Where we are
Just over seven hundred gigawatts of nominal contracted capacity sits in the UK transmission-entry-capacity queue today. The country needs perhaps a hundred and twenty gigawatts to deliver its Sixth Carbon Budget. The arithmetic is brutal: most of the queue will not connect on its original date, and a great deal of it will never connect at all. The historic position — that a TEC date was the fixed point around which a project finance model was built — was defensible when the queue was forty gigawatts and broadly first-come-first-served. It is no longer defensible.
What NESO actually changed
The headline reform is TM04+, but the substance lies in the gating regime. The queue is being converted from a first-come-first-served list into a milestone-gated cohort. Projects in the queue are now required to demonstrate evidence of progress at specified gates — land rights, planning consents, network agreements, commissioning readiness — and projects that miss gates fall back. In practice that means three things. First, the headline TEC date is no longer the right number to plan around. Second, the relevant outputs are gate-by-gate hazard rates: the probability that a project passes each gate in turn. Third, queue position itself is now endogenous: a project that misses a gate is shoved behind ones that didn't.
This is, in our view, the right reform. A queue without enforcement is a wish list. The current reform has costs — small developers report meaningful administrative burden, and gate definitions have already been refined twice since launch — but the substantive direction is correct. The market consequence is that any project finance model that does not embed a gate-by-gate hazard function will systematically over-value its inputs.
Old project finance discounted queue risk by adding twelve months to the TEC date. New project finance models the gate-by-gate hazard function, and the discount rates move accordingly. The two approaches do not produce close estimates; they produce different worlds.
The new risk model
The gate-by-gate hazard function is straightforward in principle. For each gate, the project either passes or it does not. The probability of passing depends on the evidence available — planning consent in hand, network agreement signed, EPC procurement underway — and on the comparable project history at the same node. Multiply through, and you have a connection-date distribution rather than a connection-date point estimate. Discount cash flows accordingly.
In practice it is less straightforward because the gates interact with each other and with the underlying policy environment. A project missing its land-rights gate in 2025 may be perfectly recoverable; the same project missing the same gate in 2027 may not be, because the queue cleansing rules will have tightened. Pipeline's hazard model captures these interactions explicitly and the result is, in our experience, considerably more conservative than the headline approach — typically two to four percentage points of IRR more conservative on the median portfolio, and the tails are uglier still.
Where money is being mispriced
Three places stand out. First, late-stage acquisitions of ROC-era assets, where the buyer treats the published TEC date as a fact. These transactions are routinely overpaid by hundreds of basis points of IRR when re-priced against the calibrated hazard function. Second, sleeve PPAs whose price was struck against the original TEC, with insufficient connection-slip protection. The offtaker bears the slip risk and is increasingly unwilling to do so quietly. Third, equipment forward orders placed by OEMs against developer pipelines that systematically overstate on-time connection. The OEM ends up with manufacturing slots booked against projects that slip, and the resulting cancellation economics are not pleasant.
The fix in each case is the same: replace the point TEC with a probabilistic connection-date distribution, and let the contracts price against the distribution rather than the point. This is not theoretical. In each of the three transaction categories above, we have seen Pipeline outputs change the executed price by a material margin.
What Horizon Pipeline shows
Per project, per gate: probability of passing, ninety-per-cent interval, calibration class, recent gate-pass evidence. Per portfolio: connection-date distribution, IRR distribution under the hazard function, and the marginal contribution of each project to portfolio variance. Per developer: queue exposure ranked across the portfolio. Per OEM customer: probabilistic commissioning curve. All calibrated quarterly against actual NESO gate-pass outcomes; calibration class is documented.
The single most useful thing a serious developer or financier can do, in the present queue environment, is to stop treating a TEC date as a fact and start treating it as a probabilistic forecast. This is true whether or not you use our tool to do it. Anyone who is still discounting the queue by twelve months is, in late 2026, simply not modelling the right object.