API

A static, read-only JSON API at https://kiankyars.github.io/datacenter-atlas/api/v1/. Plain files over HTTPS: no key, no sign-up and no rate limit beyond GitHub Pages fair use. GitHub Pages sends CORS headers, so browsers can fetch it from any origin.

Read before using #

• Each status is the last reported project status as of a dated source (status_as_of), not the status today.
• Locators are not building footprints or construction extents; some markers identify only a town or region, not a facility position.
• Empty or null values are unknown, never zero.
• Coverage is incomplete and varies by country.

Endpoints #

GitHub Pages serves .ndjson, .sqlite and SHA256SUMS as application/octet-stream and .zip as application/x-zip-compressed; identify formats by file extension or by media_type here, not by Content-Type.

Machine-readable contract #

api/v1/openapi.json is an OpenAPI 3.1 description of every endpoint; import it into Swagger UI, Postman or a code generator. Each response and record type also has a JSON Schema (2020-12), listed in api/v1/schemas/index.json. The schemas are generated from the dataset's schema.json and the descriptions on this page; new optional fields may appear within v1. Language models and agents can start from llms.txt.

Explorer links #

The map explorer keeps its view in the URL (q, country, status, locator, power, from, to, sort and #site=<site_id>), so a link reproduces a filtered view, for example ?country=US. The parameters, the columns of the explorer's CSV and GeoJSON exports and its optional WebMCP tools for browser agents are documented in the maintainer notes.

Response fields #

Records are typed from the source schema.json: empty CSV values become null; *_json columns become parsed JSON without the suffix; numbers, integers and booleans are JSON types. Statuses are the last reported project status on the status date. Absolute URLs in responses start with the base_url given in api/v1/index.json. The field dictionary describes every column and coded value, and the methodology explains how identifiers are formed and how long they stay valid. GeoJSON responses orient polygon rings as RFC 7946 requires (exterior rings counterclockwise); the CSV files and typed records keep the original ring order.

Site summaries #

sites.json, sites.geojson properties and country files.

site_id

Atlas site identifier.

name

Site or campus name.

country

Country or territory name.

country_iso_a2

ISO 3166-1 alpha-2 code.

latitude

Display anchor latitude (WGS 84). Not a footprint centroid.

longitude

Display anchor longitude (WGS 84).

geometry_type

Point, Polygon or MultiPolygon.

official_boundary

true when any project at the site uses an official boundary geometry.

locator_scopes

Geometry use scopes of the site (sites.csv geometry_use_scopes_json).

statuses

Distinct last reported project statuses of the site's projects.

newest_status_as_of

Latest project status date at the site.

oldest_status_as_of

Earliest project status date at the site.

project_count

Number of linked projects.

projects

Linked projects as {project_id, name, status, status_as_of}, sorted by name.

status_publishers

Sorted distinct publishers of the projects' status evidence.

power_observations

Typed power observations of the site's projects as {project_id, metric, stage, low, base, high, unit, as_of_date}, ordered by project name then metric. Planned or design values are not installed capacity.

verification_posture

Verification basis code (see vocabularies).

independent_imagery_verification

Whether imagery verification is claimed; false throughout.

url

Absolute URL of the site page.

api_url

Absolute URL of the full site document.

geometry_licenses

Sorted distinct licences of the site's geometry evidence rows (see the derived site field of the same name).

geometry_attributions

Sorted distinct attributions of the site's geometry evidence rows.

Country summaries #

countries.json and each country document.

country_iso_a2

ISO 3166-1 alpha-2 code (upper case); used in the country API paths.

country

Country or territory name.

slug

Lower-case ISO code, used in the country page URL.

site_count

Number of physical sites in the country.

project_count

Number of projects at those sites.

statuses

Number of projects per last reported project status (counts projects, not sites), most frequent first.

site_ids

Sorted site_id values of the country's sites.

url

Absolute URL of the country page.

api_url

Absolute URL of the country JSON document (api/v1/countries/<ISO>.json).

geojson_url

Absolute URL of the country's sites as GeoJSON (api/v1/countries/<ISO>.geojson).

csv_url

Absolute URL of the country's sites as CSV (api/v1/countries/<ISO>.csv).

Statistics #

stats.json: every map with its unit.

counts

Record counts of the version (see Counts). Keys come from the data snapshot manifest plus four derived counts.

by_status

Projects per last reported project status (unit: projects; sums to counts.projects).

by_country

Sites per country name (unit: sites; sums to counts.physical_sites).

by_status_month

Projects per month (YYYY-MM) of status_as_of (unit: projects).

by_geometry_type

Sites per geometry type; Polygon and MultiPolygon are counted separately here (unit: sites).

by_locator_scope

Projects per geometry_use_scope code (unit: projects). reviewed_site_locator here is one specific scope code; counts.reviewed_site_locator_projects instead counts every project with any reviewed locator.

by_verification_posture

Projects per verification posture code (unit: projects).

by_evidence_kind

Evidence rows per kind (unit: evidence rows; sums to counts.evidence).

by_evidence_license

Evidence rows per licence code; rows without a licence count as 'unspecified' (unit: evidence rows).

by_power_metric

Typed power observations per metric (unit: observations, not projects). Efficiency (PUE) and annual-energy observations are not included.

release_gates

Legacy research checks from selection-report.json; lists are replaced by their lengths.

units

The unit each by_* map counts: projects, sites, evidence_rows or observations.

Counts #

The counts object of index.json, stats.json and atlas.json.

physical_sites

Physical sites (rows of sites.csv).

projects

Projects (rows of projects.csv).

evidence

Evidence rows (rows of evidence.csv).

countries

Distinct countries and territories with at least one site.

non_us_sites

Sites outside the United States.

official_boundary_projects

Projects whose geometry is an official boundary (geometry_use_scope official_boundary).

reviewed_site_locator_projects

Projects with a reviewed locator (campus, project or site locator), not only the geometry_use_scope code reviewed_site_locator. Official-boundary and host-estate-context projects are not included.

host_estate_context_projects

Projects whose geometry only gives host-estate locality context (geometry_use_scope host_estate_context): the estate around the site, not a locator of the site. Present only in data updates that have such projects.

polygon_sites

Sites with a Polygon or MultiPolygon geometry (derived; MultiPolygon sites are included).

point_sites

Sites with a Point geometry (derived).

power_observation_projects

Projects with at least one typed power observation (derived). Efficiency and annual-energy observations do not count.

multi_project_sites

Sites with more than one project (derived).

imported_source_records

Original project records preserved by consolidation, including multi-site programs and unresolved identities.

snapshot_only_records

Multi-site or unresolved source records retained outside the counted site and project tables.

imported_source_references

Original record-local source references preserved in the evidence table.

Versions #

Each entry of versions.json.

id

Version identifier: the dataset directory name.

kind

Publication category: published for date-based updates; older snapshots retain their original archive category.

label

Human-readable version name.

checkpoint_date

Date the data snapshot was recorded.

lifecycle_reference_date

Date used to calculate observation age. Equals the update date for date-based updates; older snapshots retain their original reference date.

release_status

Published for date-based updates; older snapshots retain their original publication status.

publishable_as_final

True for published data updates. Historical snapshots retain the result of their original publication review.

counts

Record counts from the version's manifest.

manifest_sha256

SHA-256 of the version's manifest.json, which binds every original file.

files

Every file in downloads/<id>/: the original data snapshot files (derived false; they never change) and, for versions published as current, the derived formats, zip and SHA256SUMS written by the site builder (derived true; rebuilt by each site build).

download_base_url

The version's download directory; its index page lists every file with its checksum.

repository_url

The version's directory in the source repository.

added_site_ids

Sites in this version that were not in the previous one.

removed_site_ids

Sites of the previous version that are not in this one.

page_url

The version's HTML page with its changes and citation.

derived_downloads

Whether downloads/<id>/ also holds derived formats and a zip.

changes

Record-level changes from the previous version: for sites, projects and evidence the added, removed and changed IDs (with the names of the changed CSV columns), and columns added or removed per table.

Index #

api/v1/index.json.

api_version

API version of this path ("1").

name

Site name.

description

Site tagline.

base_url

Absolute URL every other URL starts with.

repository_url

Source repository.

dataset

The current data snapshot with counts, update date, observation date range and manifest hash. Observation dates are separate from the update date.

endpoints

Absolute URL templates of every endpoint; replace {site_id}, {project_id}, {evidence_id} or {iso2}.

downloads

Files of downloads/latest/ with format, media type, size, SHA-256 and pinned URL under downloads/<id>/.

license

Licence layering: Atlas data, code, OpenStreetMap-derived geometries (ODbL) and third-party terms.

attribution_url

Attribution and legal notices page.

docs_url

Human-readable API guide.

build

Source commit the site was built from (null for a local build).

title

Dataset title used in citations.

caveats

The four rules every reader of the data needs.

attribution

Attribution statement to use when reusing the data.

citation

Citation of the current version (APA style).

citation_url

All citation formats (api/v1/citation.json).

openapi_url

OpenAPI 3.1 description of the API.

schemas_url

Index of the JSON Schemas of every response.

llms_txt_url

Plain-text guide for language models and agents.

versions_url

Versions page.

downloads_url

Index of the download directories.

dataset_files_url

Download directory of the current version.

feeds

Atom feeds of new versions and new sites.

media_type_note

How GitHub Pages labels some download formats.

Examples #

curl and jq

curl -s https://kiankyars.github.io/datacenter-atlas/api/v1/sites.json | jq '.[] | {name, country, statuses, newest_status_as_of}'
curl -s https://kiankyars.github.io/datacenter-atlas/api/v1/stats.json | jq .by_status

JavaScript

const index = await (await fetch("https://kiankyars.github.io/datacenter-atlas/api/v1/index.json")).json();
const sites = await (await fetch(index.endpoints.sites)).json();
const first = await (await fetch(index.endpoints.site.replace("{site_id}", sites[0].site_id))).json();
console.log(first.site.name, first.projects.length, first.evidence.length);

Python

import json
import urllib.request

with urllib.request.urlopen("https://kiankyars.github.io/datacenter-atlas/api/v1/sites.json") as response:
    sites = json.load(response)
for site in sites[:5]:
    print(site["name"], site["statuses"], site["newest_status_as_of"])

Citation #

Kyars, K. (2026). Data Center Atlas (2026-10-27) [Data set]. Source files

Copy citation

@misc{kyars2026datacenteratlas,
  author = "Kyars, Kian",
  title  = "Data Center Atlas",
  year   = "2026",
  note   = "Data as published 2026-10-27",
  url    = "https://kiankyars.github.io/datacenter-atlas/downloads/2026-10-27-research-hundredsecond/"
}

Also as BibTeX file, RIS (EndNote, Zotero, Mendeley) and JSON with CSL-JSON and biblatex.

Example response #

The first 60 lines of api/v1/sites/vcc-site-7e0aadb2cd0ce2755f0d.json, pretty-printed (the file itself is minified):

{
  "api_version": "1",
  "dataset": {
    "id": "2026-10-27-research-hundredsecond",
    "label": "2026-10-27 · 838 sites",
    "release_status": "published",
    "publishable_as_final": true,
    "checkpoint_date": "2026-10-27",
    "lifecycle_reference_date": "2026-10-27",
    "updated_at": "2026-10-27",
    "observation_window_start": "2024-08-06",
    "observation_window_end": "2026-10-27"
  },
  "site": {
    "site_id": "vcc-site-7e0aadb2cd0ce2755f0d",
    "physical_site_stable_key": "curated:telehouse-canada-151-front-street-west-toronto",
    "name": "151 Front Street West Data Centre (Telehouse Canada), Toronto",
    "country": "Canada",
    "country_iso_a2": "CA",
    "latitude": 43.6446834278069,
    "longitude": -79.383946611135,
    "geometry": {
      "coordinates": [
        -79.383946611135,
        43.6446834278069
      ],
      "type": "Point"
    },
    "geometry_type": "Point",
    "geometry_source_entity_kinds": [
      "campus"
    ],
    "geometry_derivations": [
      "direct_geometry"
    ],
    "geometry_methods": [
      "city_of_toronto_one_address_repository_address_point_id_7929376_published_wgs84_unchanged"
    ],
    "geometry_scope_classes": [
      "official_municipal_address_point_of_operator_stated_facility_address"
    ],
    "geometry_authority_classes": [
      "official_source"
    ],
    "geometry_use_scopes": [
      "campus_locator"
    ],
    "geometry_precision_scopes": [
      "City of Toronto One Address Repository Land address point for 151 Front St W, the address Telehouse Canada gives for this data centre; the catalogue says points are positioned within the parcel or the building footprint. Used as the campus locator for the multi-tenant carrier-hotel building: it locates the building's civic address, not a floor, suite or surveyed building centre, and does not show where in the building any data hall lies."
    ],
    "horizontal_uncertainty_metres": null,
    "horizontal_uncertainty_unknown_reason": "The City of Toronto publishes no positional accuracy for One Address Repository points; the catalogue states only that points lie within the parcel or the building footprint, and the footprint and parcel were not retrieved.",
    "geometry_evidence_ids": [
      "073e9a8a-df12-5c3f-a8c6-179c863df5b7"
    ],
    "project_count": 1,
    "project_ids": [
      "b2d66094-8af2-56ec-973b-06672cba38f8"
    ],
    "project_stable_keys": [
…