Large-Load Regulatory Tracking
Where the rules for large loads are being rewritten in real time.
Large Load Registry
Every large-load interconnection request on file with Bonneville Power Administration — who is asking, for how many megawatts, at which substation, and how far the request has got. This part is built and populated today.
BPA first because it publishes load requests in the same machine-readable workbook as its generation queue. Other regions are queued behind publication quality, not interest — the same way the Queue Wire started with one ISO.
Why this exists
A large load (data center, fab etc) does not just need a wire. It needs to be allowed. What tariff it takes service under, what it must post as security, and whether it is even eligible are decided in state commission dockets and ISO tariff filings.
There are two halves to that, and they are at very different stages.
The first half is the registry: who has actually asked for power, where, and how much. That is a factual record, it is published as data in at least one region, and it is live above.
The second half is the regulatory record — the dockets and tariff filings that decide what those loads are permitted to do. That half is harder, and it is not built.
What the registry gives you
A list of large-load interconnection requests as filed: requested megawatts, the point of interconnection, county and state where reported, and the status of the request. Sorted by size, because that is the question people arrive with.
It covers Bonneville Power Administration today. Requested megawatts are as filed — an ask, not a study result and not an entitlement, and a long way from a built building.
The part still taking shape
The regulatory half is genuinely undecided, and it would be dishonest to describe it as though it were specified.
Georgia's PSC has run proceedings specifically on large-load tariffs. PJM has the Interconnection Request Application Status system and its own large-load registry. ERCOT has a large-load process. All of it is public. Almost none of it is published as data — it is docket PDFs and process pages.
What is not settled is the unit of publication. A feed of filings as they land? A short summary per docket with the operative change highlighted? A per-issue view that follows one question across several proceedings? Each implies a different pipeline and a different tolerance for being wrong, and picking one before reading a few hundred filings would be guessing.
So the honest status is: sources identified and reachable, nothing ingested yet, format undecided.
One slice that is already published as data
Local moratoria are the exception. Michael Bommarito maintains Moratorium Data 2026, an open inventory of county, city and township moratoria on data centers, battery storage, solar, wind and crypto mining — 1,291 instruments across 47 states at its September 23, 2026 refresh, under CC-BY-4.0. Site Scanner reports now carry it as an appendix: what it records for the jurisdictions containing the site, the others within 25 miles, and enacted state-level actions, with each status quoted as published.
It is someone else's research, reproduced with attribution, and it does not change the status above. The publisher calls its counts lower bounds, so an absence is not evidence of no moratorium; and a recorded instrument does not establish that a site is covered, because a pause's reach is set by its text.
The obvious risk
Regulatory text is easy to misread and expensive to misreport. Anything published here will quote the filing and link to it, and will not characterise a proceeding's likely outcome. Where a docket is ambiguous, it will be labelled ambiguous.
The same restraint already governs the registry. BPA publishes no end-use field, so a load request is never described here as a data center or anything else — even where the filer's own project name says so. The filed name is reproduced; the classification is not invented.
How this gets extended
Region by region, in the order publication quality allows — not in the order of how interesting the market is.
BPA was first because one daily spreadsheet carries both its generation queue and its load requests. PJM and ERCOT publish large-load information through different systems that need their own parsers. Several Western utilities publish PDFs or nothing at all, and a few put their queues behind a login, which is a hard stop rather than a slow one.
The registry is the part that works. The regulatory half below is still finding its shape, and this page will change as it does.
The other pipelines
Same rule everywhere: a figure ships only when it traces to a named public source. Roadmap →
Queue Wire
Which interconnection queue projects are advancing and which are not? This is a weekly diff report on generation projects.
Substation Expansion Tracker
Parsing PJM RTEP and TEAC filings for planned substation additions and upgrades.
Large-Load Regulatory Tracking
The Large Load Registry is live for BPA — 291 live requests, 50.5 GW. The regulatory-docket piece is still taking shape.
Large Load Registry is live →