The construction sector significantly contributes to global greenhouse gas emissions, creating an urgent need for efficient environmental impact assessment methods. This study evaluates machine learning (ML) models to rapidly predict the Global Warming Potential (GWP) of construction products using standardized Environmental Product Declaration (EPD) data from the ÖKOBAUDAT database. Five regression models - Artificial Neural Network (ANN), Support Vector Regression (SVR), Decision Tree (DT), Random Forest (RF), and Extreme Gradient Boosting (XGB) were trained, optimized, and compared. Data preprocessing involved careful feature selection through correlation analysis, management of missing values, and outlier removal. Model performance was assessed using the coefficient of determination R², root mean square error (RMSE), and mean absolute error (MAE). Results demonstrated that ensemble tree-based models (RF and XGB) performed best, achieving R² ~ 0.90 on held-out dataset and retaining R² ~ 0.81-0.83 on an independent external EPD validation set with mean error of almost 10% of a typical product footprint. These findings demonstrate ML’s capability to effectively approximate product life-cycle impacts, providing rapid estimates for early-stage sustainability decision-making. However, detailed Life Cycle Assessments (LCA) remain essential for critical decisions.