML), which embeds structural constraints and inductive biases into architectures to enforce realism and generalizability. This chapter concludes with a discussion of the current computational bottlenecks, emerging trends, and future research directions.