Data API
A key-authenticated read API and MCP server over our sourced datasets — live grid demand, 5-year daily history, and generation-mix trends by ISO. JSON, CORS-enabled.
Overview
Base URL https://datacenters-data.builtfor.ai/v1. All responses are JSON. Data endpoints require an API key; the index is open. Covers power (grid demand & mix across the seven U.S. ISO/RTOs — PJM, MISO, ERCOT, CAISO, SPP, NYISO, ISO-NE), data-center facilities, data-center legislation, deals & SEC filings, and AI-compute datasets. Sources: U.S. EIA, OpenStreetMap, LegiScan, SEC EDGAR, and Epoch AI — processed and refreshed by datacenters.builtfor.ai.
Authentication
Pass your key as ?key=… or an X-API-Key header. Keys are issued with a paid plan (see below).
curl "https://datacenters-data.builtfor.ai/v1/power/history?iso=CAISO&key=YOUR_KEY" # or curl -H "X-API-Key: YOUR_KEY" "https://datacenters-data.builtfor.ai/v1/power/latest"
Endpoints
| Method | Path | Description |
|---|---|---|
| GET | /v1 | API index — endpoints, ISOs, auth (no key required). |
| GET | /v1/datasets | List available datasets. |
| GET | /v1/power/latest | Current grid demand by ISO (live, hourly). |
| GET | /v1/power/history?iso=CAISO | 5-year daily demand history — GW avg / peak / min. Optional &from=YYYY-MM-DD&to=YYYY-MM-DD. Add &format=csv for a CSV download. |
| GET | /v1/power/mix?iso=CAISO | Monthly generation-mix shares (%) for an ISO. Add &format=csv for a CSV download. |
| GET | /v1/facilities?state=Virginia | Data-center locations with operator, county, footprint (sqft) & state-level ISO. Filter by state, iso=PJM, or operator=Amazon. IM3 Atlas (ODbL). |
| GET | /v1/policy?state=VA | Data-center bills — counts by state; state= (2-letter) returns that state's tracked bills. |
| GET | /v1/deals | Recent AI-infrastructure deals & SEC filings. |
| GET | /v1/compute?dataset=datacenters | Epoch AI: datacenters | clusters | companies | models. |
| GET | /v1/filings?company=Equinix&relevant=1 | Historical SEC filings archive (2022→now) mentioning "data center". Filter by company, form, year, state, relevant; paginate; format=csv. |
| GET | /v1/filings/live?q=data+center | Real-time SEC EDGAR full-text passthrough — the latest filings, beyond the stored archive. |
| GET | /v1/records?type=facility | Curated cross-dataset record index (facilities, policy, compute, deals, power). |
| GET | /v1/search?q=virginia | Keyword search across the curated record index. |
| GET | /v1/search?mode=semantic&q=... | Semantic (embeddings) search across the full corpus via Vectorize. Optional &type= to restrict. |
| GET | /v1/sources | Source provenance packets, for citations. |
| GET | /v1/index/search-documents | Embed-ready search documents (JSONL) — the input to a semantic-search pipeline. |
Example response
GET https://datacenters-data.builtfor.ai/v1/power/history?iso=CAISO&key=YOUR_KEY
{
"iso": "CAISO",
"granularity": "daily",
"units": "GW",
"range": { "start": "2021-06-19", "end": "2026-06-19", "years": 5 },
"count": 1827,
"daily": [
{ "d": "2021-06-19", "avg": 30.5, "peak": 40.3, "min": 24.1 },
…
]
}MCP server
The same data is available over the Model Context Protocol (Streamable HTTP) so an MCP client — Claude, Cursor, or your own agent — can query it as tools. Point your client at the endpoint below with your key, then ask in natural language (“What’s CAISO peak demand over the last year?”).
https://datacenters-data.builtfor.ai/mcp?key=YOUR_KEY
Tools exposed:
| Tool | Description |
|---|---|
power_latest | Current grid demand by ISO (GW). |
power_history | 5-year daily demand history for an ISO (avg / peak / min). |
power_mix | Monthly generation-mix shares for an ISO. |
list_isos | Available ISO/RTO codes. |
facilities | Data-center locations w/ operator, county, footprint; filter by state, ISO, or operator. |
policy | Data-center bills — counts by state, or a state’s bills. |
deals | Recent AI-infrastructure deals & SEC filings. |
compute | Epoch AI: data centers, GPU clusters, AI companies, model compute. |
filings | Historical SEC filings archive; filter by company, form, year, state, relevance. |
edgar_search | Live SEC EDGAR full-text search — the very latest filings. |
search | Keyword search across the curated cross-dataset record index. |
semantic_search | Meaning-based (embeddings) search across the full corpus; optional type filter. |
Indexing & search
Every dataset is also rolled into a unified, citation-ready index so the whole corpus is queryable and semantic-search-ready. Three artifacts are produced on each refresh:
| Artifact | Endpoint | Purpose |
|---|---|---|
records.json | /v1/records | Canonical typed records for dashboards/API. |
search-documents.jsonl | /v1/index/search-documents | Embed-ready documents (one per line) — the input to a vector/semantic-search pipeline. |
source-packets.json | /v1/sources | Per-source provenance & licensing, for citations. |
Semantic search is live — all 23,000+ documents are embedded (Workers AI, bge-base-en-v1.5) into a Cloudflare Vectorize index. Query it with /v1/search?mode=semantic&q=… or the semantic_search MCP tool; /v1/search without a mode is keyword. SEC filings are also queryable structurally via /v1/filings.
Get a key
API and MCP access are included with a paid plan. Subscribers get a personal key automatically — it appears on your welcome page after activation, where you can copy or regenerate it. CSV export is available now (format=csv on the history and mix endpoints). For higher rate limits or bulk data, contact us. Roadmap: native Excel export and higher-tier rate limits.
About the datacenters.builtfor.ai Data API
The Data API exposes the same sourced, regularly-refreshed datasets behind our dashboards as JSON: live U.S. grid demand by ISO, five years of daily demand history (average, peak, minimum), and monthly generation-mix trends. It is key-authenticated and available now with any paid plan. See the methodology for sourcing and the pricing page for plans.