Delays and exits in the transmission queue since 2021
How often do connection dates move, and how many projects leave the queue? We’ve followed every project in NESO’s TEC register since January 2021 to find out. Developers and investors can use this page to judge how likely a connection date is to hold when they plan or value a project. Anyone working on connections reform can see whether projects leave the queue when their securities rise. The figures on this page update with each new register.
The registers behind this page
QueueWatch holds 148 of NESO’s TEC registers, from 29 January 2021 to 27 October 2026: one a month until October 2025 and twice a week since December 2025. We follow each project from register to register, through NESO’s changes of project ID and site name.
1. Connection dates: how often and how far they move
In 2025, 17% of projects had their connection date pushed back, by a median of 24 months, and a quarter of those by 36 months or more. In 2021, more dates were pushed back (46%), but by less (a median of 10 months). In no year did more than 4% move earlier. Of the 156 projects whose connection date was pushed back in 2026 so far, 67 got the later date with their Gate 2 offer.
Connection dates pushed back and brought forward
Share of the projects we could compare
How far the pushed-back dates moved
Months
| Year | Projects compared | Pushed back | Brought forward | Unchanged | Pushed back % | Brought forward % | Median months pushed back | Lower quartile | Upper quartile |
|---|---|---|---|---|---|---|---|---|---|
| 2021 | 303 | 139 | 9 | 155 | 45.9 | 3.0 | 10 | 4 | 16 |
| 2022 | 448 | 111 | 8 | 329 | 24.8 | 1.8 | 14 | 10.5 | 23 |
| 2023 | 631 | 86 | 5 | 540 | 13.6 | 0.8 | 14.5 | 11.3 | 24.8 |
| 2024 | 1,140 | 184 | 36 | 920 | 16.1 | 3.2 | 23 | 12 | 36 |
| 2025 | 1,455 | 250 | 20 | 1,185 | 17.2 | 1.4 | 24 | 12 | 36 |
| 2026 to 27 Oct | 1,723 | 156 | 34 | 1,533 | 9.1 | 2.0 | 19.5 | 12 | 33 |
What it means: when a connection date is pushed back, it’s usually by a year or more, and the further away a project’s connection date is, the more years there are in which that date can be pushed back. A plan or valuation built on a single connection date leaves that risk out.
How we measured it
For each year, we compare each project’s connection date at the start of the year with its date at the end (for 2026, on 27 October), so we see how far it moved overall, not each move in between. Only projects that still have MW to connect and a date at both points count. A date changes with a project’s Gate 2 offer when it changes in the register that first shows the offer.
2. Exits: how much of the queue leaves each year
The queue held 551 GW at the start of 2025, and 1.9% of it left during the year. The highest share since 2021 was 9.6%, in 2023, when most of what left was offshore wind. Battery storage was the largest part of what left in 2025, 5.5 GW of the 11.9 GW, and that year 24 battery projects left, against 6 in 2021.
Capacity that left the queue
Still to connect when it left, by technology
Share of the queue that left
Share of the queue at the start of each year that left that year
| Year | Storage | Solar | Onshore wind | Offshore wind | Thermal | Hybrid | Other | All | Queue at the start of the year | Share that left |
|---|---|---|---|---|---|---|---|---|---|---|
| 2021 | 313.6 | 150 | 112.5 | 4,780 | 3,262.5 | 0 | 0 | 8,618.6 | 104,598.1 | 8.1% |
| 2022 | 55 | 50 | 478.5 | 2,000 | 0 | 0 | 2,494 | 5,077.5 | 165,062.5 | 3.1% |
| 2023 | 306.9 | 202.8 | 74.9 | 21,900 | 1,399 | 0 | 3,790 | 27,673.6 | 241,453.8 | 9.6% |
| 2024 | 3,860.8 | 817 | 282.2 | 11,765 | 1,749.9 | 4,392.4 | 2,400 | 25,267.3 | 432,020.4 | 4.8% |
| 2025 | 5,481.1 | 179 | 464.2 | 3,000 | 100 | 1,979 | 681 | 11,884.3 | 551,453.7 | 1.9% |
| 2026 to 27 Oct | 900 | 0 | 49.9 | 0 | 0 | 847 | 3,600 | 5,396.9 | 620,051.8 | 0.9% |
What it means: only a small share of the queue leaves each year, so a project shouldn’t rely on others dropping out to get an earlier connection.
How we measured it
A project has left when it drops out of NESO’s register for good. One that comes back later, even under a new project number, hasn’t left. We date each exit to the first register the project is missing from. A project missing from the latest registers could still come back, so this year’s figures may fall as new registers come out. We only count projects with MW still to connect, and the register doesn’t say why a project left. NESO started numbering projects in September 2023. For earlier registers, we follow each project mainly by its name, site and customer, so a few renamed projects may show as exits in the early years.
3. Exits and the trigger date
Every project in the queue has a trigger date, usually three to four years before it’s due to connect. From that date, leaving costs more: the charge a project pays if it leaves (its cancellation charge) goes up, and so does the security it must put up to cover that charge. If projects were leaving to avoid that, many would leave just before their trigger date. Of the 175 projects that have left the queue since 2021, only 2 left in the six months before their trigger date. The rest left earlier (108) or later (65).
When projects left, against their trigger date
Each bar counts the projects that left in a six-month window before or after their trigger date. The orange bar is the last six months before it, when leaving still cost less.
| When they left | Exits | MW still to connect |
|---|---|---|
| More than six months before | 108 | 71,697 |
| In the six months before | 2 | 218.2 |
| At or after | 65 | 12,003 |
See all bandsSee less
| Months from the trigger date | Exits | MW still to connect |
|---|---|---|
| More than 84 months before | 40 | 37,249.3 |
| 79 to 84 months before | 5 | 4,633.4 |
| 73 to 78 months before | 1 | 99 |
| 67 to 72 months before | 3 | 350 |
| 61 to 66 months before | 4 | 1,131.5 |
| 55 to 60 months before | 6 | 4,242.5 |
| 49 to 54 months before | 5 | 4,849.9 |
| 43 to 48 months before | 2 | 900 |
| 37 to 42 months before | 6 | 4,099 |
| 31 to 36 months before | 8 | 5,790 |
| 25 to 30 months before | 8 | 2,529.9 |
| 19 to 24 months before | 5 | 789 |
| 13 to 18 months before | 4 | 852 |
| 7 to 12 months before | 11 | 4,181.5 |
| 1 to 6 months before | 2 | 218.2 |
| 0 to 5 months after | 14 | 2,775.4 |
| 6 to 11 months after | 2 | 145 |
| 12 to 17 months after | 12 | 3,507.9 |
| 18 to 23 months after | 11 | 845.1 |
| 24 to 29 months after | 6 | 1,945.1 |
| 30 to 35 months after | 8 | 318.7 |
| 36 months or more after | 12 | 2,465.8 |
What it means: so far, projects haven’t rushed to leave before their charge and security step up. CMP470, the commitment fee for batteries that Ofgem is minded to approve, aims to shrink the battery queue by setting a minimum security from the day a project accepts its Gate 2 offer.
How we measured it
The register doesn’t give trigger dates, so we estimate each one from the connection date in the last register that listed the project, using the rule in CUSC Section 15 (paragraph 2.2(a)). A project’s real trigger date can differ, for example if its connection date changed after the trigger date was set, or its agreement was signed less than three financial years before it connects. Until October 2025, NESO published one register a month, so for exits before then we know the month a project left but not the day.
What if your connection date is pushed back?
QueueQuant values your project across thousands of possible futures for its connection date and the market, and tells you whether to hold, cut MW, sell or walk away, and what would change that answer.
How to cite this page
You’re welcome to quote our figures, charts and findings with this credit:
QueueWatch by QueueQuant, “Delays and exits in the transmission queue since 2021”, queuequant.com/queuewatch/delays-and-exits/ (data to 27 October 2026). Contains data from NESO’s TEC register. Supported by National Energy SO Open Data.
Supported by National Energy SO Open Data. Registers from 29 January 2021 to 11 August 2026 are copies NESO released to QueueQuant on 19 October 2026 under the Environmental Information Regulations 2004 (request FOI/26/226). Later registers are QueueWatch’s own downloads from NESO’s Data Portal.