Reversing scorched earth policies in the electricity sector: Part 4 [Nuclear options]

This article is prompted by thinking about the costs and implications of adding the Hinkley Point C nuclear power station to the GB electricity system when it is completed in the early 2030s. At the same time, Michael Cembalest’s annual report on the energy sector for JP Morgan Asset Management includes some interesting comments on the prospects for Small Modular Reactors.

Let us start with Hinkley Point C (HPC) project. It has a CfD whose current strike price is £127 per MWh, which is far above the average market price in 2025-26. If it achieves a load factor of 93%, which it should if operated properly, that is a commitment for electricity users to pay roughly £3.3 billion per year to EDF. The most reports suggest that HPC will commence operations in 2030-31 and will cost about £35 billion at 2015 prices, equivalent to about £48 billion at 2025 prices.

The project has been close to a catastrophe for EDF, which has only been able to finance the construction with the support of the French government. While the cost of electricity from HPC is very high it translates to an annual return of less than 6.5% on the cost of construction for 40 years after allowing for operating, fuel and decommissioning cost. The internal rate of return on the project will be about 5.7% in real terms. This is not a ridiculously high IRR but, given the magnitude of the cost overrun (roughly 100%), it indicates an extraordinarily generous CfD contract at the outset. Had HPC been built within its original budget of £18 billion at 2015 prices, the IRR would have been 12.6%.

What should a government do now – or when construction is completed? The choices are unpalatable. However, the UK government can issue index-linked bonds on terms that are far more favourable than paying EDF a real IRR of 5.7%. If the UK government were to buy HPC from EDF at a completed price of £48 billion financed by, say, 50-year index-linked gilts, the real IRR for such a transaction would be about 2%. The annual cost of supplying power to the grid would be reduced to about £1.75 billion per year or roughly £67 per MWh at 2025 prices.

While EDF may be reluctant to accept a price less than the book value of what it has spent on building HPC, getting the plant off its balance sheet would be a huge relief for both the company and the French government. Thus, there should be scope for a hard-nosed British government to get a better deal. EDF has the experience required to operate the plant, so any deal should include a carefully written performance contract to operate the plant for 40 years. At £67 per MWh – or less – of reliable baseload power, HPC would be expensive but not absurdly so. Of course, as with all decisions affecting public debt, the cost would be much larger if the UK government is unable to control public spending and the cost of financing public debt.

The cost of HPC is extraordinarily high – about £15 million per MW of capacity. As a comparison, Units 3 and 4 of the Vogtle Nuclear Plant in Georgia with a capacity of 2.2 GW are estimated to have cost about $27 billion including financing costs when the plant was completed in 2024. This sum was increased by the bankruptcy of Westinghouse during construction which resulted in a suspension of on-site activity for nearly a year. While it is difficult to make exact comparisons, the cost of the Vogtle units was about £12 million per MW at 2025 prices, the cost of HPC will be about 25% higher.

EDF claims that the cost of the next EPR plant at Sizewell will be lower because of learning from HPC. That seems rather optimistic. Four EPR plants have been built or are nearing completion – in Finland, China, France and England. All of them have experienced major overruns in both cost and development times. At the same time as building the plant at Sizewell, EDF is committed to a large program of building new plants in France using a substantially modified EPR2 design.

Recent experience suggests that EDF has neither the technical or human capacity to manage the construction of these projects on time and within budget. From a UK perspective the best option would be to cancel the development of the Sizewell plant, but the current government seems to have trapped itself with no immediate alternatives and a desire not to offend the French government.

The dilemma facing policymakers seeking a reliable, low carbon, type of generation is painful. There has clearly been negative learning for most Gen III/III+ reactors – i.e. costs have tended to increase over time. This was partly because of the loss of expertise following the slowdown in nuclear construction in Europe and North America after 1990, and partly because of changes in nuclear safety and regulatory rules. The only countries with a proven capability to build large Gen III+ nuclear plants close to schedule and budget are China and South Korea.[1]

The notion of relying on Chinese designs and constructors is not acceptable in most of Europe, and it is far from clear that the South Korean industry has the capacity to build more than a small number of reactors simultaneously. The construction of APR-1400 reactors at two sites in South Korea and at Barakah in the UAE between 2012 and 2024 clearly stretched KEPCO’s resources.

On the other hand, hiring KEPCO to build up to 4 units (5.6 GW) of either the APR-1400 or the modified APR+ version is the only way in which the UK would have any chance of bringing a significant amount of large nuclear capacity onto the GB electricity system after HPC is completed but before 2040. As a reference the Barakah nuclear plant consists of 4 x APR-1400 reactors built between 2012 and 2024 at a reported cost of $32 billion. This was equivalent to about £30 billion at 2025 prices or £5.4 million per MW. Even allowing a premium of 40% to allow for UK conditions, the cost would only be £7.5 million per MW or one-half of the cost of HPC.

The difficulty, of course, is that no British government is likely to be sufficiently hard-headed and free from the influence of lobbyists to make and stick to a decision to hire KEPCO to build such a project. The only merit of pushing ahead with EDF’s EPR at Sizewell is that it should avoid the shambolic mess that the bureaucracy has created over large military and civil projects. That is a very high price to pay for the incapacity of politicians and bureaucrats to take and stick to reasonable and financially responsible decisions.

The current get-out-jail card is the vision of small modular reactors (SMRs) or Gen IV reactors. The original idea of SMRs was a scaling up of the type of reactors used in nuclear submarines, aircraft carriers and other naval vessels. Such reactors could be mounted on barges, towed to remote sites and connected to the local grid. That is exactly what Russia has done in remote Arctic locations. However, what might be ok in coastal towns in Murmansk oblast is not likely to be as well received in Scarborough or Minehead. In addition, as Cembalest points out, the military are not known for cost control and economy, so the unit cost of naval reactors seems to be 2 to 3 times the cost of the most expensive large nuclear plants like Vogtle and HPC.

Stepping back from naval comparisons, the key element in the story about SMRs is the “modular” part, i.e. like system housebuilding much of the reactor can be fabricated in an offsite factory and transported to site for installation. For this model of assembling factory-built components to work, the individual reactors must be relatively small but be designed so that multiple reactors can be installed at a site to provide varying amounts of generation capacity.

As the reference to system housebuilding suggests, views of the economics of SMRs vary greatly. Cembalest quotes studies which argue that the loss of the economies of scale inherent in large Gen III+ reactors mean that the unit capital cost of building SMRs is significantly higher than the equivalent cost of building large nuclear plants.[2] Against this, it can be argued that this view is disingenuous, since projects to build large nuclear plants in Europe and North America in the last two decades have been almost uniformly disastrous due to the opposition that they attract together with the inability of most countries to manage large projects. The advocates of SMRs argue that by turning the construction of nuclear reactors into a production line, the benefits of standardised manufacturing will outweigh the classic economies of scales.

System housebuilding is not an encouraging comparison. When adopted at scale, the results have often been disappointing, except for upmarket houses designed and built to a very high standard. In the case of SMRs what is really required is a lot less talk and a lot more experimentation with public funds underwriting the construction of several designs of SMRs and running them for a reasonable period to assess their performance.

I have commented previously that advocates of nuclear power are their own worst enemy. This is all too obvious with SMRs. There are far too many designs being promoted with almost no filtering of technologies and reactor designs. Perhaps the market will eventually select a smaller number of viable models, but currently what most outsiders can see is incoherence and endless lobbying.

Several countries are funding pilot programs but commitments to move beyond the design, testing and licensing stages of projects with resources to build plants are very sparse other than in Russia and China. In large part, this is because the cost of a medium-scale (300-500 MW) pilot plant is likely to be $3 to $5 billion. Governments are reluctant to provide such money when there are several designs all claiming that their approach is the best one. Consequently, SMRs are unlikely to become a serious commercial proposition significantly before 2040.

The UK is following its familiar not-invented here strategy by supporting an SMR design proposed by Rolls-Royce that is not small, modular, or anywhere close to deployment. It is just a medium-scale conventional PWR reactor that on realistic assumptions is likely to cost at least £12 million per MW after the initial units are built.

Early SMR units are likely to be expensive. Cembalest refers to experience in China suggesting a capital cost that is two times its large plants. Still, China is the cheapest place in the world to build nuclear plants. There are SMRs under construction in the US (TerraPower’s 345 MW natrium reactor in Wyoming) and in Canada (GE-Hitachi BWRX-300 in Ontario). These initial units are expected to be very expensive – over $15 million per MW for the initial reactors but with an expectation that unit costs will fall by one-third or more for later units.

Perhaps the real lesson is that the UK is not the place to experiment with new reactor designs. TerraPower’s sodium-cooled reactor is the most radical of the designs being reviewed under the Generic Design Assessment process. It offers considerable flexibility to ramp output up or down through a molten-salt energy storage system. While the UK government is likely to fund one or more pilot units of the Rolls-Royce design for political reasons, it should not pre-empt decisions on whether to proceed with multiple SMR plants until real comparisons of operating performance and costs can be made.

[1] In broad terms Gen III/III+ reactors are a development of the standard pressurised water (PWR) reactor designs that were built from the 1970s to 1990s in Europe, Japan and North America. The Gen III+ designs have enhanced safety features that are added to prevent the failures that occurred at the Fukushima plant in Japan after it was struck by a tsunami.

[2] Readers should not pay too much attention to the levelised costs cited in Cembalest’s report. He adopts absurd assumptions about the cost of capital and, in some cases, about unit capital and other costs. In part, this is because the report seems to be confused about real vs nominal values. This is not unusual in studies prepared by US investment banks, because most US regulatory agencies are uncomfortable with working in constant prices. This reflects a preference for relying upon unadjusted financial accounts due to the legal traditions of rate of return regulation.