US interconnection queues hold 2,061 GW of generation and storage at year-end 2025, down 10% from 2,290 GW a year earlier and 21% from the 2,598 GW year-end 2023 peak. Withdrawal rates have outpaced new requests for two consecutive years, with the contraction concentrated in solar, storage, and wind while gas capacity in queues rose 86% in 2025. Only about 13% of capacity that entered the queue from 2000 to 2020 reached commercial operation, and the median time from interconnection request to commercial operation now runs 61 months.
Last updated May 27, 2026.
Active capacity is requests that have not withdrawn and have not yet reached commercial operation. The 2019 to 2020 storage jump partly reflects an LBNL methodology change: estimated hybrid-storage capacity is only included from 2020 onward. The 2025 contraction is concentrated in solar (down 19% year-over-year), storage (down 16%), and wind (down 19%); gas capacity in queues rose 86% to 253 GW and the "other" bucket (coal, nuclear, hydro, geothermal, unknown) rose 86% to 67 GW. Active capacity is not committed capacity; historical capacity-weighted completion rates run about 13%.
Source: Lawrence Berkeley National Laboratory, Queued Up: 2025 Edition data workbook, sheet 08 · As of Year-end 2014 to year-end 2025
The West non-ISO bucket covers WECC outside CAISO and is the largest single region at 567 GW. ERCOT rose from 346 GW to 408 GW year-over-year and is now the second-largest queue. CAISO fell from 273 GW to 191 GW (down 30%) as cluster 15 Interconnection Process Enhancement scoring forced resubmissions. ERCOT's own published total runs about 432 GW (late 2025) because ERCOT counts active GINR requests on a different methodology than LBNL.
Source: Lawrence Berkeley National Laboratory, Queued Up: 2025 Edition data workbook, sheet 09 · As of Year-end 2025
Capacity-weighted share of interconnection requests submitted between 2000 and 2020 that had reached commercial operation by end-2025. The US-average rate is 13.1% by capacity and 19.1% by project count. ERCOT leads at 24%, CAISO trails at 8%. This is a backward-looking cohort with enough elapsed time for most projects to have completed or withdrawn; FERC Order 2023 and the ISO-level reforms approved 2024 to 2025 are not yet visible in this dataset.
Source: Lawrence Berkeley National Laboratory, Queued Up: 2025 Edition data workbook, sheet 25 · As of Status as of end-2025
Non-ISO balancing authorities only (PacifiCorp, BPA, Duke Carolinas, Duke Progress, Duke Florida; n = 2,100+ project-level estimates). Withdrawn projects in 2018 to 2024 averaged $671/kW; completed projects in the same period averaged $194/kW. The gap is a selection effect: projects assigned high network-upgrade costs tend to drop out. LBNL slide 22 shows ISO-side completed-project costs averaged $88/kW over the same 2018 to 2024 period, about half the non-ISO BA figure.
Source: Lawrence Berkeley National Laboratory, Generator Interconnection Costs in non-ISO Balancing Authorities (February 2026), slide 18 · As of 2000 to 2024 study periods
A new power plant or storage project that wants to connect to the bulk power system files an interconnection request (IR) with the transmission provider for its location. The provider studies the request to determine what network upgrades are required, how the costs are allocated, and on what schedule the project can interconnect. The process culminates in an executed Large Generator Interconnection Agreement (LGIA), after which the developer builds and reaches commercial operation date (COD). Seven RTOs and ISOs (CAISO, ERCOT, ISO-NE, MISO, NYISO, PJM, SPP) run their own queues under FERC's pro-forma rules. About 50 non-RTO balancing authorities, mostly vertically integrated utilities, run separate queues. The LBNL dataset that this page draws on covers the seven RTOs and 50 non-ISO balancing areas, together about 98% of US installed capacity.
The pro-forma process inside an RTO has four sequential steps under FERC Order 2023: a 45-day cluster request window where new IRs are pooled, a 60-day customer engagement window, a 150-day cluster system impact study, and a 90- or 180-day facility study. Affected-system studies, where a project may impact a neighboring transmission provider, run in parallel. LBNL's 2025 Edition reports that interconnection customers in MISO, PJM, and SPP cite affected-system studies as the single largest source of timeline uncertainty.
Two service types govern the technical product. Energy Resource Interconnection Service (ERIS) gives a project the right to deliver energy on an as-available basis, with lower upgrade burdens but no firm deliverability and no capacity-market accreditation in most RTOs. Network Resource Interconnection Service (NRIS) gives firm deliverability under stress conditions; capacity-market participation in PJM, MISO, ISO-NE, and NYISO requires NRIS. NRIS is the more expensive product because it triggers more network upgrades.
ERCOT is the structural outlier. ERCOT's "Connect and Manage" model is not FERC-jurisdictional and is governed by the Texas Public Utility Commission. Under Connect and Manage, developers self-fund only direct interconnection costs at the point of interconnection; broader network costs are recovered from retail customers via the transmission provider, and the system manages resulting congestion in real time through nodal prices and Generic Transmission Constraints. The result is faster IA execution and lower up-front developer cost, with the trade-off that some projects face significant curtailment risk after they come online.
LBNL's May 2026 year-end-2025 workbook puts the active US queue at 2,061 GW across roughly 10,300 projects. The mix is solar 773 GW, storage 749 GW, wind 220 GW, gas 253 GW, and a residual 67 GW of coal, nuclear, hydro, geothermal, and other or unknown. The 2025 contraction is concentrated in renewables and storage (solar down 19%, storage down 16%, wind down 19% year-over-year), while gas capacity in queues rose 86% from 136 GW to 253 GW. Solar and storage still hold most of the active megawatts.
The regional picture changed sharply. West non-ISO (WECC outside CAISO) is the largest at 567 GW. ERCOT moved into second place at 408 GW, up from 346 GW a year earlier, the only major region with year-over-year growth. MISO fell to 382 GW from 448 GW (down 15%). CAISO fell to 191 GW from 273 GW (down 30%); LBNL attributes part of that drop to cluster 15 Interconnection Process Enhancement requests that were not resubmitted as the new scoring required. SPP holds 171 GW, Southeast non-ISO 153 GW, PJM 144 GW (down from 211 GW), NYISO 29 GW, and ISO-NE 15 GW.
The most construction-likely subset is the capacity that has executed an interconnection agreement but has not yet reached commercial operation. LBNL puts that figure at 524 GW at end-2025, up from 408 GW at end-2024. The biggest IA-executed pipelines are ERCOT (122 GW), CAISO (98 GW), West non-ISO (79 GW), MISO (74 GW), PJM (48 GW), Southeast non-ISO (45 GW), and SPP (42 GW). IA-executed status does not guarantee commercial operation; projects can still drop out from financing, EPC, permitting, or supply-chain problems.
One methodological note on ERCOT. LBNL's 408 GW year-end 2025 figure is the active-GINR count under LBNL methodology. ERCOT's own late-2025 reporting cites 432 GW of generation interconnection requests, including 176 GW storage, 158 GW solar, and 48 GW gas. The 24 GW gap is methodological (definitions of "active" versus "suspended" differ). ERCOT also reports a separate large-load queue, mostly data centers, that exceeded 233 GW in late 2025, up roughly 300% year-over-year. The large-load queue sits outside this page's scope but applies pressure to the same transmission network.
About 13% of the capacity that entered US interconnection queues from 2000 to 2020 had reached commercial operation by end-2025, weighted by megawatts. By project count, the rate is 19%. For the 2000 to 2019 cohort, LBNL reports 77% of capacity withdrawn and 10% still active as of end-2024 (Queued Up 2025 Edition). The cohort is mature enough that most projects have either completed or withdrawn, so the headline figure is a usable historical baseline. FERC Order 2023 and the ISO-level reforms approved in 2024 and 2025 are not yet visible in this number.
Completion rates vary by technology. For the 2000 to 2018 cohort, by project count, gas plants reached COD at 31%, wind at 20%, solar at 13%, standalone batteries at 11%, and solar-plus-battery hybrids at 10%. Capacity-weighted rates run lower for every technology (gas 18%, wind 15%, solar 10%, battery 5%, hybrid 7%). The pattern is partly causal (gas projects historically had firmer siting, financing, and offtake at queue entry) and partly compositional (the recent wave of renewables and storage entered queues during the years with the slowest study throughput).
Regional variation is wide. For the 2000 to 2020 cohort by capacity, ERCOT leads at 24%, ISO-NE at 18%, MISO and SPP at 14%, PJM at 13%, NYISO and Southeast at 12%, West non-ISO at 10%, and CAISO at 8%. ERCOT's lead traces to its Connect and Manage model, which lets more projects clear the IA stage even if some face curtailment later. ISO-NE's rate reflects a much smaller queue and optional process steps that allow customers to skip feasibility or facility studies for faster IA paths.
Duration is the other half of the picture. Median IR-to-COD for projects placed in service in 2025 was 61 months (about 5.1 years), down slightly from 63 months for the 2024 in-service cohort. Both more than doubled from under 24 months for the 2000 to 2007 in-service cohort. Inside the 61-month median, IR-to-IA accounts for roughly half and IA-to-COD the other half. CAISO sits well above the median on both legs; ERCOT and ISO-NE sit below. A long IR-to-COD timeline does not by itself mean a project failed; large gas plants in the 2000s ran multi-year permitting. The doubling of the median over two decades shows process throughput has not kept pace with request volume.
The 2003 interconnection process was built to study a small number of large thermal projects in series. It now has to clear a much larger volume of renewable, storage, and gas requests. The proximate causes interact, so attributing the backlog to any single factor understates the problem. The list below covers the eight that LBNL, the AEI/Grid Strategies scorecard, and the DOE i2X roadmap most consistently identify.
FERC Order 2003 finalized the standard generator interconnection procedures in 2003, the same year the modern RTO market design took its near-final shape. The process was built around a small number of large, dispatchable thermal projects studied serially. Annual new-request capacity ran about 125 GW per year through the mid-2000s. By 2023, annual new requests peaked at roughly 900 GW, and the active queue stock reached 2,598 GW. The serial-study throughput did not scale, and the 2003 design was the binding constraint until Order 2023 replaced it in 2023.
When a project triggers network upgrades, the cost is allocated to the project that caused them. The result is a strong selection effect on which projects survive. For non-ISO balancing authorities from 2018 to 2024, LBNL's February 2026 paper reports completed projects averaged $194/kW in total interconnection costs, while withdrawn projects in the same period averaged $671/kW. LBNL's ISO-side numbers from the AEI/Brattle scorecard show the same pattern (PJM completed projects $26/kW, PJM active-queue $291/kW). Network upgrades are the dominant cost component for recently withdrawn projects, well above the direct connection facilities at the point of interconnection.
Cost estimates produced at the feasibility-study stage rarely match the final number. The AEI/Brattle scorecard reports coefficient of variation in interconnection-cost estimates of 425% for MISO, 256% for NYISO, 123% for PJM, 45% for SPP, and 41% for ISO-NE. In one MISO 2017 to 2020 cluster example, the cluster Phase 1 estimate averaged $232/kW; the Phase 3 estimate for the surviving projects averaged $73/kW. The estimate can swing in either direction. Developers cite this uncertainty as a more significant problem than the absolute cost level, because it is what makes financing committed before the IA difficult.
In the pre-Order-2023 serial process, when a higher-queue project withdrew, every downstream project that had been studied assuming the first's existence had to be restudied with new cost responsibilities. Restudies could trigger further withdrawals, which triggered further restudies. The cascade was the structural weakness of PJM's pre-2023 process and the analytical basis for the cluster-study reform. Affected-system studies across ISO seams add another layer: a single seam-side project can trigger restudies in two queues at once.
Queue entry has historically been inexpensive relative to the option value of an executed IA. Developers filed many requests for a few that succeeded, and the cluster process meant queue entry was the only way to obtain a real estimate of network-upgrade cost and headroom for a given site. The 77% withdrawal rate is partly a feature of this design. The reforms in Order 2023 and ISO-level filings (study deposits raised to $55,000 to $250,000, site-control documentation required at IR, withdrawal penalties) are aimed at this margin.
Every RTO and major transmission owner the AEI scorecard interviewed in 2024 rated their interconnection-study engineering capacity as inadequate. Queue stock grew from 325 GW in 2014 to 2,598 GW in 2023, an 8x increase; study-engineer headcount did not grow on a comparable trajectory. The result is a physical throughput bottleneck on top of the process-design problems. The Order 2023 study-delay penalties ($1,000/business day for cluster studies, $2,000 for restudies and affected-system studies, $2,500 for facility studies) are intended to put a cost on this bottleneck for transmission providers.
Network-upgrade costs scale with how far a project is from existing high-voltage transmission and how loaded the nearby network already is. SPP and PJM data in LBNL's slide deck show upgrade costs averaging $150/kW for projects within 1 km of a 345 kV+ line, rising to $275/kW (SPP) and $575/kW (PJM) for projects more than 35 km away. Renewables and storage tend to site in rural areas with weaker existing transmission. The shortage of new high-voltage construction (322 miles of 345 kV+ added in 2024 against a DOE-estimated need of about 5,000 miles per year) means the network-upgrade cost burden falls more heavily on each new project. See Blizzard Power's transmission focus topic for more.
Even projects that clear the queue face supply-chain delays. Large power transformer lead times ran about 128 weeks for standard units and 144 weeks for generator step-up units in Q2 2025, up from roughly 50 weeks pre-2022. The largest units now quote up to 210 weeks (roughly 4 years), per pv magazine USA's May 2026 reporting. Prices are 4 to 6 times pre-2022 levels. The US has one domestic producer of grain-oriented electrical steel, the core raw material for transformer cores (Cleveland-Cliffs' Butler, Pennsylvania mill). Switchgear, HVDC converter equipment, and large generator step-up transformers all share similar capacity constraints. The supply-chain delay is mostly a post-IA problem, but it interacts with the queue because executed-IA capacity sits longer before reaching COD.
None of these causes is independent. The pro-forma process meant cost estimates were unstable; unstable estimates encouraged speculative requests; speculative requests overloaded study staff; staff shortages slowed studies; slow studies made the option value of queue entry larger; transmission scarcity raised the cost wedge that triggered the selection effect; equipment supply made executed-IA capacity stack up. Reforms in any single dimension face limits because the other constraints continue to bind. The reform cycle now underway works on most of these at once.
FERC issued Order 2023 on July 28, 2023, and Order 2023-A on rehearing on March 21, 2024. The rule replaces the 2003 pro-forma framework with a first-ready, first-served cluster process across three reform categories: cluster studies and timeline requirements, study-delay penalties for transmission providers, and consideration of alternative transmission technologies. Application fees are $5,000, study deposits run $55,000 to $250,000 sized to project, LGIA security is set at 20% of estimated network-upgrade costs, and site control requirements rise from 90% at request to 100% by facility study agreement.
Compliance has rolled out unevenly. Each RTO filed its own implementation plan with FERC, and the timing of approvals and the design choices vary.
PJM transitioned to cluster studies under a FERC order approved in November 2022, predating Order 2023. The Fast Lane track processed about 18,000 MW of low-cost (under $5M upgrade) projects, with final IAs issued April 18, 2025. Transition Cycle 1 (TC1) reached final IAs on November 20, 2025; Transition Cycle 2 covers about 46 GW with IAs targeted for Q1 2027. FERC approved PJM's Reliability Resource Initiative (RRI) on February 11, 2025, which selected 51 projects totaling 9.36 GW UCAP (2.11 GW from existing-plant uprates, 7.25 GW new) to add to TC2 on a fast-track basis. Cycle 1 (the first new-process cycle) closed applications on April 27, 2026 with 811 projects and 220 GW proposed, of which gas was 106 GW (48% of MW), storage 67 GW, and solar 23 GW including hybrids.
FERC approved MISO's DPP reforms in January 2024 for requests entering after January 22, 2024. The M2 study deposit doubled to $8,000/MW; withdrawal penalties scale by phase up to 100% of M2 after GIA negotiations begin. FERC initially rejected MISO's overall queue cap in 2024 and approved a revised version in January 2025. MISO's Expedited Resource Addition Study (ERAS), approved by FERC in 2025, is a 3-month GIA pathway for resource-adequacy-critical projects with a cap of 68 total ERAS requests. ERAS Cycle 2 (announced December 1, 2025) drew 6.1 GW, about 70% of it natural gas, which drew criticism from clean-energy advocates as a fuel-mix bias. MISO and FERC frame ERAS as a temporary reliability response that sunsets August 31, 2027.
FERC approved CAISO's Interconnection Process Enhancements (IPE) in October 2024. IPE caps new requests at 150% of available plus planned transmission capacity in each zone and applies a multi-factor commercial-readiness score for eligibility into the cluster study. Cluster 15 (2022, deferred to 2023) drew about 347 GW initially; after IPE scoring, 145 projects totaling about 68 GW cleared into the study phase, an 80% reduction. Cluster 16, originally an April 2024 window, was deferred indefinitely. CAISO retains a refund mechanism: developers pay network-upgrade costs upfront but are reimbursed up to a capped level over a 5-year repayment term once the project obtains firm-delivery allocation.
ERCOT is not FERC-jurisdictional and was not subject to Order 2023. The Connect and Manage model lets developers self-fund only direct facility costs at the point of interconnection; broader network costs are recovered from retail customers via the transmission provider. There is no formal cluster process; projects move sequentially based on application readiness, and modifications such as changing the point of interconnection are allowed without full resubmission. ERCOT received roughly 2,000 new generation interconnection requests in 2025, 77% of them solar plus storage. The Connect and Manage trade-off is faster IA execution and lower up-front cost against higher curtailment risk via Generic Transmission Constraints and the prospect of stranded transmission investment in regions where queued capacity does not materialize.
SPP runs an annual cluster (Definitive Interconnection System Impact Study, DISIS). The 2024 DISIS window was extended to March 2025 under a FERC waiver, and the 2025 window was deferred to April 2026. FERC approved SPP's Priority Process on November 28, 2025 (effective December 1, 2025), allowing existing power plants to expand by up to 20% under a temporary priority review. Public-interest organizations filed a protest in November 2025 arguing the process consumes transmission headroom that should be available to existing queue participants. SPP also filed a Consolidated Planning Process with FERC late in 2025; that proceeding is under review.
FERC approved NYISO's Order 2023 compliance in 2024. New Standard Interconnection Procedures launched August 1, 2024 with a transitional cluster study and a two-phase cluster approach replacing the prior class-year process. The IA target for new clusters is 1.6 years (plus 3 months for the transitional cluster). NYISO also introduced a rolling pre-application process effective May 2, 2024 that allows developers to engage with NYISO and the connecting transmission owner before formal queue entry. NYISO's active queue contracted from 78 GW at end-2024 to 29 GW at end-2025 as the new process turned over.
ISO-NE has the smallest active queue (15 GW at year-end 2025, down from 46 GW at year-end 2024) and the highest historical completion rate by count among the seven RTOs (31% for the 2000 to 2018 cohort). ISO-NE's Order 2023 compliance filing was approved and additional reforms are in development. The region's structural challenges are offshore wind cluster impacts in southeastern Massachusetts and transmission constraints in Maine. ISO-NE retains optional process steps that let customers skip feasibility or facility studies for faster IA paths, a feature no other RTO has.
The reform cycle has exposed a structural choice between two interconnection approaches. The first, used by every RTO except ERCOT, is "study and allocate": every project is studied, network upgrades are identified, costs are assigned to the project under cost-causation rules, and the upgrades are built before the project comes online. The second, used by ERCOT, is "connect and manage": developers self-fund only the direct interconnection facilities, network costs are socialized to ratepayers via the transmission provider, and the system manages resulting congestion in real time through nodal prices and Generic Transmission Constraints.
The case for study and allocate is cost causation and reliability. New projects pay for the upgrades they require, so the cost signal at queue entry reflects the actual marginal cost of integration. Upgrades are built and energized before generation comes online, which protects against the prospect of stranded transmission investment and avoids curtailment exposure after COD. Critics argue the process has become the binding constraint on buildout: 77% of requests are withdrawn, the average IR-to-COD timeline is over 5 years, and the cost-estimate volatility (425% coefficient of variation in MISO) means the cost signal itself is unreliable.
The case for connect and manage is throughput and speed. ERCOT's 24% capacity-weighted completion rate for the 2000 to 2020 cohort is the highest among the seven RTOs, and its IR-to-IA durations run shorter than other regions. The case against is that network costs do not signal at queue entry, so projects can come online and immediately face significant curtailment via Generic Transmission Constraints when local transmission is binding. Stranded transmission risk is also higher: if queued capacity does not materialize, network expansion that was built or planned on its basis is paid for by ratepayers anyway.
The reform cycle inside the seven FERC-jurisdictional RTOs is a series of attempts to move toward connect-and-manage throughput without giving up cost-causation signals entirely. Cluster studies, study-delay penalties, raised commercial-readiness thresholds, and expedited adequacy paths (MISO ERAS, SPP Priority Process, PJM RRI) all live in that space. Whether the new equilibrium is closer to the study-and-allocate baseline or the connect-and-manage extreme is the open policy question.
More articles on queue reform, ERCOT Connect and Manage, and per-RTO outcomes coming soon.
What "active capacity" means:LBNL's "active" capacity is the sum across every queue stage from feasibility study through executed interconnection agreement of all requests that have not been withdrawn and have not yet reached commercial operation. It is not committed or under-construction capacity. The 2,061 GW year-end 2025 figure includes projects at every stage of the process, and historical capacity-weighted completion rates run about 13% for mature cohorts. Active capacity overstates the megawatts that will reach commercial operation.
Capacity-weighted versus count-weighted completion rate: Chart C uses capacity-weighted rates, meaning the share of megawatts requested in the 2000 to 2020 cohort that had reached commercial operation by end-2025. The count-weighted rate (share of requests by project count) runs higher, 19.1% nationally versus 13.1% by capacity. Capacity-weighted rates are lower because larger projects are more likely to withdraw. Comparing across regions, the order is similar under both weightings, but the absolute levels differ.
LBNL versus ERCOT methodology gap:LBNL puts ERCOT's active queue at 408 GW year-end 2025. ERCOT's own reporting cites 432 GW of generation interconnection requests in late 2025. The 24 GW gap is methodological. LBNL aggregates from a snapshot of active GINR records using a specific status filter; ERCOT's published total uses a different definition of "active" that includes some suspended or in-modification requests. Where this page cites ERCOT figures, the methodology used is named.
Non-ISO scope of the $194/kW figure:The cost chart and the glance stat draw from LBNL's February 2026 paper on non-ISO balancing authority interconnection costs. The sample is five non-ISO BAs (PacifiCorp, BPA, Duke Energy Carolinas, Duke Energy Progress, Duke Energy Florida) with 2,100+ project-level cost estimates from 2018 to 2024 studies. PacifiCorp dominates the sample. LBNL slide 22 of the same report compares the non-ISO BA averages to ISO averages directly for 2018 to 2024: completed projects $194/kW non-ISO versus $88/kW ISO; withdrawn projects $671/kW non-ISO versus $443/kW ISO. The non-ISO levels are higher, but the same direction of selection effect (withdrawn higher than completed) holds on both sides.
Storage methodology change in 2020:The stacked-area chart shows storage capacity jumping from about 55 GW in 2019 to about 205 GW in 2020. LBNL began including estimated hybrid-storage capacity from 2020 onward; before 2020, only standalone-storage requests were tracked, and hybrid-storage was reported as part of the host generator's capacity. Part of the 2020 jump is a methodology inflection, so real-world storage growth that year was smaller than the 4x the chart implies. From 2020 forward, the storage series is comparable year-over-year.
Order 2023 not yet visible in the data: Most ISO-level Order 2023 compliance filings were approved in 2024 and 2025. The 2000 to 2020 cohort used for the completion-rate chart pre-dates implementation, and the IR-to-COD durations on the glance tile reflect projects placed in service in 2025 whose interconnection requests were filed under the pre-reform process. The earliest first-cycle cohorts under the new rules (PJM Cycle 1, NYISO transitional cluster, MISO post-January-2024 DPP) will not produce mature completion data until the late 2020s.
Cohort definitions and sample sizes:The capacity-weighted completion rate is for the 2000 to 2020 cohort with status snapshot at end-2025, drawn from LBNL workbook sheet 25. The 61-month median IR-to-COD is the 2025 in-service cohort, n=294, from sheet 37; the 2024 in-service cohort (n=335) median was 63 months. Sample composition shifts year to year, so single-year medians should be read alongside the multi-year trend. LBNL's workbook does not break IR-to-COD durations down by ISO in sheet 37 directly; per-ISO durations come from sheet 38.
Last updated: May 27, 2026.
This page is for informational purposes only and does not constitute investment, legal, or engineering advice. Queue figures reflect snapshots in time and change quarter to quarter as new requests enter, withdrawals occur, and IAs are executed. Per-RTO reform descriptions reflect FERC filings and public reporting as of the last-updated date.