ATLAS All The Locations of All Servers

About this dataset

ATLAS is an open dataset of 18,110 data center facilities across 116 countries, operated by 4,181 companies. Every record carries the facility name, operating company, city, state, country and street address, with geographic coordinates where they could be resolved. Free to download in twelve formats, free to use with attribution.

It was compiled from open sources: facility listings, operator websites and public directories. Nothing here is confidential. These are commercial buildings, most of which publish their own addresses.

Coordinate precision, stated plainly

Precision is not uniform, and that matters more than the headline count. Roughly 6,131 of the 18,110 records carry city-level coordinates. The rest fall back through a chain: city coordinates where available, then US state centroid, then country centroid.

A marker is therefore often not a building. It is frequently a city, and sometimes the geometric middle of a country. For seeing the shape of global infrastructure that is fine. For any analysis where a specific location matters, filter on coordinate provenance first, or you will produce a confident number that means nothing.

What the data shows

• The United States holds nearly half of everything. 8,818 of 18,110 facilities, 48.7 percent. A handful of metros hold more capacity than entire continents.
• The Netherlands has more mapped facilities than India. 444 against 342, from 17 million people against 1.4 billion. Density follows capital, connectivity and regulation, not population.
• Africa has fewer facilities than the Netherlands. Roughly 240 for 1.4 billion people, of which South Africa holds 67.
• Brazil is South America. 141 of the continent's roughly 217 facilities. Sao Paulo is a single point of failure for a continent.
• The physical layer is centralised. The top six operators alone run more than 2,300 facilities. The internet is decentralised at the protocol layer; the buildings are not.

How to use the map

• Search any facility, operator or address, then press Enter.
• Filter by country or operator to narrow the set.
• 3D Globe shows global distribution; Clusters makes points clickable.
• Robinson Projection renders a fairer world. Mercator exaggerates high latitudes, distorting exactly the comparison this data is for.
• Radius finds every facility within a distance of a point.
• Distance measures between two points in miles, kilometres or nautical miles.
• Proximity lists the nearest neighbours of any facility.
• Stats opens country, operator, quality and regional charts.

Download formats

CSV, XLSX, JSON, GeoJSON, KML, TSV, XML, SQLite, Parquet, SQL dump, YAML and a Markdown table. GeoJSON imports straight into QGIS, Mapbox, ArcGIS and Kepler.gl. KML opens in Google Earth. The SQLite build is queryable with no server. The full pre-deduplication archive of 25,757 records is published too, so the merge decisions can be audited.

Licence

Free to use, copy, publish and build upon, including commercially, on one condition: credit it as Data centers (c) Ringmast4r - Global-Data-Center-Map. No permission request needed. Dataset and licence on GitHub.

Questions

Is this every data center in the world?

No. It is the most complete open list I know of, but coverage is uneven and the long tail of small regional operators is certainly under-represented.

Why do some countries look empty?

Two different reasons, worth separating. Some regions genuinely have very little capacity. Others are under-collected because their operators publish less in the sources this was built from.

How current is it?

It is a snapshot, not a live feed. Facilities open, close, change hands and get rebranded. Verify anything time-sensitive with the operator directly.

Can I use it in a research paper?

Yes. Cite it as you would any dataset, and read the coordinate precision note above before computing exposure of any kind.

Dataset on GitHub · Privacy