Speed to power, county by county.
How long it takes to get power in a county, how much capacity is really available, and which state rule could stop you — in one sourced document, in about a minute.
The sample is the whole document, free, no email. Your own county costs nothing either — one per account, about a minute, no card.
Interconnection decides the site now
Not capital, not chips, not construction. Four things go wrong, and all four have a date attached.
The queue outlasts your schedule
Over five years from request to operation for projects energised in 2025. Of capacity that entered queues 2000–2020, 75% was withdrawn and 13% got built.
The window shuts once a year
Miss a cluster close and the next chance is twelve months out. CAISO Cluster 16 closes 15 October; ISO-NE 19 November.
The state can change the rules under you
300+ state data centre bills in the first six weeks of the 2026 session. 24 states have an approved large-load tariff.
Most of what you benchmark is phantom
Five to ten times more interconnection requests than data centres that get built. Separating real from speculative is the actual work.
We measured who publishes this. Two of fifty-six.
Most companies put customer logos here. We have one paying customer, so instead here is a finding you can check yourself.
We parsed the published interconnection queue of 56 US transmission providers. Two of them published a load-side queue. Generation interconnection is disclosed by federal rule; large load is disclosed by choice, and almost nobody chooses. That gap is why a screening question that should take an afternoon takes a month.
A document, not a dashboard login
It renders from the database when you ask for it and never changes afterwards. A figure you cited in March is still provable in June — which is the whole point if it went into a board paper. Read a real one before you decide.
The county picture
- What is queued — applications and capacity, by fuel
- What was actually built, and what was withdrawn
- The median wait, and how the county ranks nationally
- Large-load requests recorded nearby
- The deadlines that bind projects there
- A provenance page naming every source read
Not behind the price
- The interconnection deadline calendar — nobody should pay to learn a date exists
- The Load Ledger — 531 large-load interconnection requests
- 1,607 county pages and 47 state policy pages
- The Transparency Index — an index arguing for disclosure cannot be paywalled
- An API and MCP server: a key issues instantly with 1,000 free credits
Priced against the decision, not the seat
Site-selection software runs roughly $200 to $2,000 per site evaluated. Full-service consulting runs about $10,000 per site. A screening pack sits deliberately between the two, and it is billable through to a client as a disbursement — which a subscription is not.
One county is free, a one-off address report is $24, Home is $19.99 a month and Pro is $249. The full grid, with what each one includes, is on packages. The one thing that only lives here is the screening pack, because it is scoped rather than bought.
Request a Site Screening Pack
Name the counties, or just the states you are looking across — we will scope it and come back within one business day. Nothing is charged until the scope is agreed.
The $4,500 market report is the one thing that is not self-serve — the API declines it on purpose rather than queue something it cannot render. Email us and a person scopes it with you.
Three steps, no call
Pick a county
Any of the 1,575 we hold filings for. Type the name — no FIPS, no map hunting.
We render it
About a minute, then it arrives by email with a link that works for seven days without an account.
Check our working
The last page names every source read and the date each figure was true.
What we do not have
Written down here rather than discovered later, because the whole argument of this company is check every number — and that has to include ours.
- No customer logos. We have one paying customer. Anything on this page that looks like social proof is a figure you can verify, not a testimonial.
- We do not model power flow. We tell you a project exists and what happened to comparable ones. We cannot tell you whether a specific interconnection is physically feasible — that is a study, and Enverus bought a company to do it.
- We are not the largest dataset. On the generation queue, interconnection.fyi is free, daily and holds more records than we do. On large load it holds 575 requests to our 531. We say so because you would find out anyway.
- Our policy coverage is narrow — moratoriums, large-load tariffs, siting rules and interconnection reform. It is not a substitute for a legislative tracking service.
- We publish our own mistakes in a public corrections log, including one about our own figures this month.
The things people actually ask
Is the free report really free, or is it a sample chapter?
It is the full renderer — same code, same provenance page, same five pages a paying customer gets, for one county, once per account. No card and no trial clock. If you would rather not hand over an email at all, a complete report is published at /sample/.
How current is the data?
Every page and every report carries the date of the ingest it was built from, printed on the page rather than buried in a footer. If a figure is three weeks old, it says so.
What happens if a figure turns out to be wrong?
We correct it, date the correction, and publish it in a public corrections log alongside our own errors. A generated report never silently changes — version 1 stays retrievable at its own link for ever, and a newer version is offered separately.
Can an AI agent use this without a person?
Yes — that is what it was built for. A key issues instantly at POST /v1/keys with 1,000 free credits and no human approval, and the MCP server is published in the official registry as net.wattstreet/grid.
Start with the free one
Pick a county, give us an email, and a real report is generated and sent in about a minute. If it is not useful you have spent nothing and told us nothing.
One county. No card. One free report per account, ever.