Nuclear power and net zero: Too little, too late, too expensive

Prof Steve Thomas, Greenwich University, assesses the considerable obstacles to the UK government’s target for new nuclear power.

Article from Responsible Science journal, no.6
Advance online publication (original article): 9 January 2024; update published: 9 February 2024


Original article (9/1/24)


In October 2023, the British government reaffirmed the 2022 Boris Johnson target of bringing online 24 gigawatts (GW) of nuclear capacity, eight stations the size of Hinkley Point C, by 2050. [1]  As we wait for more details to be published on how the government intends to try to meet this target, this article critically considers likely proposals. While there is talk about Small Modular Reactors making a significant contribution, as I argued in my article on these in Responsible Science, no.5, [2] their rationale is based on some highly suspect assumptions about cost-savings from reducing reactor size. At most a few demonstration SMRs might be built, demonstrating only that they are far from being competitive with other options for low-carbon generation.

So, if the 24 GW target is to be met, most of the capacity will have to be in large (1.2 GW-plus) reactors. The government seems determined to drive through the Sizewell C project whatever the cost. This would comprise two reactors of the EPR-1 design used at Hinkley Point C, but that would leave a further seven to build.

To achieve the 24 GW target, at least four conditions must be met:

  1. The equivalent of eight new nuclear projects must be completed by 2050.
  2. Mature, commercial, large reactor technologies must be available.
  3. Seven sites beyond Sizewell, suitable for 3 GW stations, must be approved.
  4. Owners and financiers for eight stations, expected to cost about £250bn, must be found.

1. When could new capacity come online?

Ambitious nuclear programmes are always accompanied by the same tired rhetoric offered for more than 50 years - of cutting red tape, streamlining planning and regulatory processes, learning from past mistakes, and taking advantage of new technologies. This has never worked in the past, not because we were not trying hard enough, but because nuclear power stations intrinsically take a long time from start of planning to first power, and new technologies have proved expensive and bring their own problems. The government acknowledged this in its Impact Assessment for the Regulated Asset Base (RAB) legislation which stated that it typically took 13-17 years from a Final Investment Decision (FID) to first power. [3]  It could have added that most announced projects do not make it as far as FID. The Impact Assessment also stated that nuclear projects typically cost 20-100% more than the estimate at FID. Adding in a few years to get from project inception to FID and it is clear the whole process is likely to take 15-20 years. The Flamanville (France) and the Olkiluoto (Finland) projects will take longer than 20 years and with at least four years of construction left at Hinkley Point C, that project will take nearly 20 years if there are no more delays. Flamanville [4] and Olkiluoto [5] are about 300% over budget. Planning for any capacity that will be online by 2050 must be started by 2030.

2. Which technologies?

The EPR-1 design supplied by the French nationalised utility, EDF, is not credible for further orders. A former CEO of EDF described EPR-1 as “too complicated, almost unbuildable”. [6]  Design work has been in progress for more than a decade on its replacement, EPR-2, which is claimed to be cheaper and easier to build. EDF plans to build six EPR-2s in France, the first coming online in 2035-37. EDF has said it would not try to sell the design until an EPR-2 was operating in France. Whether the EPR-2 will live up to the claims made is irrelevant. If we must wait till the after 2035 for it to be available, EPR-2s cannot be online in the UK by 2050.

Assuming designs from Russia and China are not acceptable, that leaves us with the other two designs meant to make up the Blair programme of 16 GW by 2030, the Hitachi-GE ABWR and the Westinghouse AP1000. While these have been approved by the UK safety regulator, they are not attractive. The three reactors of the ABWR design operating in Japan use a 1986 version of it. No orders for the updated designs are in prospect and the vendor appears not to be offering it for sale.

The record of the AP1000 is almost as bad as that of the EPR with all eight orders going badly wrong. The history of the ‘AP’ designs illustrates the nuclear industry’s duplicity on reactor size. Initially it was the AP600 (about 700 MW), but this was found to be uneconomic. It was scaled up to the AP1000 (1170 MW) and this was built in China and in the USA, but to improve the poor economics, China has scaled it up to 1550 MW (CAP1400). In March 2023, Westinghouse announced its new design would be a scaled down AP1000, the 300 MW AP300.

The other candidate is the South Korean APR1400. Like the ABWR, this has been built but using a design that did not take account of the lessons from the Chernobyl disaster (a means of preventing a molten core getting into the environment) or from the 9/11 attack (a need to toughen the shell enough to absorb a hit by an aircraft). It seems unlikely that an updated design could complete the required safety review in time for an FID to be taken on a project using this technology until after 2030. The record of APR1400 projects is problematic with long delays due partly to falsification of quality control documentation in South Korea and quality issues in the UAE.

3. Where would they be built?

Eight sites were identified as suitable in the government’s siting decision of 2010. [7]  With Hinkley and Sizewell already under some sort of development, this leaves Moorside, Wylfa, Oldbury, Bradwell, Heysham and Hartlepool. There are concerns about the impact of sea-level rises for all the sites. [8]  A project for the Wylfa site underwent review by the Planning Inspectorate which recommended the project not be consented because of its environmental impact. Moorside, Oldbury and Bradwell have undergone some investigations for new nuclear capacity for projects now abandoned and this preparatory work could be utilised to speed things up.[9] Heysham and Hartlepool would need detailed assessment to determine their suitability before any project could be proposed, so they might not be available by 2030. If eight projects (including Sizewell C) need to be completed by 2050, then either the planning advice at Wylfa would need to be ignored or at least one new site would be needed – and this also assumes all planning issues at the other sites could be adequately dealt with by the end of this decade and none of these locations would be earmarked for SMRs.

4. How would they be financed and who would own them?

When electricity utilities could pass on whatever costs they incurred, they enthusiastically supported nuclear projects. Now, if nuclear projects go wrong, it will be their shareholders who bear some of the costs, so interest from utilities, particularly investor-owned ones, has evaporated. Direct government ownership is an option, although it would be an extraordinary decision to invest taxpayers’ money in nuclear projects on the basis that no other investors would be willing to take this risk. So, innovative methods of finance are required.

The finance model used for Hinkley Point C, the Contracts for Difference (CfD) model, was both a poor deal for consumers and the plant owner, EDF. The power purchase price was set in 2013, three years before the investment decision, at £92.5/MWh in 2012 money, indexed to inflation (about £124/MWh in 2023 money) with cost overruns falling on EDF. This price is more than double the price for new offshore wind. [10]  In 2013, the expected construction cost was £16bn but the latest estimate is £26bn (both in 2016 money). [11]  So EDF will have to absorb the cost overrun of at least 60% but with no increase in the price it will get for its output. This form of CfD is not an option any sane investor would back for nuclear even though, for offshore wind, it is producing impressive results and will continue to be used.

The UK government is now proposing the Regulated Asset Base (RAB) financial model. The main architecture of the scheme is known although crucial details have not been published. How far this lack of information is down to the government leaving these open for negotiation, to the government not having decided on them yet, or to the government not being willing to admit the details, is difficult to determine. There is brave talk of risk-sharing but the reality is that it will not be the government that sets the terms, it will be investors unless the government is prepared to walk away with no deal. But the government seems likely to agree to whatever it takes to lure investors in. Deepa Venkateswaran, an analyst at Bernstein, said would-be investors in Sizewell needed to be “assured a return” that was locked in at the point of investment rather than subject to change. [12]

Under RAB, it would be the investors’ income that would be fixed, not the price paid for power. The power price would be whatever it took to generate the guaranteed annual income to the owners. All electricity retailers and therefore all consumers would be required to buy their share of the output. With the Hinkley Point CfD, the owner took the risk; with Sizewell RAB, consumers take the risk.

The selling point for the RAB model has been the claim that it would reduce the cost of finance and therefore the cost of power. RAB reduces financing costs in two ways. First, because the risks will fall on consumers and taxpayers, the project would be seen by financiers as low risk to them and would attract a low interest rate. Second, the finance charges would effectively be paid by consumers as a surcharge on their bills payable from the date of FID to completion of the plant, expected to be about 15 years. Finance costs savings would be paid for by consumers as a surcharge on bills and by them, not the project owners, assuming the project risk.

Despite this, the government is struggling to find investors. It has said there are at least four companies that have pre-qualified as potential investors, [13] although pre-qualifying commits them to nothing. EDF has been forced to offer to take about 20% of the project ownership, while the government has said it would take an unspecified stake but it will be at least 20%, but probably more, enough to fill any funding gap.

The original target for RAB was UK institutional investors but given lack of interest from this source, government now seems to be relying on more controversial sources such as Middle East investment funds. [14]  It will be difficult to explain to the public why, if the Bradwell project was politically unacceptable because of the presence of Chinese money, a RAB project with, say, Saudi money is acceptable.

The government may be able to offer enough sweeteners to allow the Sizewell C project to proceed but replicating it will be more difficult. For each project, a technology, a site, and investors will have to be found. Politically it will be difficult for the government to keep taking expensive stakes in nuclear projects just because nobody else will. The scale of investment is huge, and, for example, Sizewell C alone is expected to cost about 10 times the cost of the Thames Tideway ‘super-sewer’ water project, the first major project to use the RAB model.


The electricity sector ought to be one of the first sectors to be decarbonised because of the availability of a range of viable technologies available to replace fossil-fuel generation. Boris Johnson set a target of decarbonising electricity by 2035 [15] while Keir Starmer has set a target of 2030. [16]  Given that even Sizewell C is unlikely to be online by 2035, the nuclear programme is an irrelevance in achieving net-zero. The only justification is if nuclear was the cheapest way to meet electricity demand growth by the time the first capacity could come online and the current chasm in cost between nuclear and renewables or energy efficiency measures suggests this is implausible. Judged by the requirements of time, technology availability, sites and availability of finance, the programme will fail badly. In doing so, large amounts of government time and taxpayer money will, as with previous UK nuclear programmes, be diverted away from the options that have a much higher success probability, are more cost-effective and can be deployed much quicker.

Since the original article was finalised, the UK government published its delayed Roadmap to achieve its target of 24GW of new nuclear capacity by 2050. [17]  This contained little of new substance and did not address the barriers to achieving this set out in my article above. It also made new announcements of further spending: an additional £1.3bn on the Sizewell C project; £300m on new nuclear fuel facilities for High-Assay Low-Enrichment Uranium (HALEU); [18] and £64m on Small Modular Reactors (SMRs). [19]  A new estimate for the completion cost and date of the Hinkley Point C nuclear power station was also published by Electricité de France (EDF).

The scale of the new estimates of cost and time overruns for Hinkley Point C was a shock. In the 20 months since the previous announcement, [20] the estimated completion cost had increased from £25-26bn to £31-35bn (all in 2015 money) and the completion date for the first of the two reactors had been delayed from 2027-28 to 2029-31. So expected completion is further away, and the costs and time are subject to greater uncertainty than they were in May 2022.

Apparently oblivious to this renewed demonstration that the EPR technology being built at Hinkley Point and proposed for Sizewell was “almost unbuildable” (see main article), the government announced further taxpayer support for Sizewell. It committed to increase its contribution to bring the plant to the point of Final Investment Decision (FID), projected for 2024, from £1.2bn to £2.5bn. [21]  By the end of 2022, EDF had also spent £700m on this process, [22] so adding in its contribution (assuming some further spend in 2023) might take the total to about £3.5bn. The original budget to get to an FID, set by EDF in 2016 when the agreements were signed, was £458m, less than one seventh of the latest expected cost: just to get to a position when an investment decision might be taken. This begs the question how many homes could have been insulated and how much offshore wind capacity could have been built and operational in a period of eight years with a budget of £3.5bn.

Steve Thomas is Emeritus Professor of Energy Policy at Greenwich University, UK. He has researched and written on nuclear power policy issues for 40 years. 

[Image by Kiyoshi Nakayama from Pixabay.]


[1] Science, Innovation and Technology Committee (2023).

[4] The cost estimate for Flamanville 3 at start of construction was €3.2bn, the most recent estimate (December 2022) was €13.2bn (2015 money). EDF hopes the reactor will go online in 2024.  

[5] The cost estimate at start of construction was €3bn. The final cost has been estimated to be nearly €11bn.

[9] The Bradwell project to be built by the Chinese company CGN has not been formally abandoned but EDF’s most recent annual report stated: “The project to build a nuclear power plant based on the UK HPR1000 technology reactor is unlikely to be implemented.”

[18] The announcement on HALEU, heavily couched in anti-Putin rhetoric, is even harder to comprehend. Worldwide, HALEU is used for reactors producing medical isotopes and a small number of prototype reactors. The government acknowledges that it will be difficult to deploy the type of power reactors that would need HALEU before 2050 in the UK and the outlook is no better elsewhere. So why we need to bring the new facility on-line in the early 2030s is incomprehensible. DESNZ (2024b).

[19] The government also announced it was commissioning the Office of Nuclear Regulation to carry out Generic Design Assessments (GDA) for the Holtec and GE-Hitachi Small Modular Reactors. Holtec was given £30.05m and GE Hitachi £33.6m to pay for the first two of the three of the stages of the GDA. DESNZ (2024c).

Electricity pylons at sunset

Filed under: