Demand-side statements and generation/storage queues are separate kinds of evidence. The two demand accounts below contain no capacity total; Berkeley Lab's separately dated generation and storage figures follow.
Large-load demand evidence: separate processes and categories
The two accounts below concern demand-side evidence. PJM forecast adjustments are not a load queue. Statnett's reserved and ordinary-queue categories can overlap. Neither account supplies a queue-capacity total, current headroom or a connection guarantee. Do not add these categories to each other or to the generation/storage figures below.
PJM: forecast adjustments, not a load queue
PJM says it has no load-interconnection queue. It uses large-load adjustments submitted by load-serving entities when preparing its long-term load forecast.
No queue capacity, firm connection commitment, forecast outcome or current commissioning inferred. No consultant findings or utility attachment data imported.
Only minimal independently worded attributed facts and ordinary unframed links are used. No source narrative, table, chart, marks, personal contacts, dataset or full document is copied. No open licence or recurring-monitoring grant is asserted. Reuse terms: No open licence recorded. Source site notice; this link is not a licence grant.
Statnett: queue and reservation category overlap
Statnett distinguishes queued, reserved and connected cases. It says a project with capacity reserved on special conditions can also appear in the queue for ordinary terms.
Statnett connection-case statistics; source data may be incomplete
Quantity / metric
Not recorded; this account is not a queue total
No PowerBI, customer list or download inspected/imported. No numeric total or named project; categories may overlap and must not be summed.
Only minimal independently worded attributed facts and ordinary unframed links are used. No source narrative, table, chart, marks, personal contacts, dataset or full document is copied. No open licence or recurring-monitoring grant is asserted. Reuse terms: No open licence recorded.
Download these same two accounts as JSON or CSV. Their 31 October evidence checks do not renew Berkeley Lab's separate 3 October check.
Generation and storage: what an interconnection queue is
Figures from Berkeley Lab's Queued Up 2026 Edition, covering data to the end of 2025. Each figure is shown with its metric, date and source.
Generation and storage data below: Lawrence Berkeley National Laboratory, in collaboration with GridTracker. Licence: CC BY 4.0. We link to the original and have not changed the figures. We have not added any figures together.
A power plant or battery that wants to plug into the high voltage grid has to ask the grid operator first. The operator studies what the connection would do to the grid and what upgrades it needs. Requests wait in line for those studies, and that line is the interconnection queue. Queued Up counts the projects in these lines.
What it means for data centres
We only state what the source says.
The data presented here only include resources that supply electricity to the transmission grid (i.e., generation and storage); there are separate queues for large loads and those are not included in this report.As of end of 2025. Quoted from the 2026 Edition PDF. The source does not mention data centres by name.Source
So Queued Up is a record of supply waiting to connect, not of data centre demand. It helps with one question: how much new generation and storage is applying to join each part of the US grid. It does not say how much of that is for data centres.
Headline figures
National figures, 2026 Edition, as read on the Berkeley Lab pages
These rows are the resource types the source reports separately. They do not add up to the national total, and we have not worked out a figure for the rest.
What we do not know
How much of any queue is for projects meant to power data centres. The source does not say, and we have not found a figure on the pages we read.
The size of the separate large load queues. The source says they exist and are not part of this report. We have not read a source that gives their size.
Anything outside the United States. Queued Up covers US grid operators only. Queue figures for other countries are not on this page.
Which queued projects will be built. The source reports outcomes for older requests (13% reached operation, 75% withdrawn, 10% still active for requests from 2000 to 2020), and that does not forecast the current queue.
Region by region comparisons of speed or success. Queue capacity by region is a count of what is waiting, not a measure of how fast an operator connects projects.
Where a region row lists fewer projects than the national total suggests, or where figures differ between the landing page and the PDF (for example, rounding), we have shown each as the source states it and have not reconciled them.
Limits the source states about its own data
Commercial operation dates are unknown for about 27% of operating plants in the data. Source
Dates of draft or executed interconnection agreements are unknown for about 45% of requests that have one. Source
Withdrawal dates are unknown for about 37% of withdrawn requests. Source
Hybrid storage capacity is missing or partly missing in most regions (complete in CAISO and ERCOT) and is estimated where missing. Source
Data, licence and credit
The data is compiled by Berkeley Lab in collaboration with GridTracker. The source page states: The Queued Up data file is licensed CC BY 4.0. You may use, share, or adapt the dataset as long as you attribute it to Lawrence Berkeley National Laboratory and GridTracker.Source. Licence text: CC BY 4.0. The wording on the page covers the data file. We copy only short facts and one table of figures, with attribution, and link to the file for anything more.