Estimating the Energy Consumption of Data Centres in the European Union

This repository contains the complete code, processed datasets, and outputs developed for my Master's thesis "Estimating the Energy Consumption of Data Centres in the European Union: Policy Implications for AI Deployment and Sustainability"

The project uses a bottom-up and area-based model to estimate:

• Annual energy consumption (TWh/year)
• Associated CO₂ emissions (tons/year)

Final outputs include energy and emissions estimates under three scenarios (low/mid/high), with cluster detection and national aggregation.

Repository Structure

thesis-dc-emissions/
├── input/           # Input data folders (not all files provided due to confidentiality)
├── output/          # Anonymized outputs and intermediate files
├── figures/         # Final figures and plots
├── code/            # Jupyter notebooks 
├── requirements.txt # Required libraries
└── README.md        # This file

Code Modules

code/ contains the step-by-step notebooks:

• 01_extract_urls.ipynb — Sitemap parsing and URL extraction
• 02_scraping_pipeline.ipynb — Scraping public data center pages
• 03_data_cleaning_feature_extraction.ipynb — Feature extraction and preprocessing
• 04_energy_estimation_area_based.ipynb — Energy modeling based on ITD and area class
• 05_carbon_intensity_mapping.ipynb — Joining data centers with bidding zones and CI data
• 06_emissions_calculation.ipynb — CO₂ emissions estimation per scenario
• 07_visualization_hotspots_energy_shares.ipynb — EU-level maps and scenario analysis
• 08_anonymization_geojson.ipynb — Anonymization of final GeoJSON

Data Availability

Some inputs cannot be shared due to confidentiality constraints. However, the final anonymized spatial dataset is available here: /output/data_center_emissions_anonymized.geojson

License and Usage

This repository is intended for academic research purposes only.