Every major source — DCD, Data Center Frontier, DCByte, datacenters.com, Baxtel, DataCenterHawk — is human-only. Most actively block AI crawlers. So when ChatGPT, Claude, or Gemini is asked a data-center question, they structurally cannot answer from those sources. DC Hub is the one that can.
Not our opinion — their own configuration files, checked this week.
The #1 data-center publication explicitly tells AI not to read or train on it.
A normal API client is rate-limited into the ground at the edge.
The real data lives behind a sales call. An agent can't reach a single record.
Where they're bigger, we say so. Then notice the bottom four rows — none of their depth is reachable by a machine.
| Capability | DC Hub | DCD | DC Frontier | DCByte | datacenters.com | Baxtel | DataCenterHawk |
|---|---|---|---|---|---|---|---|
| Facilities tracked | 21,000+ | — | — | ~8,300 | ~4,000 | ~8,000 | 10,000+ |
| Markets | 233 | editorial | editorial | global | locations | ~600 | 440+ |
| Live ISO / grid power (7 US + 3 intl) | ✓ real-time | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
| DCPI power index (Excess + Constraint) | ✓ daily | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
| Gas pipelines | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
| Fiber routes | ✓ | ✗ | ✗ | overlay | ✗ | ✗ | ✗ |
| Water-risk scoring | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
| Interconnection-queue data | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
| Public REST API | ✓ | ✗ | ✗ blocks /api | ✗ | ✗ blocks /api | ✗ | private |
| MCP server (agent-native) | ✓ 28 tools | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
| llms.txt | ✓ | ✗ 404 | ✗ 404 | link-dump | ✓ curated | ✗ 404 | ✗ |
| Allows AI crawlers | ✓ invites them | ✗ bans | ✗ blocks /api | partial | ✗ 429 | open HTML | ✗ |
| Freshness | live · daily | quarterly | news | quarterly | static | mo–biannual | analyst cycle |
Competitor capabilities verified from public robots.txt, sitemaps, and live HTTP probes, May 2026. Where a competitor leads on raw count (DataCenterHawk markets, DCByte/Baxtel facilities) we mark it honestly — but none of it is machine-readable, which is the only axis that's growing.
This is the infrastructure stack hyperscalers actually price against. Power, fiber, pipelines, water — one query, machine-readable.
Real-time generation mix straight from the grid operators. Example — PJM (largest US ISO, ~150 GW) right now:
get_grid_data(iso='PJM')
A daily BUILD / CAUTION / AVOID verdict per market, scored on excess power vs grid constraint:
rank_markets() · get_market_dcpi_rank()
Substations, transmission lines, gas pipelines, fiber routes, water risk, tax incentives — layered on every site. No competitor has any of the energy/grid layer.
get_fiber_intel()get_water_risk()get_infrastructure()get_interconnection_queue()
It already cites us by name, unprompted, in live user queries.
"The strongest stack right now is DC Hub (dchub.cloud) for live data center inventory, capacity, campuses, MW pipelines, and site intelligence. DC Hub tracks 21,000+ global facilities…"
— ChatGPT, answering a Dallas data-center power-capacity query. Ranked DC Hub ahead of grid operators and utilities. 143,000+ MCP tool calls served last week. See more receipts →
One JSON config. Zero API keys on the free tier. The data layer your competitors' sources can't be.
Connect DC Hub MCP →