Interactive Visualization
Explore how data center distribution has evolved across U.S. states from 2008 to 2030. Our machine learning models predict future distribution patterns based on economic and infrastructural factors including corporate tax rates, electricity prices, and grid reliability.
About This Tool
This interactive map displays the percentage share of total U.S. data centers by state, allowing you to track historical trends and explore our projections through 2030. The underlying predictions are generated using a Multiple Linear Regression model trained on state-level economic and infrastructural data.
Key Features
- Historical Data: Real data center counts from 2008-2025
- Predictive Modeling: Machine learning projections through 2030
- Interactive Timeline: Slide through 23 years of data
- State-Level Analysis: Detailed breakdown by U.S. state
Methodology
Our predictions are based on a comprehensive analysis of multiple machine learning models, including Multiple Linear Regression, XGBoost, Random Forest, and others. After extensive testing and hyperparameter tuning, the Multiple Linear Regression model demonstrated the best performance with an R² of 0.741 for count predictions.
The model incorporates features such as corporate tax rates, average electricity prices, grid reliability (SAIDI), and engineered temporal and state-specific interaction terms to capture the complex dynamics of data center location decisions.