Speed to power is now the binding constraint on a large build, ahead of land, capital, or chips, and the reason is the connection queue: the line to plug into the grid. About 2,300 GW is waiting in United States queues, roughly 77% of requested capacity historically withdraws before connecting, and the typical completed project waited about 55 months (Lawrence Berkeley National Laboratory, Queued Up, 2025). It is also the least transparent input in siting: every grid operator reports it differently, on its own cadence.
One normalized, quarterly-refreshed read on how hard it is to connect a large load in each market, and how long it takes. Per market: active queue depth, typical wait, historical withdrawal rate, and data-center share of the queue, each labelled measured or modelled and cited. Delivered as a gated API endpoint, a quarterly file drop (CSV and JSON), and a one-page quarterly brief with a plain-language change log. It sits next to our rating of 84 power markets, and it is versioned and dated like a benchmark, not a one-off report.
| Region | Markets | Depth |
|---|---|---|
| United States | 10 grid operators: ERCOT, PJM, CAISO, MISO, SPP, NYISO, ISO New England, Pacific Northwest, US Southeast, Desert Southwest | Deep. Eight measured, two modelled, all labelled. |
| Europe, live now | Great Britain, Ireland, Netherlands, Germany, Spain, France, Sweden, Norway, Finland | Queue or connection-friction reads, each labelled measured or modelled. |
| Also live | Australia | Queue read, measured. |
| On the roadmap | Italy, Poland, Denmark, and more | In build. Design partners choose the order and shape the fields. |
See the ranked cross-market view at /connection-friction. Much of the United States raw material is public. What we add is what nobody else does: cross-market normalization into one schema, a quarterly synthesis with a dated change log, integration with the PGIQ Rating, and the European primary compilation, for which there is no free equivalent.
The connection data sits inside a published governance stack, which is what separates it from a consultant's slide: a dated, public track record of every call, a pre-announced review calendar, a request-for-comment policy before any methodology change, and a baseline measurement plan that states how our calls will be graded once realized outcomes exist. An append-only point-in-time archive, with published hashes, means the history is not quietly rewritten.
We are taking a small number of founding design partners while we deepen European coverage. Europe is live now and expanding: you get the full feed at current coverage, first access to each new market and field as it ships, and a seat that shapes which European markets we build out next. Tell us a little and we will set up a 25-minute call.