US data-center project database

Source-verified data on US data-center projects

Nameplate Data tracks announced US data-center projects and checks each cost, financing, and megawatt claim against primary sources — filings, county records, and utility evidence. The purpose is to quantify which projects are actually financeable, and to give anyone working the market a verified baseline instead of announcement totals.

360+
Projects tracked across 30 states
1,200+
Source documents reviewed
200+
Cost claims we refused to book
A–D
Evidence depth grade on every project
Why it exists

Announced capacity is not a usable number.

Announced US data-center projects add up to tens of gigawatts and hundreds of billions of dollars, but most of those figures trace to a press release rather than a filing. Anyone underwriting power, credit, or land against that pipeline needs to know which projects have sites, interconnection evidence, and capital behind them. Nameplate Data checks each claim against project-specific sources and publishes what holds up, what does not, and why.

$
Repeated dollar claims

Program-level totals get smeared across sites. Trade-press estimates get cited as company disclosures. We book a dollar figure only when it traces to a project-specific source.

MW
Substituted megawatts

A contracted PPA capacity gets reported as site load. A multi-phase total gets reported as phase one. Scope substitution fails more often than dollar claims.

Anonymous queues

Public interconnection queues are aggregated and anonymous by design. Matching an announcement to grid reality takes primary-source work at the utility and county level.

How it works

From source document to a graded project

1Source and review

Filings, utility and county records, regulatory testimony, corporate disclosures, and trade press are collected and individually reviewed. Each source keeps its review status.

2Book or refuse

Every cost, financing, and megawatt claim gets a scope check and a confidence grade — or it goes to the holdout table with a written reason and the list of sources checked.

3Grade and track

Each project carries an evidence depth grade (A–D) and model-derived 0–100 indicators. Monthly snapshots freeze into an immutable ledger the model cannot rewrite.

Evidence-first

Every number traces to a document you can open

Each cost estimate, MW capture, permit, and financing event carries a confidence level and a link to the original filing or article, with its review status visible. There are no black-box scores.

Per-fact provenance — a source document behind every dollar and megawatt.
A refusal trail — the claims we chased and would not book, with reasons.
Evidence depth grades — A through D, on the same criteria for every project.
A prediction ledger — immutable monthly snapshots, graded in public.
Power plausibility Watch
2.0 GW
Claimed
On file
1.2 GW
Corroborated
Claimed 2.0 GW has no interconnect record on file — load is asserted, not substantiated.
Illustrative workbench panel — not live data. Every figure links to its source in the app.
Who uses it

For project diligence and market analysis

Market analysis

Analysts, LPs, and journalists

Track which announcements are real, which died quietly, and what state-level totals look like once every claim needs a source. The notes, holdout tables, and rollups all link to the underlying documents.

Weekly notes from the Nameplate Report
Holdout tables and evidence depth grades
State-level rollups of announced vs. sourced capacity
Project diligence

Utilities, IPPs, infra credit, and PE

Triage the pipeline, corroborate a specific project, and hand off diligence with the source trail attached. API and MCP access put the data in your own tools.

Aggregate queue-reality context per ISO
Per-project corroboration evidence and dossiers
IC memo exports, bulk export, API keys, MCP server

Start with the notes or the workbench.

The weekly notes are free. The workbench holds 360+ tracked projects with the evidence, grades, and refusals attached.