PJM Queue — Note 001

What actually survives the interconnection queue, and how long it takes

What actually survives the interconnection queue, and how long it takes

Kenneth · 2026-08-09 · Arthur Energy Partners

*Source: Lawrence Berkeley National Laboratory, Queued Up 2026 Edition (CC BY 4.0),

complete interconnection request dataset through end-2025. Analysis my own.*

The headline

PJM has withdrawn 68% of every interconnection request ever filed. Of 7,666

PJM requests in the dataset: 5,226 withdrawn, 1,192 operational, 1,134 still

active, 114 suspended.

That number is widely quoted. What follows is not.

1 · The withdrawal rate is not one number. It is eleven.

Of everything that reached a terminal outcome — built or withdrawn:

**A gas project in PJM is roughly three times likelier to get built than a

battery, and nearly five times likelier than a solar-plus-storage hybrid.**

Offshore wind at 62 withdrawals against a single completion is not a queue

statistic. It is an obituary.

The composite 68% figure conceals a spread from 34% to 98%. Anyone applying the

average to a specific project is mispricing it by a factor of two or three in

either direction.

2 · Two-fifths of the "active" queue is not active

Of 1,134 PJM requests currently marked active, **447 — 39% — were filed in 2021

or earlier.** Some entered the queue nine years ago.

Given a median time to operation of 3.2 years, a request filed in 2019 and still

sitting in "active" has already taken twice the median without reaching service.

Counting the active queue as pipeline overstates it by roughly forty per cent.

3 · Getting built takes 3.2 years, and the tail is brutal

Across 1,180 completed PJM projects, measured request date to in-service date:

The gap between the fast decile and the slow decile is six years. That spread,

not the median, is what should drive a financing assumption.

4 · Where the live queue actually sits

Active PJM requests by state:

Virginia holds more active capacity than Ohio and Indiana combined — data

centre load, visible in the queue years before it appears anywhere else.

And note Pennsylvania: second by project count, sixth by capacity. **PA's queue is

made of small projects.** Different market, different buyers, and it is invisible

if you look at gigawatts alone.

What I would want next

Withdrawal rate by cohort year — is it worsening, or is the composite figure

dragged down by a decade of dead solar? And whether the Virginia concentration is

new load or repeat filings against the same interconnection points.

Both are answerable from the same dataset. Note 002.

*Method: LBNL's complete request file, filtered to entity = PJM. Failure rate

counts only terminal outcomes, excluding active and suspended requests. Capacity

uses mw_1. Nothing here is modelled or estimated — every figure is a count from

the published data, and the file is free at emp.lbl.gov/queues if you want to

check my arithmetic.*