By Tom Glover, RWE UK Country Chair
August 26, 2026
The Onward report on the future of the UK power system has generated a lot of debate, and rightly so. It’s a serious, well-constructed piece of modelling, and it puts its finger on the question that should sit at the centre of energy policy: how do we get the cost of electricity down far enough that households and industry actually want to electrify?
That question matters. Around 90% of our emissions sit outside the power sector — in heating, transport and industry — and we’ll only shift them if clean electricity is cheap. So let me be clear up front: this is a useful contribution, and there’s plenty in it I agree with.
But there’s a difference between a useful scenario and a plan you’d build a system around. This report is the former presented as the latter. If you accept its assumptions, the results follow logically. The problem is the assumptions — because this kind of modelling largely gives you back what you put in, and this pathway is built on a stack of optimistic inputs.
Here’s where I’d push back
It’s one line, not a set of scenarios.
The central claim — savings of around £320bn between 2030 and 2050 — is a single number built on a single pathway: one carbon-price trajectory, one nuclear cost curve, one demand path, one view of gas costs. There is no year-by-year breakdown, and no sensitivity analysis showing how the answer moves when the inputs do.
That matters, because a total spread over two decades tells you nothing about where the savings actually come from, or when. Are they even across the period, or bunched at the end once higher carbon prices and assumed nuclear cost reductions do the heavy lifting? You can’t tell — and that’s precisely what you’d need to know before betting a system on it.
Real system planning isn’t about finding the pathway that looks cheapest if everything goes right. It’s about understanding how a system behaves across a range of futures — high and low demand, faster and slower delivery, higher and lower fuel and technology costs — and choosing the approach that holds up when the world doesn’t cooperate. One deterministic line, however carefully modelled, can’t do that job.
The nuclear numbers are a bet, not evidence
From the 2040s, the alternative pathway leans heavily on new nuclear to meet rising demand — including baseload demand from data centres. So the cost case depends enormously on nuclear being cheap. And here the report assumes an LCOE falling from around £150/MWh for Sizewell C to around £108/MWh for plants coming online in the mid-2040s — a drop of roughly £30/MWh — much of it delivered by SMRs.
The only SMRs in commercial operation anywhere are in Russia and China — a small floating plant in the Arctic and one high-temperature reactor online since 2023 — both state-directed, first-of-a-kind projects, and neither delivered cheaply. Not one has been built in a Western market on a comparable basis, and the nearest, the BWRX-300 in Ontario, isn’t due until the end of the decade.
And every recent data point on large, complex infrastructure — Hinkley, transmission, the wider construction supply chain, all other energy technologies currently under construction — points to costs and financing going up, not down. A sustained build programme could genuinely improve delivery over time; that’s a reasonable hope. But a hope isn’t evidence, and the report doesn’t provide enough to justify a reduction of this scale.
The sensitivity isn’t small. Hold nuclear at £150/MWh rather than letting it fall, and the extra ~7GW in the pathway needs roughly £1.8bn a year more in subsidy — every year. That single assumption is doing a large share of the work in the overall cost comparison. Lean this hard on one technology delivering quickly and cheaply and you haven’t de-risked the system — you’ve concentrated the risk. If nuclear slips, the fallback is more gas, more interconnection, or more of the renewables this particular pathway was designed to avoid.
And the new gas looks cheap too
It isn’t only nuclear. The pathway builds around 21GW of new gas, and assumes a build cost of roughly £650/kW for new CCGTs. That’s well below what the market is actually seeing. Gas turbine shortages and strong global demand have pushed capital costs up sharply — even DESNZ’s own 2025 generation-cost estimates sit higher, and several recent analyses put new-build gas at more than double the report’s figure. Understate the capex on the firm capacity the whole pathway leans on, and you understate the cost of the pathway itself. So both of the technologies this system is built around are costed at the optimistic end.
Removing carbon costs doesn’t make them disappear — it moves the bill
The pathway takes electricity generators out of the UK ETS from the early 2030s. On paper that lowers wholesale prices, because it cuts the marginal cost of gas whenever gas is setting the price. But “on paper” is doing a lot of work.
Two consequences don’t get the weight they deserve. First, the Treasury loses the ETS auction revenue — of the order of £2.8bn a year. That doesn’t vanish from the economy’s balance sheet; it has to be found somewhere. Second, and more importantly, decoupling from EU carbon pricing risks triggering the full force of the EU’s Carbon Border Adjustment Mechanism on our exporters. UK iron, steel and aluminium producers carry something like £10bn of CBAM exposure. Take our generators out of the ETS and those industries keep paying a carbon price — only now they pay it to Brussels rather than London.
That’s not a saving for UK plc. It’s a transfer of revenue offshore and a direct hit to the competitiveness of exactly the energy-intensive industries we say we want to keep. An honest cost case has to net that off — and this one doesn’t.
A few other things that deserve more scrutiny
Beyond the big items, several assumptions look hard to reconcile with how the market actually behaves:
The £94bn of wholesale savings is striking in a system with less renewable generation and more gas setting the price more often. The report should show clearly how much comes from removing carbon costs versus network, balancing, support payments or lower demand.
The demand story runs backwards. Lower prices should pull demand up, through more EVs and heat pumps — yet the pathway shows less electrification. If prices and demand move in opposite directions, the two scenarios may not be a like-for-like comparison.
Batteries effectively disappear by 2040. That’s very hard to square with current capacity market contracts and the fact that storage is one of the few merchant technologies actually being built today. Batteries cut curtailment, support security of supply and reduce reliance on gas at peak — leaving them out understates flexibility.
No new interconnectors after 2030 increases reliance on domestic firm generation and reduces our ability to export surplus. Widening cross-border spreads — especially once CBAM-type costs bite — could make existing links less economic and push up cap-and-floor payments, with knock-ons for the Irish market and UK–France trade.
A gas-heavier system is a more gas-exposed system. The report notes fuel costs rising ~£6bn under a gas-price shock, but not what that does to wholesale prices and bills once it flows through the system. That’s the number that matters for resilience.
And policy reversals aren’t free. Removing existing commitments would be read by investors as raising risk — which raises the cost of capital across the board. That effect doesn’t appear to be reflected.
Where I agree
None of this means the work is without value — the opposite. It’s a genuinely useful stress-test of where system costs sit, and it’s right that balancing, ancillary services and some network costs could be lower in a system with less variable output. Gas capacity through the 2030s looks broadly plausible — existing CCGTs will run longer than many assume, so the near-term picture doesn’t depend on a rush of new build. RWE is the largest gas generator in the UK; we’re not romantic about intermittency, and no one benefits from pretending the current trajectory has no costs.
The result is an echo of the inputs
Step back, and the pattern is hard to miss. Assume new nuclear is cheap, new gas is cheap to build, gas stays cheap to run, and carbon costs nothing — and of course the model hands you a system built on nuclear and gas. Assume instead that pollution carries a price, and that these technologies might cost what they are actually costing right now — and the same model points somewhere very different: a more balanced system that spreads the risk rather than concentrating it. The headline isn’t really a finding about the cheapest way to power Britain. It’s a reflection of the assumptions you start with.
What good looks like
If we’re going to have a serious debate about reshaping the power system — and we should — let’s have it on the right basis: a full, published and more detailed databook; a genuine range of scenarios rather than a single line; and technology-cost assumptions we can defend against recent evidence. Get those right and this becomes what it should be — a valuable input into the decision, rather than the decision itself. Indeed, this could result in a Pragmatic Power Plan (PPP…..I am a child of the 1990s power market after all!) which we could all stand behind - somewhere in the reasonable middle between the Clean Power Plan 2030 and the Cheap Power Plan.
The report is worth your time. Read it as a scenario, not a plan.
Report: Onward, “Firm Foundations: The case for cheap and reliable power”