Skip to content

Interconnection queue data

Data center and large load interconnection

Large loads now queue for grid access the way generators long have. This page collects what the public record actually says about those queues, and separates it from the generation queue, which is a different process with a different record.

GridVision AI ResearchPublished Updated

What the operators have published

410 GW
of large loads ERCOT was tracking as seeking interconnection, as of March 26, 2026
87%
of that queue described as data centers
3.6%/yr
PJM's projected summer peak growth over the next ten years, to about 222,000 MW in 2036

The ERCOT figures come from its April 2026 update to the Texas Senate Committee on Business & Commerce (ERCOT, Large Load update to the Texas Senate Committee on Business & Commerce, April 2026). The PJM figures are from its 2026 Load Forecast Report, which also projects 2.4% annual summer peak growth over twenty years, reaching 253,077 MW in 2046, and 4.0% annual winter peak growth over ten years (PJM, 2026 Load Forecast Report (January 14, 2026); PJM Inside Lines, “PJM's Updated 20-Year Forecast Continues To See Significant Long-Term Load Growth”).

A queue figure is a count of requests, not a forecast of what will be built. The generation record on this site shows what happened to requests that entered a queue: most of the capacity filed between 2000 and 2020 never reached operation. Nothing says large load requests will resolve the same way, but the same question applies — how much of this queue is one project counted once?

Large loads follow a different process

In ERCOT, large loads of 75 MW or more go through the Batch Zero process defined in PGRR145, run between the interconnecting large load entity, its transmission or distribution service provider and ERCOT, with its own forms and submission deadlines (ERCOT, Large Load Integration). That is separate from the generation interconnection process in Planning Guide Section 5, which the ERCOT queue page describes.

Elsewhere, a large load usually connects through the transmission owner and the state regulator rather than through the ISO's generator queue. So a site's power question has two halves that move on different clocks: when load can be served, and when nearby generation can reach operation.

The generation record in the states where data centers cluster

This is the half we can read node by node. For each state, the figures below are the generation interconnection requests filed there and what became of them.

StateRequestsCapacity filedReached operationWithdrewMarkets
Virginia1,334129 GW11.0%65.6%PJM, Southeast
Texas3,302711 GW8.3%25.1%ERCOT, SPP
Ohio906112 GW13.5%68.8%PJM, MISO
Illinois1,219206 GW8.2%55.7%MISO, PJM
Arizona1,046275 GW4.9%61.7%West, CAISO

Computed from Lawrence Berkeley National Laboratory, Queued Up: 2026 Edition (CC BY 4.0), requests through 2025. Shares are of requested capacity; recent filing years are unsettled.

What to check before committing to a site

  1. Which process the load itself goes through, and what stage the request is at today.
  2. What the transmission owner has said in writing about when service can begin, and on what conditions.
  3. Which generation is assumed nearby, which queue it sits in, and what stage it has reached.
  4. What the node's own record shows: how many requests were filed there, how many were built, how long each took.
  5. Which unknowns could change the answer, who owns each one, and when it resolves.

The data center power availability guide works through the evidence for each of these, and choosing a point of interconnection covers the generation half.

Want the queue record for a site you are evaluating?

Send the node and the market. We read the public queue record for it — what was filed, what was built, what withdrew and how long each took — and send back what the record does and does not show.

Ask for a node retrospective

Public-record reading only. Management review and formal interconnection studies remain authoritative.