Datang Power: The Coal Giant Betting on the Sun

I. Introduction & Episode Roadmap

The worst quarter in the history of 大唐国际发电 Datang International Power Generation did not announce itself with a bankruptcy filing or a factory closure. It arrived quietly, in a line of a financial table, in the spring of 2022. For the three months ending December 31, 2021, a company that had sold more electricity than the year before, to more customers, at government-sanctioned prices, reported a loss of roughly CNY 9.28 billion — a figure larger than the entire full-year loss, because the first half of the year had actually been profitable.³

Nothing had broken. No plant exploded. No customer left, because in the Chinese power system customers cannot leave. What happened was arithmetic: the price of coal rose, the price of electricity did not, and a company with roughly 60 gigawatts of thermal generating plant discovered that it was, in effect, a short position on coal disguised as a utility.

Four years later, the same company reported a very different set of numbers. For 2025, Datang posted revenue of about CNY 121.26 billion — actually down 1.8% — and net profit attributable to shareholders of CNY 7.39 billion, up 63.9% year on year.¹² Return on equity swung from around 10.4% to 18.2%. Operating cash flow reached CNY 37.84 billion, up nearly 45%.² In the first quarter of 2026, Datang was the only one of China's four largest listed thermal generators to grow both revenue and profit.¹³

That is a remarkable round trip. It is also the central analytical problem of this story. A company whose earnings can fall by CNY 12 billion and rise by CNY 6 billion within four years, without meaningfully changing what it owns, is not primarily an operating business. It is a spread — the gap between what it pays for fuel and what the state permits it to charge for power. The interesting question is not whether the spread widened. It plainly did. The question is whether anything structural now protects that spread from closing again.

Datang is one of the five state generators carved out of China's electricity monopoly in 2002, and one of the largest independent power producers on earth. At the end of 2025 it operated about 86,192 megawatts of capacity across 20 provinces: roughly 49.1 GW of coal, 9.5 GW of gas, 9.2 GW of hydro, 11.2 GW of wind and 7.2 GW of solar.¹ It generated about 289 billion kilowatt-hours during the year and delivered about 273 billion of them to the grid.¹ For scale, that output alone would rank among the twenty largest national electricity systems in the world.

This story runs in five movements. First, how Chinese power generation actually works — because almost nothing about Datang's profit and loss makes sense until you understand that the company does not choose its customers, and for most of its history did not choose its prices. Second, the portfolio: what actually earns the money, as opposed to what appears in the sustainability slide deck. Third, the near-death experience of 2021 and the regulatory machinery built afterward to prevent a repeat. Fourth, the clean-energy pivot — its capital cost, its funding, and the sister company that owns the assets you might have assumed Datang owned. And fifth, the rulebook rewrite of 2025 that took the guaranteed price away from new wind and solar just as the industry finished betting its balance sheet on them.

Along the way there is a chairman's chair that has changed occupants three times in two years, CNY 46 billion of perpetual bonds sitting quietly in the equity column, and a subsidiary that bought a company with negative net worth for one yuan and promptly wrote off CNY 779 million.¹ None of that appears on the cover of the annual report. All of it matters.

↑ Back to Top

II. Origins: Born From a Monopoly Break-Up

There is no garage in this story, no founder mortgaging a house. Datang's founding document is industrial policy.

The company was incorporated in 1994 as the vehicle for the northern Chinese thermal assets of what would become 中国大唐集团 China Datang Corporation, at a moment when Beijing had concluded that the Ministry of Electric Power could not build power stations fast enough to keep the lights on in a double-digit-growth economy.⁵ The state needed capital it did not have. So it did something that in retrospect looks radical: it packaged coal plants into a joint-stock company and sold shares to foreigners.

In March 1997, Datang's H shares listed in Hong Kong and London simultaneously — the first Chinese enterprise ever to list in London, and the first Chinese power company to list in Hong Kong.⁵ It is worth pausing on how strange that was. Two years earlier, most global investors could not have named a single Chinese generating company. Now a Beijing-controlled coal-fired utility was filing prospectuses under English disclosure rules and taking questions from fund managers in the City. The London listing persists to this day; the annual report still shows the H shares trading in London under the ticker DAT.²

Five years later came the event that defined the industry. In 2002, the State Council dismantled the State Power Corporation, separating generation from transmission and dividing the generating assets among five new central state-owned enterprises: 华能 Huaneng, 大唐 Datang, 华电 Huadian, 国电 Guodian, and 中国电力投资 China Power Investment. The grid went to two monopolies, 国家电网 State Grid and 中国南方电网 China Southern Power Grid.

The design intent matters more than the names. The five generators were deliberately sized so that no one of them would dominate any region — an oligopoly engineered to compete on cost and construction speed while remaining collectively answerable to the state. That structure created a peculiar competitive dynamic that persists: five very large companies, none with pricing power, all facing the same two buyers, all rewarded historically for building. Between 2002 and the mid-2010s, the Big Five did exactly what the incentive structure told them to do. They built. China's installed capacity roughly quintupled.

Datang came to the mainland market late. Its A shares listed on the Shanghai Stock Exchange in December 2006, making it the first Chinese company simultaneously listed in Hong Kong, London and Shanghai.⁵ The delay was less a strategic choice than a function of A-share market conditions and regulatory queueing in the mid-2000s, but the outcome shaped the shareholder register in a way that still matters.

That register is worth reading closely, because it explains the governance reality better than any policy statement. At the end of 2025, China Datang Corporation directly held 35.34% of the shares. Add the H shares held through its wholly owned Hong Kong arm — 3.28 billion shares, about 17.7% of the total — and the parent group controlled roughly 53.04% of the company.² Behind it sat a cluster of provincial and municipal state investors: 河北建设投资集团 Hebei Construction Investment Group at 6.93%, 天津市津能投资 Tianjin Jinneng Investment at 6.57%, 北京能源集团 Beijing Energy Group at 1.31%.² The free float, spread across roughly 181,000 retail and institutional holders, is a genuine minority in every sense.

So the "founder" here is the Chinese state, and the shareholders sitting alongside you are largely other arms of the Chinese state. This is not a criticism; it is a structural fact with investment consequences that recur throughout this story. A company owned this way will be asked to do things — keep a city's heating running, absorb a distressed asset, build capacity ahead of demand — that a purely commercial operator would refuse. Sometimes the state pays for those services. Sometimes it does not.

The investable story does not really begin in the 1990s. It begins when the building stopped being automatically profitable — and to see why, you have to understand the machinery that sets Datang's prices.

↑ Back to Top

III. How Chinese Power Generation Actually Works

Picture a factory that makes exactly one product, sells it to exactly one buyer, cannot store any of it, and for most of its history was told by a government agency what price it would receive. That is the business Datang is in, and every inflection point in this story flows from it.

The monopsony. Datang's electricity goes to provincial grid companies or into provincial wholesale markets. There is no alternative channel, no direct-to-consumer option, no export. The commercial relationship that dominates most industrial businesses — negotiating with customers — barely exists. What exists instead is a negotiation with policy.

For two decades, that negotiation had a simple form: the National Development and Reform Commission set a benchmark on-grid tariff for coal power in each province, and generators received it. When coal was cheap, generators made money. When coal was expensive, they did not. The tariff moved rarely, and always politically, because retail electricity prices in China have long been treated as an input to industrial competitiveness rather than as a market price.

In October 2021, in the middle of the crisis described later in this story, the NDRC widened the band within which coal tariffs could float around the benchmark to plus or minus 20%, and pushed nearly all industrial and commercial users into market trading. This was the single most consequential regulatory change of the decade for companies like Datang — and it is worth being precise about what it did and did not do. It allowed prices to rise when coal was scarce. It did not guarantee they would.

The evidence for that caveat is now abundant. Market transactions accounted for roughly 89.3% of Datang's on-grid volume in the first half of 2026, and the company's average realized on-grid tariff was about CNY 430.92 per megawatt-hour including tax, down 3.05% year on year.²⁴ In the first quarter of 2026 the decline was steeper, about 4.8%.¹⁴ Marketization, once demanded by generators as protection against cost spikes, now cuts the other way: with capacity growing faster than demand, competitive auctions push prices down.

Regional dispersion. A second feature that surprises newcomers is how much prices vary by province. In 2025, Datang realized about CNY 635 per MWh in Beijing, roughly CNY 489 in Guangdong, and about CNY 399 in Shanxi — a spread of nearly 60% between the highest and lowest.¹ Provinces set their own market rules, their own coal contracting norms, and increasingly their own renewable pricing mechanisms. A generator's geographic footprint is therefore not a rounding error. It is a material determinant of realized price.

The capacity payment. The most important structural change since the crisis arrived in January 2024, when China began paying coal plants a fixed monthly fee simply for existing and being available. The mechanism is easiest to understand by analogy: instead of paying a firefighter only for hours spent fighting fires, you also pay a retainer for being on call. Coal plants, as renewables grow, are increasingly on call.

The national framework benchmarks a coal plant's fixed costs at CNY 330 per kilowatt of capacity per year and reimburses 30% or 50% of that through capacity payments, depending on the province — roughly CNY 100 to CNY 165 per kilowatt.⁶ Analysis published by Carbon Brief estimated the mechanism distributed about CNY 107 billion nationally in 2024, boosting revenue at a typical 600 MW plant by 4.7% at the lower rate and 7.9% at the higher one, with the national floor scheduled to rise to at least 50% in 2026.⁶

For a company with 49 GW of coal capacity, this is not a rounding error either — it is a multi-billion-yuan annual revenue stream that arrives regardless of whether the plants run. It converts part of a volatile merchant business into something resembling an availability contract. Management is explicit about chasing it: the 2025 annual report lists "actively seeking capacity tariff policy support, ensuring full recovery of coal capacity payments" as a formal mitigation measure against price risk.¹

But note what the mechanism is and is not. It is a subsidy for keeping coal plants alive during a renewables build-out, granted uniformly to compliant coal units across the industry. Datang receives it. So does Huaneng, Huadian, GD Power and every provincial generator with a compliant unit. It improves the sector's floor; it does not differentiate any company within it. Independent analysis has also questioned whether the scheme achieved its stated purpose, noting that between 70% and 100% of coal capacity in most provinces qualified and that there is little evidence coal operating hours fell as a result.⁶

The demand backdrop that governs everything. One national statistic frames the whole industry's next decade. China's total electricity consumption passed 10 trillion kilowatt-hours for the first time in 2025, reaching 10.37 trillion, up 5.0% — with July alone exceeding one trillion kWh, the first time any country has crossed that threshold in a single month.¹ Demand, in other words, is still growing healthily.

The problem is that supply is growing faster, and in a different direction. National installed capacity rose 16.1% during 2025 to 3.89 billion kilowatts, with 550 GW added in a single year — more than 80% of it wind and solar.¹ Non-fossil capacity reached 61.7% of the national total.¹ Most striking of all, wind, solar and biomass together accounted for 97.1% of the increase in national electricity consumption.¹

Read that last figure slowly, because it is the most important number in this story that has nothing to do with Datang specifically. Essentially all incremental Chinese electricity demand is now being met by renewables. A coal fleet in that environment does not lose absolute volume quickly, but it stops growing, its running hours decline, and its economics depend increasingly on payments for availability rather than for energy. That is the transition, expressed in one statistic.

Two other mechanics complete the picture. Utilization hours — the number of hours a year a plant actually runs — are falling across China as renewables take load: the national average across all plants above 6 MW dropped 312 hours in 2025 to 3,119, with coal units down 269 hours to 4,346.¹ And heating, a regulated public service bundled into many northern plants, is priced below cost by design, a point that will matter greatly when we open the segment accounts.

Which is exactly where to go next: with the rules understood, what does Datang actually earn money on?

↑ Back to Top

IV. The Portfolio: What Actually Drives Revenue and Profit

Open Datang's 2025 annual report to the segment tables and something counterintuitive happens. The company that most investors classify as a coal-fired utility — 57% of its nameplate capacity burns coal — reports that coal generated slightly less than half of its power-segment pre-tax profit.¹

The precise numbers are worth laying out, because the shape of this table is the shape of the investment case. Pre-tax profit by generation type in 2025 ran as follows: coal units including heat, CNY 5.62 billion, up 119.1% year on year; gas, CNY 466 million, up 36.8%; hydro, CNY 2.14 billion, up 6.5%; wind, CNY 2.94 billion, up 38.5%; and solar, CNY 771 million, up 9.8%.¹

Add the three genuinely clean sources — hydro, wind and solar — and they contributed about CNY 5.85 billion, marginally more than coal. Hold that thought, because it cuts against the standard narrative in both directions. The bull version ("Datang is still a coal company, the green stuff is decoration") is wrong on profit. The other bull version ("the transition is done") is wrong too, for a reason visible in the same tables: coal's contribution more than doubled in a single year purely because fuel got cheaper, while wind and solar — the assets that supposedly de-risk the company — grew far more slowly and remain vulnerable to a different force entirely.

How the revenue splits. Of about CNY 119.79 billion in core operating revenue in 2025, electricity sales were CNY 105.09 billion, or roughly 88%, down 2.3% year on year — and management attributed the decline squarely to lower on-grid tariffs, not lower volume.¹ Heat sales added CNY 6.58 billion. Everything else — the coal business, fly ash, aluminum, transport, chemicals — came to CNY 8.12 billion.

The heating problem. Now the number that rarely appears in broker notes. Datang's heat sales in 2025 generated CNY 6.58 billion of revenue against CNY 10.05 billion of cost — a gross margin of negative 52.7%.¹ That is a loss of nearly CNY 3.5 billion on the year, and it improved by 4.7 percentage points from 2024.

Heat is not a business Datang chose. Combined heat and power plants in northern China supply district heating to cities under tariffs set by local governments, which are politically constrained because the customers are households in winter. The company delivers a public service at a structural loss, and the loss is embedded inside the "coal including heat" segment. Strip it out conceptually and the coal generation business is considerably more profitable than the segment table suggests — and the company as a whole is carrying a permanent, policy-mandated drag of roughly CNY 3 billion a year. That is one concrete price of state ownership, and it is quantifiable.

Why capacity share overstates the green story. The 2025 report proudly notes that clean energy reached 42.99% of installed capacity, up 2.62 points, after adding 5,120 MW of clean capacity during the year.¹ Read the definition carefully: that figure includes gas turbines. Hydro, wind and solar together are about 32% of capacity. Investors comparing Datang to a renewables developer should use the narrower number.

More importantly, capacity is a poor proxy for either output or revenue. Datang's coal units ran 4,306 hours in 2025; wind ran 2,212 and solar 1,333.¹ A megawatt of solar therefore produces roughly a third as much electricity per year as a megawatt of coal. And it sells that electricity for less: in Hebei, Datang's coal fleet realized about CNY 456 per MWh while its solar realized CNY 384; in Shanxi the gap was CNY 393 against CNY 336.¹ Lower running hours multiplied by a lower price is why 32% of capacity does not become 32% of revenue.

What renewables do offer is a superior margin structure once built — no fuel, minimal variable cost — which is precisely why wind delivered CNY 2.94 billion of pre-tax profit off 13% of capacity. The honest framing is that Datang's clean fleet is already a meaningful profit contributor, but its economics depend almost entirely on realized prices, and realized prices are now set by a mechanism that changed in 2025.

The coal hedge, examined. The outline of any thermal generator's defence against fuel spikes is to own coal mines. Datang's version of this is smaller and more indirect than commonly assumed, and worth stating precisely. The company purchased 119.77 million tonnes of coal in 2025 at a unit cost of CNY 742.4 per tonne of standard coal delivered into the boiler, excluding tax — down CNY 128.94 per tonne, or 14.8%.¹¹⁰ Its own mining interests are modest by comparison: the most significant, 同煤大唐塔山煤矿 Tongmei Datang Tashan Coal Mine, is an equity-accounted associate rather than a consolidated subsidiary, carried at about CNY 6.91 billion and paying CNY 456 million of dividends receivable at year-end.¹ A smaller consolidated coal business, 内蒙古宝利煤炭 Inner Mongolia Baoli Coal, took a CNY 18.7 million goodwill impairment in 2025.¹

Meanwhile CNY 17.21 billion of coal came from the parent group under a related-party framework agreement with a CNY 24 billion annual cap.¹ So the "captive" hedge is best described as partial and largely intra-group: Datang buys much of its coal from itself in the broad sense — from the same state that owns it — but the listed company captures only a slice of the mining margin. Against 119.77 million tonnes of annual burn, an equity stake in one large mine is a hedge in the way that owning a small vineyard hedges a restaurant's wine list. It helps at the margin. It does not neutralize the exposure.

Myth versus reality, three times over. The first consensus belief about this company is that it is a pure coal play whose renewables are cosmetic. The segment accounts refute that: clean generation already earns roughly as much pre-tax profit as coal does. The second belief, the mirror image, is that the transition is largely complete because "clean energy" exceeds 40% of capacity. That figure counts gas turbines, and capacity is not output. The third is that the coal business is hedged by captive mining. It is hedged at the margin, through an associate stake and a small consolidated miner, against a burn of nearly 120 million tonnes.

Two smaller line items round out the picture, both worth a sentence rather than a paragraph. Datang reports only two segments — power and "others" — and the others bucket, which contains the coal trading and mining activities, fly ash, aluminum and related businesses, produced CNY 6.29 billion of revenue and CNY 1.32 billion of pre-tax profit in 2025, about a tenth of group pre-tax profit on 5% of revenue.¹ It is small, it is more profitable per yuan of revenue than generation, and it is not a growth engine.

The second is research spending, which at CNY 17.4 million of expensed R&D is essentially a rounding error against a CNY 121 billion revenue base, with the bulk of technical spending capitalized into projects instead.¹ For a company whose future depends on operating an increasingly complex mix of thermal, hydro, wind and solar assets inside volatile markets, that is a data point about what kind of business this is: an owner and operator of infrastructure, not a technology developer.

The whole segment picture, then, reduces to a single sentence an investor can hold onto: Datang is a fuel-cost story with a growing, price-exposed renewables option attached and a loss-making public heating obligation stapled to the side. The fuel-cost story is what nearly killed it.

↑ Back to Top

V. The Near-Death Experience: The 2021 Coal Crisis

In the autumn of 2021, factories in Guangdong were operating on rationed schedules, traffic lights failed in parts of the northeast, and Chinese social media filled with photographs of office workers climbing stairs in unlit towers. A country that had spent two decades building the largest electricity system in human history was, briefly, running short of power.

The cause was not a shortage of generating plant. China had plenty of plant. The cause was that running it had become financially ruinous.

Three forces collided. Post-COVID industrial demand rebounded harder than planners expected. Domestic coal supply had been constrained by years of safety and consolidation campaigns and by decarbonization-driven mine closures. And import channels tightened. Thermal coal prices roughly doubled. Because coal-fired tariffs were still effectively capped, the entire cost increase landed on generators' income statements.

The numbers at Datang describe a business being crushed from the inside. Revenue for 2021 actually rose 8.16% to about CNY 103.41 billion — the company sold more electricity than ever. Operating costs rose 33.49% to CNY 103.73 billion. Fuel costs alone reached CNY 65.40 billion, an increase of CNY 23.52 billion in a single year.³ The company reported a net loss attributable to shareholders of CNY 9.26 billion, against a CNY 3.04 billion profit in 2020.³

The quarterly path is the part that should stay with investors. Datang earned profits above CNY 800 million in each of the first two quarters of 2021, lost CNY 1.62 billion in the third quarter, and then lost CNY 9.28 billion in the fourth.³ Coal-price pass-through in this system is not linear; it is a dam that holds and then bursts. Long-term contract coal covers a portion of supply at administered prices, so the marginal tonne — bought on the spot market in a panic — sets the pain, and the accounting recognizes it all at once, with year-end impairments layered on top.

This was industry-wide, not company-specific. Central state-owned enterprises' coal-power businesses lost about CNY 101.7 billion in 2021 according to figures disclosed by the state assets regulator, and China Datang Corporation as a group reported a total loss of CNY 21.66 billion.⁴ It was the sector's first collective loss since 2009.

Here is the analytical core of this episode, and it is the "why not" of the investment case stated in its purest form. A generator with no ability to set output prices, no meaningful fuel hedge, and a fixed obligation to supply is structurally short its main input. Nothing about Datang's management, engineering or asset quality caused the 2021 loss. Nothing about better management would have prevented it. It was a design feature of the market Datang operates in.

Which is precisely why the regulatory response matters more than the operational one. The state did not bail out generators with cash. It changed the rules: it widened the tariff float band in October 2021, pushed industrial users into market pricing, strengthened long-term contract coal discipline, and — three years later — introduced the capacity payment that pays coal plants for availability rather than output.

Did it work? The 2022–2025 evidence. The recovery was real but slow. Datang was still loss-making in 2022. By 2023, net profit attributable to shareholders had crawled back to CNY 1.37 billion, then CNY 4.51 billion in 2024, then CNY 7.39 billion in 2025.² The 2025 result came with revenue down 1.8% — this was entirely a cost and mix story. Unit fuel cost fell 14.8%; the coal-and-heat segment's pre-tax profit more than doubled; gross margin widened by 4.39 percentage points to 18.32%; and per-kilowatt-hour profit improved by about CNY 0.015.¹¹⁰¹⁵

So: structural repair or a favorable turn in the coal cycle? The honest answer is roughly two-thirds cycle, one-third structure, and investors should resist management framing that implies otherwise.

The cyclical part is unambiguous. Coal at CNY 742 per tonne of standard coal is not a policy achievement; it is a market price, and the same market took it far higher in 2021. The structural part is genuine but narrower than the recovery in profits implies: capacity payments now cover a meaningful share of coal plants' fixed costs regardless of dispatch, the tariff band can absorb some input shocks, and long-term contract coverage has improved. Those changes raise the floor. They do not restore the ceiling — indeed, market-clearing tariffs are now falling.

The most useful stress test is management's own risk disclosure, which is franker than the earnings headlines. In the 2025 annual report the company warned that tightening domestic production controls, Indonesian export quotas, restricted Russian and Australian supply and Middle East conflict were pushing the 2026 coal market toward a "tight balance" with increased price volatility.¹ That is the company telling investors, in its own filing, that the input which produced both the 2021 disaster and the 2025 recovery remains outside its control.

The response to that vulnerability — the strategic answer to "what if coal does it again?" — has been to buy assets that do not burn anything.

↑ Back to Top

VI. The Clean Energy Pivot: Capital Deployment and Asset Injections

Every large Chinese generator now tells a version of the same story: coal is the past, wind and solar are the future, and the transition is proceeding on schedule. Datang added 5,120 MW of clean capacity in 2025 alone — more than the entire generating fleet of many mid-sized countries — and its clean-energy ratio climbed to 42.99% from 40.37% a year earlier.¹⁵ Since 2024, total capacity has grown from about 79.1 GW to 86.2 GW.⁵¹

The strategic logic is sound and needs no defending: shift earnings away from a commodity-price-exposed core toward assets with no fuel cost. The questions worth asking are about execution, funding and — most of all — corporate structure. Because the structure of the Datang group makes the pivot considerably more complicated than the capacity numbers suggest.

The sister company you might have assumed was a subsidiary. 中国大唐集团新能源股份有限公司 China Datang Corporation Renewable Power, listed in Hong Kong as 1798.HK, is the group's dedicated wind and solar platform. At the end of 2025 it operated 19,751.71 MW — 14,353 MW of wind, 4,888 MW of solar and 510 MW of storage.²⁵ That is a larger renewables fleet than the one inside Datang Power itself.

It is not, however, owned by Datang Power. It is a separately listed sibling under the same parent. The group's long-standing internal architecture, described in Chinese financial press for years, assigns thermal assets to Datang International, wind to Datang Renewable, and hydro to 桂冠电力 Guiguan Power.¹⁸ Three listed vehicles, one parent, three overlapping mandates and one obvious governance question: when the group develops a new wind farm, which listco gets it, and on what terms?

That question is not academic, and Datang Renewable's own results make it urgent in a different way. In 2025 the renewables platform earned net profit of CNY 1.81 billion, down from CNY 2.62 billion a year earlier, despite a larger fleet — with the company attributing the decline to deepening electricity market reform and intensifying competition.²⁵ A pure-play renewables operator inside the same group, with better financing costs than most peers, saw profits fall by nearly a third while its capacity grew. Any investor underwriting Datang Power's renewables build-out as a straightforward de-risking exercise should sit with that data point for a while.

The injection question, tested against history. The primary M&A mechanism in this corner of the market is not open-market acquisition. It is parent-to-listco asset injection: the state parent sells assets it owns to the listed vehicle it controls, at a price set by professional appraisal, approved by a board the parent effectively appoints.

Datang has done this at scale before, and the historical record is the most useful evidence available for judging future injections. In December 2017, Datang Power agreed to acquire 100% of three thermal subsidiaries from the parent — Datang Heilongjiang, Datang Anhui and Datang Hebei — for CNY 18.13 billion in cash.¹⁸ The appraised values carried substantial premiums to book: Heilongjiang's net assets of CNY 3.21 billion were valued at CNY 5.88 billion, an 83% uplift; Anhui's CNY 4.51 billion became CNY 7.80 billion, up 73%; Hebei's CNY 2.98 billion became CNY 4.44 billion, up 49%.¹⁸ And the assets were not, at the time, performing: for the first nine months of 2017, only Heilongjiang was profitable, with Anhui and Hebei losing money and the three together posting a combined loss of roughly CNY 505 million.¹⁸

State the finding plainly, because the outline of this episode asked for a verdict and the evidence supports one: in the largest injection in the company's recent history, the listed company paid cash, at premiums of roughly 50% to 80% over book value, for assets that were collectively loss-making at the time of transfer. The stated rationale was group strategic coordination and honoring commitments to avoid intra-group competition.¹⁸ Those are real considerations under Chinese listing rules. They are also not the same thing as value creation for minority shareholders.

And it still happens. In September 2025, Datang Anhui Power Generation — one of the very entities acquired in 2017 — agreed to buy the remaining 50% of 安徽电力股份有限公司 Anhui Electric Power from 淮南矿业 Huainan Mining Group for a symbolic price of one yuan.¹⁶ The target had, as of April 30, 2025, total assets of about CNY 790 million against liabilities of about CNY 2.47 billion — owners' equity of roughly negative CNY 1.69 billion and an asset-liability ratio of 314%.¹⁷ It had lost money in 2023 and lost more in 2024.

Buying a negative-equity company for one yuan sounds like a bargain until you follow the accounting. Because the target's net assets were negative, the transaction generated goodwill of about CNY 1.18 billion, of which CNY 779 million was immediately impaired — a 66% write-down rate — reducing consolidated net profit by roughly CNY 589 million.¹⁶¹⁷ The 2025 annual report confirms the impairment.¹ The rationale disclosed was social responsibility: Anhui Electric Power is Huainan's sole municipal and industrial heating supplier, serving roughly 350,000 square metres of residential heating.¹⁶

Look at the fourth-quarter 2025 numbers with that in mind. Datang earned about CNY 2.2 billion in each of the first three quarters of 2025, then just CNY 674 million in the fourth — while fourth-quarter profit excluding non-recurring items was CNY 1.17 billion.² The gap is where the deal landed. For the full year, non-recurring items reduced attributable profit by roughly CNY 467 million.²

That single transaction is the clearest available illustration of what state ownership costs a minority shareholder in cash terms. It also shows the machine is still running: in December 2025, sister company Guiguan Power announced the acquisition of two Datang Tibet clean-energy subsidiaries for CNY 2.03 billion, adding 915 MW of operating clean capacity and 1,415 MW of hydro under construction.²⁷

What it costs to build. Renewables in this sector are funded overwhelmingly with debt, and Datang is no exception: investing cash outflow ran at about CNY 26.5 billion in 2025, against operating cash inflow of CNY 37.8 billion.¹ The parent group's own pipeline gives a sense of the scale of commitment ahead — CNY 56.2 billion of budgeted construction projects as of September 2025, including the Zhala hydropower station in Tibet, of which CNY 22.4 billion had been spent.²⁶

Which raises the question that governs the entire pivot: what price will all this new capacity receive for its output? In February 2025, Beijing changed the answer.

↑ Back to Top

VII. The 2025 Rulebook Rewrite: Document 136

For fifteen years, building a wind farm in China involved a comforting piece of arithmetic. The state set a feed-in tariff. The grid was obliged to buy a guaranteed share of output at that price. A developer could model twenty years of revenue with a spreadsheet and a calculator, take the model to a bank, and borrow against it. Renewables were, financially speaking, an annuity wearing a turbine.

On February 9, 2025, the National Development and Reform Commission and the National Energy Administration published Document 136, and the annuity ended. From June 1, 2025, all electricity from newly commissioned wind and solar projects must be sold through market transactions rather than at an administratively fixed price.⁷⁹

What replaced it. The mechanism is best understood as a contract-for-difference, closely resembling the auction system used in the United Kingdom. Each province determines an annual quota of new clean capacity. Projects compete in auctions to win a "sustainable price" for a defined share of their output — the settlement contracts cover the same guaranteed proportion as the old offtake rules, 85% for solar and 70% for wind. Output above the quota, or from projects that fail to win, sells at whatever the market pays: bilateral power purchase agreements or the provincial spot market.⁷⁸

In plain terms: the state stopped guaranteeing a price to everyone and started auctioning a guarantee to some. For the winners, revenue looks much like before. For everyone else, a wind farm becomes a merchant power plant with weather instead of fuel.

Early evidence. The first auctions ran in September 2025. Shandong's inaugural auction cleared wind at CNY 0.319 per kWh and solar at CNY 0.225 per kWh — and industry observers noted the solar price sat below the level generally thought necessary to finance new projects.⁷ The International Energy Agency subsequently cut its China clean-energy forecast through 2030 by 129 GW.⁷

Implementation is incomplete. As of October 15, 2025, eighteen provinces had finalized their rules, ten had published drafts only, and three — Jiangsu, Tianjin and Tibet — had published nothing.⁷ For a company operating in twenty provinces, that is not a resolved policy environment; it is a patchwork, and provinces have every incentive to design mechanisms that suit local industrial electricity costs rather than generator returns.

Why a 20-province footprint makes this harder, not easier. Diversification usually reduces regulatory risk. Under Document 136 it multiplies the number of rulebooks a generator must master. Each province sets its own quota size, auction design, settlement proportion and treatment of existing projects, which means Datang's renewable revenue will be determined by twenty separate administrative processes running on different timetables. A developer concentrated in one favorable province may well earn better returns than a national operator averaging across good and bad designs. Scale, so often an advantage in this industry, offers little protection here.

Testing management's framing. Here the comparison between what the company says and what independent analysts say is genuinely informative. Datang's public tone about the reform is constructive: the annual report's mitigation language emphasizes optimizing bidding strategy, participating in green power and carbon trading, and extracting environmental value from renewable output.¹ But the same document's risk section is blunter than the strategy section, warning that after Document 136 all new-energy tariffs are marketized, that large volumes of renewable capacity entering the market may drag overall market prices lower, and that green power trading price volatility will hurt the stability of renewable project returns.¹

That is not a company describing a stabilizing reform. That is a company describing margin compression, phrased carefully. Independent analysis reaches similar conclusions from the outside, questioning whether the mechanism delivers the price stability regulators intended.⁸ And the measurable outcomes so far point the same way: Datang Renewable's profits fell in 2025 despite fleet growth, and Datang Power's realized tariff has declined in each of the last several reporting periods.²⁵²⁴

There is a deeper strategic irony worth naming. Document 136 arrived precisely when the Big Five had finished orienting their capital plans around renewables. The policy that made the build-out bankable was withdrawn after the build-out commitments were made. That sequencing is a recurring feature of investing in Chinese regulated industries, and it belongs in any risk assessment of the sector: the rules that underwrite a capital cycle can be rewritten mid-cycle, by the same authority that wrote them, without compensation.

Which makes the question of who is steering the company, and on whose behalf, considerably more than a governance formality.

↑ Back to Top

VIII. Current Management, Ownership, and Capital Allocation Discipline

On June 26, 2026, the board of Datang International met and elected a new chairman: 宋波 Song Bo, aged 53, holder of an MBA and a senior engineer's title, a career man of the Datang system who had served as party secretary and deputy general manager at the group's Yunnan and Gansu power companies and as party secretary and chairman at Datang Jilin.¹⁹ The announcement gave no reason for the change. It simply recorded that 李霄飞 Li Xiaofei would no longer serve as chairman.¹⁹

Li Xiaofei had held the job for eight months. He was elected on October 28, 2025, at 48 years old, arriving from senior roles at Datang Xinjiang Power Generation and, notably, as general manager of the group's coal company.²⁰ He replaced 李凯 Li Kai, who had taken the chair little more than a year earlier and whose disclosed compensation was CNY 280,000 for the year.²¹

Three chairmen in roughly two years, none of the transitions accompanied by a stated cause. It would be a mistake to read this as scandal, and equally a mistake to read it as nothing.

What governance actually looks like here. Datang International sits under 国务院国资委 SASAC, the State-owned Assets Supervision and Administration Commission of the State Council, through its parent. Senior executives at central state-owned enterprises are appointed through a party-managed cadre system and rotated across the group and, at times, across the industry. The chairman is simultaneously the party committee secretary — Song Bo holds both titles.¹⁹

The investment consequence is specific and worth stating without moralizing. There is no meaningful equity-based incentive alignment of the kind Western investors take for granted. A chairman paid a few hundred thousand yuan a year, with no material stock ownership, rotating on a multi-year cadence set by an external appointing authority, is not optimizing for the share price over a decade. He is executing group and state priorities: supply security, capacity targets, employment, decarbonization milestones, and — when policy demands it — absorbing a bankrupt municipal heating company for one yuan.

This should shape how investors read every management statement in this story. The usual analytical tools — does the CEO deliver on guidance, is the narrative consistent, is capital allocated with discipline — remain useful, but they measure institutional behavior rather than individual conviction. The person changes; the institution's incentives do not.

The 2025 governance housekeeping. One structural change did occur during 2025 that deserves a mention: the company completed a board election cycle and abolished its 监事会 supervisory board, amending its articles accordingly, in line with China's revised Company Law which shifts the supervisory function to the board's audit committee.¹ This is a market-wide change rather than a Datang-specific reform, and it modestly concentrates oversight in a board the controlling shareholder dominates.

Capital allocation: the one genuinely encouraging trend. For years, Datang's dividend was an afterthought. For 2023, the company distributed about CNY 139 million — a rounding error against a CNY 1.37 billion profit, and a payout ratio near 10%.²⁸ For 2024 it lifted the dividend to CNY 0.0621 per share, roughly CNY 1.15 billion.²⁸ For 2025, the board proposed a total distribution of about CNY 2.74 billion, or CNY 0.148 per share including an interim payment of about CNY 1.02 billion already made in the autumn — a payout ratio the company puts at 46.91% of profit attributable to ordinary shareholders.²

That is a threefold rise in two years and, importantly, the introduction of an interim dividend where none existed. It is the clearest evidence in this story of a shift in capital-allocation behavior toward minority holders, and it aligns with a broader push by Chinese regulators for listed state enterprises to improve shareholder returns — a campaign the annual report explicitly nods to under the heading of "market value management."¹

Two caveats keep this from being an unambiguous win. First, the payout ratio is calculated after deducting the coupon paid to perpetual bondholders, which flatters it relative to a simple dividend-to-net-profit calculation. Second, the record before 2023 was erratic enough that two good years do not establish a policy. Peers set a demanding benchmark: 华能国际 Huaneng Power International grew 2025 net profit 42.2% to CNY 14.41 billion on revenue of CNY 229.29 billion, and Chinese listed hydro and renewable state enterprises commonly distribute far more of their earnings than Datang has historically managed.¹¹

The credibility test. Has management's story shifted opportunistically with the coal cycle? Reading the filings in sequence, the answer is a qualified no, with one important asterisk. The explanation of the 2021 collapse was straightforward and unhedged: fuel costs rose, the company reported it, and the stated remedy was to lean harder on long-term contract coal and supply-security obligations.³ The 2025 attribution is equally direct — profit rose because coal got cheaper — and the company did not dress the recovery in transformation language.¹¹⁵ The clean-energy narrative has also remained consistent in direction, if not in emphasis.

The asterisk is disclosure quality rather than honesty. Datang holds interim and annual results briefings by telephone; the 2025 interim call was held on August 29, 2025 with then-chairman Li Kai, board secretary and chief accountant 孙延文 Sun Yanwen, and an independent director attending.²² For the 2026 interim results, the company scheduled its call for the morning of August 31, 2026, following publication of the half-year report on the evening of August 28, with investors required to register and submit questions in advance — by August 24, a full week before the call.²³

Pre-submitted questions, vetted in advance, produce a very different information environment from open analyst Q&A. Investors accustomed to unscripted exchanges on Western earnings calls should calibrate accordingly: the live version of this story is not very live, and the most reliable source of insight remains the filings themselves, particularly the risk disclosures that management writes with more candor than the results headlines.

Those filings lead directly to the least glamorous and most consequential part of the story: how all this is financed.

↑ Back to Top

IX. The Balance Sheet: What the Pivot Is Actually Funded On

Every utility is a leveraged bet on the durability of its cash flows. The interesting question is never whether the leverage exists — it always does — but where it is recorded, what it costs, and who is really lending.

At the end of 2025, Datang reported total assets of about CNY 333.53 billion, total liabilities of about CNY 233.95 billion, and an asset-liability ratio of 70.15%, down 0.87 percentage points from a year earlier.² Equity attributable to shareholders was CNY 80.35 billion.² That is a leveraged balance sheet by any standard, though not unusual for a Chinese generator, and the direction of travel is mildly favorable.

The perpetual bond question. Then there is the item that changes how the whole picture should be read. Sitting inside equity, under "other equity instruments," were CNY 46.30 billion of 永续债 perpetual bonds.¹ These are securities with no fixed maturity that, under Chinese accounting standards, may be classified as equity rather than debt because the issuer can defer redemption and, in principle, defer coupons.

Economically, they behave far more like debt. Datang paid CNY 1.55 billion of interest to perpetual holders in 2025, and that payment is deducted before earnings per share, which is why reported EPS of CNY 0.3155 is materially lower than net profit divided by shares outstanding.¹² The company redeemed CNY 2.25 billion of perpetuals during the year to optimize its capital structure.¹⁰

The analytical point is simple and important: an investor who takes the 70.15% asset-liability ratio at face value is understating leverage. Reclassify the perpetuals as debt — as many credit analysts do — and the ratio moves substantially higher, while attributable common equity shrinks toward CNY 34 billion. This is a legitimate accounting treatment, disclosed clearly in the filings, but it is exactly the kind of judgment an investor should adjust for rather than accept.

What the debt actually costs — and this is the genuinely impressive part. Datang's comprehensive financing cost fell to 2.33% in 2025, down 30 basis points year on year and the lowest level of the entire 14th Five-Year Plan period, according to the company.¹ It issued 17 tranches of medium-term notes, ultra-short-term commercial paper and corporate bonds during the year, raising CNY 36.5 billion.¹ The pricing tells the story: mid-term notes issued in 2025 carried coupons between roughly 1.81% and 2.48%, against 2.95% to 3.15% on paper issued in 2022 and 2023, and ultra-short-term paper issued in 2026 priced as low as 1.48%.²

Finance costs consequently fell 16.2% to CNY 4.40 billion even as the asset base grew.¹ The company's disclosed interest coverage ratio improved to 3.63 times from 2.46 times.² Note that different data providers compute coverage differently — some report figures well above this using EBIT and excluding capitalized interest — so investors comparing Datang to peers should insist on a like-for-like definition rather than accepting screen values. On the company's own consistent basis, debt-service capacity improved markedly in one year, driven by both higher earnings and cheaper money.

Falling funding costs are the closest thing to a genuine structural tailwind in this story, and they are not evenly available. Access to sub-2% funding in the interbank market is a function of the parent's AAA credit standing and implied state support, not of Datang's operating performance. Several of its 2025 and 2026 issues were labelled 能源保供特别债 energy supply-guarantee special bonds — an instrument category created for policy purposes — alongside carbon-neutrality green notes and technology-innovation perpetuals.² The state is, in effect, subsidizing the liability side of the transition.

Who is really lending. A significant slice of the financing runs through the family. Under related-party agreements, Datang's subsidiaries kept daily deposits of up to CNY 17.25 billion at 大唐财务公司 Datang Finance Company, and owed it CNY 11.97 billion at year-end.¹ Through the group's capital arm, Datang did CNY 5.23 billion of sale-and-leaseback financing, CNY 4.51 billion of factoring, CNY 1.65 billion of direct leasing and CNY 1.15 billion of entrusted loans during 2025.¹

None of this is hidden — it is disclosed, capped and approved — but an activist investor would note that a listed company financing itself substantially through its controlling shareholder's finance arm has fewer degrees of freedom than one funding itself in open markets, and that deposits placed with a related finance company are an exposure most minority holders never think to price.

The parent's own constraint. China Datang Corporation is rated AAA domestically, but its leverage is rising: the group's asset-liability ratio reached 73.92% at the end of September 2025, up from 70.50% a year earlier following a consolidation of new entities, with net debt to EBITDA around 8.4 times and negative free cash flow.²⁶ A parent under balance-sheet pressure with CNY 56.2 billion of construction budget in flight is a parent with an incentive to move assets — and the funding of them — into listed vehicles.²⁶

Does the new capital earn its cost? This is the question that decides the story, and the honest answer is that the disclosure does not permit a confident conclusion. Datang does not publish project-level returns for new wind and solar. What can be observed: cheap debt at roughly 2%, a national auction system that in its first outing cleared Shandong solar below the level observers considered financeable, and a sister renewables company whose profits fell in 2025 while its fleet grew. Cheap funding raises the odds that incremental renewable investment clears its cost of debt. It says nothing about whether it clears the cost of equity, and the early Document 136 evidence points the wrong way.

Whether that matters competitively depends on what everyone else in the industry is doing — and they are all doing the same thing.

↑ Back to Top

X. Competitive Landscape: Where Datang Sits in the Big Five

For most of the past decade, Chinese financial media treated Datang as the underachiever of the Big Five — slower to recover from the coal shock, thinner on margins, later to clean energy than 华能国际 Huaneng Power International or 华电国际 Huadian International. In 2022, while some peers clawed back toward break-even, Datang was still posting losses.

That characterization is now out of date, and the evidence for saying so is unusually clean, because in the first quarter of 2026 the four largest listed thermal generators reported within days of each other under identical policy conditions.

Huaneng International reported revenue of about CNY 56.78 billion, down 5.9%, and net profit of CNY 4.48 billion, down 9.8%. Huadian International reported revenue of CNY 30.47 billion, down 9.4%, with profit down 9.9% to CNY 1.79 billion. 国电电力 GD Power reported revenue of CNY 39.17 billion, down 1.6%, and profit down 22.8% to CNY 1.40 billion. Datang reported revenue of CNY 30.27 billion, up 0.2%, and net profit of CNY 2.89 billion, up 29.3%.¹³¹⁴

Datang was the only one of the four to grow either line — and it out-earned Huadian, a company of almost identical revenue scale, by more than 60%.¹³ Across full-year 2025, the pattern was the same in relative terms: Datang posted both the smallest revenue decline and the fastest profit growth among the flagship listed companies of the five groups, with its per-kilowatt-hour profit improving by about CNY 0.015 as fuel costs fell.¹⁵ Huadian's 2025 profit, by contrast, rose just 1.4% to CNY 6.07 billion on revenue down 14%.¹²

Now the harder question: what does that prove? Less than it appears. Consider the mechanisms that could explain the gap.

The first is fuel cost management, and here Datang has a genuine, measurable result: its unit standard-coal cost fell 14.8%, and the coal-and-heat segment's profit more than doubled. But every generator in China bought cheaper coal in 2025. The differentiator, if there is one, lies in procurement mix — how much long-term contract coal versus spot, how much imported, how efficiently blended — and Datang's disclosure on this is qualitative rather than quantitative. The company describes strategies ("buy in the off-season, burn in peak season," smart blending, daily monitoring) but does not publish the contract-versus-spot split that would let an outsider verify skill against luck.¹

The second is hydrology. Datang's hydro fleet ran 3,859 hours in 2025, up 343 hours year on year, with Yunnan hydro up more than 1,400 hours and Chongqing up 331.¹ Reporting on the first quarter of 2026 attributed part of Datang's outperformance to favorable water conditions in Chongqing and Yunnan.¹³ Rainfall is not a competitive advantage. It is weather, and it reverses.

The third is portfolio mix — a fleet slightly less coal-heavy than some peers, with a meaningful hydro base and a growing renewables book. This is real, durable and partly the result of deliberate capital allocation. It is also the kind of advantage measured in single percentage points of margin, not in the 30-point profit gaps of a single quarter.

Regional footprint. Datang's thermal concentration sits in the Beijing-Tianjin-Hebei corridor and along the southeast coast.¹ Does the northern position confer an edge? The 2025 regional gross margins argue against it: the Beijing-Tianjin-Hebei region delivered an 11.78% gross margin, while Inner Mongolia produced 26.54%, Sichuan-Chongqing 40.59% and Jiangxi 26.34%.¹ Being close to the capital's demand center means high realized tariffs, but also high coal transport costs, heavy heating obligations and older units. The margin data suggests the northern base is at best neutral, and the company's better economics come from mine-mouth coal in Inner Mongolia and hydro in the southwest.

The evidence bar, applied. Set against peers, virtually every tailwind Datang enjoys is policy-uniform. The capacity payment goes to all compliant coal units. The widened tariff band applies nationally. Long-term contract coal discipline is a state mandate. Document 136 applies to everyone. Low-cost bond funding reflects central SOE status shared by Huaneng, Huadian, GD Power and 华润电力 China Resources Power alike.

That leaves a short list of things that are actually Datang-specific: a somewhat more diversified fuel mix, a hydro book that had a good year, a demonstrated ability to cut unit fuel cost faster than peers in 2025, and the lowest reported funding cost the company has ever achieved. The first is structural. The second is weather. The third is unproven over a cycle. The fourth is real but shrinking as a differentiator, since all central generators are refinancing into the same low-rate environment.

An investor should therefore treat the recent outperformance as evidence that Datang is no longer the laggard — a meaningful update — without concluding that it has acquired a moat. In a policy-uniform oligopoly, relative performance is mostly a function of asset mix, fuel procurement and hydrology, all of which mean-revert. And it is worth remembering that the sector's collective challenge remains formidable: analysis of the Big Five's emissions trajectory has long questioned whether their coal-heavy fleets can peak emissions on the national timetable without stranding capital.²⁹

Which is a natural cue to step back and ask what kind of industry this actually is.

↑ Back to Top

XI. Industry Structure: Porter's Five Forces and 7 Powers, Applied

Strategy frameworks are usually applied to companies with strategies. Applying them to a Chinese state generator is clarifying in a different way: it shows precisely which levers exist and which do not.

Buyer power: near-absolute. Datang sells to provincial grid companies and provincial markets. There is one channel, no substitutes for it, and no possibility of building a direct customer relationship. Even in the marketized world of 2026 — where about 89% of Datang's volume clears through market transactions — the "market" is a provincially designed auction whose rules, participants and price caps are set by regulators.²⁴ The buyer is, in the end, the same state that owns the seller.

Supplier power: high, and historically the company's single largest exposure. Coal producers, many of them also state-owned, supply an input that represents the majority of the cost base. The 2021 episode demonstrated the mechanism in its purest form. Long-term contracts and administrative price guidance moderate the extremes, but the underlying dependency remains.

Rivalry: intense but non-price in the traditional sense. The five groups do not compete on brand or product — a kilowatt-hour is a kilowatt-hour. They compete for construction approvals, provincial quotas, grid connection slots, renewable resource concessions and, increasingly, for market share in auctions where bidding aggressively simply lowers everyone's realized price. Datang's own risk disclosure describes competition driving long-term contract tariffs "toward the low end" and warns that the flood of new renewable capacity entering the market may drag overall prices down further.¹

Barriers to entry: very high, and double-edged. Building a gigawatt of thermal capacity requires billions in capital, land, water rights, environmental permits, fuel logistics and a grid connection agreement. No private entrant will replicate Datang's fleet. But the same barriers that keep others out keep Datang in: 49 GW of coal plant cannot be sold, redeployed or wound down quickly. High entry barriers protect incumbents in growing industries; in shrinking or transitioning ones, they trap capital. The first reduction is now visible — coal capacity edged down from 49,134 MW at the end of 2025 to 48,734 MW by June 2026 as older units retired.¹²⁴

Substitutes: the substitute is inside the house. Wind, solar and storage are displacing coal-fired output, and Datang owns increasing quantities of all three. That is the correct strategic response, but it means the company is funding its own cannibalization with borrowed money — which is only value-accretive if the new assets earn better returns than the old ones. Document 136 made that a genuinely open question.

Through Hamilton Helmer's 7 Powers, the audit is short. Counter-positioning: none — Datang's business model is the industry standard, and it has no structural advantage an incumbent rival cannot copy. Brand: none in any commercial sense; electrons do not carry logos. Network economies: none; a customer gains nothing from other customers using Datang's power. Switching costs: none, because there is no switching — the grid dispatches, and the buyer never chooses a supplier by preference. Cornered resource: partial and weak — hydro sites and prime wind resources are genuinely scarce and Datang holds some, but the allocation of new resources is administrative rather than proprietary. Process power: limited — thermal efficiency varies modestly across operators, and Datang's fleet is credibly modern, with 20 ultra-supercritical units totaling 15,320 MW and all 100 of its in-service coal units upgraded to ultra-low emission standards.¹

That leaves scale economies, and specifically the financial kind. Datang's true power is that it can borrow enormous sums at sub-2.5% because the state stands behind it. In a business where the asset is capital equipment with a 30-year life and the return is a regulated spread, cost of capital is the whole game. It is a real power. It is also one shared with four other companies of similar scale and identical ownership.

The sober conclusion is that this is a utility, not a compounder — a business whose returns are set by policy and commodity cycles rather than by accumulated competitive advantage. That framing is essential to holding the bull and bear cases in proportion.

↑ Back to Top

XII. Bull vs. Bear: The Investment Case

Strip away the narrative and the debate over Datang comes down to a single disagreement: whether the events of 2024 and 2025 changed the character of the business or merely its weather.

The bull case, stated at its strongest.

Start with the floor. Before 2024, a coal plant that did not run earned nothing while still paying for staff, maintenance and debt service. Now it collects an availability fee covering a growing share of fixed costs — a payment scheduled to rise nationally in 2026.⁶ That does not eliminate fuel-price risk, but it changes the worst case. A repeat of 2021's conditions would still hurt; it would hurt from a higher base, with a wider tariff band to absorb some of the shock and better long-term contract coverage.

Second, the earnings mix is genuinely broadening. Hydro, wind and solar contributed just under half of power-segment pre-tax profit in 2025 — a fact that surprises most investors who think of Datang as a coal company. Each additional gigawatt of clean capacity shifts the earnings base toward assets with no fuel bill.

Third, the financing position has improved materially: cheaper debt, better coverage, falling interest expense, and an asset-liability ratio moving in the right direction.

Fourth, capital returns have begun to normalize — a threefold dividend increase in two years, and an interim payment introduced where none existed.

Fifth, the downside is genuinely cushioned by state ownership. Central generators do not fail. When 2021 threatened the sector's solvency, the state changed the rules rather than let the companies collapse. For an equity holder, that is a real, if unquantifiable, floor.

Finally, the operational base is competitive: a modern coal fleet, national coverage across 20 provinces, and — on the most recent evidence — the best profit trajectory among the large listed thermal generators.¹³¹⁵

The bear case, stated with equal force.

The most damaging observation is that 2025's profit surge was a fuel-cost event. Revenue fell. Volumes rose barely. What changed was the price of coal, which management does not control and which its own 2026 risk assessment expects to tighten.¹ Strip out the fuel windfall and the underlying business faces falling realized tariffs — down about 4.8% in the first quarter of 2026 and 3.05% in the first half — with dropping utilization hours as renewables take load.¹⁴²⁴

Second, capital allocation answers to policy, not to shareholders. The clearest evidence is not rhetorical but arithmetic: the acquisition of a negative-equity heating company for a symbolic price, generating a CNY 779 million goodwill impairment that consumed roughly two-thirds of what would otherwise have been a normal fourth quarter.¹¹⁶ A company that can be directed to absorb losses for social reasons cannot be underwritten purely on its earnings power.

Third, related-party history sets a low bar for future injections. The 2017 precedent — cash paid at appraisal premiums of roughly 50% to 80% over book for a collectively loss-making portfolio — is the single best guide to how the next injection may be priced.¹⁸ With the parent's leverage rising and a large construction pipeline in flight, further transfers are a live possibility.²⁶

Fourth, leverage is understated as reported. The CNY 46.3 billion of perpetual bonds carried in equity is the largest single accounting judgment in these accounts, and it is the difference between a 70% asset-liability ratio and something considerably less comfortable.¹

Fifth, the renewables build-out is being executed into a pricing regime nobody has yet seen through a full cycle. The Shandong auction cleared solar below what observers considered financeable; only some provinces had finalized rules by late 2025; and the group's dedicated renewables platform saw profits fall by nearly a third in 2025 despite adding capacity.⁷²⁵

The activist stress test. What would a skeptical investor demand at the annual meeting? Four things, all reasonable and none currently provided. Disclose the return on invested capital of the renewable projects commissioned in the past three years, so minorities can judge whether the pivot creates value. Publish an explicit policy on related-party asset acquisitions — including independent valuation and a minority-approval mechanism — before the next injection rather than after. Quantify the annual cost of the loss-making heating obligation and disclose whether any compensation is received for it, since a roughly CNY 3.5 billion gross loss deserves its own line of explanation. And commit to a multi-year dividend policy, because two good years after a decade of erratic payouts is a trend, not a policy.

So does Datang win from here? The evidenced answer is narrower than either camp would like. Datang wins if coal prices stay moderate, if capacity payments continue to rise, if its funding-cost advantage persists, and if renewable returns stabilize under provincial CfD mechanisms. Three of those four are policy variables set by others. The one genuinely company-specific claim — that Datang manages fuel procurement and portfolio mix better than peers — has one strong year of evidence behind it and no disclosed data allowing verification over a cycle.

The case breaks if coal tightens as management itself warns, if market tariffs keep sliding as capacity outruns demand, or if the parent decides the listed company is the right home for assets it no longer wants.

↑ Back to Top

XIII. Risk Radar

The risks worth tracking here are unusually concrete, because in this industry they arrive through identifiable mechanisms rather than vague macro channels.

Coal price volatility remains the dominant swing factor. The mechanism is well established by now: a rise in the delivered price of standard coal flows into the cost line within a quarter, while output prices adjust slowly, partially and politically. Capacity payments and the modest mining stake blunt the blow; they do not stop it. Management's own forward view for 2026 is the tightest of any risk it discloses, citing production controls, Indonesian export quotas, restricted supply from Russia and Australia, and Middle East conflict pushing international prices higher.¹

Tariff and policy risk now runs in both directions. For a decade the fear was that regulated tariffs would be too low. The new fear is that competitive markets will be worse. With clean capacity growing faster than electricity demand and Document 136 pushing renewable output into those same markets, Datang's realized price has fallen in consecutive periods. Any further reform — a tightening of capacity payment eligibility, a provincial decision to cap market prices during a cost-of-living squeeze — would land directly on earnings.

Refinancing and cost-of-capital risk is currently benign and structurally fragile. Today Datang funds itself near 2%. That reflects abundant domestic liquidity and implied state support, not the credit quality of a leveraged thermal generator. A rise in Chinese benchmark rates, or a narrowing of the implicit guarantee for central SOE subsidiaries, would raise the cost of rolling both conventional debt and perpetuals — and perpetual instruments typically carry step-up coupons that penalize non-redemption.

Execution risk sits in two places. One is the physical build-out: connecting new wind and solar in resource-rich regions where the company itself warns that inadequate grid access and insufficient peaking resources may push curtailment rates higher.¹ Curtailed electricity is capital earning nothing. The other is transactional: each parent-group acquisition carries the possibility of another goodwill charge, and the market has now seen what that looks like.

Hydrology is the underrated variable. Hydro is about 10.7% of capacity but delivered CNY 2.14 billion of pre-tax profit in 2025 — a contribution well above its weight, boosted by strong water in the southwest.¹ Rainfall in the upper Yangtze basin is volatile and, on a multi-decade view, climate-sensitive. A dry year in Yunnan and Sichuan would remove several hundred million yuan of high-margin profit with no offsetting cost saving.

Two second-layer items deserve a mention. First, working capital: accounts receivable stood at about CNY 20.77 billion at year-end 2025, with the top five debtors accounting for CNY 19.65 billion — a concentration that reflects the grid monopsony and, for renewable operators across China, a long history of delayed subsidy settlement.¹ Second, the auditor. 天职国际 Tianzhi International issued a standard unqualified opinion on the 2025 accounts, and no credit rating adjustments were made during the reporting period.² Given the goodwill judgments and the discount rates of 7.13% to 11.01% used in impairment testing, both are worth continuing to watch rather than assume.¹

↑ Back to Top

XIV. KPIs the Article Should Track Going Forward

Most utility scorecards drown in metrics. For this company, three numbers carry nearly all the signal.

First, the fuel-to-tariff spread. Track the unit cost of standard coal delivered into the boiler alongside the average realized on-grid tariff, both of which Datang discloses regularly. Everything else in the income statement is downstream of the gap between them. A falling coal cost with a falling tariff — the 2025 and 2026 pattern so far — can still expand profit, but only while the cost falls faster. Watching the two together, rather than either alone, is the single most efficient way to know whether the earnings recovery has legs.

Second, clean energy's share of segment profit, not of capacity. The capacity ratio is the number management leads with, and it is inflated by the inclusion of gas. The one that matters is what share of power-segment pre-tax profit hydro, wind and solar deliver, and — critically — whether that share holds up as more of the fleet sells into Document 136 market pricing. If clean energy's profit share rises while its capacity grows, the pivot is working. If capacity grows while profit share stalls, the company is converting cheap debt into low-return assets.

Third, the dividend payout ratio, read alongside perpetual issuance. Payout is the cleanest available signal of whether capital allocation is genuinely tilting toward minority shareholders. But it must be read with the perpetual bond balance: distributing more cash to ordinary shareholders while simultaneously issuing equity-classified perpetuals to fund capex is a different act from distributing out of genuine free cash flow. Watch both lines together.

All three are disclosed by the company itself rather than requiring estimation. Unit fuel cost and realized tariff appear in the operating discussion and the quarterly on-grid electricity announcements; segment profit by generation type appears in the annual report's power-segment table; the dividend proposal and perpetual balances appear in the annual results and the equity statement. The discipline is not in finding them but in insisting on all three together, because any one in isolation can be made to tell a flattering story.

↑ Back to Top

XV. Business and Investing Lessons

A regulated business is only as strong as the regulator's willingness to let prices move. This is the enduring lesson of 2021, and it generalizes far beyond China. Any business whose input price is set by a market and whose output price is set by an authority carries an embedded short position that no amount of operational excellence can hedge. Investors routinely underwrite regulated utilities as bond proxies precisely because prices are administered — the 2021 experience shows that administered prices are protective only while the administrator chooses to make them so. The correct question is never "is the tariff regulated?" but "what happens to the tariff when the input cost doubles, and how fast?"

State ownership installs a floor and a ceiling. The floor is genuine: when the sector's losses reached national significance, the rules changed. Few private companies enjoy a regulator who redesigns the market to restore their solvency. But the same ownership that supplies the floor caps the ceiling. A company that can be instructed to acquire an insolvent heating utility, to run capacity below cost for public benefit, or to build ahead of demand for supply security is not optimizing for return on capital. Investors are buying a policy instrument with a share price attached, and should size the position and the expected return accordingly.

Energy-transition capex deserves the same scrutiny as any other capital cycle. The most common analytical error in this sector is to treat renewable investment as self-evidently good because it is green. Capital deployed into wind and solar obeys the same arithmetic as capital deployed into anything else: it creates value only if returns exceed the cost of capital. When the subsidy that underwrote those returns is withdrawn mid-cycle — as Document 136 did — the arithmetic changes without the narrative changing at all. The discipline is to demand project-level returns rather than capacity milestones, and to notice when a company reports the second because it does not wish to report the first.

Finally, read where the risks are written, not where the results are. In this story, the most useful analysis was found in the risk section of the annual report, where the company describes a tightening coal market, falling market tariffs and rising curtailment in plain language — while the results headlines celebrate record profit. Both are true. Only one is forward-looking.

↑ Back to Top

XVI. Epilogue: What to Watch Through 2030

Datang enters the second half of the decade in a fundamentally different position from the one it occupied in the winter of 2021 — profitable, cheaper to finance, less coal-weighted, and paying real dividends. It also enters it with more unresolved questions than at any point in its history.

The first will be answered province by province. Document 136's mechanisms are still being written, and the numbers that emerge from provincial auctions between now and 2030 will determine the economics of every wind farm and solar plant the Big Five commission in that window. If sustainable prices clear at levels that finance new projects, the renewables pivot becomes what management describes. If they clear at Shandong's opening levels, the industry will discover it has built a large fleet of assets earning less than the cost of the capital raised to build them. Nothing in the current disclosure lets an outsider predict which.

The second is a governance question with a financial answer. The parent group is carrying rising leverage, a substantial construction pipeline, and three listed platforms with overlapping mandates.²⁶ Whether further assets flow into Datang International — and at what valuation, and whether minority shareholders get a meaningful vote — is the most consequential capital-allocation decision the company does not control. The 2017 precedent and the 2025 Anhui transaction are the two data points investors have. Both argue for watching the terms closely rather than assuming fairness.

The third is the longest-dated. China's commitment to peak carbon emissions before 2030 and reach neutrality before 2060 — the 双碳 dual carbon goals — implies that the terminal value of 48.7 GW of coal-fired plant is a policy variable, not an engineering one. The capacity payment mechanism currently pays those plants to stay available, effectively converting them from energy suppliers into insurance policies for a grid dominated by intermittent generation. That is a rational transitional design, and it may persist for a decade or more. It may equally be tightened once storage costs fall far enough that batteries can provide the same service more cheaply. The first coal retirements have already begun.

Watch the interim results due in the days ahead, and the ones after that, for the two lines that matter most: what Datang paid for coal, and what it received for power.²³ Everything else in this story is commentary on the distance between them.

References

1. 大唐国际发电股份有限公司 2025年年度报告 (2025 Annual Report) — 大唐国际发电股份有限公司, 2026-03 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

2. 大唐国际发电股份有限公司2025年年度报告摘要 — 上海证券报, 2026-03-28 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

3. 电力燃料费用大幅增加，大唐发电去年亏损92.64亿元 — 21世纪经济报道, 2022-03-30 ↩↩↩↩↩

4. 越过巨亏200亿这道坎 — 界面新闻 ↩

5. 公司概况 (Company Profile) — 大唐国际发电股份有限公司 ↩↩↩↩↩

6. Guest post: China's 'capacity payments' boosted coal-plant revenue by up to 8% — Carbon Brief ↩↩↩↩

7. Analysis: Only half of Chinese provinces finalise key 'Document 136' renewable rules — Carbon Brief ↩↩↩↩↩↩

8. Explainer: How China's renewable pricing reforms will affect its climate goals — Carbon Brief ↩↩

9. China to switch from FITs to market-oriented renewables pricing — pv magazine, 2025-02-12 ↩

10. 大唐发电2025年报解读：归母净利增63.91% — 新浪财经, 2026-03-27 ↩↩↩

11. 华能国际2025年净利增42% 燃料成本下降促火电增利 — 新浪财经, 2026-03-25 ↩

12. 华电国际电力股份有限公司2025年年度报告摘要 — 华电国际, 2026-03-26 ↩

13. 四大电力最新业绩披露：仅大唐发电逆势上涨 — 广东省水力和新能源发电工程学会, 2026-05 ↩↩↩↩↩

14. 大唐发电：一季度归母净利润28.93亿元，同比增加29.26% — 新浪财经, 2026-04-28 ↩↩↩

15. 五大发电集团旗舰上市公司2025年业绩大起底 — 新浪财经, 2026-04-16 ↩↩↩↩

16. 1元接手负资产！大唐发电收购亏损电力企业50％股权 — 每日经济新闻, 2025-10-29 ↩↩↩↩

17. 1元接手负资产！大唐发电子公司收购亏损电力企业其余50％股权 — 证券时报, 2025-10 ↩↩

18. 大唐发电获注181亿电力资产 或为央企重组做铺垫 — 21世纪经济报道, 2017-12-08 ↩↩↩↩↩↩

19. 宋波任大唐发电董事长 — 中国能源新闻网, 2026-06-29 ↩↩↩

20. 大唐发电"换帅"：48岁李霄飞接任董事长 — 腾讯新闻, 2025-10-29 ↩

21. 大唐发电董事长及执行董事李凯离任 — 新浪财经, 2025-10-29 ↩

22. 大唐国际发电股份有限公司关于召开2025年度中期业绩说明会的公告 — 上海证券报, 2025-08-19 ↩

23. 大唐国际发电股份有限公司关于召开2026年度中期业绩说明会的公告 — 上海证券报, 2026-08-18 ↩↩

24. 大唐发电(00991.HK)上半年累计完成上网电量约1,294.693亿千瓦时 — 新浪财经, 2026-07-20 ↩↩↩↩↩

25. 大唐新能源(HK1798) 公司动态与财务数据 — 同花顺金融服务网 ↩↩↩↩

26. 中国大唐集团有限公司2026年度第三期中期票据信用评级报告 — 中证鹏元, 2026-04-07 ↩↩↩↩↩

27. 大唐集团筹划！组织架构或将生变 — 广东省水力和新能源发电工程学会, 2026-01 ↩

28. 大唐发电(601991) 分红融资 — 同花顺金融服务网 ↩↩

29. China's Big 5 power producers face uphill battle in meeting peak emissions targets — S&P Global Commodity Insights, 2021-06-07 ↩