What nuclear procurement and project delivery can learn from aerospace

The conventional explanation for nuclear power’s cost and schedule failures runs roughly as follows: nuclear is uniquely complex, burdened by regulation, politically contested in some regions, and too capital-intensive for private markets to bear. The implication is that the solution lies in regulatory simplification, public financing, and political will.

This explanation is not wrong. But it feels incomplete and obscures the more actionable diagnosis.

Aerospace and defence programmes operate under comparable technical complexity, comparable regulatory scrutiny, and, in the case of military systems, far greater political sensitivity. Yet the best-performing aerospace programmes consistently deliver at a level that the nuclear industry, at least in the West, has failed to match for decades. For instance, in France, Naval Group is expected to deliver the sixth and final Barracuda-class nuclear attack submarine to the Marine nationale in 2029, approximately one year ahead of its contracted 2030 schedule. A nuclear submarine programme, delivering early.

The difference is not primarily technical but structural. It lies in how contracts are written, how risks are allocated, how supply chains are designed, and how owner capability is built and maintained throughout the life of a programme. All to say, how project delivery is designed.

These are procurement architecture questions. And the nuclear industry might not have asked all the relevant ones.

The scale of a systemic problem

The data on Western nuclear construction performance is well-documented, but the pattern bears restating because it is so consistent.

The Flamanville EPR in France began construction in 2007 with an estimated cost of €3.3 billion and a four-year build timeline. By the time the reactor approached commercial operation, the cost had exceeded €13 billion, and the project was almost fifteen years late.

In Finland, the Olkiluoto 3 EPR was originally budgeted at €3.2 billion. The final cost exceeded €11 billion, a 3.7-fold increase, and the reactor came online thirteen years behind schedule. The project ended in multibillion-euro litigation between the Finnish utility TVO and contractor Areva over responsibility for the overruns.

In the United States, the Vogtle expansion in Georgia was contracted at an estimated $14 billion for two AP1000 reactors. The final cost exceeded $25 billion,  nearly double the estimate,  with the project running six years behind schedule. The overruns were so severe that Westinghouse Electric, the world’s leading nuclear technology vendor, filed for bankruptcy in 2017 after being unable to absorb losses from its fixed-price construction contract.

In the United Kingdom, Hinkley Point C was approved in 2016 with an estimated cost of £18 billion. By 2026, current price estimates had risen to approximately £48 billion, with the first unit not expected to generate power until 2030 at the earliest.

These are not isolated failures attributable to bad luck or unique local conditions.
They are the consistent output of a particular procurement architecture and constrained project delivery.

An inconvenient comparison

The instinctive objection to comparing nuclear with aerospace is that aerospace programmes also overrun. The F-35 Joint Strike Fighter, one of the most expensive weapons programme in history, saw its total acquisition cost rise from an initial estimate of $233 billion to over $485 billion by 2023, with the programme running more than a decade behind schedule.

The objection is valid as far as it goes. But it obscures the more illuminating comparison.

The F-35 is a cutting-edge fifth-generation combat aircraft developed over four decades in parallel with the operational doctrine it is meant to enable.
A nuclear power station, by contrast, is a large civil infrastructure project built around reactor designs that have been in commercial operation for decades. The physics has not changed. The regulatory framework, while demanding and being more stringent following the Fukushima accident, is not new. The argument that nuclear is simply “too complex to deliver reliably” collapses when one examines South Korea’s record.

South Korea has been building nuclear reactors sequentially since the 1970s. Its domestic APR1400 units have been constructed at an overnight cost of approximately $2,300 per kilowatt, roughly a quarter of the cost per kilowatt at Vogtle. The four-unit Barakah project in the UAE, built by a South Korean-led consortium, demonstrated learning rates that Western programmes cannot approach: Unit 4 reportedly cost around 40% of what Unit 1 cost, reflecting the compounding effect of building the same design with the same workforce in sequence.

The variable that separates South Korea from the West is not technical capability or regulatory environment. It is programme architecture: standardised design, a stable supply chain, a contractor with genuine serial construction experience, and an owner that maintains deep technical capability throughout.

The contract structure and project delivery problem

The most consequential design choice in any major nuclear programme is the contract structure and how project delivery is designed; and Western nuclear has persistently made the wrong one.

The lump-sum turnkey or fixed-price EPC model transfers virtually all construction risk from the project owner to the contractor in exchange for cost certainty. In theory, this protects the owner.
In practice, for projects as complex and first-of-a-kind as the EPR or AP1000, it has produced a sequence of contractor bankruptcies, protracted litigation, and projects stranded mid-construction.

The mechanism is straightforward. A contractor bidding a fixed price on a novel, highly complex project must either embed massive risk contingencies, making the bid uncompetitive, or underbid and absorb losses as complexity materialises.
The incentive structure is perverse: contractors have every reason to bid low to win work, and every reason to fight over variations rather than solve problems. Areva’s fixed-price contract at Olkiluoto ended in years of litigation with TVO. Westinghouse’s fixed-price contracts at Vogtle and V.C. Summer bankrupted the company.

The White & Case analysis of major energy procurement is unambiguous on this point: there is now “a growing reluctance among contractors to sign up to EPC contracts on a lump-sum turnkey basis”, given the complexity of modern energy projects. The legal and commercial framework that was supposed to provide certainty delivered the opposite.

Aerospace and defence procurement has, by necessity and hard experience, developed more sophisticated approaches. Development risk and production risk are typically separated. Incentive fee structures align contractor reward with delivery outcomes rather than variation claims.
The owner maintains sufficient technical capability to manage the scope and validate contractor performance.
Where fixed-price elements exist, they are applied to well-defined, lower-risk work, not to the totality of a novel, first-of-a-kind programme.

The UK’s emerging approach for future nuclear projects, including the Regulated Asset Base model for Sizewell C and alliance contracting models being explored for AMR programmes, represents a belated recognition that the LSTK model cannot work for novel, complex infrastructure. The transition is happening. It is happening more slowly than it should.

Supply chains designed for accountability

The aerospace industry organises its supply chain through a tiered structure in which accountability is explicit and traceable at every level. Tier 1 suppliers, like large integrators working directly with the OEM, are responsible for the performance and quality of their own supply base. Tier 2 suppliers are held to clear specifications by their Tier 1 customers. Traceability, quality management, and regulatory compliance flow up and down the chain.

Nuclear procurement has historically pursued a different objective: lowest cost.
The result is a supply chain in which nuclear quality assurance requirements are sparsely met, certified suppliers are few, and accountability for quality failures is diffuse.
Counterfeit or substandard components have entered reactor supply chains from manufacturers insufficiently subject to nuclear-grade inspection regimes.

The deeper problem is that lowest-cost procurement in a sector with sparse NQA-certified suppliers does not actually deliver lowest cost. It delivers deferred cost in the form of rework, inspection failures, and regulatory hold-ups. NQA requirements are associated with significant cost escalation when applied without sufficient clarity or experience, but the alternative, inadequately qualifying suppliers, results in a different, less controllable form of cost escalation.

Redesigning nuclear supply chains for accountability rather than lowest cost means developing a smaller number of deeply qualified suppliers with genuine programme relationships, applying tiered accountability logic, and maintaining owner visibility into supply chain performance rather than delegating it entirely to the prime contractor.

The owner’s capability problem

Perhaps the most underappreciated factor in Western nuclear construction failures is the erosion of owner capability.

In the decades between the last Western new build programmes and the current generation of projects, nuclear utilities lost, through retirement, restructuring, and institutional neglect, most of the in-house capability needed to manage the delivery of a major nuclear project. The IAEA has identified knowledge management and expertise retention as among the most critical challenges facing the sector, noting that new recruits can fill vacant seats but “cannot replace lost knowledge” accumulated over decades of operational experience.

The consequences materialised with particular clarity at Vogtle. When Westinghouse’s bankruptcy forced the project owners, Georgia Power and its co-owners, to take over direct management of the construction programme, they were responsible for managing a massively complex project for which they were, as observers noted at the time, fundamentally “ill-prepared”. The owner’s technical capability had atrophied during the years when nuclear construction had seemed like someone else’s problem.

The contrast with aerospace and defence procurement is instructive. The government customer maintains dedicated technical teams, integrated project teams, programme offices, and technical authorities throughout the life of a programme.
The UK established a dedicated Submarine Delivery Agency in 2018 precisely to create an enduring owner capability function for its submarine programme, with an explicit mandate to develop and retain the expertise needed to manage complex, long-duration delivery.

The principle is clear: the owner’s technical authority is not optional overhead. It is the mechanism by which the procurement can be managed at all. Nuclear new build programmes require this principle to be applied consistently, building and sustaining integrated owner teams with genuine technical depth from pre-FEED through construction and commissioning, not deploying a programme management office at contract signature and hoping for the best.

The serial build imperative

The final lesson from aerospace and from South Korea is the discipline of serial construction.

Aerospace’s most successful programmes achieve cost reduction and schedule reliability through the systematic application of learning rates: the same design, the same processes, the same supplier relationships, repeated and refined across successive production runs. Complex, technically demanding systems can be built reliably at scale when the programme architecture supports learning rather than constantly restarting from scratch.

Western nuclear has done the opposite. Every new build project in recent decades has been, in important respects, a first-of-a-kind effort: a new site, a modified design, a reconstituted supply chain, a fresh contractor team. Accumulated learning from one project is not systematically transferred to the next. MIT research on nuclear construction costs estimates that the overnight cost of the nth AP1000 unit, after ten to twenty builds, could converge to less than half the cost of the first unit.

South Korea’s reported cost per kilowatt for domestic units is the result of building the same design, with the same team, multiple times. Unit 4 at Barakah demonstrably benefited from three prior builds on the same site; the learning rate was real and measurable.

The implication for nuclear policy is uncomfortable: a fleet strategy is not merely a preference but an economic and operational necessity. Individual bespoke projects, each treated as unique, will continue to perform the way Western nuclear has performed. Programmes structured for serialisation, standardised designs, committed pipelines, and continuity of the contractor and supply chain will perform as South Korea has.

What needs to change

The nuclear industry’s cost and schedule failures are not a mystery.
They are the predictable output of a particular set of procurement choices: contracts that transfer risk to contractors who cannot bear it; supply chains organised for lowest cost rather than accountability; owner organisations that have allowed their technical capability to atrophy; and programmes structured as one-off events rather than the first unit in a series.

None of these problems is technical. All of them are tractable. The aerospace and defence sector has, through decades of expensive experience, developed approaches to each of them, not perfectly, but substantially better than current nuclear industry practice.

The current pipeline of Western nuclear projects, EPR2 in France, Sizewell C in the UK, the emerging SMR programmes across Europe and North America,  offers a genuine opportunity to apply these lessons before they are learned again at enormous cost.

The programme’s architecture determines its outcome. Getting the procurement right is not a detail. It is the work.