Worlds colliding – reshaping of grid connections – data centres to be treated the same as generation assets 7 August 2026

Ofgem proposes a fee equivalent to 2.5% to 7.5% of average project capital expenditure, estimated to be £9.5m per MW and so this fee equates to £237,500 to £712,500 per MW. For example, a data centre developer will have to provide (or secure) £23.8 to £71.3m for a 100 MW data centre.

Ofgem has compared other markets which have introduced a similar fee, such as Spain which is €40,000 per MW and certain US states which is US$50,000 to US$70,000 per MW. However, Ofgem has gone on to dismiss such comparators on the basis that they have different markets. Since Ofgem’s analysis, Spain has in fact changed their system to a capacity reservation fee. The capacity reservation fee is characterised as an advance payment for grid charges and is a monthly variable payment based on certain characteristics of the demand projects (increasing the longer the demand asset is not consuming energy from the grid). This has proved to be more effective in deterring speculative projects and the hoarding of demand capacity. As a result, 1 GW of demand projects have withdrawn their transmission connection applications.

Anecdotally, in May 2026 Denmark introduced a three-month temporary pause on new grid connections due to demand exceeding the network’s capacity.

Ofgem believes the fee would create a meaningful deterrent for speculative developers whilst imposing only a modest impact on viable projects, particularly given that the funds are returned once the project reaches energisation.

The commitment fee would be ‘secured’ through existing mechanisms such as:

• letters of credit from a qualified bank;
• performance bonds from a qualified company (being at least investment grade); and
• cash deposits.

The proposed commitment fee will apply to existing and future projects. Projects that have energisation dates within six months of the fee being introduced will be exempt. If the commitment fee increased, then a project would be allowed to self-terminate without forfeiting the fee.

Hybrid projects will only be required to provide a commitment fee as regards the data centre capacity. A non-data centre project which changed to a data centre would then be subject to the commitment fee, but may lose its position in the queue. Whereas a project which changed from a data centre to a non-data centre project or reduced its capacity would be liable to forfeit the full amount of the fee or the proportion of reduction where it has reduced capacity.

It should be noted that under the Gate 2 to Whole Queue Process for generators/batteries, notwithstanding that such projects were granted a “protected” connection, 62% of them across the three transmission networks had their connection dates delayed.

Queue management milestones

The second major reform proposed is to introduce data centre sector-specific readiness milestones designed to demonstrate continuous project progression. Projects would need to provide evidence at various points in development that they remain credible and capable of delivery.

The following are the proposed sector specific key milestones for projects with at least 10MW capacity:

M0.5.Dc: Compute Offtaker Evidence 

Within six months of signing a connection agreement, projects must:

• select a commercial pathway, namely a self-operation or lease/sale pathway; and
• provide non-binding evidence of a compute customer, such as a memorandum of understanding or heads of terms agreement, for at least 20% of compute capacity for at least one year.

M2.Dc: Long-Lead Procurement 

Projects must demonstrate procurement commitments for critical electrical equipment such as transformers and switchgear. Evidence may include contracts, invoices or supplier commitments.

M6.Dc: Financial and Technical Capability 

Projects must demonstrate:

• financial strength using either investment-grade credit ratings or committed financial support; and
• technical readiness through recognised certifications such as EN 50600, TIA-942 or Uptime Institute Tier Certification.

Projects operating under a lease-or-sale model must also provide binding compute offtaker agreements at this stage.

Projects that fail to provide required evidence of satisfaction of the relevant milestone by the applicable deadline could lose their queue position. This creates a chicken and egg scenario, as often hyperscalers and compute offtakers are unwilling to contract without connection certainty, which is now a milestone to the connection in the first place.

Connection security

There is currently a variance as regards the security to be provided by generation and battery projects as opposed to demand customers. Generation and battery projects are required to provide security based on the User Commitment Methodology whereas demand customers are generally required to provide security for the full value of the connection costs. Under the User Commitment Methodology, a distinction is made between works specifically attributable to the customer (generally a higher proportion being so allocated to the customer but reducing over the course of the works) and those which are transmission reinforcement works (being reduced to the extent that other customers are benefiting from such wider works). In principle Ofgem is in support of a proposal to bring customer demand in line with generation and battery projects as regards the security regime and assumes that it will proceed. The proposal was first submitted in July 2023 and a draft final modifications report was issued on 18 June 2026 and is awaiting final approval. This should, in effect, reduce the amount of security required to be provided under connection offers to demand customers, such as data centres and will alleviate to a small extent the burden of a commitment fee.

Financing and Capital Commitment

From a financing and investment perspective, the ‘Curate’ package could be regarded as a trade, namely, one that creates a materially more credible connection queue, but which is bought with a large forfeitable capital commitment imposed at the least bankable point in the development timeline of a project. A 100 MW data centre project must secure £23.8m to £71.3m from offer/acceptance, to energisation; a 500 MW campus, well over £100m. There is also a disproportionate impact on long dated connections, with Ofgem’s own modelling showing a 170bps reduction in the IRR for a 2035 connection based on a 7.5% commitment fee. Typically, senior construction debt is not available at such an early stage, so the burden to fund and secure the commitment fee will likely fall on sponsor equity, group balance sheet, or a bilateral SBLC/bonding line, rather than on project-specific debt. It is possible that well-capitalised sponsors, hyperscalers, listed platforms and large infra/private credit funds may end up absorbing the commitment fee and inadvertently leading to greater structural consolidation in the market, but independent developers and JV platforms may not as easily, and therefore, a potential pre-FID market outcome may be earlier partnering with balance sheet capital and more outright earlier sales of consented sites. In any case, we would expect the commitment fee quantum to become a live negotiating point on deals, focussed specifically on equity commitments, sponsor support undertaking and pre-development documentation, long before these are picked up in debt finance documentation. Lenders and investors will be concerned with who posted the fee collateral, in what form, and whether it is transferable on a change of control, asset or share sale as well as whether it has been fully or partially forfeited following a capacity reduction, together with any forfeiture consequences if any milestones are missed.

The subtler effect of these reforms may be seen in terms of values more generally. Grid queue positions already trade freely at a premium in the UK market; OFGEM is now proposing pricing that option explicitly and requiring it to be collateralised and regulated. Schemes that have already cleared the commitment fee and the M0.5.Dc, M2.Dc and M6.Dc milestones should therefore become more, not less, valuable. Diligence and valuation practices will need to catch up.