中国核电 China National Nuclear Power (601985.SS): The State's Bet on Atomic Cash Flow
I. Cold Open & Roadmap
On the last day of July 2026, China's State Council did something it had not done in a single sitting for years: it approved four nuclear power projects at once — eight reactors, more than ¥170 billion of committed capital, all in one meeting.1 Two of those projects went to a single listed company: units 1 and 2 at Zhuanghe in Liaoning, and units 3 and 4 at Jinqimen on the Zhejiang coast.1 There was no roadshow, no competitive bid, no press conference with a CEO in a hard hat. A committee met, a document was issued, and roughly ¥85 billion of thirty-year infrastructure landed on one company's balance sheet.
That company is 中国核能电力股份有限公司 China National Nuclear Power Co., Ltd. — ticker 601985 on the Shanghai Stock Exchange, known to everyone in the market simply as 中国核电. It is the listed power-generation arm of 中国核工业集团 China National Nuclear Corporation (CNNC), the state conglomerate that built China's first atomic bomb. And it is, by a wide margin, one of the two most important nuclear generators on earth by capacity under construction.
Here is the scale. At the end of 2025 the company controlled 26 operating reactors totalling 25.00 GW, and another 19 units under construction or approved-but-not-yet-started totalling 21.86 GW.2 Against a national nuclear fleet of 62.48 GW, that is roughly forty percent of China's installed nuclear capacity sitting inside one listed vehicle.2 Its reactors generated 200.8 billion kilowatt-hours in 2025 — about forty percent of the 485.2 billion kWh China's entire nuclear fleet produced.2 For global context: the world had 436 grid-connected reactors and 398 GW of nuclear capacity at end-2025, so this single Shanghai-listed company owns something on the order of one-sixteenth of all operating nuclear capacity on the planet, and is building a pipeline nearly as large as its existing fleet.2
Now the numbers that make the story interesting rather than merely large. In FY2025 the company reported revenue of ¥82.08 billion, up 6.22% year-on-year. Consolidated net profit was ¥17.97 billion, up 8.56%. But net profit attributable to shareholders of the listed company was ¥9.30 billion, up 6.00%.2 Read those three lines again. More than forty-eight percent of the profit this business generates does not belong to the people who own the stock. Weighted average return on equity was 8.21%, down 1.23 percentage points from the prior year. Earnings per share actually fell 2.16%, to ¥0.453.2
And the quarterly path was worse than the annual headline suggests. Attributable profit ran ¥3.14 billion in Q1 2025, ¥2.53 billion in Q2, ¥2.34 billion in Q3, and ¥1.30 billion in Q4.2 The year did not grow into strength; it decayed into weakness. Then in Q1 2026, revenue fell 6.65% and attributable profit fell 34.19%.3 Management attributed the drop to a heavier refuelling-outage schedule, lower on-grid volumes, and falling market-based electricity prices in both nuclear and renewables.3
That last item is the spine of this entire story. In 2025, the share of the company's nuclear output sold at market-negotiated prices rather than a government-set tariff jumped to 71.36%, up 20.58 percentage points in a single year.4 The realised comprehensive nuclear tariff fell 5.16% to ¥0.3937 per kilowatt-hour including VAT.4 For reference, the national benchmark tariff the NDRC set for new nuclear units back in June 2013 was ¥0.43/kWh.5 Thirteen years and an enormous capital cycle later, this company is selling power for roughly eight percent less than the benchmark price of 2013 — in nominal terms, before inflation.
So the central tension of this episode is not "will they build the reactors." They will; the state has decided. The tension is this: a company can hold an essentially unassailable regulatory moat and still watch its margin per unit of output get repriced downward — because the same state that grants the licence also writes the pricing rules. Nuclear is a business whose cost is fixed for sixty years at the moment concrete is poured, and whose revenue is now being handed over to a spot market the government is deliberately building.
There is also a second engine running alongside the reactors, and it is running fast — wind and solar, 33.64 GW of it at end-2025, larger in nameplate megawatts than the nuclear fleet itself.2 Whether that engine is a diversification triumph or an expensive distraction is one of the sharper questions in this story, and the 2025 numbers give a surprisingly blunt answer.
The roadmap from here: the origins of a bomb-making ministry that became a ratepayer-facing utility; the boom-bust-boom approval cycle that governs everything; the economics of the core reactor fleet and the tariff transition now reshaping them; the renewables arm and what its profit line actually shows; capital allocation, the parent relationship, and who really funds the buildout; management under state stewardship; the competitive structure; the risk radar; and finally the bull and bear case with the handful of metrics that will settle it.
Subscribe on LinkedInII. From Bomb-Makers to Ratepayers: Origins
The institutional ancestor of this company was not a utility. It was called the Second Ministry of Machine-Building — a deliberately dull name for the organisation that ran China's nuclear weapons programme from the mid-1950s. Its mandate was not returns on capital. It was a device, a delivery system, and a national deterrent, pursued under conditions of extreme secrecy and material scarcity.
That heritage matters for a specific and non-sentimental reason: it explains the operating culture that a public-market investor is buying into seventy years later. Weapons programmes select for engineers who treat schedule and safety as absolute and cost as a variable to be reported afterward. They select for vertical integration, because you cannot outsource a fuel cycle to a supplier who might be cut off. They select for institutional patience, because the payoff horizon is measured in decades. Every one of those traits survives in the listed company today — in its 90%-plus equipment localisation rate, in its ten-year framework contracts for uranium, and in a capital-allocation process where the approving authority is a State Council meeting rather than an investment committee.
The civilianisation began in earnest in the 1980s, and its physical expression was Qinshan. The 秦山核电站 Qinshan Nuclear Power Plant on the Zhejiang coast connected to the grid in 1991 — mainland China's first indigenously designed and built commercial reactor. By modern standards it was tiny: a 300-megawatt-class unit, roughly a quarter the size of a single Hualong One reactor today. But Qinshan was never really about the megawatts. It was proof that China could design, build, license, and operate a civil power reactor without a foreign licensor holding the blueprints — and it seeded the engineering cadre, the regulatory apparatus, and the supply chain that the entire subsequent buildout depended on.
Qinshan also supplies one of the quieter economics lessons in this business. On the Q3 2025 earnings call, management noted that Qinshan Unit 1 received regulatory approval in 2021 from the National Nuclear Safety Administration to operate for an additional twenty years beyond its original licence.6 Second-generation units carry forty-year operating licences and third-generation units sixty years, and after safety upgrades and evaluation, extension is available.6 A twenty-year extension on a plant whose capital cost was amortised decades ago is close to free cash flow — and it is the single most underappreciated form of value creation in nuclear. It is also, notably, not something management markets aggressively.
The 2015 listing: capital markets as a state financing channel
The listed company itself is young. China National Nuclear Power Co. went public on the Shanghai exchange in 2015, raising roughly ¥16.25 billion — at the time one of the A-share market's largest listings in five years.7 CNNC, which had held roughly 97% of the equity, retained control at approximately 70.33% post-listing. China Three Gorges, COSCO, and China Aerospace Investment each took stakes of about 1%.7
What is worth pausing on is where the money went. IPO proceeds were earmarked for named reactor units already under construction — Fuqing, Sanmen phase 1, Changjiang in Hainan, and Tianwan units 3 and 4.7 This was not a founder monetising a business, nor a growth company funding an expansion it might otherwise not afford. It was the state opening a new financing channel for a buildout that had already been decided upon. The equity market was being asked to co-fund infrastructure whose returns would be set by policy.
That framing has an uncomfortable corollary that runs through the rest of this story. If minority shareholders are a financing channel rather than owners exercising control, then the terms on which capital is raised — and the terms on which assets are subsequently injected, and stakes in subsidiaries subsequently sold — become the central governance question. It is not an abstract concern. As of the end of 2025, CNNC's direct stake had come down to 55.67%, or 11.45 billion shares, having increased by 240.7 million shares during the year.2 The 全国社会保障基金理事会 National Council for Social Security Fund appeared as the second-largest holder with 1.44 billion shares, or 7.02%, essentially the entire position built during 2025.2 国新产业资本管理 Guoxin Industrial Capital Management — the investment arm of state holding company China Reform Holdings — took its stake to 761.6 million shares, or 3.70%, after adding 693.6 million shares in the year.2
Asked about Guoxin on the Q3 2025 call, the company confirmed that Guoxin had won a board seat by shareholder vote and was "actively participating in corporate governance," describing it as a long-term strategic shareholder.6 That is a genuinely interesting development. It is also worth being clear-eyed about what it is not: another arm of the state taking a board seat is not the arrival of independent shareholder oversight. Meanwhile, Hong Kong Central Clearing — the conduit for Stock Connect foreign money — reduced its holding by 292.0 million shares over the year, to 1.31%.2 Domestic state capital bought; the offshore marginal buyer sold.
Which raises the obvious question: what changed in the intervening decade that made a decided-upon buildout stop, and then restart at twice the speed?
III. The Industry's Boom-Bust-Boom Cycle & Why It Matters Now
On March 11, 2011, a magnitude 9.0 earthquake struck off the coast of Japan, and within days three reactor cores at Fukushima Daiichi had melted down. China's response was measured in days, not months. The State Council froze all new reactor approvals and licensing pending a nationwide safety review of the existing fleet and every site under construction.8
The formal freeze lasted roughly eighteen months, ending in October 2012 when the State Council approved a comprehensive safety plan and allowed projects to resume.8 But the number that matters is not when the freeze ended — it is what happened afterward. Through the 2010s, China averaged only a handful of new grid connections a year, and the approvals that did come through were restricted to Generation III-plus coastal designs. The chill outlasted the freeze by most of a decade.
For an investor, this is the single most important piece of pre-history in the story, because it establishes the mechanism by which this company's growth rate is set. It is not set by demand. It is not set by capital availability. It is set by a political risk appetite that can be switched off by an event on the other side of the East China Sea and stay off for years. No amount of operational excellence at Qinshan or Fuqing would have shortened that pause by a single month.
The freeze also left a structural legacy visible in the map of this company's assets to this day. Approvals since have gone to coastal sites running Generation III-plus designs, and the pattern held through the July 2026 batch, where all four approved projects — in Liaoning, Zhejiang, Guangdong and Shandong — sat on the coast.1 The company's own fleet reflects it exactly: every plant sells into a coastal provincial grid, in Zhejiang, Jiangsu, Fujian or Hainan.4 It concentrates the fleet in exactly the provinces that are also building the most solar and running the most advanced spot markets, which is one reason the tariff pressure discussed later in this story lands harder than a national average would suggest. The safety decision of 2011 and the price problem of 2025 are, in a roundabout way, the same decision.
2019 and the regime change
The inflection came around 2019, when approvals resumed at scale. Since then China has approved dozens of reactor units representing well over ¥1 trillion of direct investment — a genuine regime change from post-Fukushima caution to an energy-security-driven buildout.9 The driver was not a nuclear-safety epiphany. It was the arithmetic of an economy importing the overwhelming majority of its oil and a large share of its gas through sea lanes it does not control, in an era of deteriorating relations with the country that patrols them. Nuclear fuel is compact, storable, and — through CNNC's own fuel-cycle assets — domestically manufacturable. In an energy-security framework, that is worth paying for even if the plant-level IRR is unremarkable.
The 十五五 15th Five-Year Plan, approved in March 2026, set a target of approximately 110 GW of operating nuclear capacity by 2030.10 Against 62.48 GW installed at end-2025, that implies adding roughly 48 GW in five years — nearly doubling the fleet.2 World Nuclear Association data suggests China's operating capacity is expected to surpass the United States around 2030 to become the largest in the world.1 Longer term, the direction of travel points toward roughly 169 GW and about ten percent of generation by 2035, though the formal 2035 policy anchor submitted in November 2025 was framed as non-fossil sources exceeding thirty percent of total energy rather than a hard nuclear number.10
The July 31, 2026 approval batch — four projects, eight units, over ¥170 billion — is the visible expression of this rhythm.1 Three listed operators shared it: this company took four Hualong One units, 中国广核电力 CGN Power took two units of an upgraded Hualong One 2.0 at Huizhou, and 国家电力投资集团 State Power Investment Corporation took two 国和一号 CAP1400 units at Laiyang.1 Industry commentary put the total investment multiplier from these projects across the supply chain at roughly ¥5 trillion.1
What the cycle actually means for the equity
Here is the honest reading. A reactor supercycle guarantees volume. It does not guarantee returns. Approvals give the company a decade of visible capacity growth, which is genuinely rare and genuinely valuable — but every one of those units is a fixed-cost commitment made today against a revenue stream that will be priced by market rules being written concurrently, and that have already moved against the industry once.
There is also a subtler point about approval-driven growth that the market tends to under-weight: it is not a competitive win. On the Q3 2025 call, an investor asked directly whether this company's share of new approvals would decline now that 华能集团 China Huaneng Group and SPIC had obtained nuclear operating licences. Management's answer was pure cooperative boilerplate — that Huaneng's entry "helps improve nuclear management and safety standards," that all parties should achieve complementary advantages through cooperation.6 What management did not say is that it competes for these units, because it does not. Allocation is administrative. Consistent with that, the World Nuclear Association notes that other utilities including Huaneng, Datang, Huadian and Guodian may hold minority stakes of up to 25% in nuclear projects.10 Indeed, the Zhuanghe project approved in July 2026 is a joint venture with Datang.1
That structure is a shield against competition and a constraint on control at the same time. It also has a direct financial consequence, which we will come to when we look at where the money for these reactors actually comes from. First, the core engine itself.
IV. The Core Engine: Nuclear Power — Segment Economics
Walk through a nuclear plant's income statement and it looks less like a power company and more like a toll road with a physics problem attached. Almost everything is decided before the first kilowatt-hour is sold.
The capital is enormous and sunk: at end-2025, gross fixed assets on the consolidated balance sheet stood at ¥532.32 billion, with accumulated depreciation of ¥179.76 billion and a net book value of ¥351.98 billion — 46.98% of total assets.4 Depreciation on that base is a fixed charge that does not care about the electricity price. Fuel is the one genuinely variable input, and management has been consistent that it has run at roughly twenty percent of operating cost since the IPO, procured through ten-year framework agreements under which the company buys natural uranium and contracts out fuel assembly fabrication.3 Asked on the FY2025 call about forecasts of uranium reaching $200 per pound, management's answer was structural rather than promotional: the framework-contract approach means spot-price volatility has limited effect on fuel cost, and controlling shareholder CNNC holds China's only domestic nuclear fuel manufacturing capability.3
There is one more fixed levy worth knowing because it rarely appears in Western nuclear comparisons: since October 1, 2010, pressurised-water reactors in commercial operation for more than five years pay a spent-fuel disposal fund of ¥0.026 per kilowatt-hour, charged to current production cost.6 That is about 6.6% of the FY2025 realised nuclear tariff, permanently.
So: fixed capital, fixed depreciation, semi-fixed fuel, a per-kWh waste levy, and a plant that runs flat out because the marginal cost of the next kilowatt-hour is close to zero. The company's fleet averaged over 8,000 utilisation hours in 2025 for the first time in its history, against a national nuclear average of 7,809 hours and a national all-sources average of 3,119 hours.32 It also reported 22 units scoring a full WANO composite index.3 Operationally, this fleet runs well. That is not in dispute.
Which means that in a business where volume is maximised and cost is fixed, essentially all of the earnings variance lives in one line: price.
The tariff transition, and why it is the whole story
For most of the last decade, nuclear power in China was sold largely at a government-set tariff, anchored to the ¥0.43/kWh national benchmark established in 2013 with regional adjustments where local coal-fired benchmarks were lower.5 It was, functionally, a regulated return.
That has now substantially ended. In 2025, market-traded volume from the company's nuclear fleet reached 134.05 billion kWh, up 54.13% year-on-year, against total nuclear on-grid volume of 187.85 billion kWh.4 The market-traded share therefore rose to 71.36% from 50.78% — a 20.58 percentage-point move in twelve months.4 The realised comprehensive nuclear tariff fell to ¥0.3937/kWh from ¥0.4151, down 5.16%.4 The blended tariff across all generation, including renewables, was ¥0.3899/kWh, down 6.27%.4
On the FY2025 call, an investor asked the question directly: is the comprehensive tariff going to keep declining as marketisation advances? Management's answer was, to its credit, mechanical rather than evasive. From 2022 to 2024, they said, power market prices held high and nuclear market prices stayed firm. From 2025, spot-market construction accelerated, provinces moved into formal or continuous settlement trial operation, and — driven by falling primary energy prices and rapid renewable capacity expansion — market prices weakened, with nuclear market tariffs coming under pressure alongside.3
That is the correct causal chain, and it deserves to be spelled out plainly because it is counterintuitive. Nuclear is not losing share to wind and solar at the capacity level; China is building all three simultaneously. What is happening is that in any given hour, the marginal price in a spot market is set by the cheapest available generator. When a province has installed enormous quantities of solar — nationally, solar capacity reached 1,201.73 GW at end-2025, up 35.4% in one year, and solar generation grew 39.8% — there are a great many hours in which the clearing price is very low.2 A baseload plant that must run through those hours takes that price. Nuclear's competitive position is unchanged; its revenue per kilowatt-hour is being set by somebody else's cost curve.
Asked about the 2026 outlook, management fell back on a formulation it used repeatedly across both calls: the national energy authority is studying a nuclear tariff policy, and the specifics will follow official publication.3 In the meantime, the company said it would optimise its trading structure, improve personnel capability, and safeguard unit output.3 Those are real levers, but they are second-order ones. There is no mitigation plan here that offsets a 5% annual price decline; there is a hope that policy will provide a capacity payment or a mechanism price. Investors asked about exactly that — when nuclear, as a zero-carbon baseload source, might receive a capacity subsidy — and got the same deferral to future official policy.3
Myth vs. reality: three things the consensus gets wrong
Myth: this company is a mid-sized nuclear operator with roughly a sixth of the domestic market. Reality: it controls about forty percent of China's installed nuclear capacity and produced about forty percent of its nuclear electricity in 2025.2 The confusion usually comes from conflating the listed company with the group, or from stale figures. The scale here is genuinely first-tier globally.
Myth: nuclear in China is a protected, regulated-return business. Reality: as of 2025, roughly seven-tenths of this company's nuclear output cleared at market prices, and the realised tariff sits below the 2013 nominal benchmark.45 The protection is on the volume side, not the price side.
Myth: nuclear counts as green power in China, so it will capture the clean-energy premium. Reality: it does not. Asked on the Q3 2025 call about green certificates and carbon trading eligibility, the chairman's office gave an unusually candid answer — nuclear has to date not been included in China's green certificate and green power system, making it "the only non-fossil energy source excluded from that system."6 Industry bodies and operators are pushing for inclusion, and the state has begun exploring how to recognise nuclear's low-carbon attributes, but there is no timetable.6 This is a live, quantifiable option that is currently valued at zero, and it cuts both ways: it is upside if granted, and it is evidence of how little pricing leverage the industry has that it has not been granted already.
Technology and the shape of the moat
The workhorse design is 华龙一号 Hualong One (HPR1000), the third-generation pressurised-water reactor that CNNC and CGN unified in 2013 after years of parallel development. Management stated on the Q3 2025 call that Hualong One is a Chinese-owned-IP design with an equipment localisation rate above 90%, and that there is no technology gap creating import dependence.6 For a country worried about supply-chain chokepoints, that number is the point of the entire programme.
The cautionary counterexample sits inside the company's own portfolio. Sanmen was the world's first AP1000 deployment, licensed from Westinghouse. First concrete was poured in April 2009; Unit 1 connected to the grid in June 2018 and entered commercial operation in September 2018.10 Nine years from first concrete to commercial operation on a design marketed as simpler and faster to build. That is the real-world price of first-of-a-kind technology transfer, and it is the strongest available argument for why standardising on a domestic design that has now been built repeatedly is worth more than any individual reactor's thermal efficiency.
The "second curve" management wants you to believe in
There is one more strand of the nuclear story that management pushes harder than the numbers currently justify, and it deserves examining precisely because of the gap. Asked on the FY2025 call about returns from multi-purpose nuclear applications, management described it as a deliberate new frontier: beyond steam and heat supply, nuclear energy has roles in seawater desalination, hydrogen production, nuclear-powered shipping, isotope production, and integrated smart-energy services for industrial parks. Nuclear, they said, is evolving from pure electricity generation toward comprehensive energy supply, and the company is cultivating multi-purpose utilisation as its "second curve" of profit growth — with results, in their words, already beginning to show.3
The disclosed reality is early-stage. Jiangsu Nuclear's 和气一号 steam-supply project ran for more than 11,000 hours cumulatively without incident, and Hainan Nuclear's steam project entered operation during 2025.4 The scale is quantifiable: Jiangsu Nuclear supplied 3.41 million tonnes of steam in 2025, which the company converts to a generation-equivalent of 1.024 billion kilowatt-hours — roughly half a percent of nuclear output.4 On isotopes, Qinshan placed commercially reactor-produced carbon-14 and lutetium-177 on the market and the company launched an isotope brand, 和福一号.43 Technical services are marketed under a brand called 八方核护, with an explicit push to take those services international.4
The concept is legitimate — industrial steam sold under bilateral contract escapes the spot market entirely, which is exactly the right instinct given everything above. But an investor should hold management to the arithmetic. A "second curve" that is currently one two-hundredth of core output, with research spending essentially flat, is a direction of travel, not a profit engine. If it becomes one, it will show up as steam and services revenue growing several-fold from a tiny base over multiple years — a slow, checkable claim.
In Hamilton Helmer's 7 Powers vocabulary, the durable advantage here is a cornered resource — but the resource is the operating licence and the approved site, not the technology. Anyone in China who wants to build a Hualong One can buy one; nobody can grant themselves a site or a licence. That distinction determines exactly what the moat protects: it protects the company absolutely from competitive entry, and not at all from the licensor's own pricing decisions.
The peer check
The most useful test of whether this is an industry problem or a company problem is CGN Power, the listed vehicle of 中国广核集团 China General Nuclear — a company of comparable size operating a larger reactor fleet, and dual-listed on HKEX (01816.HK) and Shenzhen (003816.SZ).11 In FY2025, CGN reported revenue of ¥75.70 billion, down 4.11%, and attributable net profit of ¥9.77 billion, down 9.9%.12 Its average market electricity price fell about 8.8% year-on-year while its market-traded share rose about 5.2 percentage points.12
Set that against this company's +6.22% revenue and +6.00% attributable profit and the conclusion is clear on both counts.2 First, the margin squeeze is structural and industry-wide, not a management failure — the larger, more mature fleet with less capacity growth went backwards. Second, the growth this company reported was bought with new capacity and volume, not with price. Strip out the added reactors and renewables megawatts and the underlying per-unit economics moved in the same direction as CGN's.
That is a legitimate and repeatable strategy for as long as the approval pipeline keeps delivering. It also means the second engine — the one adding megawatts fastest — deserves a hard look.
V. The Second Engine: New Energy — Growing Fast, Strategic Diversification
If you had only the capacity table from the FY2025 annual report and no other information, you would conclude this had quietly become a renewables company that happens to own reactors.
At December 31, 2025, controlled renewable capacity in operation stood at 33.64 GW — comprising 10.63 GW of wind and 23.01 GW of solar, plus a further 1.65 GW of standalone battery storage — against 25.00 GW of nuclear.2 Another 7.93 GW of wind and solar was under construction.2 Renewable capacity grew 13.67% in the year, with more than 4 GW added.3 Generation grew far faster than the fleet: renewable output rose 31.29% to 43.62 billion kWh, and on-grid volume rose 31.34% to 42.90 billion kWh.2 In the first nine months of 2025 the gap was even starker — renewable on-grid volume grew 34.82% against 11.44% for nuclear.6
The strategic logic is sound on paper and worth stating properly. Nuclear projects take the better part of a decade from approval to commercial operation and consume vast capital before producing a yuan of revenue. Wind and solar projects are built in months to a couple of years, are financeable in smaller increments, can be geographically dispersed — the company's renewable projects span more than thirty provinces and regions, all selling into local grid companies — and turn cash faster.4 Redeploying nuclear's steady operating cash flow into shorter-cycle assets is a defensible way to smooth the capital cycle and diversify away from a single tariff mechanism.
And then you look at the profit line
Here is where the story turns, and it turns hard.
For FY2025, management disclosed segment results on the earnings call: nuclear generated revenue of ¥67.11 billion, up 4.02%, with attributable net profit of ¥8.96 billion, up 18.53%. New energy generated revenue of ¥14.97 billion, up 17.30% — and attributable net profit of ¥346 million, down 71.60%.3
Read that against the capacity figures. A business with more nameplate megawatts than the entire nuclear fleet, growing generation at over thirty percent, contributed under four percent of the company's attributable profit — and its contribution collapsed by more than seventy percent in a single year.
Management's explanation was multi-causal and, notably, did not hide behind a single excuse: the renewables industry faces the twin pressures of falling tariffs and curtailment difficulties; newly commissioned units drove up depreciation; income tax preferences reduced; asset impairments increased; and the attributable ownership percentage declined.3 That last item is the one investors should sit with. On the Q3 call, the company had already flagged that two REIT-like issuances and a market-oriented debt-to-equity swap in the prior year, plus capital increases bringing in strategic investors, had diluted the listed company's share of new-energy profits.6
The tariff data confirms the pricing half of it. The realised new-energy comprehensive tariff fell 11.21% in 2025 to ¥0.3732/kWh from ¥0.4203 — more than double the rate of nuclear's decline.4 Renewables are further along the same marketisation road, and the destination is visible from here.
Margin structure tells the rest. In Q3 2025, the company disclosed a nuclear gross margin of 43.12% against a new-energy gross margin of 34.03%, with cost increases driven mainly by depreciation on newly commissioned units.6 The gross-margin gap is real but not catastrophic; the collapse happens below the gross line, through depreciation, tax, impairment, and minority dilution.
The impairment question
Renewables at this company were not all built. A material portion was bought, and the balance sheet records the price. On the Q3 2025 call, an investor asked pointedly whether the new-energy projects that 中核汇能 CNNC New Energy — the renewables subsidiary — had acquired at a premium were meeting original expectations, and how much goodwill would be written off at year-end.6 The answer at the time was that goodwill stood at approximately ¥5.7 billion, "the overwhelming majority" from new-energy acquisitions, tested annually.6
The FY2025 audit answered the question. Goodwill impairment was designated a Key Audit Matter. Gross goodwill at year-end was ¥6.08 billion against an impairment provision of ¥663 million, leaving a carrying value of ¥5.42 billion.4 Of the subsidiaries tested, 103 showed recoverable amounts above carrying value and no impairment; 11 — including Linze Suyuan New Energy and Tongyu Qiangfeng Wind Power — showed recoverable amounts below carrying value and were written down.4 Fixed-asset impairment was also flagged as a Key Audit Matter, with a ¥576 million provision against the ¥532.32 billion gross fixed-asset base.4 The auditor, 立信 BDO China Shu Lun Pan, issued a standard unqualified opinion.2
An eleven-out-of-114 failure rate on acquired renewable assets is not a crisis. It is, however, an early and independently audited signal that the M&A-led portion of the renewables build was done at prices that assumed a tariff environment that no longer exists. Investors should expect this line to be tested again each year-end, and management said as much.6
Where this leaves the second engine
Asked directly on the FY2025 call whether the new-energy segment would swing into a loss, management gave the most candid answer of either transcript: the renewables industry broadly is experiencing "revenue growth without profit growth, with both volume and price falling and profit space sharply narrowed," and 2026 policy volatility was already showing through in Q1 results.3 The stated response was benchmarking against peers, cost-reduction programmes across production and marketing, financial costs, asset revitalisation and daily operations, with the goal of keeping full-year new-energy operations "basically stable."3 Note what that goal is: stability, not growth. On the Q3 call, management had put it in capital-allocation terms — nuclear projects will proceed at the pace of national approvals, while new-energy projects "will be prudently adjusted in line with industry trends," with investment decisions made strictly against return requirements.6
The honest assessment is this. The renewables arm is a genuine second engine measured in megawatts and kilowatt-hours, and a marginal one measured in profit attributable to the people who own the stock. It bought volume growth at the cost of depreciation, leverage, minority dilution, and now impairment. Whether it becomes a second profit engine depends on the renewables tariff environment stabilising — and the 11.21% decline in realised price says it has not yet.
The disclosure caveat matters too: segment profitability is communicated through earnings-call commentary rather than a granular audited segment note, which means investors are relying on management's own attribution of costs between two businesses that share a balance sheet and a treasury function. That is a lower standard of evidence than the nuclear fleet's operating data, and it should be weighted accordingly.
Which brings us to the balance sheet itself — because both engines are being fed by the same capital machine, and the way that machine works is the least-understood part of this company.
VI. Capital Allocation, the Group Relationship & Deal Discipline
Start with the single most revealing pair of numbers in the FY2025 accounts.
Cash generated from operations was ¥37.41 billion, down 8.13% from the prior year.2 Cash paid to acquire fixed assets, intangibles and other long-term assets — capex — was ¥93.68 billion, up from ¥90.71 billion in 2024.4
Capex ran at two and a half times operating cash flow. It did not decline. Any framing of this company as a mature utility harvesting cash from a completed asset base is, on the evidence, premature by several years.
So where does the other ¥56 billion come from? Two places, and both have consequences for the equity.
Debt, and the one thing that has gone genuinely right
Long-term borrowings rose to ¥363.18 billion at end-2025 from ¥304.30 billion a year earlier.4 The debt-to-asset ratio was 69.01%, up 0.74 percentage points, against total assets of ¥749.26 billion, up 13.57%.2 Interest coverage was 2.74 times, improved from 2.64 times.2
Asked on the Q3 2025 call about rising leverage and debt-rollover pressure, management was specific about the structural constraint and the mechanics: SASAC imposes a loan-risk red line, the company operates below 70%, projects are funded on a standard 20% equity / 80% bank debt structure, and new project starts inevitably push the ratio up.6 The mitigations described were optimising loan structure and issuing perpetual bonds or other equity-like instruments to hold the ratio at a reasonable level.6 Financing cost was stated as running between 2% and 3%.6
That last figure is where management has delivered something concrete and verifiable. The bond schedule in the annual report tells the story cleanly: legacy paper from 2021 and 2023 carries coupons of 3.17%, 3.25%, 3.30%, 3.44%; the 2025 vintage prices at 1.75% for the first tranche of technology-innovation bonds, 1.79% for a green super-short-term note, 1.95% and 1.97% for Jiangsu Nuclear Power paper, 2.14% for Sanmen.4 On the FY2025 call management explained the driver — existing loans have essentially been replaced once over through refinancing, multi-channel funding and rate optimisation, producing a marked annual decline in finance costs, with the stock of legacy loans "basically replaced through one cycle."3
This is real value creation, achieved without a press release. On roughly ¥363 billion of long-term borrowings, a 100 basis point reduction in average cost is worth around ¥3.6 billion a year pre-tax — comparable to the entire attributable profit contribution of the renewables segment several times over. It is also, importantly, a diminishing well: management's own framing that the replacement cycle is largely complete means the incremental benefit from here is smaller than the benefit already booked.
The minority-interest machine
The second funding source is the one that deserves far more attention than it gets. In FY2025, the company received ¥19.52 billion from equity investments, of which ¥18.52 billion came from minority shareholders subscribing into subsidiaries.4 In 2024 the equivalent figure was ¥26.93 billion.4
Follow that through the balance sheet. Minority interests stood at ¥113.51 billion at end-2025, up from ¥99.15 billion — against equity attributable to the parent of ¥118.72 billion.42 The minority pool is now within five percent of the size of the listed shareholders' equity.
This is the mechanism that explains the gap between ¥17.97 billion of consolidated net profit and ¥9.30 billion attributable.2 The company funds reactors and renewable projects substantially by selling equity stakes in the project companies themselves — to other state entities, to insurers, to debt-to-equity-swap vehicles, to REIT-like structures. The listed company consolidates the assets, the revenue and the debt; it keeps a shrinking proportional claim on the profit.
An activist would frame this bluntly: minority-interest financing is dilution that never shows up in the share count. The listed entity's earnings per share can stagnate while its balance sheet, generation volume and press releases all grow impressively — which is precisely what happened in 2025, when revenue rose 6.22%, consolidated profit rose 8.56%, attributable profit rose 6.00%, and EPS fell 2.16%.2
To be fair to management, the alternative is worse. Funding ¥93.7 billion of annual capex through the listed equity would mean serial rights issues; funding it entirely with debt would breach the SASAC ratio. Project-level equity from state co-investors is a rational solution to a hard constraint. But the trade-off should be named, because it is the difference between a growth story and a growth story that accrues to somebody else.
Asset injections: the real "M&A"
There is no outbound acquisition story here. The recurring transaction pattern is related-party: as reactors complete construction under CNNC's development entities, they move into the listed company. This is the mechanism through which the pipeline becomes earnings, and every injection is effectively an acquisition from the controlling shareholder.
The governance question is whether injection pricing has favoured the parent or minority shareholders over time. It is a hard question to answer from public filings, because valuations rest on appraisals of assets with no market comparables — and because a parent holding 55.67% both sets the price and votes on it.2 What can be said is that the incentive structure is unambiguous and the check on it is disclosure quality and regulator scrutiny, not shareholder consent.
For evidence on whether group-level deal discipline exists at all, the nearest available data point sits one entity away. In December 2024, CNNC-affiliated listco 中核科技 CNNC Technology proposed acquiring 中核西仪 China National Nuclear Xi'an Instruments to consolidate the group's nuclear-equipment manufacturing.13 In July 2025, the deal was terminated.[^14] That is a genuine, if modest, signal: within this group, transactions do get killed when terms do not work. It is not proof that reactor injections into 601985 have been priced fairly, but it is better than the absence of any such precedent.
Returns to shareholders: better than the reputation
The distribution record is stronger than the sector's SOE reputation implies. Cumulative dividends since listing had exceeded ¥24.2 billion by Q3 2025, with the payout ratio at or above 35% every year.6 For FY2025 the company paid its first-ever interim dividend of ¥0.02 per share, worth over ¥400 million, and proposed a final dividend of ¥0.16 per share, calculated at approximately ¥3.28 billion against 20.57 billion shares outstanding excluding treasury.62 Total FY2025 distribution was guided at over ¥3.6 billion for a payout ratio above 39%.3
It also completed its first-ever buyback. By end-September 2025 the company had repurchased 21.42 million shares for over ¥194 million; the programme closed on April 27, 2026 having repurchased 54.01 million shares for approximately ¥483 million — 96.75% of the plan's ceiling.632 On the FY2025 call management said it would "explore normalising the buyback mechanism," conducting repurchases as market conditions and the cash position allow.3
Asked whether 2026 distribution policy would prioritise the ≥35% ratio or absolute stability of the amount, management chose the latter framing: despite significant investment pressure and capital requirements, it would try to keep the total dividend amount relatively stable, with the 2026 plan determined by full-year profit.3 That is a meaningful, and slightly worrying, tell. Defending an absolute dividend amount while attributable profit falls means the payout ratio rises mechanically — the opposite of the sell-side framing that "capex has peaked, so both ROE and dividends rise from here."
That thesis is testable and, so far, unsupported. Capex went up in 2025, not down. Operating cash flow went down. ROE went down 1.23 points. The three-to-four first-concrete pours a year that management guided to for 2026 and 2027 imply the capital cycle extends well beyond this reporting period.6 The falsification test is simple and each annual report settles it: watch whether the capex line in the cash flow statement actually turns.
The valuation question nobody can settle cleanly
Historically the A-shares have traded at a marked premium to CGN Power's H-shares — one 2018-vintage comparison put the gap at roughly 24x earnings and 2.6x book against roughly 8.6x and 1.3x — despite CGN generating more nuclear electricity.14 Two explanations compete. One is genuine differentiation: a larger relative pipeline and a renewables arm that CGN's listed vehicle does not match in scale. The other is structural: A-share liquidity, domestic index inclusion, and restricted foreign access to the H-share market.
The 2025 evidence tilts toward the second. The renewables arm delivered under four percent of attributable profit and its contribution fell by more than seventy percent.3 A growth premium underwritten by a segment earning ¥346 million rests on a thin foundation. Meanwhile the foreign conduit was a net seller through the year while domestic state funds bought.2 Investors paying a premium should be explicit about which of the two explanations they are underwriting.
All of which depends, ultimately, on who is making these decisions and what they are optimising for.
VII. Management: State Stewards, Not Owner-Operators
The FY2025 and Q1 2026 earnings briefing, held by telephone and webcast, drew 166 participants — Changjiang Securities, CITIC, CICC, China Merchants, Tianfeng, E Fund's peers across the mutual-fund complex, insurance asset managers, private funds.3 The Q3 2025 briefing had drawn 123 institutional investors on the call plus 314 attendees on the Shanghai Stock Exchange's roadshow platform, with 23 questions answered live.6 By SOE standards, this is a company that shows up and takes questions.
Who does the answering matters. Chairman 卢铁忠 Lu Tiezhong, listed as age 50 in the FY2025 annual report, has chaired the board since 2021 and is an engineer by training who has also held directorships across CNNC-affiliated entities including CNNC Operations & Maintenance Technology and a nuclear innovation vehicle in Xiong'an.4 Director and General Manager 邹正宇 Zou Zhengyu, age 57, has served as general manager since 2023 and simultaneously serves as executive director and general manager of both 中核技术投资 CNNC Technology Investment and China Nuclear Power (UK).4
That second detail is the structurally important one. The senior executives of the listed company hold executive roles at affiliated group entities. This is normal in Chinese SOE practice and it is not evidence of wrongdoing. It does mean that the people negotiating on behalf of minority shareholders in a related-party asset injection may hold formal fiduciary positions on both sides of the group.
What they are paid, and what that implies
Compensation is modest to the point of being analytically significant. Lu Tiezhong received pre-tax remuneration of ¥1.79 million from the company in 2025.4 Zou Zhengyu received approximately ¥1.76 million. The Chief Accountant received ¥1.47 million, a deputy general manager ¥1.57 million, the Board Secretary ¥1.42 million.4 Independent directors received ¥144,000 each.4 Several directors received nothing at all from the listed company, with the annual report marking them as receiving compensation from related parties — that is, they are paid by CNNC.4
Set ¥1.79 million — roughly $250,000 at prevailing rates — against the compensation of chief executives running comparably sized Western utilities, and the ratio is not close. Management here does not hold meaningful personal equity; the annual report shows nil or negligible shareholdings for the chairman and general manager.4
The analytical conclusion is not "management is underpaid" or "management is virtuous." It is that the incentive system contains no shareholder-return component of consequence. Nobody at the top of this company gets materially richer if the stock compounds. Careers in the CNNC system advance through safe operations, on-schedule construction, and delivery against national energy policy — and the fleet's operating record suggests those incentives work extremely well for what they target. They simply do not target return on capital, and investors should not expect them to produce it as a by-product.
The credibility test: how do they explain bad news?
The way to assess management credibility is to compare what they said across two calls, six months apart, as conditions deteriorated. On this test the record is mixed, and interestingly so.
Where they were concrete. On finance costs, management gave a specific mechanism and a specific limit — the replacement cycle is largely done, further reductions will be smaller.3 On fuel, they explained the ten-year framework structure and why spot uranium moves matter less than headlines suggest.3 On the profit decline in the first nine months of 2025, they named the causes without hedging: renewables revenue growth without profit growth, plus dilution from REIT-like issuances, a debt-to-equity swap, and capital increases.6 On a 59.96% jump in selling expenses — a question designed to catch them out — they gave the arithmetic directly: an increase of ¥27 million on a prior-year base of ¥45 million.6 On outages, they explained the actual maintenance taxonomy: routine overhauls of roughly 30 days, five-year overhauls of about 45 days, ten-year overhauls of about 60 days, with seven outages in Q1 2026 and five completed by April 30.3 On the 2026 production target, the answer to whether it could be maintained despite a heavy outage schedule was a single sentence: the outage schedule was already built into the annual plan.3 Terse, but falsifiable.
Where they were not. On every question touching the price of electricity — the variable that actually determines the equity's value — the answer routed to future government policy. Nuclear tariff mechanism: the energy authority is studying it, await official release.3 Capacity payments for zero-carbon baseload: same.3 Green certificates and CCER eligibility: the company is closely tracking developments, engaging expert teams, conducting methodology research, timetable subject to national policy.6 Long-run tariff trend: an accurate description of why prices fell, followed by trading-structure optimisation as the response.3
This is not evasion in the sense of misleading anyone. It is an honest reflection of a company that genuinely does not control its own price. But investors should register what it means: when the largest variable in the P&L moves against them, management has no company-specific lever to pull and does not pretend otherwise. The mitigation levers they do control — cost programmes, trading optimisation, financing costs, the "集约化、标准化、数智化" intensification/standardisation/digitalisation strategy repeated in both calls — are worth real money but are an order of magnitude smaller than the tariff swing.63
One further piece of evidence on priorities: in the first nine months of 2025, R&D expense grew 0.14%.6 For a company that markets a strategic-emerging-industries agenda spanning perovskite solar cells, medical isotopes — Qinshan placed commercially reactor-produced carbon-14 and lutetium-177 on the market in 2025, and the company launched an isotope brand called 和福一号 — hydrogen production, desalination and nuclear-powered shipping, a flat research budget is a useful reality check on how much of that agenda is currently funded versus aspirational.43
A small but telling procedural detail sits in the front matter of the FY2025 annual report. Three directors did not attend the board meeting that approved the annual accounts — including General Manager Zou Zhengyu — each recorded as unable to attend due to work commitments, each voting by proxy.2 It is a legal and unremarkable practice, and it would be overreading to build a thesis on it. But at a company where minority shareholders have no other lever, the seriousness with which the board treats its own formal proceedings is one of the few observable proxies for governance culture available to an outside investor. A chief executive delegating his vote on the annual accounts is worth noting in that ledger.
The governance conclusion
The ultimate controller is the State Council's 国务院国有资产监督管理委员会 State-owned Assets Supervision and Administration Commission. With CNNC holding 55.67% directly and the National Council for Social Security Fund and Guoxin holding another 10.7% between them, well over two-thirds of the register is state capital.2 There is no plausible activist pathway. A dissident shareholder cannot win a proxy contest, cannot force an asset sale, cannot replace a director.
The check on capital discipline here is therefore not the market. It is regulatory disclosure requirements, exchange rules, the audit, and SASAC's own return-on-capital metrics. That is a weaker check than shareholder pressure, but it is not nothing — the Key Audit Matter designations on goodwill and fixed-asset impairment, and the resulting ¥663 million write-down, are precisely the kind of independent constraint that produces uncomfortable numbers management would not volunteer.4
Given all that, how strong is the competitive position actually?
VIII. Competitive & Structural Analysis
Run this business through Porter's five forces and something unusual happens: four of the five forces come back almost inert, and the fifth is doing all the work.
Threat of new entrants: near zero, with an asterisk. You cannot enter this industry by raising capital or hiring talent. You need a nuclear operating licence and a state-approved coastal site, both granted administratively. That said, the barrier is not literally absolute — the Q3 2025 call included an investor question premised on Huaneng and SPIC having obtained nuclear operating licences, and the World Nuclear Association records four additional utilities permitted minority project stakes of up to 25%.610 Entry happens by state decision, which means the barrier protects incumbents exactly as long as the state wishes it to. That is a different risk profile from a patent or a network effect.
Rivalry among incumbents: structurally muted. The three operators do not compete for customers. Each plant sells into its designated grid company — Qinshan Unit 1 to State Grid Zhejiang, Qinshan units 2 and 3 and Fangjiashan to State Grid's East China branch, Jiangsu Nuclear to State Grid Jiangsu, Fuqing and Zhangzhou to State Grid Fujian, Hainan Nuclear to Hainan Power Grid, Sanmen to State Grid Zhejiang — with settlement monthly.4 There is no share to win. Even the growth allocation is administrative, as the July 2026 batch demonstrated when all three operators received units simultaneously and this company's Zhuanghe project was structured as a joint venture with a fourth party.1
Supplier power: contained by vertical integration. The critical input is nuclear fuel, and the controlling shareholder holds China's only domestic fuel manufacturing capability.3 A standalone operator facing a tight uranium market would have a genuine problem; this one has a parent that makes the fuel assemblies and a ten-year contracting structure that absorbs spot volatility. Equipment supply is similarly de-risked by the 90%-plus Hualong One localisation rate.6
Threat of substitutes: real, but not in the way it looks. Nothing substitutes for nuclear baseload at the capacity-planning level — the state has decided it wants roughly 110 GW by 2030 and coal, gas and renewables are all being built alongside rather than instead.10 But substitution is happening hour-by-hour in the spot market, where 1,201.73 GW of solar and 640.01 GW of wind with near-zero marginal cost set clearing prices during high-output periods.2 The substitution is not of the asset; it is of the price.
Buyer power: rising, and this is the force that moves. The buyer used to be an administrative tariff schedule. Increasingly it is a market — 71.36% of nuclear volume in 2025.4 And the ultimate buyer, the state, is simultaneously the entity designing the market and setting the rules by which it clears.
Through the 7 Powers lens, the picture is consistent with the Porter read. The company has a cornered resource in licences and sites. It has some scale economies in construction — management's stated cost strategy is design optimisation, equipment localisation, application of accumulated construction management experience, and standardisation to reduce unit capital cost on subsequent units.6 It has no meaningful branding, no network economies, no switching costs — a kilowatt-hour is a kilowatt-hour — and no counter-positioning, because it is the incumbent. Crucially, none of these powers operates on the price side. The moat is a fortress around the volume and nothing at all around the margin.
Why this wins from here
The bull mechanism is honest and mechanical: the growth is contracted rather than hoped for. Nineteen units at 21.86 GW were under construction or approved at end-2025, and management stated on the Q3 call that these are expected to become capable of commercial operation over the following five to six years.6 By January 1, 2026, Zhangzhou Unit 2 had already entered commercial operation, taking the operating fleet to 27 units and 26.21 GW.4 Two more were guided for the second half of 2026 — Tianwan Unit 7, and the Hainan small modular reactor, which had completed cold testing and was in the commissioning phase.3 Three new units in one year is meaningful capacity growth for a company this size, and it does not depend on winning anything.
Add the fleet's operating record — record utilisation hours, 22 units at full WANO composite score — the demonstrated financing-cost reduction, the possibility of licence extensions on older units, and the currently zero-valued option of nuclear being admitted to the green certificate system, and there is a genuine case.36
Why it may not
Every element of that case is a volume or cost argument. None of it addresses price. If the realised tariff continues to fall at even half the 2025 rate while capex stays near ¥93 billion annually and project-level minority equity keeps funding it, the arithmetic produces exactly what 2025 produced: more electricity, more assets, more consolidated profit, and flat-to-down earnings per share.24
The second structural problem is that the state occupies every seat at the table. It is the controlling shareholder, the licensing authority, the market designer, the tariff setter, the tax authority, and — through the grid companies — the counterparty. When those interests conflict, minority shareholders have no mechanism to influence the outcome and no precedent suggesting they would prevail.
Which is a useful way to think about the specific risks that follow.
IX. Risk Radar
Tariff marketisation is the primary risk, and it is structural. Nuclear's market-traded share nationally has moved from roughly 30% in 2020 to around 60% and above today.12 For this company the 2025 figure was 71.36%, and management told investors on the Q3 call to expect market volumes to increase further in 2026 absent major provincial rule changes, with each province's 2026 trading scheme expected around year-end 2025.46 The mechanism is not cyclical mean reversion. Renewable capacity is still being added at 20-35% annual rates nationally, spot markets are still being extended province by province, and each increment pushes more baseload volume into hours priced by zero-marginal-cost competitors.2 Nothing in the current policy framework arrests this; a mechanism price or capacity payment could, but none has been published.
Policy risk has already crystallised once. In November 2025 the Ministry of Finance eliminated the VAT refund mechanism for newly approved nuclear units.12 CITIC Securities estimated that VAT refunds account for roughly 2% of revenue but approximately 10% of net profit for representative nuclear operators.12 That gearing — a small revenue item that is a large profit item — is what makes it dangerous. It applies to newly approved units, so it does not hit the existing fleet, but it degrades the economics of precisely the pipeline that constitutes the growth story. Note also the direction of the FY2025 nuclear segment result: management attributed the segment's 18.53% attributable profit growth partly to increased VAT refunds alongside higher on-grid volume and lower income tax expense.3 The tailwind that flattered 2025 is being withdrawn for future capacity.
Tax preferences are rolling off independently. On the Q3 call, management explained the increase in income tax expense as partly reflecting some nuclear and new-energy projects exiting the "三免三减半" three-year exemption plus three-year half-rate preferential period.6 This is a known, dateable, mechanical headwind on projects as they mature — and it will recur.
Execution risk on a near-doubling of the fleet. Eighteen units and 20.65 GW remained under construction or approved-pending-construction after Zhangzhou 2 entered service.4 Management described its construction control framework — the "六大控制七个零" six controls and seven zeros programme — as broadly under control.6 But Sanmen's nine-year first-concrete-to-commercial-operation timeline is a live precedent inside this same company, and three to four first-concrete pours per year for the next two years means the exposure is expanding, not contracting.106
Capital structure and refinancing. Leverage at 69.01% sits just below the SASAC constraint, interest coverage is 2.74 times, and roughly ¥56 billion of annual capex is funded from sources other than operating cash flow.24 The near-term picture is benign: financing costs of 2-3%, new issuance at 1.75-2.27%, and a maturity ladder that includes a large 2026 cluster of mid-term notes and perpetuals originally issued at 3.17-3.44%, which will refinance downward at current rates.46 The risk is not today's cost — it is that the strategy explicitly depends on continued access to cheap capital and to project-level equity from other state entities. Both are conditional on China's low-rate environment and on state co-investors continuing to find nuclear project equity attractive at a time when the tariff outlook is deteriorating. Neither is contractual.
Goodwill and asset impairment. The ¥5.42 billion net goodwill balance, overwhelmingly from renewables acquisitions, faces an annual test against falling tariffs and curtailment.46 Eleven subsidiaries failed the 2025 test.4 The write-down was modest relative to profit, but the direction of travel in renewable pricing means the 2026 test will be applied to the same assets under worse assumptions.
Curtailment, and the mechanism price that partially offsets it. Management named curtailment — 消纳困难, difficulty getting renewable output absorbed by the grid — as one of the twin pressures on the new-energy segment alongside falling tariffs.3 Curtailment is the risk that a wind or solar farm is physically able to generate and is instructed not to, which converts a fixed-cost asset into a stranded one for those hours. There is a partial policy backstop: in Q1 2026, roughly 3.69 billion kilowatt-hours of the company's renewable output, about 36% of the total, fell under a mechanism-price arrangement, and management confirmed that the "renewable sustainable development price-difference settlement" — an off-market top-up where the market clearing average falls below the mechanism price — had been fully received.3 That is a genuine, functioning support scheme, and it is worth tracking because it covers only a portion of volume and is itself set by policy that can change.
A new intermediary in the offtake chain. In 2025 the company routed 29.86 billion kilowatt-hours through electricity retailing agents, of which 27.70 billion was nuclear and 2.16 billion renewables.4 That is roughly one-seventh of nuclear on-grid volume now reaching end-users through a commercial intermediary rather than a direct grid settlement. It is a natural consequence of marketisation and creates both an opportunity — the ability to structure longer-dated bilateral contracts, which management cited as an industry response to price pressure — and a new layer of counterparty and margin exposure that did not exist under an administered tariff.12
Concentration and disclosure. Every kilowatt-hour is ultimately sold to a state grid company, settled monthly.4 Counterparty credit risk is minimal; negotiating leverage is also minimal. On disclosure, the material gap is renewables segment profitability, which reaches investors through call commentary rather than a granular audited segment note — a lower evidentiary standard for what is now the second-largest part of the asset base.
Tail risk. Nuclear carries an irreducible low-probability, high-consequence safety risk. The company disclosed that its plants carry nuclear insurance plus commercial property all-risk, construction all-risk and third-party liability cover across construction and operating phases.6 That is standard industry practice and appropriate. It is worth noting proportionately and not dwelling on: the 2011 freeze demonstrated that the operative risk to this equity from an accident anywhere in the world is regulatory, not physical.
What is conspicuously absent. No credit rating adjustments were reported during 2025 for the company or its bonds.2 The audit opinion was unqualified.2 There is no going-concern language, no restatement, no auditor change. On the accounting-signal dimension, the file is clean — the judgments requiring scrutiny are the impairment assumptions on renewables and the appraisal values in related-party injections, not the integrity of the reported numbers.
X. Bull vs. Bear Synthesis & KPIs to Watch
The bull case, stated at its strongest
This is one of very few equities in the world offering contracted, non-competitive, decade-long capacity growth in an essential commodity. Eighteen units are already approved and building; the state has committed to roughly 110 GW nationally by 2030 and continues to approve at a normalised pace, with eight units cleared in a single July 2026 sitting.4101 Three units enter service in 2026 alone.43
The operating asset is excellent: record utilisation hours, a fleet-wide safety record with 22 units at full WANO composite score, fuel supply secured through a vertically integrated parent, and older units eligible for twenty-year licence extensions that are close to pure margin.36 Financing costs have fallen materially and demonstrably. Shareholder returns are improving at the margin — a first interim dividend, a completed first buyback, a stated intention to normalise repurchases, and a payout above 39% for FY2025.63
And there is a genuine free option: nuclear's exclusion from China's green certificate system is anomalous, is being actively contested by the industry, and if reversed would deliver a per-kilowatt-hour uplift on 188 billion kilowatt-hours of annual output with no incremental capital.62
The bear case, stated at its strongest
The company sells a commodity into a market the state is deliberately liberalising downward while simultaneously subsidising the competitors that set the clearing price. Realised nuclear tariff fell 5.16% in 2025; renewables tariff fell 11.21%; market-traded share leapt over twenty points in a single year and management expects it to rise further.46 There is no company-specific defence against this, and management has not claimed one.
Meanwhile the profit that reaches listed shareholders is being diluted structurally. Consolidated profit grew 8.56% and attributable profit grew 6.00% while earnings per share fell 2.16%, because the buildout is funded substantially by selling project-level equity — ¥18.52 billion of it in 2025 — to co-investors who take an ever-larger share of the output.24 Capex rose to ¥93.68 billion against ¥37.41 billion of operating cash flow; the "capex has peaked" thesis has not yet been evidenced by a single data point.42 Return on equity fell 1.23 points to 8.21%, and the quarterly profit path deteriorated through 2025 before Q1 2026 attributable profit fell 34.19%.23
The second engine has not yet become a profit engine: renewables delivered ¥346 million of attributable profit on 33.64 GW of capacity, down 71.60%, with ¥5.42 billion of goodwill from acquisitions now impairing at the edges.34 Policy has already turned once with the VAT refund cancellation, worth roughly a tenth of net profit for representative operators on new capacity.12 And the governance structure means minority shareholders finance a state programme without any mechanism to influence how it is executed or priced.
The clinching comparison is the peer: CGN Power, running a larger reactor fleet with less capacity growth, saw revenue and profit both decline in 2025.12 That is what this business looks like when the volume tailwind stops covering the price headwind.
The three KPIs that settle it
Everything above reduces to three trackable series. Investors should not need models — they need to watch these each reporting cycle.
1. Realised nuclear tariff per kilowatt-hour, alongside the market-traded percentage of on-grid volume. Disclosed in the annual report's electricity marketing section. This is the single variable that determines whether volume growth converts into shareholder value. The 2025 marks were ¥0.3937/kWh and 71.36%.4 If the tariff stabilises — or if a nuclear mechanism price or capacity payment is published — the bull case has its foundation. If it continues declining while market share of volume rises toward saturation, the compounding case erodes regardless of how many reactors get built.
2. New-unit grid connections and commercial-operation dates against the announced pipeline, plus annual first-concrete pours. Management guided to Tianwan 7 and the Hainan small modular reactor in the second half of 2026 and three to four first-concrete pours per year.36 Actual delivery against those specific, dated commitments is the cleanest available test of execution — and Sanmen's history establishes that slippage in this industry is measured in years, not quarters.
3. Capital expenditure in the cash flow statement, read against the dividend payout ratio. These two lines together settle the "capex peaks, returns rise" thesis. In FY2025 capex was ¥93.68 billion and payout was above 39%.43 A rising payout funded by a falling profit base is not the same thing as a rising payout funded by a falling capex base — and only the cash flow statement distinguishes them.
XI. Epilogue & Closing Reflections
Stand back from the numbers and the picture is of a company caught precisely in the middle of two transitions at once.
It is mid-buildout: 18 units under construction against 27 operating, capex still running at two and a half times operating cash flow, three to four new construction starts a year for the foreseeable future.46 And it is mid-tariff-transition: seven-tenths of its output already priced by a market that did not meaningfully exist for nuclear five years ago, with the remaining administered share shrinking and no replacement mechanism yet published.43 Its two engines run at different speeds and, currently, in different directions — the reactor fleet delivering nearly all the profit on slower volume growth, the renewables arm delivering nearly all the volume growth on almost none of the profit.3
The larger lesson generalises well beyond one Shanghai-listed utility, and it is worth stating carefully because it cuts against a common investing instinct. Regulatory moats are usually described as the strongest kind — unassailable, durable, immune to competitive attack. This company has as pure a version of that moat as exists anywhere: no competitor can enter, no customer can switch, no technology can displace it within its licensed sites. And in 2025 it earned an 8.21% return on equity, down more than a point, with earnings per share going backwards.2
The reason is simple and easily forgotten. A moat granted by an authority is only as valuable as that authority allows it to be. The same state that made competitive entry impossible also decided that seven-tenths of the output should clear in a spot market flooded with subsidised zero-marginal-cost supply, and also decided in November 2025 to withdraw a VAT refund worth roughly a tenth of profit on future units.12 Barriers to entry protect against competitors. They offer no protection whatsoever against the entity that erected them.
What to watch over the next few reporting cycles is therefore less about reactors and more about language and cash. On the language side: whether management's answer to the tariff question ever changes from "the energy authority is studying it" to a concrete mechanism with a published price — and whether the green-certificate exclusion, which the company itself has flagged as an anomaly, is ever resolved.36 On the cash side: whether capex finally turns, whether the renewables segment's profit contribution recovers or continues shrinking against a growing asset base, and whether attributable profit and earnings per share start moving in the same direction as consolidated profit.
Until those change, the honest characterisation is this: China National Nuclear Power is an operationally excellent, structurally protected, state-directed builder of essential infrastructure, whose shareholders currently receive a diluted and policy-determined slice of the returns on capital they help fund. Whether that is an attractive proposition depends entirely on the price paid and on a judgment about which way the state's pricing decisions break next — and that judgment, unlike the reactors, is not something any amount of due diligence can make contracted.
References
-
中国核能电力股份有限公司 2025年年度报告摘要 (FY2025 Annual Report Summary) — 巨潮资讯网 cninfo, 2026-04-30 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩
-
中国核电 2025年度暨2026年一季度业绩说明会会议记录表 (FY2025 + Q1 2026 earnings-briefing transcript) — 东方财富 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩
-
中国核能电力股份有限公司 2025年年度报告全文 (FY2025 Full Annual Report), 2026-04-30 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩
-
Notice on improving the nuclear power tariff mechanism (NDRC, 2013-06-15) — China Energy Portal ↩↩↩
-
中国核电 2025年三季度业绩说明会会议记录表 (Q3 2025 earnings-briefing transcript), 2025-10-31 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩
-
中国核电IPO获有条件通过 融资规模或被缩减 — 中国日报财经 China Daily Caijing, 2015-05-14 ↩↩↩
-
中国核电 2025年科技创新公司债券募集说明书 (2025 bond prospectus) — 上海证券交易所, 2025-11-05 ↩
-
Nuclear Power in China — World Nuclear Association country profile ↩↩↩↩↩↩↩↩
-
中广核电力股份有限公司 投资者关系活动记录表 (CGN Power investor relations record), 2026-01-17 ↩
LinkedIn
WhatsApp
Reddit
Hacker News
Email
Kindle
RSS Feed
Spotify
Apple Podcasts
Amazon Music
Audible
YouTube