Dongfang Electric: China's Power-Equipment Giant Rides the AI Energy Supercycle

I. Cold Open & Roadmap

Sixty kilometres off the coast of Fujian, in the wind-scoured water of the Taiwan Strait, stands a machine that should not exist yet. Its three blades each stretch 153 metres — longer than a football pitch, heavier than a fully loaded Boeing 747 at 83.5 tonnes apiece — sweeping a circle 310 metres across. The nacelle sits 185 metres above the waves. Rated at 26 megawatts, it is the largest single wind turbine ever erected anywhere on Earth, and it was built to survive a direct hit from a Pacific typhoon.¹²

The company that built it is not Vestas, not Siemens Gamesa, not GE. It is 东方电气 Dongfang Electric, and its headquarters are roughly 1,800 kilometres inland, in Chengdu, in the Sichuan basin — a place chosen not for logistics or talent but because, in the 1960s, Chinese military planners judged it far enough from the coast to survive a war.

In the same twelve-month stretch that the 26MW prototype went up, something stranger happened. In November 2025, three squat, grey, 50-megawatt heavy-duty gas turbines were craned onto flatbeds in Deyang, trucked across China to the port of Lianyungang, and shipped to Kazakhstan — the first time a complete Chinese-designed heavy-duty gas turbine had ever been exported as a finished unit.³ Four months later, in March 2026, the same product line took an order for twenty units from a Canadian customer, worth roughly RMB 4 billion, destined in part for data-centre power supply in North America.⁴ Citi flagged the order to clients on 4 March 2026; the Hong Kong shares jumped more than 13% in a session.⁵

Hold those two images together. A Mao-era machinery complex, founded in 1958 and dispersed inland under a defence-industrial doctrine explicitly designed to make China self-sufficient in the event of nuclear war, is now selling turbines into the supply chain that powers American and Canadian artificial-intelligence infrastructure — the single hottest bottleneck in global energy capital equipment. How does that happen?

That is the story. It runs in four movements.

First, the origin: a state-planned industrial vehicle built to absorb technology, not to earn returns for shareholders — a legacy that still shapes how capital is allocated inside this company today.

Second, the scaling: two decades riding China's electrification, when the country added more generating capacity than any nation in history and three domestic champions were handed the work.

Third, the pivot that actually matters to an investor: the deliberate, decades-long localisation of nuclear island equipment and, later, heavy-duty gas turbines — the two highest-barrier, highest-margin categories in the entire power-equipment world, and the two where Western incumbents had held a closed shop for a century.

Fourth, the present: a five-segment conglomerate that generated roughly RMB 78.6 billion of total operating revenue in 2025 and booked more than RMB 117 billion of new orders, whose shares became a proxy for the "AI needs electricity, China builds electricity machines" trade, ran from RMB 18.29 to RMB 45.38, and then gave back a large chunk of it.⁶⁷⁸

The interesting question is not whether Dongfang Electric has done impressive engineering. It plainly has. The question is whether that engineering has produced a durable economic advantage — evidence of pricing power, share gains, sustainable margin, repeatable export wins — or whether the market has paid a theme-stock premium for a handful of headline units. This piece tests that, using the company's own filings, its order book, its segment gross margins, and what management has actually said when things went wrong.

Start where the machines were first cast: in a river valley in Sichuan, in a country preparing for a war that never came.

↑ Back to Top

II. Origins: Built for a Nation, Not a Market (1958–1990s)

On 13 October 1958, in the town of Deyang in central Sichuan, ground was broken on a hydroelectric equipment factory.⁹ There was no market study behind it. There was no customer list. The Great Leap Forward was underway, China had almost no capacity to build its own generators, and the state had decided it needed one.

Six years later the logic hardened into doctrine. Facing what Beijing read as simultaneous threats from the United States and, after the Sino-Soviet split, the Soviet Union, 三线建设 the Third Front programme began in 1964: a colossal, secretive relocation of China's strategic industry into the mountainous interior, far from any coastline an invader could reach.⁹ Steel mills, aerospace plants, machine-tool works and power-equipment factories were dispersed into Sichuan, Guizhou, Shaanxi and Gansu — often deliberately tucked into valleys, sometimes into caves. Economists have spent decades arguing the Third Front destroyed enormous value by ignoring transport costs and agglomeration. That is probably true. It also built industrial capability in places that would otherwise never have had it.

Dongfang Electric is one of the surviving children of that doctrine. Around 1966 three further plants joined the Deyang works: Dongfang Boiler in Zigong, Dongfang Turbine in Deyang, and an electrical machinery plant in Leshan.⁹ Together they formed something rare in the Chinese economy of that era — a complete power-island manufacturing cluster, capable of making the boiler, the steam turbine and the generator that together constitute a thermal power plant. By 1978, Deyang was recognised as China's third heavy-machinery centre, after Harbin in the northeast and Shanghai on the coast.⁹

That trio was not an accident either. The Chinese state deliberately created three overlapping power-equipment groups — 哈尔滨电气 Harbin Electric in the north, 上海电气 Shanghai Electric on the coast, and Dongfang in the west — and gave each a broadly similar mandate in boilers, steam turbines and generators. This was not competition in the Western sense. It was redundancy engineering applied to industrial policy: if one region were lost, the other two could keep the lights on. The commercial consequence, which persists to this day, is that China's power-equipment market has three domestic champions of comparable scale chasing largely the same customers, which structurally caps pricing power in the commoditised parts of the business.

The other inheritance matters more for investors, and it is worth stating plainly. Dongfang Electric was created as a technology-absorption vehicle for the state. Its purpose was to take foreign designs — Soviet, then Western — digest them, localise them, and eliminate China's dependence on imports. Return on invested capital was not the objective function. Capacity, capability and self-sufficiency were.

That origin explains behaviours that look irrational through a purely shareholder lens and rational through a state-capability lens: a willingness to fund research programmes for over a decade before first revenue, a tolerance for low-margin volume business that keeps factories and skills alive, and a capital structure with a controlling shareholder that answers to 国务院国有资产监督管理委员会 SASAC, the central-government body that supervises China's largest state enterprises. It also explains the governance trade-off that runs through every section that follows: minority shareholders are riding alongside a majority owner whose objectives overlap with theirs, but are not identical to theirs.

By the early 1990s, the state began asking these industrial children to at least look like companies. In December 1993 the Deyang electrical machinery business was restructured into a joint-stock company.¹⁰ What happened next would give a Third Front factory something none of its founders could have imagined: foreign shareholders.

↑ Back to Top

III. Scaling with China's Electrification (1990s–2000s)

There is a particular kind of business luck that comes from being positioned in front of the largest infrastructure build in human history. Dongfang Electric had it.

Between the mid-1990s and roughly 2010, China added electricity generating capacity at a pace no country had ever attempted. Power demand compounded at double digits for years on end. The Three Gorges project alone required a fleet of 700MW-class hydro turbines that had never been built at that scale anywhere. Coal-fired capacity was thrown up across the country in a decade-long construction frenzy. Somebody had to make the boilers, turbines and generators — and China's planners had already decided that somebody would be domestic.

The capital markets arrived just in time to fund it. The company issued H-shares in Hong Kong on 31 May 1994 and began trading there on 6 June 1994 under the code 1072; it then issued A-shares domestically on 4 July 1995 and listed on the Shanghai Stock Exchange on 10 October 1995 under 600875.¹⁰ For a Third Front enterprise to be raising equity from Hong Kong investors within thirty years of being built to survive a Soviet invasion is one of the sharper ironies in Chinese industrial history.

The listings were, in practice, a funding mechanism rather than a change of control. The state retained the majority. What the market provided was capital for capacity — and capacity was the binding constraint of the era. Over the following decade the group's constituent factories expanded, and by the end of the 2000s Dongfang Electric had become one of the largest steam-turbine manufacturers in the world by unit volume, alongside its two domestic siblings. In 2007 the group's listed vehicle was consolidated and renamed Dongfang Electric, replacing the original electrical-machinery identity, with the Chengdu headquarters serving as the organisational centre of a genuine conglomerate spanning thermal, hydro and, increasingly, wind.⁹

It is worth being honest about what this era did and did not prove. It proved Dongfang Electric could manufacture heavy rotating machinery at enormous scale, to schedule, in a domestic market that was effectively reserved for it. It did not prove the company could win business on merit against GE, Siemens or Mitsubishi in an open contest, nor that it could earn attractive returns when the growth stopped.

And the growth did stop, or at least normalise. China's power demand growth decelerated. Coal capacity approached saturation, then overshot it. The steam-turbine business that had carried the company became what it remains today: large, cyclical, competitive and structurally low-margin, with three domestic players bidding against each other for orders from a concentrated set of state-owned utility buyers who know exactly what things cost.

That is the trap the company had to escape. A manufacturer of commoditised heavy equipment in a maturing market has only two exits: consolidate the industry, or move up into categories where the barriers are higher and the competition is thinner. Consolidation was not on offer — Beijing wanted three champions, not one. So Dongfang Electric took the second road, and it took it into the two hardest technical categories in the entire sector.

The first was nuclear.

↑ Back to Top

IV. The Pivot That Matters: Nuclear Localization & the Hualong One Bet (2000s–2020s)

On 23 October 2017, a component roughly the size of a small apartment building was shipped out of a plant in Guangzhou. It was a steam generator — the heat exchanger that sits inside a nuclear island and transfers energy from the intensely radioactive primary loop to the clean secondary loop that actually spins the turbine. This particular unit was destined for the fifth reactor at the Fuqing plant in Fujian, and it was the first third-generation nuclear steam generator that China had designed and built itself.¹¹

To understand why that mattered, you need to understand what a steam generator actually is in economic terms. Think of a nuclear plant as two completely separate water circuits that must exchange heat without ever exchanging a single molecule. The steam generator is the wall between them — thousands of thin alloy tubes, welded to tolerances measured in fractions of a millimetre, inside a pressure vessel that must contain enormous heat and pressure without leaking for sixty years.¹¹ There is no maintenance shutdown that fixes a failed steam generator cheaply. It is, in the most literal sense, a component you cannot afford to get wrong, which is why the global list of firms certified to build them has historically been very short.

For decades that list did not include anyone Chinese. China's early reactors were imported: French designs from Framatome/Areva, later Westinghouse AP1000s, with the highest-value nuclear island components coming from abroad. Every reactor built that way exported margin and, more importantly, kept China's build rate hostage to foreign supply chains and foreign export politics.

The decision to break that dependency was made at state level, not board level, and it played out over roughly two decades. Production of the Fuqing steam generator began in August 2014 at DEC's Guangzhou heavy machinery subsidiary; it took three years to build and qualify.¹¹ That timeline is the entire economic story of nuclear equipment in one data point: three years of work, on one component, for one reactor, by a supplier who had to be certified before the first weld.

What China got out of it was 华龙一号 Hualong One (HPR-1000) — the domestically developed Gen-III reactor design jointly evolved by 中国核工业集团 CNNC and 中国广核集团 CGN, of which Dongfang Electric is a core equipment supplier, providing the steam generator, the pressuriser and the turbine-generator set.¹¹¹² The first Hualong One unit entered commercial operation in 2021, and the design has since become the most widely deployed third-generation reactor platform on the planet by unit count, with a domestic fleet of units in operation, under construction and approved.¹³

The scale of the pipeline behind it is what makes this segment strategically valuable rather than merely prestigious. As of early 2026 China operated 58 reactors totalling roughly 56.4 GW, with a further 33 units and more than 35 GW under construction.¹³ In April 2025 the State Council approved ten new reactors — eight Hualong One units and two CAP1000s — across five coastal sites, an investment estimated at around RMB 200 billion and the fourth consecutive year of ten-plus approvals.¹⁴ On 31 July 2026, the State Council approved eight more units: three pairs of Hualong One reactors at Jinqimen, Taipingling and Zhuanghe, plus two CAP1400 units at Laiyang.¹³

Here is why an investor should care more about nuclear than about the headline revenue it generates. In 2025, Dongfang Electric's nuclear equipment line produced roughly RMB 5.66 billion of revenue, up 16.1% — meaningful, but only about 7% of the group.⁷ Its gross margin, however, was 24.1%, the highest of any product line the company reports, against a blended company gross margin of 15.94%.⁷¹⁵ Nuclear is not where the volume is. It is where the profit density is, and where the customer relationship lasts sixty years rather than six.

Now the necessary scepticism. Management and much of the domestic sell-side describe Dongfang Electric as holding the leading share of China's nuclear equipment market, with industry estimates around 43%.¹² That claim deserves qualification rather than acceptance. "Nuclear equipment" is not one market; it is a bundle of distinct components with distinct suppliers. Dongfang is genuinely strong in steam generators, pressurisers and conventional-island turbine-generators. Shanghai Electric, by contrast, dominates reactor internals and main coolant pumps, with domestic share in reactor internals reported at close to 95%.¹² Harbin Electric competes across much of the same conventional-island territory. A single blended "market share" number for nuclear equipment obscures more than it reveals, and an investor should treat "highest market share in nuclear" as a directional claim about a segment leadership position, not a verified monopoly.

What is not in dispute is the strategic logic. Nuclear equipment is high-barrier, certification-gated, multi-decade business with structurally better economics than coal turbines. Having proved the model once, the company applied the identical playbook — a decade-plus of state-funded R&D aimed at a category Western firms had closed off — to something even harder.

↑ Back to Top

V. Five Segments, One Conglomerate: Mapping the Current Business

Read Dongfang Electric's annual report and you meet five reporting segments, which is at least two more than most investors have patience for. It helps to think of the company not as five businesses but as one core and four satellites orbiting it at various distances.

The core is Clean & Efficient Energy Equipment: nuclear, gas, hydro and thermal/coal. This is the profit engine, the technology franchise, and the reason the company exists. Around it sit Renewable Energy Equipment (wind and solar), Engineering & Trade (turnkey EPC contracting and international trade), Modern Manufacturing Services, and Emerging Growth Industries (hydrogen, storage and marine equipment).

In 2025 the group reported total operating revenue of roughly RMB 78.6 billion, up 12.8%, with net profit attributable to shareholders of RMB 3.83 billion, up 31.1%.⁶ New effective orders reached RMB 117.25 billion, up 15.9% — the second consecutive year above the RMB 100 billion threshold.¹⁵

The order mix tells you where the centre of gravity actually sits. Energy equipment accounted for roughly RMB 78.9 billion of those orders, up 20.3%; manufacturing services RMB 26.0 billion, up 24.2%; and emerging industries RMB 12.3 billion — down 15.4%.⁷ Read that last number carefully. The segment that generates the most excited commentary shrank its order intake by double digits in a year when the rest of the company grew strongly. That is a useful early calibration of how much weight the "emerging growth" story should carry.

Break the revenue down by product line and the picture sharpens further. In 2025 coal-fired equipment generated about RMB 24.5 billion, up 20.9%. Wind generated about RMB 18.2 billion, up 48.3%. Nuclear was RMB 5.66 billion, gas RMB 5.63 billion — down 20.9% — and hydro RMB 3.90 billion, up 36.7%.⁷

Pause on that. In the year the market re-rated this stock as an AI-power and clean-energy play, the largest single revenue line was still coal, and the gas turbine line — the crown jewel of the growth narrative — actually declined. Both facts are entirely reconcilable with the bull case, because gas turbine revenue recognition lags order intake by years and the Canadian and Kazakh orders had not yet flowed through. But an investor buying the narrative should know that in 2025 they were buying a company whose revenue was still overwhelmingly built on conventional thermal and price-war wind, with the differentiated businesses representing a modest slice of the top line.

The margins are where the segments really separate, and they are the single most useful table in the company's disclosure. In 2025, nuclear earned a 24.1% gross margin; coal power, remarkably, earned 20.7%, up 4.3 percentage points year on year; hydro 12.1%, down 5.1 points; gas 12.0%, up 4.7 points; and wind — 2.8%, up 5.1 points from a substantially worse level the year before.⁷

Three conclusions fall out of that immediately. First, nuclear is the quality business and it behaves like one: high margin, stable, slow. Second, the coal business is not a uniform disaster — its margin actually improved sharply in 2025, which says the pricing environment for thermal contracts signed a couple of years earlier had recovered. Third, wind at a 2.8% gross margin is not a business, it is a factory-utilisation programme. At that level, after selling, R&D and administrative costs, the wind line is almost certainly not contributing meaningfully to operating profit, and may be diluting it.

That sets the proportions for everything that follows. Nuclear, gas and hydro get the deepest treatment because that is where the differentiated economics live. Wind gets real but shorter treatment, because scale without margin is not a moat. Hydrogen and marine get a clearly-labelled optionality pass, because that is what the order book says they are.

Start with the engine.

↑ Back to Top

VI. The Core Engine: Nuclear, Gas & Hydro Equipment — Industry Structure and Competitive Position

For roughly a century, the heavy-duty gas turbine has been the most exclusive club in industrial manufacturing. The membership list has essentially been four names: GE (now GE Vernova), Siemens Energy, Mitsubishi Power, and Ansaldo. Not four hundred. Four.

The reason is metallurgy and time. A heavy-duty gas turbine burns fuel and drives a compressor and turbine assembly at temperatures well above the melting point of the alloys the blades are made from. Dongfang's G50 runs a turbine inlet temperature around 1,600°C.¹⁶ The blades survive because they are single-crystal castings, internally hollow, riddled with microscopic cooling channels, and coated in ceramic thermal barriers that hold a film of cooler air against the metal surface. Getting that right is not a matter of reading a paper. It requires an accumulated body of proprietary process knowledge — casting yields, coating chemistry, cooling-hole geometry — that can only be acquired by making thousands of blades badly before you make them well.

That is why the incumbents' position was so secure. Any entrant faces years of negative cash flow, enormous test-rig capital expenditure, and a customer base that will not buy an unproven turbine because a failure means a plant offline. Rational shareholders kill projects like that. Which is precisely the point about who funded this one.

Dongfang Electric's G50 — a 50MW F-class heavy-duty gas turbine, the first with fully Chinese intellectual property — entered commercial service in March 2023 after a development programme the company describes as spanning thirteen years and generating well over a hundred patents.¹⁷ For context on the patience involved: a Western industrial that started that programme in 2010 would have spent the entire 2010s explaining to investors why a project with no revenue deserved another year of funding. Dongfang's controlling shareholder did not have to make that argument. That is the clearest, most testable evidence in this entire story that the state-ownership model can buy something a public-market capital structure struggles to: time.

Then the world changed underneath the product. Global electricity demand growth, largely dormant in developed markets for two decades, reawakened — driven substantially by data centres and AI training and inference clusters. Buyers wanted dispatchable power they could build fast, and gas turbines are the only technology that fits. The incumbents' order books filled and their delivery windows stretched out for years. Whatever one thinks of Chinese equipment, an entrant with available slots suddenly had a genuine commercial proposition: reported delivery cycles for the G50 of roughly thirteen months against multi-year waits from the majors.¹⁶

The commercial proof points arrived quickly. In Kazakhstan's Zhambyl region, three G50 units were supplied for a 50MW-class combined-cycle project — the first overseas application of a Chinese-developed F-class heavy-duty gas turbine, with units loaded in Deyang and shipped out in November 2025 in what was the first complete-unit export of a Chinese heavy gas turbine.³ Then, in early March 2026, the order from a Canadian customer for twenty 50MW gas turbine generator sets, valued at approximately RMB 4 billion, aimed at local power supply, data-centre power and related energy infrastructure, with the first batch of roughly ten units targeted for delivery by end-2026 and the balance in 2027.⁴

That Canadian order is the hinge of the current investment narrative, and it deserves both credit and scrutiny. On the credit side: it is the first time Chinese heavy-duty gas turbine technology has been bought by a customer in a high-income Western market that had every alternative available to it, and the buyer is spending real money. On the scrutiny side: it is one order, from one customer, not yet operating. The gross margins being circulated for it in Chinese brokerage commentary — figures in the 40–50% range — are sell-side estimates, not company-disclosed segment economics, and should be treated as such.¹⁸

Behind the orders, management is building physical capacity, which is the more meaningful commitment. Reported plans take annual gas turbine output from roughly ten G50-equivalent units today to around twenty-five G50 units plus three larger G200-class units by end-2027, and on toward thirty-five G50 plus ten G200 by end-2029 — close to a four-fold capacity expansion, with the 200MW-class G200 targeted for launch in 2027.¹⁸ The board has also approved a first-phase capability-enhancement project at Dongfang Turbine to support it.¹⁵ Capacity investment of that scale before the orders are fully in hand is a genuine expression of confidence — and also a genuine operating-leverage risk if the orders do not materialise.

Now the evidence test the bull case has to survive. What proof exists that a G50 is reliability-competitive with a GE or Siemens machine? The honest answer, as of August 2026, is: very little, because there has not been time to generate it. GE Vernova, Siemens Energy and Mitsubishi sell against fleet databases containing millions of cumulative operating hours across hundreds of installed units, which is what allows them to underwrite long-term service agreements with confidence and to price availability guarantees. Dongfang's fleet is young and small. Its commercial success so far is measured in units shipped, not in years of field performance. Availability, hot-section life, forced-outage rates and overhaul intervals for the G50 fleet are not publicly disclosed in any form comparable to what the incumbents publish.

This matters commercially, not just technically. In heavy gas turbines, roughly the more profitable half of the lifetime economics sits in aftermarket parts and long-term service agreements. An OEM wins those by convincing a customer that its machine will run. Until the G50 has multi-year fleet data, Dongfang is competing largely on price and delivery slot — which are real advantages in a shortage, and considerably weaker ones when capacity catches up. An investor should treat every current gas turbine win as evidence of a delivery-window arbitrage first and evidence of technical parity second, until the fleet data says otherwise.

It is also worth noting that Dongfang is not China's only horse in this race, and not the lead horse in every class. The national 300MW-class F-class heavy-duty gas turbine programme — a different and larger machine — rolled its first prototype off the line in Shanghai Lingang on 28 February 2024 and achieved first ignition in October 2024, under a joint venture led by 国家电投 State Power Investment Corporation with Harbin Electric, Dongfang Electric and Shanghai Electric as partners.¹⁹²⁰ In the largest frame sizes, Dongfang is a participant in a national consortium, not a solo champion. Separately, Dongfang has long been Mitsubishi Power's Chinese licensee partner, with cumulative orders through the partnership reaching 150 units by October 2024 — a reminder that part of its gas turbine business remains a licensed relationship with an incumbent rather than an indigenous challenge to one.²¹

Then there is the coal reality check, which belongs in the same breath rather than in a footnote. Coal-fired equipment remains the single largest revenue line in the company. Its economics are entirely a function of when the contract was signed: these are fixed-price, multi-year orders, so a contract booked in an oversupplied, price-war environment locks in a poor margin that shows up in results two or three years later. Management said exactly this when it explained the first-half 2024 profit decline, attributing part of it to coal power projects sold in the period carrying low gross margins.²² By 2025 the coal margin had recovered to 20.7%, which suggests the worst-priced vintage had rolled off — but the mechanism has not gone away, and China commissioned 78 GW of new coal capacity in 2025, the highest annual level in a decade, with roughly 291 GW still in the construction pipeline.²³ The same overcapacity that generates orders also generates the price pressure that makes those orders unprofitable.

Hydro is the quieter third leg, and it is where the overseas story began. Dongfang supplied the first hydro-generating unit for the 720MW Karot Hydropower Station on Pakistan's Jhelum River, about 65 kilometres from Islamabad — the first hydropower investment project under 一带一路 the Belt and Road Initiative and a flagship of the China-Pakistan Economic Corridor, commissioned in June 2022 with four 180MW turbines expected to generate roughly 3.2 billion kilowatt-hours a year.²⁴²⁵ In 2025 hydro revenue grew 36.7% but its gross margin fell 5.1 points to 12.1%, a reminder that even the flagship businesses are subject to project-mix swings.⁷

The takeaway for the core engine is a split verdict. Nuclear is a proven, high-barrier, high-margin franchise with a state-guaranteed domestic pipeline. Gas is a credible technological breakthrough that has produced early commercial wins but not yet a demonstrated moat. Coal is a large, cyclical, contract-vintage-driven business that giveth and taketh away. Hydro is solid and lumpy. The mix is improving — but it is improving from a base that is still mostly conventional.

Which brings us to the part of the company that grew fastest in 2025 and earned almost nothing for it.

↑ Back to Top

VII. Renewable Energy: Wind, Solar, and a Brutal Margin Environment

The 26MW offshore turbine is a magnificent piece of engineering, and it is worth understanding what it does not prove.

The machine's numbers are genuinely startling. Its blades passed static testing in May 2025, were shipped to the Fujian offshore test base in early August, and the unit was installed for testing in September 2025.¹²⁶ At an average wind speed of 10 metres per second, a single unit is designed to generate around 100 gigawatt-hours a year — enough to supply a small city from one tower. A prototype is slated for the Changle Offshore I (North) wind farm in the Taiwan Strait, alongside eighteen 16MW machines.¹

The engineering logic behind ever-larger offshore turbines is sound. Offshore wind's dominant costs are not the turbine itself but the foundation, the installation vessel, the subsea cable and the maintenance vessel trip — all of which are roughly per-tower costs. Double the capacity per tower and you halve those costs per megawatt. This is why the industry has raced up the size curve for fifteen years.

But there is a difference between a flagpole and a franchise, and the gap between them is visible in one number: 2.8%.

That was Dongfang Electric's wind equipment gross margin in 2025 — on RMB 18.2 billion of revenue that grew 48.3%.⁷ It is worth dwelling on how unusual that combination is. The segment grew revenue by nearly half and still earned a gross margin that would be considered thin for a commodity distributor, let alone a manufacturer of complex machinery. And 2.8% was an improvement of 5.1 percentage points, which means the prior year was worse.

The cause is well documented and industry-wide. China's onshore and offshore wind OEM market is contested by 金风科技 Goldwind, 明阳智能 Mingyang Smart Energy, 远景能源 Envision Energy and a long tail of others, bidding into centralised procurement auctions run by state power groups where price is the dominant award criterion. The result has been years of falling bid prices per kilowatt, passed straight through to OEM margins. When a buyer's tender is scored primarily on price and every bidder can meet the technical specification, the equipment becomes a commodity regardless of how sophisticated it is. That is what a 2.8% gross margin looks like from the inside.

So what is Dongfang actually getting for its wind business? Three things, all real, none of them a moat.

First, factory absorption. A heavy-machinery company with fixed plant and a skilled workforce benefits from volume even at low margin, because the alternative is idle capacity. Wind revenue nearly doubling in two years keeps a lot of steel moving.

Second, a seat at the offshore table. Offshore wind in China is less brutally commoditised than onshore, involves higher engineering content, and is more closely tied to the marine and hydrogen optionality discussed later. A company that can build the world's largest turbine has credible standing in the deep-water segment where the technical bar is higher.

Third, political and strategic alignment. Under 双碳 the dual carbon goals — peaking emissions before 2030 and carbon neutrality before 2060 — a central SOE energy-equipment maker is expected to be present in renewables. Absence would be a strategic liability regardless of margin.

What the wind business is not, on current evidence, is a source of earnings power or competitive advantage. There is no visible pricing premium for the 26MW machine, no disclosed cost position superior to Goldwind's or Mingyang's, and no evidence that Dongfang's scale in wind gives it purchasing or service advantages the leaders lack. The honest framing is that wind is a large, fast-growing, near-breakeven revenue line that the company participates in for strategic and industrial reasons, and that any improvement in its margin — say from 2.8% toward high single digits — would represent a meaningful earnings surprise precisely because nothing in the current data suggests it is coming.

A sceptical investor would go further and ask the diworsification question: is RMB 18 billion of revenue at 2.8% gross margin worth the working capital, warranty exposure and management attention it consumes, when the same engineering talent could be pointed at nuclear and gas turbines earning three to eight times the margin? That question does not have a clean answer at a company whose controlling shareholder has national energy objectives as well as financial ones. But it is the right question, and it is one Western industrial conglomerates have been forced by activists to answer for far less egregious margin gaps.

The offshore business does, however, connect to something genuinely novel — and it happens at sea, which is where the export story picks up too.

↑ Back to Top

VIII. Overseas Expansion & the Belt and Road Angle

For most of its history, Dongfang Electric's international business was straightforward: sell equipment to countries that could not build it themselves, often financed by Chinese policy banks, often as part of a state-to-state relationship. That model built power plants across Asia, Africa and the Middle East. It also produced modest, lumpy, politically contingent revenue.

The Engineering & Trade segment is the vehicle for a more ambitious version of that: exporting complete engineering-procurement-construction packages rather than crated components. The difference matters commercially. Selling a turbine makes you a supplier competing on unit price. Delivering a working power station makes you the integrator who chose the supplier — capturing more of the project value and, critically, locking in the equipment specification.

Karot is the template. Chinese capital financed it, a Chinese developer built it, and Chinese equipment — including Dongfang's first generating unit — went inside it.²⁴ For Pakistan, it delivered 720MW of hydro capacity into a chronically power-short grid. For the Chinese suppliers, it delivered a reference project in a market Western contractors had largely written off as too difficult.

The 2025 numbers show this business is real but still modest. International contracts exceeded RMB 14 billion during the year, about 12% of total new orders, while overseas revenue was around RMB 5.14 billion, or roughly 6.5% of the group total.¹⁵ Management highlighted first-ever overseas contracts in nuclear, in 50MW gas turbines and in pumped-storage units during the year, and describes the company as supplying equipment and services to nearly 90 countries and regions.¹⁵⁶

Read those two percentages together and you get the shape of the export story precisely. Orders at 12% versus revenue at 6.5% means the overseas book is growing faster than the delivered revenue — international business is being won faster than it is being recognised. That is the correct pattern for a genuine internationalisation, and it is the single cleanest quantitative evidence available that the export narrative is more than anecdote. It is also a reminder of scale: even at 12% of orders, more than seven-eighths of this company's demand still comes from China.

The geopolitical framing is where the analysis gets genuinely contested. Western nuclear new-build has spent two decades demonstrating an inability to deliver on schedule or budget — Hinkley Point C and the Vogtle expansion being the canonical examples. Chinese reactor and equipment delivery is marketed globally on the opposite proposition: speed, cost discipline, and a domestic supply chain that has stayed hot because it has never stopped building. There is real substance to that claim; a supply chain that builds ten reactors a year retains skills that one which builds one reactor a decade cannot.

But the comparison is not clean, and an honest analysis should say so. Chinese reactors are built under different regulatory regimes, different labour cost structures, different land acquisition rules and different financing costs than a British or American project. Some of the schedule advantage is manufacturing and construction competence; some of it is simply that the state can clear obstacles a private developer cannot. Investors should credit Chinese suppliers with genuine execution capability while resisting the inference that the same speed and cost would survive transplantation into a Western regulatory environment.

And then there is the risk that cuts in the other direction — the one that the market has largely ignored while celebrating the Canadian order.

A Chinese central SOE, supervised by SASAC, supplying grid-connected generating equipment into North American data-centre infrastructure sits at the intersection of three things Western governments have shown increasing sensitivity about: critical infrastructure, Chinese state ownership, and AI capacity. The equipment is not a semiconductor, and gas turbines are not obviously dual-use in the way that networking gear is. But the security argument writes itself: control systems embedded in generating assets that serve strategic computing infrastructure, supplied by a company answering to the Chinese state. Nothing has been restricted as of August 2026. That is not the same as nothing being restrictable.

The asymmetry is what makes this a genuine risk rather than a theoretical one. Belt and Road markets are additive and defensible; North American orders are the ones carrying the valuation premium, and they are precisely the ones exposed. An investor who is bullish on the Canadian breakthrough is, by construction, taking a position on Western trade policy toward Chinese energy equipment over the next five years.

Before turning to how the company is governed and how it allocates capital, there is one more part of the portfolio to size honestly — the part where the marketing runs furthest ahead of the income statement.

↑ Back to Top

IX. Emerging Growth: Hydrogen and Marine — Real Optionality, Sized Correctly

On 10 December 2025, the China Classification Society issued an Approval in Principle for a floating offshore platform that combines wind power generation with direct seawater electrolysis — producing hydrogen from seawater without a desalination step first. Dongfang Electric's Dongfu Research Institute developed it, working with the team of academician 谢和平 Xie Heping, whose underlying seawater-hydrogen research was published in Nature in 2022. It was, by the certifier's account, the world's first floating wind-to-hydrogen platform to obtain such certification.²⁷

The engineering problem this addresses is worth explaining, because it is elegant. Seawater is a terrible electrolysis feedstock: the chloride ions corrode electrodes and produce toxic chlorine gas, so conventional systems desalinate first, which adds cost, weight and complexity — all of which are punishing on a floating platform hundreds of kilometres offshore. The approach validated here uses a membrane-based process that lets water molecules migrate out of seawater into a clean electrolyte, leaving the salts behind, effectively performing the separation for free as part of the process.

The strategic rationale is equally clear. As China's offshore wind moves into deeper, more distant waters, the cost of getting the electricity home rises steeply — long subsea HVDC cables are expensive and take years to permit and lay. Converting energy to hydrogen at the point of generation and shipping it turns a transmission problem into a logistics problem. The design point cited alongside the certification is a 17MW system with a 262-metre rotor producing roughly 68 million kilowatt-hours a year.²⁷

Now the sizing, which is the whole point of this section. This is pre-commercial. An Approval in Principle is a certifier's judgment that a design concept is fundamentally sound and safe — it is a green light to proceed to detailed engineering, not a commissioned asset producing revenue. There is no disclosed cost per kilogram of hydrogen, no offtake agreement, no commercial fleet. Meanwhile the Emerging Growth Industries segment that houses these ambitions booked RMB 12.3 billion of new orders in 2025 — down 15.4% year on year, the only segment to shrink.⁷

So the correct framing is: real optionality, strategically coherent with the core, and a rounding error on today's earnings. If offshore wind-to-hydrogen scales during this decade, Dongfang Electric will have arrived early with certified technology and the marine engineering capability to build the platforms. If it does not — and the history of hydrogen is littered with technically successful demonstrations that never crossed the cost threshold — the write-off is small relative to the group.

An investor should assign this segment approximately the weight its order book does, which is to say: a modest, non-zero call option, not a growth driver. The far more important question for the next five years is who is making the capital allocation decisions, and whether their track record justifies trusting them with a four-fold gas turbine capacity expansion.

↑ Back to Top

X. Current Management, Ownership, and Capital Allocation

Chinese central-SOE governance is genuinely different from the Western model, and pretending otherwise leads investors to ask the wrong questions. There is no founder here, no charismatic capital allocator, no proxy fight waiting to happen. There is a listed operating company sitting beneath a parent group that answers to the state.

At the top of the listed entity, 罗乾宜 Luo Qianyi serves as chairman, having taken over the group chairmanship in a transition through 2025 after a career that included running 中国机械工业集团 Sinomach, one of China's other large state machinery groups.⁶²⁸ 张彦军 Zhang Yanjun, a Xi'an Jiaotong University alumnus who became general manager of the parent group in March 2024, was appointed president of the listed company on 16 April 2024, and is the executive who fronts the company to investors — he delivered the remarks at the 2025 annual results briefing held in Hong Kong.²⁹¹⁵

That briefing is itself a data point about how this management team engages. More than a hundred institutional investors and analysts attended, including JPMorgan, Citi, HSBC, UBS, Hillhouse and BOC International, with discussion covering gas turbine industry development, the coal power market outlook, hydropower expansion and international market expansion.¹⁵ For a company whose entire founding logic was insulation from the outside world, holding a substantive investor briefing in Hong Kong for global institutions is a meaningful cultural distance travelled.

Ownership. The controlling shareholder, China Dongfang Electric Corporation Group, held 51.37% as of the first quarter of 2026.³⁰ That figure has moved around, and the movements are informative. As of September 2025 the parent held 1,772,966,194 shares, or 52.29%; a placement of 68 million new H-shares at HK$15.92 on 24 September 2025, raising gross proceeds of approximately HK$1.083 billion, passively diluted the parent to 51.27%.³¹ The subsequent drift back up to 51.37% is consistent with the parent buying shares in the open market — modest in scale, but a signal of a parent adding to rather than harvesting its position.

Capital markets activity. Two transactions in 2025 deserve an investor's attention, and they should be read together rather than separately.

In April 2025 the company completed an A-share private placement raising RMB 4.123 billion — described as the largest such placement in the A-share power equipment sector since 2021. Of that, RMB 2.527 billion was used to purchase minority stakes in four subsidiaries — Dongfang Turbine, Dongfang Boiler, Dongfang Electric Machinery and Dongfang Heavy Machinery — from the parent group.³² Then in September the H-share placement raised roughly HK$1.07 billion net, with the company indicating approximately half toward R&D and half toward expanding sales channels.³¹

The subsidiary buy-in is the transaction worth interrogating. Buying out minority interests in operating subsidiaries is genuinely accretive to the listed company's economics: it means more of the profit those factories generate flows to listed-company shareholders rather than leaking to minority holders. That is a real benefit. It was also a related-party transaction, purchasing assets from the controlling shareholder using money raised from public investors. The structure is common in Chinese SOE restructurings and is not evidence of anything improper. But an activist investor would want independent valuation evidence that the price paid to the parent was fair, and would note that minority shareholders' recourse in such a transaction is limited when the counterparty controls more than half the votes. This is the concrete, non-theoretical form that "SOE governance opacity" takes.

Incentives. Dongfang Electric operates a restricted A-share incentive scheme dating from 2019. The clearest evidence that it has teeth rather than being decorative came in 2025, when the company repurchased and cancelled 17,334 restricted A-shares granted to a participant who had become ineligible following an organisational change, reducing total shares from 3,117,499,457 to 3,117,482,123.³³ The absolute amounts are trivial. The behaviour is not: an incentive plan that mechanically claws back shares when a participant no longer qualifies is functioning as designed, and companies that follow through on immaterial clawbacks tend to be the ones that follow through on material ones.

Against that, a small counter-signal. Three senior managers — Wang Jun, Hu Xiukui and Dan Jun — completed a disclosed share reduction plan by early February 2026, selling a combined 30,000 shares for roughly RMB 750,000, and in April 2026 senior vice president Li Jianhua disclosed a plan to sell up to 15,500 shares.³⁴³⁵ These are rounding errors against a company capitalised in the tens of billions and are explicitly attributed to personal funding needs. They are worth noting only for completeness and for the fact that they occurred after the shares had run hard, not as evidence of anything more.

The capital allocation record. Strip away the noise and the defining capital allocation decision at this company was funding the gas turbine programme through more than a decade of losses to reach commercialisation in 2023.¹⁷ There is no other single fact in this story that better distinguishes what this ownership model can do. A shareholder-driven industrial facing quarterly scrutiny would very likely have killed that programme somewhere around year six, when it had consumed enormous capital and produced no revenue. The state-backed structure did not have to. Today that programme has produced an exportable product entering a global shortage.

That is a genuine, testable structural advantage — and it comes with a matching structural cost, which is that the same patience can fund things that should have been killed. Wind at a 2.8% gross margin is arguably the other side of the same coin.

Shareholder returns. For FY2025 the board proposed a cash dividend of RMB 5.30 per 10 shares, totalling approximately RMB 1.833 billion and representing a 47.84% payout of net profit attributable to shareholders — an increase from RMB 4.03 per 10 shares for FY2024.¹⁵ A payout near half of earnings from a company simultaneously funding a four-fold gas turbine capacity expansion is a reasonable balance, and the direction of travel on dividend per share is upward. Notably, Dongfang Electric has grown almost entirely organically rather than through acquisition — there is no meaningful M&A track record to benchmark, which itself distinguishes it from the serial-acquirer industrial conglomerate playbook and removes an entire category of value-destruction risk.

The credibility test. The most useful window into how this management explains itself came when results disappointed. In the first half of 2024, revenue rose 12.21% to RMB 32.93 billion but net profit attributable to shareholders fell 15.52% to RMB 1.691 billion, with gross margin down 2.07 points to 15.36%.²²

Management's explanation named four specific causes, quantified three of them, and deflected none. First, the coal power projects recognised in the period were mostly contracts carrying low gross margins. Second, shares in 川能动力 Sichuan New Energy Power acquired through a non-monetary asset exchange had fallen since the start of the year, producing a RMB 181 million exchange loss, against RMB 109 million of investment income from the prior-year holding — a RMB 290 million year-on-year profit swing. Third, R&D expense rose RMB 277 million year on year. Fourth, foreign exchange losses cost RMB 98 million.²²

That is a good answer. It is specific, arithmetically checkable, and it names an unforced error — the equity stake that fell in value — rather than hiding behind macro conditions. Compare it to the far more common corporate formulation of "challenging market conditions and increased competitive intensity." Two years later, that explanation also proved predictive: the coal margin subsequently recovered to 20.7% as the poorly-priced contract vintage worked through, which is what management implicitly said would happen.⁷ Narrative consistency between the explanation given in a bad half-year and the results delivered two years later is one of the more reliable available proxies for management honesty.

That improvement is visible in the numbers — and so is something else.

↑ Back to Top

XI. The Numbers That Tell the Story — And the Stock's Wild Ride

Start with the five-year arc, because it is the cleanest way to see what has actually happened here. Revenue moved from roughly RMB 36.2 billion in 2020 to roughly RMB 78.6 billion in 2025 — more than doubling in half a decade at a company whose end market, Chinese power generation, is often described as mature.⁶ That is not the growth profile of a sleepy state industrial. It is the growth profile of a company positioned in front of a capacity build-out.

The profit line took a detour on the way. After the first-half 2024 stumble, 2025 delivered net profit attributable to shareholders of RMB 3.83 billion, up 31.1% — with total profit up 23.2% and, most tellingly, net profit excluding non-recurring items up 61.2%.¹⁵⁷ That last figure is the one to focus on. Non-recurring gains and losses — asset disposals, fair value swings, government grants — are noise. Core profit growing at nearly double the pace of headline profit means the operating business genuinely improved rather than being flattered by one-offs.

The interim data supported the same read. First-half 2025 revenue rose 14.03% to RMB 38.15 billion, with net profit up 12.91% to RMB 1.910 billion — steady, unspectacular, and directionally consistent.³⁶

Then the first quarter of 2026 introduced a complication worth being precise about. Revenue grew 5.68% to roughly RMB 17.2 billion — a marked deceleration — while net profit attributable to shareholders jumped 37.41% to RMB 1.585 billion.³⁷⁷ But net profit excluding non-recurring items rose only 11.49%, and the gap was substantially explained by a RMB 325 million year-on-year increase in fair-value gains.⁷ New orders in the quarter grew just 2.34% to RMB 36.6 billion.⁷ Operating cash flow was negative RMB 2.70 billion, though improved by RMB 595 million versus the prior year.³⁷

The plain-English translation: in the first quarter of 2026 the underlying business grew respectably but not spectacularly, order intake growth flattened sharply against a very strong comparison base, and the headline profit number was materially better than the underlying one because of investment gains. Negative operating cash flow in the first quarter is normal for heavy equipment manufacturers, where working capital builds ahead of milestone payments, but the combination of decelerating orders and non-operating profit support is exactly the pattern a sceptical investor should watch.

The stock. Over the twelve months to early August 2026, the A-shares traded between RMB 18.29 and RMB 45.38 — a range in which the high is nearly two and a half times the low.⁸ The shares changed hands around RMB 27.95 in early August 2026, meaning roughly a 38% decline from the peak.⁸

What happened is not mysterious. Dongfang Electric became the cleanest listed proxy for a specific idea: artificial intelligence requires enormous quantities of electricity, dispatchable generating equipment is the global bottleneck, Western OEMs are sold out for years, and here is a Chinese manufacturer with available capacity and a fresh North American order. That is a genuinely compelling narrative, and narratives of that quality attract capital far faster than fundamentals change. Chinese commentary at the time noted the valuation had reached roughly 38 times earnings, described as high within the sector.³⁷

Then it corrected, as theme trades do — not because the thesis was disproved, but because the price had moved much further than the earnings. Foreign institutional interest has continued: JPMorgan added to its H-share position repeatedly through mid-2026, including approximately 1.113 million shares at around HK$21.39 in early July.³⁸ That is evidence of real institutional conviction, though it should not be over-read; a large bank's aggregated holdings include client and market-making positions and are not a clean signal of house view.

The analytical discipline here is to separate two things that moved together. The durable improvement is real: order intake above RMB 100 billion for two consecutive years, core profit growing faster than headline profit, gross margins expanding across four of five product lines, an overseas order book growing faster than overseas revenue. The theme premium was also real, and a portion of it has now been given back. Both statements are true simultaneously.

The KPIs. Rather than tracking twenty metrics, three carry most of the information about whether this thesis is working.

First, new order intake and backlog specifically in gas turbines and nuclear. Total order intake is a poor guide because coal and wind orders can mask everything. What matters is the differentiated lines: whether gas turbine units under contract keep climbing beyond the Canadian, Kazakh, Iraqi and domestic orders already secured, and whether nuclear equipment orders track China's approval cadence. This is the direct read on whether the technology breakthrough is converting into a durable business.

Second, overseas orders as a percentage of total. At 12% of new orders in 2025 against 6.5% of revenue, the internationalisation is real but early.¹⁵ If that order share keeps climbing toward the high teens and revenue share follows, the export story is structural. If it plateaus near 12%, the Canadian order was an event rather than a channel.

Third, blended gross margin. At 15.94% in 2025, the group margin is the single-number summary of the mix shift from low-margin coal and wind toward high-margin nuclear and gas.¹⁵ Every element of the bull case eventually shows up here. If the mix shift is genuine and gas turbine economics are as good as claimed, this number must rise over the coming years. If it does not, the mix shift is not doing the work the narrative says it is.

Which sets up the argument.

↑ Back to Top

XII. Bull vs. Bear: Why Wins From Here, Why It Might Not

The bull case rests on four legs, and the first is the strongest.

Dongfang Electric has achieved genuine technological entry into two categories that were closed for decades. Nuclear steam generators and heavy-duty gas turbines are not markets you enter with capital and ambition; they are markets gated by certification, process knowledge and customer risk aversion. The company got through both gates — the first with a domestically built Gen-III steam generator in 2017, the second with a commercially deployed indigenous F-class gas turbine in 2023.¹¹¹⁷ Whatever one concludes about durability, the entry itself is a fact, and it is rare.

Second, the demand environment. The global shortage of dispatchable generating equipment driven by AI data-centre load growth is not a Chinese narrative; it is visible in the multi-year delivery windows of every Western OEM. Shortages reward whoever has capacity, and Dongfang is adding capacity aggressively while incumbents are constrained.

Third, the export channel. Belt and Road relationships give access to markets where Western suppliers are absent, slow or unwilling — and 2025 proved the channel works for products beyond hydro, with first overseas contracts in nuclear, gas turbines and pumped storage.¹⁵

Fourth, the ownership model. Patient state capital funded a thirteen-year gas turbine programme to completion, and the parent has been adding to its stake rather than reducing it. The incentive clawback and the specificity of management's explanation for the 2024 miss are, on the evidence, governance signals pointing the right way.

The bear case is not the mirror image, and it is more textured than "it's a Chinese SOE."

The order book still contains substantial low-margin conventional business, and the mechanism that damaged 2024 earnings is structural rather than one-off: fixed-price contracts signed during coal overcapacity show up as poor margins years later. China commissioned 78 GW of new coal capacity in 2025, its highest annual level in a decade, with 291 GW still in the pipeline and 161 GW of new proposals in a single year — while coal generation itself declined.²³ Building capacity into falling utilisation is the definition of an environment that produces bad contract pricing.

Wind, at 2.8% gross margin against Goldwind, Mingyang and Envision, is a structurally thin-margin category with no visible path to differentiation. This is a fifth of group revenue producing close to nothing.

The gas turbine moat is unproven. The fleet is young, field data is scarce, and the current advantage is substantially a delivery-slot advantage — which is a function of the incumbents' backlog rather than of Dongfang's superiority. GE Vernova, Siemens Energy and Mitsubishi are all expanding capacity in response to the same demand. When their delivery windows normalise, the Chinese entrant's proposition narrows to price, and price competition against firms with decades of amortised development and installed-base service revenue is a much harder fight.

Governance opacity is real but should be stated precisely. The issue is not that the company misbehaves; nothing in the record suggests it does. The issue is that minority shareholders own a slice of a company whose controlling holder has objectives — energy security, employment, national technology capability — that will occasionally outrank return on capital, and whose related-party transactions cannot be effectively contested.

And the geopolitical tailwind may reverse. The Canadian order that drove the re-rating is precisely the exposure most vulnerable to trade or security restriction.

The activist stress test requires an adjustment, because the standard tool does not apply. With the state holding 51.37%, no conventional activist campaign is possible — no proxy contest, no board seat, no forced spin-off.³⁰ Saying so honestly is more useful than pretending otherwise.

But the questions an activist would ask still have force, and three stand out. Why does a company earning 24% gross margins in nuclear operate a wind business at 2.8% — and what would the group's return on capital look like if that capital were redeployed? What independent valuation supported the RMB 2.527 billion paid to the controlling shareholder for subsidiary minorities? And how much of the reported profit improvement is operating versus investment gains — a live question after a first quarter in which headline profit grew 37% and core profit grew 11%?³²⁷

None of these will produce a campaign. All of them are the right lens, and the third is genuinely actionable for an investor tracking quarterly disclosure.

The sharper question is simpler: is the AI-power-equipment re-rating durable earnings growth or a theme premium? The evidence supports a split verdict. Order intake above RMB 100 billion for two consecutive years, core profit up 61% in 2025, and a real overseas order channel are durable. Thirty-eight times earnings for a heavy-equipment manufacturer with a 15.94% blended gross margin, negative first-quarter operating cash flow and flat order growth was not. Foreign institutional buying in the H-shares tells you sophisticated investors see something here; it does not tell you what price they think is right.

↑ Back to Top

XIII. Frameworks: Porter's Five Forces and 7 Powers, Applied

Frameworks earn their keep when they force you to be specific. Applied to Dongfang Electric's core nuclear, gas and hydro equipment business, both do.

Porter's Five Forces.

Barriers to entry are extremely high — arguably as high as anywhere in industrial manufacturing. Certification alone can take years, capital intensity runs to billions, and the process knowledge is tacit. The evidence for this is the company's own history: thirteen years and a state balance sheet to enter the 50MW gas turbine market, three years to build a single qualified steam generator.¹⁷¹¹

Supplier power is moderate. The critical inputs are large forgings, single-crystal castings and specialty alloys — genuinely scarce capabilities with a limited global supplier base. But Dongfang's group structure is heavily vertically integrated, encompassing its own heavy machinery, boiler and turbine subsidiaries, and the April 2025 transaction increased its economic ownership of exactly those units.³² Integration blunts supplier power.

Buyer power is high and concentrated. This is the force that most constrains returns. Domestic customers are a handful of enormous state power groups and nuclear operators — CNNC, CGN, SPIC and the big five generators — who buy through competitive tenders, know the cost structure intimately, and can play three domestic champions against each other. Coal and wind margins are the direct evidence. Nuclear escapes some of this only because the qualified supplier list is short enough to limit the buyer's alternatives.

Substitution threat is low for the specific function. Dispatchable, high-inertia generating capacity for baseload and grid stability has no near-term substitute at scale; batteries address duration, not the multi-day, multi-season role these machines play. The substitution risk is at the technology-choice level — solar plus storage displacing new gas or coal builds — rather than at the equipment level.

Rivalry is intense but oligopolistic. Domestically, three players of comparable capability. Globally, four heavy gas turbine houses plus a new entrant. Notably, in 2025 Shanghai Electric out-earned Dongfang on revenue — roughly RMB 126 billion versus RMB 77.6 billion — but Dongfang led the sector on net profit at roughly RMB 3.97 billion versus Shanghai Electric's RMB 3.09 billion, on a lower debt-to-assets ratio of 70.39% against a sector average of 72.92%.³⁹ Smaller and more profitable than its largest domestic rival is a genuinely favourable competitive position, even if the blended gross margin of 15.94% still sits slightly below the peer average.³⁹

Helmer's 7 Powers identifies which of these translate into durable excess returns.

Cornered resource is the strongest claim, and it applies to nuclear rather than gas. The certified capability to manufacture Gen-III nuclear steam generators is held by a very small number of qualified firms; certification is not purchasable and cannot be replicated on a competitor's timetable. The 24.1% gross margin in nuclear versus 15.94% blended is the price signal confirming it.⁷¹⁵

Process power — advantage embedded in accumulated manufacturing know-how rather than in a patent — applies to both nuclear components and gas turbine hot sections. It is the classic power of heavy manufacturing, and it is real, but it is also the power most vulnerable to a determined, well-funded challenger, which is exactly what Dongfang itself was.

Switching costs favour incumbency once a utility standardises on a platform: spare parts, trained technicians, service contracts and regulatory qualification all accrue to the installed supplier over a sixty-year asset life. This is a power Dongfang holds domestically and largely lacks internationally, which is the honest limit on the export story.

Scale economies are partial. Manufacturing scale helps; but with three domestic competitors of similar size, no one has decisive scale advantage in China.

What Dongfang conspicuously does not have is network economies, branding in the sense of price premium, or counter-positioning. And the crucial open question is the one the gas turbine business poses directly: is thirteen years of R&D building a new cornered resource, or did it merely close an entry gap that will narrow again as GE Vernova, Siemens and Mitsubishi expand capacity into the same shortage? The distinguishing evidence would be pricing power and repeat orders from customers with alternatives. Right now, the evidence is one Western order and a fast delivery slot. That is a promising start and not yet a power.

↑ Back to Top

XIV. Risk Radar

Five risks are material to the economics of this specific company. Each has an identifiable mechanism, which is what separates a risk from a worry.

Coal and thermal overcapacity. The mechanism is contract vintage. Equipment orders are fixed-price and take two to four years to deliver, so the margin recognised today reflects the competitive environment of two to four years ago. When China's coal build-out runs ahead of demand — 78 GW commissioned in 2025 against declining coal generation, with 291 GW in the pipeline and record new proposals — utilities have leverage over equipment pricing, and that shows up in Dongfang's income statement well after the fact.²³ The 2024 first-half miss was this mechanism in action.²² The 2025 recovery to a 20.7% coal margin shows it is cyclical rather than permanent — and cycles repeat.⁷

Wind OEM price war. Structural margin compression across the Chinese wind OEM industry, driven by price-led centralised procurement, keeps a fifth of revenue at near-zero gross margin.⁷ There is no company-specific fix; the constraint is industry structure, and it will only resolve through consolidation or a change in how state buyers award tenders.

Gas turbine execution and reliability. The most consequential company-specific risk. If early G50 units in Kazakhstan or Canada suffer availability problems, forced outages or hot-section durability issues, the damage extends far beyond warranty cost: it would validate every Western procurement officer's instinct to stay with GE Vernova or Siemens Energy, and it would kill the export narrative that carries the valuation. Fleet data is thin precisely because the fleet is new, so this risk cannot be diligenced away — it can only be observed over time. The company is simultaneously scaling capacity roughly four-fold, which raises manufacturing quality risk during ramp.¹⁸

Geopolitical and export control. This one cuts both ways, which is what makes it interesting. Chinese state backing opens Belt and Road markets that Western suppliers do not reach; it also makes a Chinese central SOE supplying generating equipment into North American AI infrastructure a plausible target for future trade or national-security restriction. The Canadian order is both the highest-visibility win and the most restrictable asset.

Regulatory and policy dependence. The highest-margin segment is the one least driven by market demand. China's nuclear build-out pace is set by State Council approvals within five-year plan frameworks — ten reactors in April 2025, eight more on 31 July 2026.¹⁴¹³ Those approvals have run at ten-plus units annually for several consecutive years, which is a favourable environment, but the cadence is a policy variable, not a demand variable. A slowdown in approvals under the 15th Five-Year Plan would hit Dongfang's best business directly, with no market mechanism to offset it.

Two categories deliberately excluded: generic cybersecurity and inflation risk. Neither is a distinguishing driver of this company's economics relative to the five above, and listing them would dilute rather than inform.

One second-layer observation worth flagging: with a debt-to-assets ratio of 70.39%, the balance sheet carries substantial leverage — though for a heavy equipment manufacturer with large customer advances and progress billings, a high reported liability ratio partly reflects contract liabilities rather than financial debt.³⁹ Combined with negative first-quarter operating cash flow, it means working capital discipline during the gas turbine capacity ramp is worth monitoring.³⁷

↑ Back to Top

XV. Durable Business & Investing Lessons

Three lessons generalise beyond this company.

Patient capital can buy things public markets struggle to fund — and the bill comes due elsewhere. The single most important asset Dongfang Electric owns today is a gas turbine platform that took over a decade of funded development before generating meaningful revenue.¹⁷ No listed Western industrial with quarterly earnings pressure and an activist on the register would likely have survived that programme intact. The state-backed structure could, and did. That is a genuine structural advantage of this ownership model, and investors should credit it rather than dismissing SOEs as uniformly inefficient. The trade-off is symmetrical: the same patience that funds a thirteen-year turbine programme also tolerates a wind business earning a 2.8% gross margin, and the same controlling shareholder that provides the patience is the one whose related-party transactions cannot be challenged.⁷ You do not get to choose one side of that coin.

Second-mover catch-up in complex capital equipment is achievable but measured in decades, not product cycles. Dongfang's nuclear localisation took roughly twenty years from strategic decision to indigenous Gen-III steam generator. Its gas turbine journey took thirteen years to commercialisation in 2023 and another two-and-a-half to first Western order in 2026.¹¹¹⁷⁴ The practical implication is a discipline about time horizons. When the same company presents a certified but pre-commercial floating wind-to-hydrogen platform, the correct base case is not "revenue in three years" but "this is a call option that may pay off across a decade, if at all."²⁷ Applying the company's own demonstrated development timelines to its newest technologies is the most useful calibration tool an investor has here.

Segment mix matters more than headline growth. This is the practical takeaway. Dongfang Electric's 2025 revenue grew 12.8%, which tells you almost nothing about the quality of that growth.⁶ What matters is that nuclear earned 24.1% gross margins while wind earned 2.8% — a nearly nine-fold spread within one company.⁷ The same top-line number could represent a company shifting decisively toward its high-margin frontier businesses or one whose growth is coming disproportionately from its worst. In 2025 wind grew 48.3% and nuclear grew 16.1%, meaning the fastest-growing line was also the least profitable — while the blended margin still improved because coal and gas margins recovered.⁷ Only by decomposing it do you learn what actually happened. For any diversified industrial, the mix shift, not the growth rate, is the value driver.

The final lesson is about narrative discipline. Dongfang Electric in 2026 is neither the pure AI-power play that drove the shares to RMB 45 nor a lumbering state coal-turbine maker. It is a conglomerate in the middle of a genuine but incomplete mix shift, with one proven high-barrier franchise, one promising and unproven one, and two large conventional businesses that pay the bills and periodically hurt. Holding all four of those facts at once is harder than holding a slogan, and considerably more useful.

↑ Back to Top

XVI. Epilogue: What to Watch Next

Four things will settle the argument over the next several years, and none of them require guessing.

The 15th Five-Year Plan's nuclear cadence. China approved ten reactors in April 2025 and eight more on 31 July 2026, extending a multi-year run of double-digit annual approvals.¹⁴¹³ Dongfang's most profitable business is a direct function of that cadence continuing. If the plan sustains or raises the pace, the nuclear equipment pipeline is visible for a decade. If it slows, the highest-margin line slows with it, and there is no market mechanism to compensate.

Whether the North American gas turbine breakthrough repeats. The Canadian order for twenty units is either the first of a series or an isolated event.⁴ The tell will be a second and third Western or allied-market order — and, just as importantly, the absence of any restriction on the first. Watch the delivery of the initial batch targeted for end-2026 as closely as the order flow: on-time delivery and clean commissioning would be the first real operating evidence, as distinct from commercial evidence, that this product works in a demanding market.

The margin trajectory as legacy contracts roll off. The blended gross margin of 15.94% is the summary statistic for the entire mix-shift thesis.¹⁵ Nuclear, gas and hydro carry higher margins than coal and vastly higher than wind. If the differentiated businesses genuinely grow faster than the conventional ones, this number must rise. Several years of a flat blended margin would mean the mix shift is being offset by price competition in the very categories the story depends on.

Whether the re-rating proves durable. The shares travelled from RMB 18.29 to RMB 45.38 and back to roughly RMB 28 within a year.⁸ That volatility is not incidental to the story; it is the market publicly debating exactly the question this article has tried to test — durable earnings growth or theme-stock premium. The resolution will come from the first three items on this list, and it will come slowly, in order books and gross margins rather than in headlines.

There is a certain symmetry in where this leaves things. A factory built in a Sichuan river valley in 1958, explicitly to make China independent of foreign technology in the event that the world turned hostile, now finds its most exciting product exposed to precisely the risk that the world turns hostile in the other direction. The Third Front planners were solving for a war that never arrived. Their successors are selling into a computing boom nobody could have imagined. Whether the machines they build turn out to be a durable franchise or a well-timed shortage trade is a question the fleet data, the order book and the gross margin line will answer over the next several years — and not before.

References

1. Dongfang installs 26 MW offshore wind turbine at Dongying test base — Windtech International ↩↩↩

2. World's First 26 MW Offshore Wind Turbine Rolls Off Production Line — Offshore Wind, 2024-10-14 ↩

3. China's Self-Developed F-Class 50 MW Heavy-Duty Gas Turbine Enters Overseas Market for the First Time — New Energy Era ↩↩

4. 东气G50燃机斩获海外大额订单，订单达到20台 — 新浪财经, 2026-03-09 ↩↩↩↩

5. 东方电气盘中逆市涨超13% 花旗称公司已取得20台50MW燃气轮机订单 — 新浪财经, 2026-03-04 ↩

6. 东方电气股份有限公司2025年年度报告摘要 — 上海证券报, 2026-04-01 ↩↩↩↩↩↩

7. 东方电气(600875)2025年报及2026年一季报点评：业绩高增长 规划进行自主燃机扩产 — 中财网研报, 2026-05-12 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

8. Dongfang Electric Corp Ltd Class A (600875) stock price and 52-week range — Investing.com ↩↩↩↩

9. 从"三线建设"中扎根，拓宽电力盘面，东方电气何以多点开花？ — 界面新闻 ↩↩↩↩↩

10. 东方电机(600875) 证券基本资料/公司资料 — 新浪财经 ↩↩

11. Dongfang Electric Delivers China's First Third-Generation Nuclear Steam Generator to Hualong One Reactor — Yicai Global, 2017 ↩↩↩↩↩↩↩

12. 核电崛起：2.5万亿利益链谁主沉浮？ — 中国核能行业协会 ↩↩↩

13. China approves construction of eight more reactors — World Nuclear News, 2026 ↩↩↩↩↩

14. Ten new reactors approved in China — CNNC, 2025-04-29 ↩↩↩

15. 东方电气召开2025年度业绩说明会 — 东方电气股份有限公司 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩

16. Breakthrough in High-End Overseas Expansion: China's Indigenous G50 Gas Turbine Lands Major Gas Turbine Sales in Canada — Bowzon Turbine, 2026-03-07 ↩↩

17. China's first F-class heavy-duty gas turbine put into commercial use — CGTN, 2023-03-10 ↩↩↩↩↩↩

18. 东方电气燃气轮机业务分析 — 新浪新闻 ↩↩↩

19. 国之重器！300兆瓦级F级重型燃气轮机首台样机总装下线 — 新华网, 2024-02-29 ↩

20. 我国自主研制的300兆瓦级F级重型燃气轮机在上海点火成功 — 中国政府网, 2024-10 ↩

21. Orders for Gas Turbines through Partner Firm in China Reaches 150 Units — Mitsubishi Power, 2024-10-16 ↩

22. 东方电气：2024年上半年净利润16.91亿元 同比下降15.52% — 新浪财经, 2024-08-30 ↩↩↩↩

23. Built to peak: coal power expansion runs out of room in China — Centre for Research on Energy and Clean Air ↩↩↩

24. First Hydropower Station of China-Pakistan Economic Corridor Starts Operation — SASAC, 2022-06-01 ↩↩

25. 720 MW Karot hydropower project in Pakistan begins operating — Renewable Energy World ↩

26. Dongfang Installs 'World's Largest Single-Capacity' Offshore Wind Turbine for Testing — Offshore Wind, 2025-09-03 ↩

27. DEC secures world's first CCS AIP for wind-powered hydrogen production platform — Shanghai Metal Market, 2025-12 ↩↩↩

28. 东方电气集团董事长，悄然换人！ — 证券时报 ↩

29. 西安交通大学校友张彦军任中国东方电气集团总经理 — 西安交通大学校友网, 2024-03 ↩

30. 东方电气股份有限公司2026年第一季度报告 — 上海证券报, 2026-04-30 ↩↩

31. 东方电气股份有限公司关于股东权益变动的提示性公告 — 上海证券报, 2025-09-25 ↩↩

32. 东方电气完成41.23亿元定增 进一步夯实能源装备领域领先地位 — 中证网, 2025-04-14 ↩↩↩

33. Dongfang Electric to Repurchase and Cancel Restricted Shares — TipRanks ↩

34. 东方电气三位高管减持计划实施完毕 合计减持3.0万股套现约75万元 — 新浪财经, 2026-02-09 ↩

35. 东方电气两高管拟减持合计不超1.63万股 — 新浪财经, 2026-04-09 ↩

36. 东方电气(01072.HK)2025年上半年净利润同比增长12.91% — 瑞财经 ↩

37. 东方电气2026年一季度营收174.7亿元，归母净利润同比增37.41% — 经济观察网, 2026-05-06 ↩↩↩↩

38. JPMorgan increased its stake in Dongfang Electric (01072.HK) by 1.113 million shares — Futu News, 2026-07 ↩

39. 东方电气的前世今生：2025年营收775.83亿居行业第二，净利润39.66亿领先同行 — 新浪财经, 2026-04-25 ↩↩↩