In July 2022, the Greater London Authority sent an interesting note to property developers in Ealing, Hillingdon and Hounslow. Data centres built along the M4 corridor had taken up the remaining electricity capacity in the area
It stated that major new applicants to the distribution network would wait years for a connection; some were told it could take until 2035. City Hall then spent years convening SSEN, National Grid and the affected boroughs to free up capacity. In fact, its own update documents put a number on what had been stuck: stalled sites comprising at least 11,690 permitted new homes have now been unlocked with City Hall support, 3,315 of them since March 2024 through SSEN’s capacity allocation study.
To be fair, the attribution was contested from the start. Hillingdon’s cabinet member for residents’ services named the coverage as scaremongering and stated “power capacity has been reserved for developments that may never take place”, and the GLA later clarified to TechUK that there was no ban on new housing. Funnily enough, that council complaint from 2022 is the exact problem the national numbers now describe.
The data centre connection queue keeps building
Ofgem wrote to the industry in November 2025 about the demand connection queue. Between November 2024 and June 2025, total contracted offers in that queue rose from 41 GW to 125 GW. Demand projects face no “needed” filter under the reformed connections process (generation does), so no rule asks whether the volume matches any forecast of what the country will use. Ofgem itself says the queue now exceeds even its most ambitious demand forecasts.
NESO then asked projects in the transmission-level demand queue to identify themselves. Data centres made up over half of the responses by both count and capacity, at 50.8 GW. Within that, projects with evidence of a final investment decision total 21.6 GW across 71 projects.
Translating GW into housing equivalents
Now, what does 21.6 GW mean in housing terms?
Network operators size new connections using After Diversity Maximum Demand, and SPEN’s calculator figures, published through Scottish Government research, put a 300-home development at 1,074 kVA with heat pumps, rising to 2,076 kVA once each home also has a standard EV charger. That is 6.9 kVA per home at the heaviest end, which is the all-electric home that new build heat standards point towards. Divide the financially committed data centre queue by that figure, and you get the connection capacity of roughly 3.1 million homes. The government’s target for this Parliament is 1.5 million, at 370,000 homes a year. In turn, what this means is that the data centre projects with financing behind them have, on their own, reserved grid capacity equivalent to about two Parliaments of housebuilding, and the full 50.8 GW queue works out to around 7.3 million homes, or close to twenty years of delivery at the government’s rate. The comparison is one of scale, to be clear, since reserved capacity sits at particular substations and a data centre in Slough does not delay a house in Leeds. West London is what it looks like when both want the same substation.
Speculative intent vs genuine demand
I believe the real problem lies in those numbers. Of the data centre projects NESO surveyed, 32% have secured an off-taker; 68% have not, often pending a firm grid connection date. The 21.6 GW with financial commitment is under half of the 50.8 GW total. NESO’s own caveat on its summary is that the responses represent developer intent, and intent is what housebuilders are queuing behind. Ofgem’s letter says the queue contains credible projects alongside less viable ones that crowd out those with genuine merit and block progress for the projects behind them. Therefore, the constraint on housing connections in places like West London is only partially real demand. The remainder? Capacity that was booked and may be released years later, after the housing scheme it delayed has been refinanced or abandoned….
The financial burden of data centres on new developments
For anyone with a site, two figures from the same Scottish Government research show where these effects may land. A modelled 60-home social housing scheme on a constrained network needs a new primary substation at £4m to £6m, and whether the developer carries that alone depends on whether the DNO judges the location a high development area. In Dunfermline, later connections pay £250,000 per MVA towards reinforcement that an earlier scheme triggered. This means the question to put to the DNO at appraisal stage is what capacity is already reserved at the relevant substation and by whom, because the answer decides whether your scheme connects on the existing network or part-funds the next asset. A connection date is a queue position, and the queue is full of projects that may move, vanish, or otherwise not take their place.
Regulatory reform is crucial to move forward
Ofgem’s November guidance commits to curating a viable demand queue: reviews of entry and membership requirements, a possible progression commitment fee, DNO processes updated to ask applicants for evidence of project maturity, and closer alignment between distribution and transmission requirements.
There is give on the other side too, since NESO found 81% of queued projects would accept a phased or non-firm connection in exchange for an earlier date. For housebuilders, that reform matters more than any single data centre application, simply because a readiness test at queue entry separates the 21.6 GW with financing from the capacity that is only claimed. The GLA’s accounting shows the real cost of getting it wrong: 11,690 permitted homes in three boroughs that waited.
The post Housebuilders are queuing behind data centres that may never be built appeared first on Planning, Building & Construction Today.
