License the whole index
Funds, developers and utilities use the Data Center Index directory as a data feed: 612 data center campuses and 266.6 GW of tracked capacity across 45 US states, refreshed daily, with a source URL on every figure.
What the licence covers
- Every field, every row. Campus name, operator, tenant, city, state, coordinates, status, capacity, planned and operational MW, estimated cost, expected completion, and the public companies linked to the build.
- Full history, not a snapshot. Every status and capacity change is dated, so the index can be reconstructed as it stood on any past day.
- A source URL on every figure. Each capacity is labelled filed (a permit, utility or SEC filing states it), stated (the developer or press states it) or estimated (modeled), carrying the date of the newest source behind it.
- Coverage on request. Name a market and it is swept and sourced.
- Delivery that fits the stack. CSV, GeoJSON, a scheduled drop, or the JSON API and MCP server.
How the data is built
Every row starts from a named source: an SEC filing, a permit or planning record, a utility or commission filing, a company project page, or reporting that quotes one of those. Aggregator listings are never the source of a fact. Changes go through a proposal queue that records the URL, the quote and what the quote measures, so a phase figure cannot be filed as a campus total.
Accuracy is measured rather than asserted: a random sample is re-verified against an independent second source, per-field error is reported with a confidence interval, and every row a past audit corrected is re-read in the next one. Automated checks run nightly and fail on a figure without a date behind it.
Two upstream sources are redistributed under CC BY 4.0, Epoch AI's Frontier Data Centers and the AI Data Center Index (SheetGenius, Inc.), and are credited on every export.
Talk to us
Send an enquiry from this page or email hello@datacenterindex.ai.