PJM is the grid operator for a region that includes Virginia, Ohio and Indiana. Voltane has named it as its next market after Texas. This page applies the method behind Voltane's Texas time to power estimate to PJM's public interconnection history and prints what comes out.
Voltane's parcel screening is Texas today. PJM is planned, not live. Nothing on this page is an estimate for a PJM parcel or site. It describes what happened to past projects, by state.
What is measured
- Months from an interconnection request to a signed interconnection agreement, where the public file has the agreement date.
- Months from an interconnection request to operation.
- The completion rate: the share of past requests that reached operation, among those that either reached it or withdrew. Requests still in the queue are counted apart.
Each estimate is a range, not a date: a low, a middle and a high value, taken from the lower quartile, the median and the upper quartile of past projects.
Read these first
- This is generation interconnection history, not large-load history. The public record covers power plants and storage asking to connect. A data center or another large load does not go through the generation queue, and its timeline can differ.
- The process has changed. PJM moved from a serial queue to study cycles in 2023, so older history may not predict the new cycles.
- A state is not a site. A group says what happened to past projects in that state. It does not study a substation, a line or a request.
- An estimate is not a promise. It is not available capacity, a queue position or a date a utility has agreed to.
PJM's own words on the change are quoted in the notes at the end of this page, and its page is linked under Sources.
ERCOT and PJM side by side
Both rows use the same request years and the same definitions wherever the public file allows. The last column says where each market's clock stops, because the two are not measured at exactly the same point.
| Market | Requests in the fitted years | Reached operation, of those resolved | Months to a signed agreement, middle | Months to power, low | Months to power, middle | Months to power, high | Backtest median error, months | Clock stops at |
|---|---|---|---|---|---|---|---|---|
| ERCOT (Texas) | 907 | 45.1% | 22.4 | 38.8 | 50.6 | 62.3 | 11.0 | Approval to energize or synchronize where listed, else operation |
| PJM | 3,074 | 15.9% | 37.8 | 26.9 | 53.4 | 70.6 | 14.0 | Operation |
Estimates from past projects. Low, middle and high are the lower quartile, the median and the upper quartile of projects that got there. "Reached operation" is energized projects divided by energized plus withdrawn ones; requests still in the queue are in neither. Unknown means the file has no value. Data (JSON)
PJM by state
Virginia, Ohio and Indiana are the PJM states Voltane has named first, so each is shown on its own where the history is deep enough, and the rest of PJM is one group. Indiana had too few finished projects in the fitted years to stand alone, so it is counted with the rest of PJM, and the group's name says so.
| Group | Requests | In operation | Withdrawn | Still in the queue | Reached operation, of those resolved | Months to a signed agreement, middle | Months to power, low | Months to power, middle | Months to power, high |
|---|---|---|---|---|---|---|---|---|---|
| Virginia | 667 | 99 | 467 | 101 | 17.5% | 40.5 | 28.9 | 52.1 | 76.1 |
| Ohio | 364 | 65 | 211 | 88 | 23.6% | unknown | 37.4 | 66.4 | 81.9 |
| Rest of PJM (with Indiana) | 2,043 | 243 | 1,478 | 322 | 14.1% | 34.9 | 25.3 | 49.6 | 67.4 |
| PJM-wide | 3,074 | 407 | 2,156 | 511 | 15.9% | 37.8 | 26.9 | 53.4 | 70.6 |
Estimates from past projects, the same request years as the table above. Months to power run from the queue date to the operational date. A group with too few dated agreements shows unknown for the agreement column. Data (JSON)
The backtest
An estimate like this should be tested against what happened. We fitted the groups on older requests only, then compared what they predicted with newer requests the method had not seen, using the same windows as the ERCOT backtest. The result is printed here whether it flatters the method or not, and for PJM it does not: very few of the newer requests have reached operation.
| Measure | Result |
|---|---|
| Requests in the test set | 1,794 |
| Of those, in operation and counted | 27 |
| Of those, withdrawn | 1,538 |
| Of those, still in the queue | 227 |
| Median error | 14.0 months |
| Actual results inside the predicted range | 74% |
| Predicted share in operation within 48 months | 9.1% |
| Actual share in operation within 48 months | 0.9% |
| Fitted on requests through | 2019-12-31 |
| Tested on requests from | 2020-01-01 |
| Data as of | 2025-12-31 |
| Source | Berkeley Lab "Queued Up" interconnection queue data file, 2026 edition (PJM requests through 2025-12-31, CC BY 4.0), compiled from PJM's public queue |
Median error is the typical gap, in months, between the predicted middle value and what happened, among the test projects that reached operation. The predicted range runs from the low to the high estimate. Data (JSON)
- Cohort hold-out, the same windows as the ERCOT backtest. Groups and their quantiles are fitted on PJM generation requests filed 2015-01-01 to 2019-12-31 (outcomes as known at the as-of date), then compared with requests filed 2020-01-01 to 2021-12-31. For each test project that reached operation, the error is its actual months from request to operation minus its group's predicted p50; coverage is the share of those projects whose actual months fall inside the group's p25 to p75 band (50% would be ideal). Groups are Virginia, Ohio, Indiana and the rest of PJM; a state with fewer than 30 energized training projects is merged into the rest of PJM.
| Group | Test requests | In operation and counted | Predicted low | Predicted middle | Predicted high | Actual middle | Actual inside the range |
|---|---|---|---|---|---|---|---|
| Rest of PJM (with Indiana) | 1,107 | 19 | 25.3 | 52.5 | 68.2 | 38.5 | 74% |
| Virginia | 458 | 6 | 25.8 | 53.4 | 77.4 | 48.8 | 100% |
| Ohio | 229 | 2 | 39.3 | 68.3 | 82.0 | 24.4 | 0% |
Months. "Actual inside the range" is the share of the group's counted test projects that landed between the predicted low and high. A row with a handful of counted projects says very little. Unknown means the file has no value. Data (JSON)
What this does not tell you
- Fitted on PJM generation interconnection history, not large-load history. A data center or another large load does not go through the generation queue; no public project-level large-load history exists.
- PJM changed its interconnection process. "The Transition Date was July 10, 2023." PJM describes the change as "transitioning from a serial "first-come, first-served" queue approach to a "first-ready, first-served" cycle approach." (PJM, Interconnection Process Reform page, read 2026-11-09.) Every project counted here asked to connect under the old serial queue, so this history may not predict the new cycles.
- Censoring: 227 of the 1,794 test requests were still in the queue at the as-of date. They are neither failures nor successes and are left out of the error and coverage numbers. The test projects that are counted finished within 48 to 72 months, so the actual medians here run short and will rise as slower projects finish.
- Check without censoring: the model gave the 1,794 test requests a 9.1% average chance of operation within 48 months; 0.9% of them made it. The estimate ran optimistic for those request years.
- Small test set: 27 of the counted test projects asked to connect in 2020; none of the requests filed in a later test year had reached operation by 2025-12-31.
- The training projects' outcomes are as known today, not as known at the cut-off, so this is a hold-out by request year, not a forecast that was made in the past.
- Energized here means the operational date in the Berkeley Lab file. ERCOT's figures use ERCOT's energization or synchronization approval where its report lists one, which usually comes a few months before operation, so the two markets are not measured at exactly the same point.
- Agreement dates are published for few PJM projects (247 of 7,666 requests in the file, none after 2022-09-30). The months to a signed agreement rest on that subset; a group with fewer than 30 dated agreements shows unknown.
- PJM's queue lists small projects too: 41% of the energized projects in the fitted years were under 20 MW. ERCOT's public list covers larger generators, so the two are not the same mix of projects.
- A state is not a site. A group's numbers say what happened to past projects in that state, on many utilities' systems. Solar, wind, storage and gas are pooled.
Use it in practice: Texas solar site selection · battery storage screening · reading ERCOT grid data