Cooling towers play a critical role in fab energy and water performance, yet they are often operated with limited visibility into how well they are actually performing. This presentation looks at how combining physics-based models with real-time data and machine learning can uncover hidden inefficiencies in cooling tower operation. Using multiple case studies, it shows how targeted adjustments to airflow, water flow, and controls can improve reliability, reduce energy and water use, and extend equipment life without major capital upgrades.