Ofgem - proposed data centre connection reforms

Ofgem’s proposed data centre connections reforms are intended to improve queue discipline by deterring speculative projects and forcing earlier evidence of deliverability. The core proposals are:

• a substantial returnable commitment fee, potentially reaching tens of millions of pounds for larger schemes, and
• new data centre-specific milestones requiring evidence of customer demand, procurement commitment, financial capability and technical maturity.

The reforms are likely to achieve part of their objective: cheap optionality should become much harder to maintain. However, the practical effect may be to narrow the market towards hyperscalers, major operators and very well-capitalised developers with established customer relationships. The retrospective nature of the proposals also means existing projects, unless close to connection, may need to reassess quickly whether they can justify putting further capital at risk.

The key policy question is whether Ofgem’s deterrence model strikes the right balance. A regime designed to clear speculative capacity from the queue may also reduce competition, constrain development models and do little to channel capital into wider grid reinforcement. For many developers, powered land may therefore become less of a managed development strategy and more of a high-stakes commitment, with limited ability to hedge once the fee has been posted.

Ofgem’s proposed changes to Data Centre connections: Is powered land for data centre use about to become a game of Blackjack?

The need for reform is clear. Ofgem says the data centre connections queue has grown rapidly and now represents around 73 GW of demand across roughly 315 projects. Its concern is that the current regime makes it relatively cheap to secure and retain queue positions, encouraging speculative projects, distorting investment signals and delaying viable schemes.

Ofgem’s stated objective is to improve “queue health”, not to predetermine how much data centre demand should exist. The aim is to deter speculative and non-viable projects while imposing the lowest feasible burden on viable developers, as part of the broader “Curate, Plan and Connect” reform framework.

What is being proposed?

In broad terms, Ofgem proposes two linked changes: a substantial commitment fee and a sector-specific milestone regime.

First, in-scope data centre projects would have to secure a potentially returnable commitment fee from acceptance of the connection offer until energisation. The commitment fee will be refundable if the project proceeds, but will otherwise be forfeited if the milestones are not met. Ofgem proposes a fee of around £237,500/MW to £712,500/MW for projects above the 40 MW threshold, meaning a 100 MW project could need to secure roughly £23.8 million to £71.3 million at the outset. The fee is intended to be returned at energisation but forfeited if the project terminates or fails to comply with the rules. A mid-point outcome of the consultation would put the cost at £500,000/MW or £50m for a 100MW connection offer.

Second, Ofgem proposes additional queue management milestones tailored to how it sees data centre projects being developed and financed. These would sit alongside the existing queue management regime which focuses on planning, land acquisition and construction milestones. The new milestones would test customer demand, procurement commitment, financial strength and technical maturity as the project progresses.

Importantly, the proposal would be retrospective. Only projects with a connection date in the near future (6 months is the proposal) would be excluded. All other projects would be caught. This means that those who already have a connection offer will need to decide quickly whether they want to risk additional capital to retain their already sunk investment costs or lose their investment value.

The milestones

M0.5.Dc – Pathway selection

At M0.5.Dc, six months after connection signature, a project must choose its commercial pathway and submit non-binding compute offtaker evidence. Ofgem proposes two pathways: self-operation, for developers intending to build and operate the facility themselves, and lease or sale, for projects intended to be leased, sold or transferred to a third-party operator.

The purpose of this milestone is to require an early demonstration that there is a credible end-use case behind the grid capacity being reserved. For now, Ofgem is only asking for non-binding evidence. For the ‘sale or lease pathway’ heads of terms or an MoU, rather than a firm contract, would be sufficient at this stage. For the ‘self-operation pathway’ a form of self-certification is proposed. Even so, based on our experience advising sell-side and buy-side of powered land deals this is a very high bar. This milestone alone is likely to prevent any form of speculative application. It clearly favours the hyperscale and established data centre operators over other forms of developer / promotor. It may create a catch 22 situation – a customer may be reluctant to engage if the developer does not have a power connection, but conversely the developer will not now be able to accept a power connection offer without that customer engagement.

M2.Dc – procurement commitment

At M2.Dc, projects must provide long-lead electrical equipment procurement evidence. Ofgem’s view is that procurement of equipment such as transformers or switchgear is a meaningful indicator that a project has moved beyond concept stage and is committing real resources towards delivery.

That may be a sensible seriousness test, but it may also be one of the more operationally challenging milestones. In practice, forcing evidence of long-lead procurement at a fixed stage could constrain procurement strategy, reduce flexibility in contractor engagement and package design, and lock projects into early technical or commercial decisions before the wider delivery model is fully settled. Ofgem itself is consulting on whether the requirement should be limited to transformers and switchgear or extended more broadly, which underlines that calibration will matter. However, in our experience forcing developers into this level of capital commitment to retain their connection adds a significant financial hurdle and raises the stakes considerably.

M6.Dc – finance, technical capability and, in some cases, a firm customer

At M6.Dc, the project must submit financial capability evidence and technical capability evidence, and projects on the lease or sale pathway must also provide binding compute offtaker evidence. In effect you need to show the “deal is done”.
For financial capability, Ofgem proposes either an investment-grade credit rating or legally committed credit support tied to the project. For technical capability, it proposes evidence linked to recognised data centre standards or certifications, such as EN 50600, TIA-942 Design Certification or Uptime Institute Tier Certification of Design Documents. For lease or sale projects, this is also the point where non-binding customer engagement is no longer enough: Ofgem is currently minded to require a binding commitment covering at least 20% of compute capacity for at least one year, although it is consulting on variants.

Taken together, M6.Dc, on paper, is the milestone that most clearly separates projects with a live route to delivery from those still relying on optionality. In practice, taking into account the other milestones, it is likely that meeting the other milestones will have required this level of commitment already. So, whilst this milestone looks to be the most challenging and outside of the control of the developer, in reality if a developer has managed to meet the other milestones this one is more likely to be capable of satisfaction.

Will the proposals achieve Ofgem’s objectives?

In one sense, probably yes. A very large up-front financial commitment which is not returned if the project does not meet onerous milestones is plainly capable of discouraging speculative projects from entering or remaining in the queue.
But the bigger question is what sort of market that leaves behind. In practice, the proposals are likely to favour hyperscalers, major co-location operators, and a small group of well-capitalised developers with strong end-user relationships. The 40MW threshold will also push towards smaller deployments potentially spread more widely, potentially reversing the trend towards bigger and bigger campuses . This regime will be much harder for developers whose model depends on securing land, planning and power first, and tying in the end-user later. Many of the schemes we are seeing being built out today started life this way; viewed through that lens the future impact on pipeline could be significant.

There is also a concern with the commitment fee structure itself. Ofgem’s preferred option requires 100% of the fee to be secured at the outset and kept in place until energisation, resulting in limited ongoing incentive to self-terminate after that point. That means the regime may be good at deterring entry, but less effective at encouraging timely exit.

In commercial terms, once the fee is posted, the developer may have little incentive to walk away early (contrary to one of Ofgem’s stated goals – to encourage developers to terminate the project if it is not viable) even if the project starts to struggle, because doing so crystallises the loss. A tiered model under which part of the fee could be recovered depending on when the project is terminated might better support rational early exit and release capacity sooner. Ofgem appears to have rejected this on the basis that they prefer simplicity. An alternative approach requiring non-refundable periodic payments has been rejected even though this could have created a valuable income stream for investment purposes and delivered a less “blackjack or bust” outcome

Additional challenges and likely market responses

There are also practical questions around how the reforms would work in reality. The long-lead procurement milestone may be more awkward in practice than it appears on paper, because procurement sequencing often evolves with design development, contractor strategy and financing structure.

Similarly, Ofgem’s IRR analysis assumes that the fee itself does not create wider knock-on cost or timing effects beyond the cost of carrying the security. That is a significant assumption. In reality, tighter milestone pressure and earlier procurement commitments could affect financing, transaction structures, design flexibility and delivery timing in ways that are not captured by a narrow IRR model.

A further possible consequence is that some market participants may place greater emphasis on private wire arrangements where that is feasible. If conventional grid access becomes more financially demanding and more binary, alternative power structures may become correspondingly more attractive. Perhaps this is an intentional unstated outcome Ofgem would like to see. However, this would not benefit the wider demand market save for reducing the queue.

The likely overall impact

The most likely outcome is a more concentrated market for powered land and grid-backed data centre development. Queue discipline may improve, but the pool of participants is also likely to narrow towards those who are either the end-user themselves or have unusually strong capital backing and customer access. Despite Ofgem’s stated aims of not wanting to hinder competition, this seems an inevitable outcome.

That may solve an immediate queue-management problem, but it may also be a missed opportunity. The consultation is primarily a deterrence model. It does not do much to channel capital into grid reinforcement, establish a longer-term market framework for demand growth, or encourage deeper partnership between the data centre sector and the energy sector. In essence it is focused on clearing out the queue. It does not provide, in a meaningful way, a long term plan to ensure UK grid can deliver for the demand and related opportunities of current and future compute power.

Conclusion

Ofgem is right that the current position is unsustainable. Cheap optionality cannot continue to clog the queue.

But the proposed answer may do more than deter speculation. The combination of heavy up-front underwriting, staged milestone pressure and an all-or-nothing fee structure may mean that only the biggest players — or a very small number of developers with strong end-user relationships — can sensibly afford to participate. For everyone else, powered land starts to look less like a development strategy and more like Blackjack: expensive to enter, difficult to hedge and carrying a real risk of losing everything.