Methodology & Data Sources

All data in this dashboard comes from publicly available sources. Below is a complete list of every data source used, what we extract from it, and how frequently it updates.

Important Caveats

  • AI-specific data is estimated. Data center operators do not publicly distinguish AI from non-AI workloads. The AI share figures are derived from IEA modeling and academic research, not direct measurement.
  • Company reports use different methodologies. Scope boundaries, carbon accounting methods, and water measurement vary between providers. Direct comparisons should be interpreted cautiously.
  • Per-query figures are from Google only. The 0.24 Wh/query figure is specific to Google Gemini on Google infrastructure. Other providers and models will differ.
  • Projections are scenarios, not predictions. IEA 2030 projections represent modeled scenarios under stated policies, not forecasts.
  • Model efficiency data is approximate. The Hugging Face AI Energy Score measures GPU energy only, not full-stack (networking, cooling, storage).

Energy Consumption

International Energy Agency (IEA)

Global data center electricity consumption figures and 2020-2030 projections. Base case and scenario analysis.

Annual
Data used: Total DC energy (TWh), AI share projections, growth rates
www.iea.org/reports/energy-and-ai/energy-demand-from-ai

Google Cloud — Measuring Environmental Impact of AI Inference

Per-query energy, carbon, and water measurements for Gemini Apps. Full-stack methodology including accelerators, host systems, and overhead.

Research paper (2025)
Data used: 0.24 Wh/query energy, 0.03g CO2e/query carbon, 0.26ml/query water
cloud.google.com/blog/products/infrastructure/measuring-the-...

Hugging Face AI Energy Score

Standardized energy efficiency ratings for AI model inference. Benchmarked on NVIDIA H100 GPUs with standardized datasets.

Biannual updates
Data used: Wh per 1,000 queries by model, 1-5 star ratings
huggingface.github.io/AIEnergyScore/

Carbon Emissions

Nature Sustainability — Environmental Impact of AI Servers

Peer-reviewed projections for AI server environmental impact in the USA (2024-2030). Covers carbon, water, and mitigation scenarios.

Research paper (2025)
Data used: AI carbon: 24-44 Mt CO2eq (US), water: 731-1,125M m3
www.nature.com/articles/s41893-025-01681-y

EPA Greenhouse Gas Reporting Program (GHGRP)

Facility-level greenhouse gas emissions data for large emitting facilities including data centers.

Annual submissions
Data used: Facility emissions by industry type
www.epa.gov/enviro/greenhouse-gas-restful-data-service

IATA — Aviation Emissions

Global aviation CO2 emissions for industry comparison context.

Annual
Data used: ~920 Mt CO2 annual aviation emissions
www.iata.org/en/programs/environment/climate-change/

Water Usage

Digiconomist / Patterns Journal

First research to isolate AI-specific water footprint from general data center operations. Peer-reviewed, open access.

Research paper (2025)
Data used: 312.5-764.6 billion liters global AI water (2025)
digiconomist.net/the-hidden-carbon-and-water-costs-of-ai/

UN FAO AQUASTAT

Country-level water withdrawal data for comparison context.

Annual
Data used: National water withdrawal volumes
www.fao.org/aquastat/en/

USGS — Water Use in the United States

Municipal and per-capita water consumption data for US cities.

Every 5 years
Data used: City daily water consumption
www.usgs.gov/mission-areas/water-resources/science/water-use...

Comparison & Context

World Bank — Electricity Consumption

Country-level total electricity consumption for national scale comparisons.

Annual
Data used: Country electricity TWh
data.worldbank.org/indicator/EG.USE.ELEC.KH.PC

Cambridge Bitcoin Electricity Consumption Index

Real-time Bitcoin mining energy consumption estimate.

Live
Data used: ~160 TWh annual Bitcoin energy
ccaf.io/cbnsi/cbeci

EPA Greenhouse Gas Equivalencies Calculator

Conversion factors for making emissions relatable (trees, miles driven, homes powered).

Static reference
Data used: CO2 per mile, trees per ton CO2, kWh per home
www.epa.gov/energy/greenhouse-gas-equivalencies-calculator

EIA — US Energy Information Administration

Average US household electricity consumption for opportunity cost calculations.

Annual
Data used: 10,500 kWh/year per US household
www.eia.gov/tools/faqs/faq.php?id=97&t=3

Company Data

Google Environmental Report 2025

Energy, water, carbon, and renewable energy data for Google operations and data centers.

Annual
Data used: Company energy/water/carbon/PUE/renewable %
sustainability.google/reports/google-2025-environmental-repo...

Microsoft Sustainability Report

Scope 1, 2, and 3 emissions, water usage, and renewable energy commitments.

Annual
Data used: Company energy/water/carbon/PUE/renewable %
www.microsoft.com/en-us/corporate-responsibility/sustainabil...

Amazon Sustainability Report

AWS and Amazon overall environmental footprint including data center operations.

Annual
Data used: Company energy/water/carbon/PUE/renewable %
sustainability.aboutamazon.com/reporting

Meta Sustainability Report

Data center energy, water, and emissions data for Meta operations.

Annual
Data used: Company energy/water/carbon/PUE/renewable %
sustainability.meta.com/
Dashboard data last curated: June 2025. Static data files are in src/data/.