In 2013, I visited the Headquarters of Hitachi in Tokyo, with Sheffield University’s VC, Sir Keith Burnett. Sheffield had recently built the Nuclear Advanced Manufacturing Centre (NAMRC) – part of the High Value Catapult Network, funded to support the UK’s ambitious new nuclear build programme at the time. Our mission was to try and persuade Hitachi that NAMRC could help them build a UK supply chain and develop the nuclear skills they would need to deliver their project of building two new nuclear reactors at Wylfa, on Anglesey. The last nuclear power station to be built in the UK, Sizewell B, was finished in 1995; in the intervening years much national nuclear capability in the nuclear sector had been lost, and, at a time when there were high ambitions for the UK’s new nuclear build programme, NAMRC had been established to help rebuild that national capability.
In Tokyo, we met the head of Hitachi’s nuclear division, and the head of Hitachi Europe. We were very courteously received (perhaps helped by the presence of the UK’s Ambassador to Japan, who had been helpful in facilitating the meeting). We had a very interesting discussion about their reactor design – the Advanced Boiling Water Reactor – and their ambitions for rolling out this technology across the world. But on our primary mission, in the most polite way, we were completely rebuffed. It was Hitachi’s technology, the UK government wasn’t paying for it, so why would they want our help to build a UK supply chain? They had a perfectly good one of their own.
In the event, Hitachi is not going to build any new nuclear reactors in Wylfa, or anywhere else in the UK. In 2019, Hitachi mothballed its plans, definitively pulling out in 2020, citing a failure to reach a financially satisfactory agreement with the UK government. It seems highly likely that, in making this decision, Hitachi reflected on the experience of the UK’s first new nuclear project, the two European Pressurised Water Reactors (EPRs) being built at Hinkley Point C, in Somerset – a dismal story of cost overruns and long delays.
In 2013, I described the Hinkley Point deal as being spectacularly bad for the UK (“The UK’s nuclear new build: too expensive, too late”). What I hadn’t anticipated was how bad a deal it would turn out to be for EDF, the French state electricity company that is building Hinkley Point C. The deal guaranteed a fixed price for the electricity generated at HPC, but the construction risk was to be shouldered by EDF. In 2013, the cost was estimated at £16 bn – the final real terms cost is likely to be at least double this.
The UK taxpayer doesn’t have to contribute to this cost overrun directly – but EDFs experience means that no other operator is prepared to build a nuclear power station on this basis (including EDF – a second EPR installation at Sizewell will be financed in a different way, with the UK state taking a substantial equity stake in the project). So it’s important to understand where these cost overruns come from.
Part of the story must be the way the UK regulates large infrastructure projects in general, and nuclear projects in particular. This issue has been comprehensively addressed in a recent government report chaired by John Fingleton. The 2025 Nuclear Regulatory Review identifies a number of places where the UK system has introduced delays and additional costs.
This begins with a very slow process – up to four and a half years – for a nationally significant infrastructure project to receive a Development Consent Order. It incorporates exhaustive studies of environmental impacts (31,406 pages, in the case of Hinkley Point C) and slow timescales for environmental permitting. The UK regulatory authorities required extra control and instrumentation systems compared to the French authorities in their approval of the predecessor EPR at Flamanville. These extra systems added hundreds of millions to the project cost. Most notoriously, works on the intake cooling water system to protect fish cost £700 m, including an acoustic deterrent costing £50m – the so-called “fish disco”.
But can this be the whole of the story? These examples account for cost increases measured in hundreds of millions – but the actual cost overruns are an order of magnitude bigger than that. Here’s how the cost estimates changed with time (all figures here corrected for inflation and expressed in 2015 £s):
- 2016, projected cost: £18 bn.
- 2019, projected cost: £21.5-£22.5 bn. “Cost increases reflect challenging ground conditions which made earthworks more expensive than anticipated, revised action plan targets and extra costs needed to implement the completed functional design, which has been adapted for a first-of-a-kind application in the UK context.“
- 2022, projected cost: £25-26bn. “18 months delay due to COVID“
- 2024, projected cost: £31-£34 bn. “The cost of civil engineering and the longer duration of the electromechanical phase (and its impact on other work)“
To some extent, one might suspect that the British regulators might offer a convenient excuse for EDFs failures – and one sees a pushback against that in the Office for Nuclear Regulation’s somewhat snippy response to EDF’s 2024 announcement: “In relation to the volume of additional steel and concrete required at Hinkley Point C, we do not recognise our regulatory requirements as being the principal factor in these increases, as they are broadly similar to the requirements in France.”
In fact, EDF’s own recent record of building new nuclear reactors in France isn’t brilliant. The EPR that EDF built in Flamanville itself (a single unit, in contrast to the pair under construction at Hinkley) suffered from very substantial delays and cost overruns. It started construction in 2007 with an estimated cost of €4.3bn, but ended up with a total construction cost of €13.2 bn, and a total project cost of €19.3 bn.
It’s interesting to see a French perspective on the difficulties EDF has had building new reactors, both in France and the UK. As a fully French State owned entity, EDF is subject to the oversight of France’s Court of Auditors, which has issued two very interesting – and rather critical – reports this year. One, Le modèle économique d’Électricité de France, is an overall view of EDF’s business model, while La filière EPR : une dynamique nouvelle, des risques persistants is a deep dive into the EPR programme. The Flamanville programme was described as an operational failure, and EDF criticised for “inadequate project monitoring, lack of vigilance on the part of the supervisory authorities, and their delay in addressing the loss of technical expertise and quality culture within the nuclear industry.”
For France, the issue of understanding the origins of cost overruns at both Flamanville and Hinkley C has become critical, because of the 2022 decision that six further EPRs are to be built in France (with an option for a further 8 later). The EPR report is blunt about the challenges this poses – “The loss of technical skills and quality control within companies in the nuclear sector, as well as EDF’s lack of industrial control over the sector, have been identified as major causes of the operational failure of the Flamanville EPR construction project. The diagnosis of the sector’s weak industrial capabilities was made belatedly, in response to the difficulties encountered. This loss of skills is explained in particular by a gap of about fifteen years between the launch of the construction sites of Civaux 2 (the most recent French reactor in service) and Flamanville 3. Most of the supply chain companies involved in the construction site were affected, particularly the subcontractors, as EDF had not assessed the capabilities of the industrial sector before making the decision to launch the construction of a new type of reactor, nor had it been able to control the consequences of this lack of foresight on a construction site of this scale.”
Even in France, the issue comes down to a loss of national capability in the nuclear sector.
What does one mean by “national capability”? Put simply, it’s a combination of people who know how to do things, organisations with the hard and soft infrastructure that enables those people to get stuff done, and networks which can effectively bring together organisations and people with the different expertise and capacity needed to achieve the overall goal.
In the UK, the 1990’s saw a loss of national capability – the government owned civil nuclear sector was essentially run down to zero, and capability in the wider electricity generation sector was reduced through privatisation and subsequent acquisitions by overseas companies.
The attitude in the 2000s was essentially that this didn’t matter – the view was that the UK was a rich country, and if there was some technology it needed, it could buy it in. In the 2010s, the UK didn’t seem so rich any more, so the approach was to buy the technology in, but to get someone else to pay for it. The scale on which this approach has failed can be measured by the distance between the 2014 ambition of constructing 18 GW of new nuclear power stations, and the 2025 reality in which only 3.2 GW is under construction.
How does one regenerate national capability? Learning happens by doing. Our previous model of buying the technology in from foreign vendors, and getting overseas investors to pay for it, while the role of UK organisations is to police what they do, does perhaps come naturally to a country with a specialisation in knowledge intensive business services. Someone has to write those 31,406 pages of environment impact reports, and negotiate those complex contracts. But it’s ownership that gives control – and it’s natural that EDF, as major funders as well as contractors for Hinkley Point C, will feel a responsibility to their owner – the French state – to maximise the benefits of the learning by doing for the French nuclear industry, rather than supporting the development of a UK one.
The UK now needs major investment in technological infrastructure. If this is going to get built in an affordable and timely way, we do need to move some of the regulatory road-blocks. But this won’t be enough – we need to relearn how to make and build things: we need to take control and rebuild lost national capabilities.