How China Runs 38 Nuclear Reactors as One Construction System

Most of the world cannot build a single nuclear reactor on time and on budget. One organisation is managing 38 of them at once.

That organisation is China’s State Nuclear Power Technology Corporation (SNPTC), and its approach carries a name that sounds like a procurement memo: fleet-based nuclear construction. The idea behind it is far more consequential than the label suggests.

At the World Nuclear Association’s World Nuclear Symposium in London (9-11 September 2026), SNPTC president Ma Yuanhua argued that this construction model is not a Chinese advantage to be admired from a distance. He framed it as an industry-wide necessity. The global nuclear sector has committed to tripling output by 2050, and the question has shifted. It is no longer whether the world should build more reactors, but whether it can physically deliver them at the pace required.

This piece gives you three things: the operational blueprint SNPTC actually uses, the four structural bottlenecks that apply to every country attempting large-scale builds, and what the answers mean for uranium demand forecasts. This is the mechanism, not a general tour.

What “fleet-based construction” actually means in practice

The term sounds administrative. The logic underneath it is not.

Fleet-based construction is the deliberate treatment of many concurrent reactor builds as a single integrated production system rather than a collection of separate projects. Most nuclear programmes build one reactor, learn hard lessons, then start the next one almost from scratch. SNPTC does the opposite. It treats every site as one node in a shared factory line.

Small modular reactor technology represents a parallel construction philosophy: rather than achieving scale through fleet replication of large units, SMRs pursue factory-built standardisation at the individual unit level, a model that sidesteps some of the certified welder and site-specific regulatory bottlenecks that constrain large-reactor fleet approaches.

The load-bearing idea is design reuse. SNPTC has reported a standard design reuse ratio exceeding 80% across its projects. That means detailed engineering, procurement specifications, and construction sequences are replicated rather than reinvented at each site.

Here is why that matters. The expensive, error-prone phase of a reactor is the engineering. Locking in 80% of the design before breaking ground means that phase happens once and is then stress-tested across dozens of units, rather than improvised afresh every time a new project begins.

The rest of the model builds on that foundation. SNPTC’s fleet approach rests on four operational pillars:

• Standardised design reuse: a fixed reference design replicated across sites
• Centralised procurement: bulk purchasing across all concurrent projects
• AI-assisted and modular construction: digitally sequenced, factory-built components
• EPC-led commissioning: an engineering, procurement and construction model with high owner involvement

Centralised procurement is where the scale pays off directly. Buying for 38 sites at once means bulk purchasing at a scale no single-project builder can match, and it connects to a cross-project manpower matrix that pools specialist labour rather than bidding for it site by site.

The performance targets tell you how tightly the system is meant to run. SNPTC targets a 56-month construction timeline using the fleet approach, a 95% first-time pass rate for key processes, and a first-time equipment acceptance pass rate above 93%.

“Fleet-based construction is an industry necessity, not an optional strategy,” Ma Yuanhua, SNPTC president, told the World Nuclear Symposium.

From procurement to commissioning: how the workflow connects

The point of the model is continuity. SNPTC links design, procurement, construction, and commissioning into one uninterrupted workflow, using modular construction and digitally assisted sequencing to keep each stage feeding the next.

Commissioning follows an EPC-led model with substantial owner involvement: 40% to 50% for new-build projects and 50% to 60% for expansion projects. That ratio matters because it signals where quality accountability sits. The owner is not handed a finished plant to inspect at the end; it is embedded in the process throughout, which reduces the risk of defects surfacing only at start-up.

AI tools and early-warning systems for high-risk components are the practical expression of this. Rather than absorbing the chaos of running dozens of projects at once, the model manages that complexity by flagging problems on long-lead items before they cascade into schedule delays.

The four bottlenecks that determine whether the world can replicate this model

SNPTC’s leadership identified four challenges to delivering multiple nuclear projects safely, on time, and on budget. The important framing is that SNPTC presents these as industry-wide barriers, not China-specific ones. They are a linked chain: you cannot fix supply chain without fixing workforce, and you cannot fix either without design standardisation holding the whole system together.

The first is design standardisation. A fixed reference design is what makes serial production possible, but frequent changes disrupt construction while fully static designs risk obsolescence. SNPTC manages this through a Design Change Board using tiered approval to prevent design drift. The global cautionary tale sits in France, which cut construction times from roughly 80 months for early units to under 60 months through standardisation, only to watch the same stress-corrosion cracking discovered in 2021 propagate across its uniform fleet. Standardisation multiplies quality, and it multiplies flaws.

The second is project management at scale. Coordinating personnel, equipment, and commissioning across concurrent programmes requires replicable frameworks. In China’s second-series AP1000/CAP1000 builds, stronger project integration reduced milestone durations by 48%, shortened intervals by 42%, and lowered schedule variance by 60%. The Shidao Bay CAP1400 unit reached grid connection in roughly 5.3 years, against about 9 years for first-series units.

The third is supply chain and long-lead equipment. Procurement timelines for critical components risk failing to keep pace with simultaneous demand. WNA analyses warn that uranium conversion services are expected to be “very tight,” and that nuclear-grade structures and components face capacity constraints.

The nuclear fuel supply chain faces the same serial-production logic as reactor construction: without coordinated investment in conversion, enrichment, and fabrication capacity running years ahead of reactor commissioning, the fuel cycle becomes the binding constraint even when construction timelines are met.

The fourth challenge, workforce, is the most acute near-term constraint, and it deserves room of its own.

The certified welder problem: a concrete limit on construction speed

The global nuclear industry employed roughly 1.1 million to 1.2 million people across 2023-2024. The IAEA anticipates the sector will need 4 million professionals by 2050, and training pipeline demand rose 280% in 2024 alone. Around 46% of surveyed nuclear companies already report hiring difficulties causing operational bottlenecks.

The sharpest shortage is in certified nuclear-grade welders and non-destructive testing (NDT) technicians, the specialists who verify weld integrity. The numbers are stark. The United States faces a shortfall of 480,000 welders and Japan lacks 250,000. Australia has 67,000 welders in total, yet fewer than 5,000 meet nuclear-grade qualification, and a third of those are nearing retirement.

These are not abstract labour statistics. If a country cannot certify enough welders, it cannot close the containment structure on a reactor, regardless of how good the design is or how much capital has been committed.

Nuclear-grade welding qualification takes years to achieve and cannot be accelerated by simply increasing training intake, which builds a structural lag between demand and supply. SNPTC’s labour pooling mechanism partly mitigates this, but it only works within a single-jurisdiction fleet. It does nothing for a country starting a new programme from a low base.

Why China builds reactors for a fraction of Western costs

Start with the number that produces the shock. A Chinese reactor costs roughly $2,500 to $3,000 per kW, about one-third of recent US and French capital costs.

A single 1.4 GW reactor costs an estimated $3 to $4 billion in China, against approximately $20 billion in the United States for equivalent capacity.

The instinctive explanation is cheap labour. That is not the primary driver. The cost advantage rests on institutional continuity and the economics of serial production, and it breaks down into three structural forces:

• State financing and cost of capital: around 70% of Chinese reactor costs are covered by state-backed loans, sometimes at interest rates as low as 1.4%
• Indigenisation: substituting expensive imported equipment and labour with domestic alternatives, cited as the primary driver of declining costs
• Continuous build and serial production: sustained institutional stability that prevents engineering capability from decaying between projects

That 1.4% interest rate is the figure that matters most to anyone comparing Chinese and Western nuclear economics. At that cost of capital, time on site is cheap. At Western commercial rates, every month of delay compounds into hundreds of millions of dollars of additional financing cost.

Speed then compounds the advantage. China typically builds reactors in five to seven years, about half the time of modern Western projects. Because financing costs accumulate over the construction period, a shorter build is not just faster; it is structurally cheaper.

Indigenisation is what makes the pricing durable rather than a one-off. By replacing foreign suppliers with domestic capacity, China reportedly halved construction costs by the early 2000s and has held them broadly stable since. The lesson for anyone assessing a Western or emerging-market programme is uncomfortable: without concessional state financing and an unbroken build pipeline, matching these economics is close to impossible.

What the fleet model means for uranium demand and the 2050 target

The engineering story ends at a fuel-market question. If reactors get built at the pace the fleet model implies, someone has to supply the uranium.

The IAEA’s high-case scenario projects global nuclear capacity rising from 377 GW at the end of 2024 to 992 GW by 2050, roughly tripling output. The low case reaches 561 GW. Those figures are the demand engine that uranium investors are pricing in, and the upper scenarios quietly assume the fleet model scales beyond China to other jurisdictions.

The fuel numbers follow. The WNA’s World Nuclear Fuel Report 2025 puts global reactor uranium requirements at roughly 68,920 tU in 2025, projected to reach between over 107,000 tU in the lower scenario and over 204,000 tU in the upper scenario by 2040.

Here is the central tension. The construction model determines the demand trajectory, but mining capacity determines whether fuel can actually be delivered. For indicative context, supporting 1,200 GWe of global capacity would require roughly 250,000 tU per year, against current annual primary production estimated near 60,000 tU in 2025 (both figures unverified, used for context only).

The gap between what is mined today and the upper-scenario demand is not a distant abstraction. The speed at which China and others can certify the fleet model in new jurisdictions is the variable that decides how quickly that gap widens, and whether mining investment decisions made now prove early or late.

SNPTC has stated its intent to make the fleet model transferable internationally, which is significant. It is the difference between the upper demand scenarios being achievable and being theoretical. Three variables are worth tracking:

The uranium supply gap widens fastest in scenarios where fleet-model adoption spreads beyond China, because each new jurisdiction that replicates the 56-month construction timeline compresses the lead time between final investment decision and first fuel load, leaving mining capacity expansion with less runway to respond.

1. The pace of fleet-model adoption outside China
2. Progress on the four bottlenecks: workforce, supply chain, design standardisation, and project management
3. Uranium mining capacity expansion relative to the demand trajectory

Past performance does not guarantee future results. Financial projections are subject to market conditions and various risk factors.

What the SNPTC blueprint changes for countries building nuclear from scratch

The fleet model is proven in China. It is also contingent on conditions most other countries do not yet have. The useful takeaway is a diagnostic lens rather than a verdict.

Some elements of the model travel well. Others do not:

International reactor deployment initiatives launched in 2025 and 2026 are the institutional expression of this ambition: governments and multilateral bodies have begun formalising the financing and regulatory coordination frameworks that fleet-model adoption outside China would require, even if workforce and supply chain gaps remain unsolved.

• Transferable: design standardisation frameworks, EPC-led commissioning structures, and Design Change Board governance
• China-specific: state financing at 1.4%, a single-jurisdiction regulatory environment, and more than 30 years of continuous build

SNPTC has explicitly said it intends to collaborate internationally on technical standards, supply chain resilience, advanced construction methods, and talent development. That signals the model is designed to be exportable rather than kept proprietary, which is meaningful for whether the upper demand scenarios hold.

The core insight sits underneath all of it. Tripling nuclear output by 2050 is a construction management challenge as much as a capital or policy one, and the fleet model is the only demonstrated framework for meeting it at scale. Its 56-month delivery target is the benchmark against which other national programmes can reasonably be measured.

For you, the practical value is the four-challenge framework SNPTC itself uses. Whether you are tracking a uranium miner, an energy infrastructure company, or a national programme, you can now assess delivery credibility against four dimensions rather than trusting announced capacity targets alone. The framework reveals exactly which capability gaps a country needs to close before a buildout can reach Chinese-scale efficiency.

This article is for informational purposes only and should not be considered financial advice. Investors should conduct their own research and consult with financial professionals before making investment decisions.

The construction model is proven; the transfer is the test

SNPTC has demonstrated something the rest of the world has not attempted: that fleet-based nuclear construction is operationally viable at scale. The 38-unit programme is the proof point, not a projection.

The four bottlenecks, workforce, supply chain, design standardisation, and project management, are not unique to China. They are simply hardest to solve outside the institutional framework that concessional financing, single-jurisdiction regulation, and three decades of continuous build have built.

So watch the right signals. Not reactor announcements, which are political targets, but certified workforce growth, design freeze decisions, and long-lead equipment contract timelines. Those three variables are what separate a credible nuclear programme from an ambition, and they are now yours to track against the same framework China uses to run its own.