Who builds it.
Who pays for it.
A living public ledger of the AI data-center buildout. Every figure is receipt-backed, every claim has a status, and missing information stays visible.
The buildout at a glance
Coverage and disclosure patterns in the current public release.
Facility status
Core-field disclosure
Recent records
A sample from the ledger. Select any row to inspect receipts and claim states.
Facility ledger
Filter by geography, utility, or project status. Click a row for field-level receipts.
Residential rates by utility territory
Nominal and inflation-adjusted changes belong side by side. Driver notes report what rate cases cite; they do not assign causation.
| Loading utility rates… |
Unknowns are part of the result
The ledger counts missing disclosure instead of converting it to zero or dropping the record.
“A negative result describes the method, not the source, until every capability warranted by the evidence has been tried.”
Every negative records its method chain. Older negatives without one remain explicitly reopenable.
Five claim states
Verified — directly supported by the cited receipt.
Estimated — derived with a stated method.
Disputed — material sources conflict.
Undisclosed — the holder does not publish it.
Unknown — not established by the search performed.
Correction ledger
Contradictions and self-corrections remain in the public files. The current release contains 83 correction records, preserving what changed and why.
Method, provenance, and downloads
Built for a reader who needs to verify a number without asking the project team.
Map the claims
Record what advocates, industry, utilities, regulators, economists, labor, residents, and parties assert.
Search against each position
Look for evidence each side would least want found, including flat rates and delivered jobs.
Bind facts to receipts
Store exact locators, source URLs, archive copies, hashes, and claim states.
Preserve uncertainty
Log failed methods, contradictions, disclosure gaps, and later corrections.