Nuclear EPC Market Analysis
Who builds nuclear power plants, how much have they built, and what does the COP28 tripling target mean for the market? An analysis of the market gap to fill.
The world has never built nuclear this fast
The COP28 Declaration to Triple Nuclear Energy calls for 1,200 GWe of global nuclear capacity by 2050 — up from ~400 GWe today. That means roughly ~800 GWe of new build in 25 years.
For context: the fastest nuclear build era in history (1972–1989) averaged 16.4 GWe/year. Meeting the tripling target requires ~32 GWe/year — roughly double the peak historical rate, sustained for 25 years.
This report examines the existing nuclear EPC landscape and asks: if every established nuclear EPC replicates its entire historical construction record in the next 25 years, how much of the target can they cover? The answer — roughly 57% — reveals a 345 GWe open market gap that opens up for new entrants to fill.
Who builds nuclear power plants
Historical capacity built by EPC (GWe)
Approximate, deduplicated. Source: IAEA PRIS, WNA, company reports.
| EPC Contractor | Country | Role | Units | GWe | Key Reactor Types |
|---|---|---|---|---|---|
| Bechtel | USA | EPC contractor | 150 | 76 | Various (Westinghouse, GE) |
| EDF / Framatome | France | AE + NSSS | 66 | 72 | PWR 900/1300/N4, EPR |
| GE Vernova Hitachi | USA / Japan | NSSS + partial EPC | 83 | 72 | BWR-1 to 6, ABWR |
| Rosatom | Russia | Full EPC + NSSS | 75 | 65 | VVER-440/1000/1200 |
| CGN | China | Full EPC + NSSS | 27 | 30 | CPR-1000, Hualong One |
| CNNC | China | Full EPC + NSSS | 30 | 28 | CNP-300/600, Hualong One, CAP1000 |
| AtkinsRéalis | Canada | NSSS + EPC | 41 | 24 | CANDU-6, CANDU MONARK |
| MHI | Japan | NSSS + EPC | 24 | 22 | PWR (Mitsubishi design) |
| KEPCO / KHNP | South Korea | Full EPC + NSSS | 32 | 32 | OPR-1000, APR-1400 |
| Westinghouse | USA | NSSS vendor | 12 | 14 | AP1000 |
| Toshiba / IHI | Japan | NSSS + EPC | 10 | 10 | BWR, ABWR |
| Aecon-Kiewit | Canada | Lead EPC contractor | 5 | 5 | CANDU refurb, BWRX-300 |
| Others | Various | Domestic / support | 31 | 16 | IPHWR, Ansaldo, Doosan |
| Deduplicated global total | ~647 | ~400 | Since 1954. IAEA PRIS. | ||
From ~400 GWe to 1,200 GWe by 2050
439 reactors across 31 countries. Built over 70 years since the first grid-connected reactor in 1954.
Endorsed by 30+ countries at COP28 in December 2023. WNA's World Nuclear Outlook targets even higher: 1,446 GWe.
Twice the total capacity ever built — in one third of the time. The peak historical rate (1972–1989) was 16.4 GWe/year; ~800 GWe in 25 years requires ~32 GWe/year — double the fastest pace ever achieved.
The scale challenge: Between 1972 and 1989 — the fastest nuclear build era in history — the world averaged 16.4 GWe/year. Meeting the COP28 target requires 32.9 GWe/year from 2025. That's double the peak historical rate, sustained for 25 years.
What if every EPC replicates its entire history?
A generous thought experiment: if every nuclear EPC builds as much capacity in the next 25 years as they have built across their entire history — how much of the ~800 GWe target is covered?
Replication scenario vs. ~800 GWe target
| EPC | Historic GWe | Share of target | Assessment | Commentary |
|---|---|---|---|---|
| Bechtel | 76 | 9.5% | Challenging | Largest Western EPC. Vogtle experience. Poland build. Constrained by NSSS vendor availability. |
| EDF / Framatome | 72 | 9.0% | Challenging | Narrowed geographic focus to France, Nordics, Benelux, and UK. Withdrew from bids in Poland, India, and Canada. Rebuilding domestic capability for EPR2 programme (6–14 units). Sizewell C ongoing. |
| GE Vernova Hitachi | 72 | 9.0% | Ambitious | $40B Japan-US deal (Mar 2026) for BWRX-300 builds in Tennessee and Alabama. Darlington SMR under construction in Canada. Multiple international deployments planned. |
| Rosatom | 65 | 8.1% | Uncertain | Active pipeline (~20 units abroad), but sanctions and geopolitics cloud outlook. |
| CNNC + CGN | 50 | 6.2% | Likely exceeds | Building 6–10 units/year domestically incl. Hualong One and CAP1000 (localised AP1000). 8+ CAP1000 units under construction or approved. Only player at pace to surpass historical total. |
| KEPCO / KHNP | 32 | 4.0% | Plausible | Strong recent execution (Barakah). Won Czech tender. IP settlement likely restricts some markets. |
| MHI / Toshiba-IHI | 32 | 4.0% | Ambitious | Major Japan-US nuclear deal ($80–100B total). MHI, Toshiba, and IHI named as supply chain partners. MHI developing next-gen reactor. Japanese restart accelerating. |
| AtkinsRéalis | 24 | 3.0% | Plausible | Refocusing on CANDU. Won $2.85B Cernavoda contract (Romania, 2 new units). Developing CANDU MONARK (Gen III+). CAD $304M government loan. Turkey and Poland exploration. |
| India (NPCIL) + others | 15 | 1.9% | Plausible | India targets 22 GW by 2031. Active PHWR + imported designs. Argentina: 3 reactors. |
| Westinghouse | 14 | 1.7% | Plausible | Massive global demand. 6 operating, 14 under construction (incl. China CAP1000), 5 under contract. Poland (3 units), Bulgaria (2), Ukraine (up to 9), 10 more planned in US. $80B Japan-US partnership. |
| Aecon-Kiewit | 5 | 0.6% | Plausible | Lead EPC for Darlington BWRX-300 (North America's first grid-scale SMR, $1.3B CAD). Construction licence issued Apr 2025. 3 more units planned. Also assisting Estonia deployment. |
| Total (all EPCs replicate) | 458 | 57% | ||
| Remaining gap | 345 | 43% | Market space for new entrants | |
Six implications for the market
Even if every existing nuclear EPC replicates its entire historical construction record in just 25 years, it would only cover roughly half the ~800 GWe target.
The peak historical build rate (1972–1989) was 16.4 GWe/year. The tripling target requires ~32 GWe/year — double the fastest pace ever achieved, sustained for 25 years.
Massive scaling of existing EPCs and entry of new players is required. O&G EPCs, modular construction firms, and new national programmes will be essential.
CNNC + CGN is the only player currently building at a rate that could plausibly exceed its historical total within 25 years — 6–10 units/year, ~8 GW/year — including the CAP1000 (localised AP1000).
Western EPCs (Bechtel, EDF/Framatome) face capability gaps from decades of dormancy. Workforce rebuilding is the critical bottleneck — not reactor technology.
Read our competence transfer report arrow_forwardCountries without domestic nuclear EPC capability risk being deprioritised. EDF's 2025 withdrawal from non-European markets to focus on France proves the point: when domestic demand kicks off, international customers lose priority. Building domestic EPC capacity is a more robust path to nuclear deployment.
This gap is exactly the market opportunity that the NNC concept is designed to address by leveraging existing O&G EPC and yard expertise towards the nuclear industry.
Read about the NNC Concept arrow_forwardSources
International Atomic Energy Agency. Power Reactor Information System (PRIS). IAEA.
World Nuclear Association. World Nuclear Outlook 2025 and reactor database.
Declaration to Triple Nuclear Energy. COP28, December 2023.
Company reports and investor presentations: Bechtel, EDF, GE Vernova Hitachi, Rosatom, CGN, CNNC, AtkinsRéalis, MHI, KEPCO/KHNP, Westinghouse, Toshiba/IHI, Aecon-Kiewit.
Compiled March 2026. Data reflects approximate, deduplicated capacity figures. Some overlap exists between NSSS vendors and EPC contractors (e.g. Westinghouse reactor technology underpins ~185 GW of global capacity, but direct EPC on only ~14 GWe).