Summary

  • Ofgem's TMO4+ package screened generation and storage against strategic need, but transmission-connected demand sits outside the Clean Power 2030 technology scope and therefore satisfied the same criterion automatically, facing only a readiness test on land rights or a development-consent route.
  • The asymmetry has numbers attached: more than 300 GW of projects left the queue in December 2025 while roughly 99 GW of transmission-connected demand stayed, with 11.8 GW of that demand scheduled for delivery by 2030 and 86.7 GW by 2035.
  • One anonymised 60 MW data-centre scheme was quoted an estimated 2033 connection year because transformer upgrades were needed. That year, not the wholesale power price, belongs in the model.

The gate that faced one way

Ofgem approved the TMO4+ package on 15 April 2025: CMP434, CMP435, CM095, three connections methodologies and the licence changes that carry them (Ofgem summary decision document). Paragraph 1.2 explains why the package existed — new connection offers for generation and demand were stretching into the 2040s. Paragraph 6.23 puts a figure on the demand side at decision time: 42 GW, of which 17 GW sat at transmission and 25 GW at distribution, spanning industrial load and data centres.

The machinery commenced on 10 June 2025, after the statutory 56-day standstill and a short extension of transitional reliefs to the end of 9 June (Ofgem extension decision). Before that, Ofgem's January 2025 cut-over decision had already paused responses to most new applications from 29 January, while letting new directly connected demand projects keep using the transitional offer route until 21 March, and leaving distribution applications open but no longer routed to the system operator for assessment (Ofgem cut-over decision).

What the test actually asked

The Gate 2 criteria, dated 30 April 2025, set the price of admission: readiness evidence through secured land rights or a qualifying Development Consent Order route (NESO Gate 2 Criteria Methodology). Then section 6.3 does the decisive work. Transmission-connected demand is listed as outside the Clean Power 2030 Action Plan technology scope, which means it automatically satisfies strategic-alignment criterion (d). Generation and storage had to argue that they were needed. Demand only had to argue that it could happen.

That is not a loophole. It is a scope definition, and scope definitions decide who gets asked the hard question.

The methodology even holds out the possibility of advancement through the CMP435 gated application window — but eligibility and a request are not an earlier energisation, and no clause in the document makes them one.

The arithmetic of the deletion

On 15 April 2025 the queue held more than 750 GW of projects, roughly four times the requirement for 2030 and twice the requirement for 2050 (NESO on the reform). By 16 October it exceeded 800 GW, with queue-position notifications expected in early December and an explicit admission that implementation had taken longer than anticipated (NESO in October).

Then the reckoning. On 8 December 2025 NESO published a prioritised pipeline of 283 GW of generation and storage and approximately 99 GW of transmission-connected demand, and said more than 300 GW of projects in the previous queue would not progress (NESO results announcement). The demand table splits 11.8 GW for delivery by 2030 and 86.7 GW by 2035. Read that table the way an operator reads a timetable: supply was cut at a stroke; demand was re-timed.

Demand kept arriving while the rules were rewritten

Ofgem's demand connections update of 6 November 2025 tracks contracted demand offers from 41 GW in November 2024 to 125 GW in June 2025 (Ofgem demand update). The components are given as 97 GW transmission and 29 GW distribution — which sum to 126 GW, one gigawatt above the stated total, a discrepancy visible on Ofgem's own page. Ofgem attributes a significant share of the growth to data centres without quantifying it, confirms that demand projects outside embedded demand face readiness requirements but no needed filter, and announces further demand reform.

Readiness without need, in one sentence.

And the arithmetic slip is worth holding onto: these are administrative counts of applications, not metered load, and they do not close perfectly even in the regulator's own summary.

What a queue position costs

National Grid Electricity Distribution's November 2025 data-centre study, prepared by Regen, records developer testimony that a 60 MW site received an estimated connection year of 2033 for want of transformer upgrades, with a connection quotation of roughly £10 million (NGED data-centre impact study). The same study's August 2025 pipeline table lists 19 accepted offers totalling 1,711 MW, nine unaccepted offers at 140 MW and 22 enquiries at 1,275 MW across its licence areas, and it states plainly that these quantities and its connection-year modelling are scenario assumptions rather than guaranteed offers. Its modelling puts accepted-offer sites online by 2032 at the latest — a different population from the anonymised 2033 case, and the document does not reconcile the two.

NGED's connections reform webinar of 12 June 2025 fills in the mechanics: distribution evidence submissions opened on 20 May and closed on 29 July, alongside NESO's 8–29 July transmission window; the reformed pipeline would initially be ordered by transmission queue-entry date as identified through NESO countersignature; and contracted energisation dates are distinct from the earlier non-firm access dates that some accelerated projects hold (NGED reform webinar).

Non-firm access is the grit in the system.

It lets a hall energise ahead of its firm date while carrying curtailment risk. For a training cluster running flat out, that is not a discount; it is an outage waiting for a windy Tuesday.

The only figure that isolates data centres

NESO's written evidence to the Environmental Audit Committee, published 21 May 2026, reports that 2025 pipeline data showed approximately 72.8 GW of data-centre connection requests through 2039; separately, nearly 59 GW awaiting transmission connection across 173 data-centre projects, including 40 projects of 500–1,500 MW; and it cautions that not all requested demand is expected to connect (NESO written evidence).

Look at the shape rather than the headline. 173 projects carrying 59 GW averages near 340 MW each, with a fat tail of forty projects in the half-gigawatt to 1.5-gigawatt band.

That is the signature of a small number of very large campuses — and it explains why one connection date can decide whether a regional fibre build has an anchor tenant at all.

Dates, not megawatts

The offer timetable is the next observable fact. NESO said protected projects scheduled to connect in 2026 or 2027 would begin receiving formal offers with confirmed connection dates during December 2025 and into 2026, with the remaining offers finalised by the third quarter of 2026 (NESO on the pipeline).

Why an internet correspondent is writing about this

NATIONAL GRID UK LIMITED sits in the BTW directory as a RIPE NCC member with a number-resource record (directory entry). That one card contains the joke of British digital infrastructure: the company that decides when a data hall may run also holds address space, and nobody in the networking trade thinks of it as an internet company at all.

The stake

Three consequences follow.

Capacity is now rationed by date rather than price: a developer holding a 2033 offer cannot buy an earlier year, only accept non-firm access or move the load.

An investor pricing powered land is pricing a queue position whose value depends on rules Ofgem has said it will revisit.

And the aggregate demand totals cover all demand — industrial, electrolytic, digital — so nobody should quote 99 GW as if it were server capacity. The figures here are as the regulator and the system operator published them; the demand totals are not data-centre totals, and the single 2033 case is one developer's testimony rather than a typical outcome.

Conclusion

Britain built a filter that asks supply whether it is needed and asks demand whether it is possible.

Supply got smaller and better. Demand got larger and scheduled.

Until demand faces a genuine need test, the scarcity that governs British digital infrastructure is a connection date — and whoever holds an early one holds leverage that appears in no hosting price list.