Mapping the world's data center infrastructure against the carbon, water, and land footprints of the electricity.Based on the UNU-INWEH's 2026 global assessment of AI's environmental cost.
In 2025, the world's data centers consumed an estimated 448 TWh of electricity, if data centers were a country, they'd rank 11th globally by electricity use. AI workloads alone accounted for roughly a fifth of that, a share projected to double by 2030 as total data center demand climbs toward 945 TWh. But the environmental cost of that electricity isn't uniform: a kilowatt-hour drawn from a hydropower-heavy grid and one drawn from a coal-heavy grid carry very different carbon, water, and land footprints — and "low-carbon" doesn't automatically mean "low-water" or "low-land." This project maps where that infrastructure actually sits, and what powering it costs depending on where it's hosted.
Carbon, water, and land footprint intensities vary by an order of magnitude across the world's major data center hubs, driven entirely by each country's electricity mix. Toggle between footprints below, the colors show each country's footprint intensity for that dimension relative to the global average.
| Country | Data Centers | Carbon vs. global average |
|---|
What the infrastructure lens measures, and where the numbers come from.
| Label | Column | Description | Source |
|---|---|---|---|
| Data Centers | data_centers | Number of PeeringDB-registered data center facilities in the country. | World Bank Data360 (PeeringDB) |
| Networks | total_networks | Sum of networks (ASNs) present across all data center facilities in the country — a proxy for interconnection density. | PeeringDB API |
| Exchanges | total_ix | Sum of internet exchange points present across all data center facilities in the country. | PeeringDB API |
Drag any of the three lenses across the map. Each aggregates one metric for the data points beneath it — move them between regions to compare data centers, networks, and exchange points side by side.