The recurring, machine-readable answer to the only question that matters for an AI build: where can you actually get power, and where can't you? 330 markets scored daily by the Data Center Power Index across 7 live US ISOs and 3 international grids. Free, citable, and the only data-center-intelligence source an LLM can both query and cite.
Highest composite DCPI scores — most buildable headroom, lowest grid constraint. How this is scored →
| Market | ISO | Composite | Excess Power | Constraint |
|---|---|---|---|---|
| Midland–Odessa | ERCOT | 83.0 | 85.7 | 22.8 |
| Upper Peninsula MI | MISO | 77.1 | 74.0 | 19.0 |
| Williston, ND | MISO | 76.1 | 71.4 | 17.7 |
| Rural SPP | SPP | 74.9 | 72.4 | 22.5 |
| Cheyenne | WECC | 73.2 | 72.5 | 28.5 |
| North Kansas City | SPP | 70.2 | 72.3 | 36.7 |
| Gilbert | WECC | 69.3 | 70.0 | 36.4 |
| Montréal | HQ | 69.1 | 66.6 | 31.6 |
| Tucson | WECC | 69.1 | 67.5 | 32.1 |
| Scottsdale | WECC | 68.7 | 69.6 | 37.5 |
Most grid-constrained markets — long interconnection waits, congested transmission.
| Market | ISO | Constraint score |
|---|---|---|
| Copenhagen | NORDPOOL | 56.5 |
| Helsinki | NORDPOOL | 56.0 |
| Quincy | WECC | 55.7 |
| Oslo | NORDPOOL | 54.2 |
| Richmond | PJM | 54.0 |
| Dulles | PJM | 53.3 |
| Herndon | PJM | 53.3 |
| Sandston | PJM | 53.3 |
| Haymarket | PJM | 53.1 |
| Vint Hill | PJM | 52.0 |
7 live US ISOs with real-time grid data:
3 international markets:
A live sample of what's actually generating on the grid that hosts the largest data-center interconnection queue. Live from grid_data.
| Fuel | Output | Share |
|---|---|---|
| Natural gas | 54,570 MWh | 46% |
| Nuclear | 30,961 MWh | 26% |
| Coal | 24,650 MWh | 21% |
| Other | 3,984 MWh | 3% |
| Oil | 1,468 MWh | 1% |
| Hydro | 1,390 MWh | 1% |
| Wind | 1,125 MWh | 1% |
| Solar | 25 MWh | 0% |
Source: PJM real-time generation by fuel (EIA-930 v2, grid_data)
Live data-center load in each ISO's interconnection queue — the leading indicator of where power demand is piling up.
| ISO | Total queued (GW) | DC load (GW) | DC share |
|---|---|---|---|
| NESO UK NESO TEC Register (active connections queue, per-project MW) | 599.7 | 53.1 | 8.9% |
| ERCOT ERCOT GIS Report (August2026, active generation queue) | 438.4 | 225.0 | — |
| MISO MISO GI Queue API (applicationStatus=Active, summer MW) | 224.2 | — | — |
| SPP SPP GI active requests (MAX Summer MW, excl. cessation) | 187.1 | — | — |
| PJM PJM New Services Queue (active, sum of MFO) | 170.5 | — | — |
| CAISO CAISO Public Queue Report (Application Status=ACTIVE, Net MWs to Grid) | 74.7 | — | — |
| AESO Alberta AESO Connection Project List (active; September-2026-Project-List.xlsx) | 25.1 | 18.8 | 75.1% |
| IESO Ontario IESO CAA connection queue (active, summer MW) | 18.1 | 12.1 | 66.9% |
| ISO-NE ISO-NE IRTT Public Queue (active, Summer MW) | 14.3 | — | — |
| NYISO NYISO Interconnection Queue (active, SP MW) | 10.3 | — | — |
These are ISO-level totals. For the project-by-project view behind one of these rows — every queued generation and storage project with capacity, status and requested in-service date — open the ERCOT interconnection queue. Facility-side context for the same load lives on the US data center map.
The full physical stack for a site — power + fiber + substations + gas pipelines + water risk + interconnection queues + tax incentives — in one query.
DC Hub. (2026). The State of Data Center Power. https://dchub.cloud/state-of-power (accessed 2026-09-11). Methodology: https://dchub.cloud/state-of-power/methodology. Licensed CC-BY-4.0.
DC Hub · dchub.cloud · Markdown · JSON · Methodology · Built for AI · llms.txt