Doosan Enerbility (034020): Half Reactor Maker, Half Bobcat

Doosan Enerbility is the only company in South Korea that can forge and machine the inside of a nuclear reactor. It is also, on its own numbers, mostly a bulldozer company — one that keeps its books in US dollars. Both of those things are printed in the same half-year filing, 427,562 characters of it in Korean, alongside a customer list on which a single unlisted buyer accounts for more than half the contract book. This is what the company actually is, read from its own disclosures.

Method note on currency. Every money figure in this report is shown in US dollars, converted once at ₩1,345.06 / $1 — the USD/KRW rate our own price history records for 7 September 2026, the same date as every share price and market capitalisation below. Share prices are the exception to the dollar rule: they are shown in won first, because that is what you will see on a Korean quote screen, with the dollar equivalent in brackets. Where a Korean filing states its own dollar amount and its own exchange rate, we use the filing’s figures and do not re-convert.

Every conversion here is checkable. We print the won figure beside or beneath each dollar figure, so any number in this report can be re-derived by dividing that won amount by 1,345.06. We spell this out because a currency conversion is the one class of figure our pre-publication checks cannot verify: it appears in no filing, so there is nothing for a machine to match it against.

🏢 What Is Doosan Enerbility, and What Does It Actually Make?

Doosan Enerbility is the only company in South Korea that manufactures nuclear steam supply systems — reactor pressure vessels, steam generators and reactor coolant pumps — and it is also, through a 48.23% stake it consolidates in full, the parent of the world’s best-known compact-equipment brand, Bobcat.

Founded on 20 September 1962 as Hyundai Yanghaeng, taken into state hands in October 1980 and renamed Korea Heavy Industries & Construction, privatised into the Doosan group in March 2001 and renamed again in March 2022, the company is headquartered at 22 Doosan Volvo-ro, Seongsan-gu, in Changwon, South Gyeongsang province. It listed on the Korea Exchange on 25 October 2000. Chairman and CEO Park Jeong-won runs it alongside COO Jeong Yeon-in and CFO Park Sang-hyun. The parent company employed 6,233 people at the FY2025 year-end — 4,806 permanent and 1,427 on fixed-term contracts. For general background and an entity anchor, see Wikipedia’s Doosan Enerbility page; every figure in this report comes from the filings, not from there.

(Sources: DART half-year report rcpNo 20260814003782, “I. Overview of the Company” and consolidated note 1; employee counts from the FY2025 annual report, rcpNo 20260320001246.)

The description in the filing’s own words is broader than “nuclear.” Doosan casts and forges the heavy steel that reactors are made from, designs and builds combined-cycle gas plants, supplies steam turbines and generators, runs desalination and water-treatment plants, executes power-plant EPC contracts end to end, and makes offshore wind turbines. On top of that sit 61 consolidated subsidiaries at 30 June 2026 — two of them separately listed in Korea — of which the two that matter are Doosan Bobcat (48.23%) and Doosan Fuel Cell (30.33%).

📊 Where Does the Revenue Actually Come From?

More than half of it — 52.25% in the first half of 2026 — comes from Doosan Bobcat’s compact construction equipment, and only 45.49% from the power, nuclear and turbine business the stock is famous for.

Basis: DART half-year report, rcpNo 20260814003782, section II “Revenue performance” (매출실적). “H1 2026” is the six-month cumulative column (제64기 반기), not the three-month column. FY2025 and FY2024 are full audited years. The filing states these are net segment revenues after consolidation adjustments, and the percentages are the filing’s own. Won amounts before conversion: Bobcat ₩4,694,956m / ₩8,791,773m / ₩8,550,978m; Enerbility ₩4,087,322m / ₩7,788,916m / ₩7,292,841m; Fuel Cell ₩172,667m / ₩418,124m / ₩344,341m; total ₩8,985,938m / ₩17,057,876m / ₩16,233,055m. Converted by us at ₩1,345.06/$1 throughout, so year-on-year movements are won-basis movements.

Two things follow from that table, and neither survives a summary.

First, the mix has barely moved. Bobcat was 52.68% of revenue in FY2024, 51.54% in FY2025 and 52.25% in the first half of 2026. Whatever the nuclear order book is doing, it has not yet changed what this company mostly sells.

Second, and this is the detail that reframes the American story: Doosan Bobcat reports its own consolidated accounts in US dollars. DART returns Bobcat’s FY2025 balance sheet with the currency field set to USD — total assets of $8,558,454 thousand, against Doosan Enerbility’s own statements in won. A dollar-functional subsidiary inside a won-reporting parent has a mechanical consequence you can see in the comprehensive income statement: of Doosan’s $597.4M of total comprehensive income in H1 2026, $316.3M — 52.95% — was foreign-currency translation gain, not operating performance. (Source: DART fnlttSinglAcnt API, corp code 01032486, FY2025 consolidated, currency field USD; and rcpNo 20260814003782, consolidated statement of comprehensive income, six-month cumulative: total ₩803,531m of which ₩425,465m foreign operations translation.)

Splitting the same revenue by geography rather than by segment makes the point a second way. The Americas were 40.8% of segment-note revenue before consolidation eliminations in H1 2026 — and 93.8% of that was Bobcat.

⚠️ Different basis from the segment table above — do not mix the two. This is the consolidated segment note’s revenue from contracts with customers, before consolidation eliminations, six-month cumulative, which totals ₩9,060,113m. The revenue-mix table earlier in this report uses the filing’s separate “revenue performance” schedule, which is net of consolidation adjustments and totals ₩8,985,938m — the same figure as the consolidated income statement. The ₩74,175m gap is intersegment elimination. Both tables are correct; they answer different questions. Source: rcpNo 20260814003782, consolidated note 20, “disaggregation of revenue from contracts with customers.” Columns may not sum exactly because each cell is rounded after conversion at ₩1,345.06/$1.

Now put a second window on it. A single half-year is a snapshot; the change is the story.

Same source and basis as the table above (rcpNo 20260814003782, consolidated note 20, six-month cumulative, before eliminations). Won amounts: Europe ₩197,470m → ₩614,565m; Americas ₩133,487m → ₩224,153m; Middle East ₩1,722,759m → ₩1,582,513m; Korea ₩1,294,037m → ₩1,389,499m; Asia ₩361,655m → ₩406,742m. Percentage changes are won-basis and computed by us; both periods are converted at the same ₩1,345.06/$1 so the dollar columns move with the won, not with the exchange rate.

Read the two windows together and the direction is unmistakable. The European share of the Enerbility segment’s revenue nearly tripled, from 5.32% to 14.57%, in twelve months — the Czech contract beginning to be recognised. The Americas went from 3.60% to 5.31%: growing, but from a base so small that even after a 67.9% jump it is a third the size of Europe. The power business is, in revenue terms, a Korean and Middle Eastern company acquiring a European line.

The segment table and the geography table agree with each other, which is reassuring but not independent — both are built from the same revenue ledger. So here is a third cut, from a different part of the document entirely. The word 미국 (United States) appears 63 times in the 427,562-character half-year report. We read all 63 and classified each one by which business it belongs to. The distribution tracks the revenue split almost exactly.

Classification performed by us on the full text of DART half-year report rcpNo 20260814003782, retrieved via the DART document API and read in position order. Every one of the 63 occurrences was assigned to exactly one row; 34 + 21 + 4 + 4 = 63. This is our reading of the filing, not a figure the filing publishes.

Read that table next to the revenue tables above and it says the same thing twice. The overwhelming majority of the American content in this filing is a compact-equipment business: dealer networks, emissions rules, customs paperwork and an intra-group merger. Of the 21 mentions belonging to the power segment, 17 are entity names in a schedule — 13 in the consolidated subsidiary list and 4 repeating those same entities in the parent-only investment list. The two that describe a live commercial position are about gas turbines — the filing lists “USA (4.6GW)” in its gas-turbine supply record and calls entry into “the United States, the home market of the gas turbine” a proof of the technology. That is a real achievement, and it belongs to the turbine business rather than the nuclear one.

🗓️ How Did a 1962 Machinery Maker Become Korea’s Only Reactor Builder?

Because for twenty years it was not a private company at all — it was the Korean state’s heavy-industry champion, nationalised in 1980 and only sold to the Doosan group in 2001.

The short version, which is all a shareholder needs: founded 20 September 1962 as Hyundai Yanghaeng, taken into state ownership in October 1980 under the heavy and chemical industry restructuring and renamed Korea Heavy Industries & Construction, listed on the Korea Exchange on 25 October 2000, privatised into the Doosan group in March 2001, and renamed Doosan Enerbility by resolution of the annual general meeting on 29 March 2022. Doosan Corporation has been the largest shareholder since 26 March 2002 and, in the filing’s own words, there has been no change since. (Source: DART half-year report, rcpNo 20260814003782, sections I.1 and I.2.)

That single fact — two decades as a state enterprise — is why a privately held company is Korea’s sole nuclear equipment maker, and it has a consequence that runs through the rest of this report. Doosan does not win nuclear export contracts. The Korean consortium wins them, and Doosan supplies the consortium.

One structural change since is worth recording because it affects what you own: in January 2025 the Czech subsidiary Doosan Skoda Power converted to a joint-stock company and listed on the Prague Stock Exchange, with Doosan retaining 67.00%.

🌍 Is Doosan Actually a Global Leader, or Just a Korean One?

Korean, on the company’s own published figures: Doosan holds 100% of Korea’s domestic nuclear primary-equipment market but only 3.0% of the overseas one, and its overseas turbine shares are single-digit — 9.3% in steam turbines and 9.0% in gas turbines.

Market-share claims about this company get repeated in English coverage until they acquire a precision they never had, so we went to the source. The half-year report (rcpNo 20260814003782) uses the word 점유율 — market share — 21 times, and prints four quantified tables: power and desalination equipment, castings and forgings, construction, and fuel cells. Three of those four belong to the business the stock is actually valued on. The company declines to give a figure only where it says it cannot see competitors’ sales — Bobcat’s compact equipment, industrial vehicles and hydraulics — and, separately, for its golf-course operation.

Here is the whole disclosure, transcribed. Read the basis column before the numbers; the domestic and overseas rows answer different questions, and the first column is a half-year figure against two full years.

Source: DART half-year report, rcpNo 20260814003782, section II.4 “Business overview”, sub-section “(2) Market share etc.” for the Doosan Enerbility segment and the equivalent sub-section for the Doosan Fuel Cell segment. Every percentage above is the filing’s own; we have computed none of them and have translated the basis notes rather than paraphrasing them. The construction row’s own underlying figures, in the filing’s own unit of 억원 (hundred-million won), are 323,492 of domestic public-sector contract value against 658 of Doosan public-sector orders in H1 2026 — $24.050bn against $48.9M when converted by us at ₩1,345.06/$1. The filing sources the fuel-cell denominator of 1,477MW to Korea’s Electric Power Statistics Information System and prints the share as 50%; 731 ÷ 1,477 is 49.49%, so read the 50% as the filing’s own rounding. Doosan Fuel Cell was 1.92% of group revenue in the same period. The filing omits a share figure for Bobcat’s compact equipment, industrial vehicles and hydraulics, stating in each case that it cannot verify competitors’ sales.

Two things in that table are worth more than the headline number.

The domestic turbine rows have swapped places. Doosan’s share of the domestic steam-turbine market went 98.2% → 43.4% → 0.0% across the three periods, while its domestic gas-turbine share went 52.0% → 53.4% → 96.5%. We cannot tell you from this table which of those is a competitive event and which is an ordering-calendar artefact: the basis is “primary equipment ordered by KEPCO, KHNP and IPPs,” so a 0.0% appears both when a supplier loses every award and when there are no awards to win. The filing marks the difference in one place and not the other — the desalination row carries an explicit note saying there were no biddable orders in the period, and the steam-turbine row carries no such note — which tells you the company does flag order droughts, but it does not license us to conclude that this one was a loss. It is a question to put to management, not an answer.

The second is that the fastest-moving business here is not the nuclear one. Doosan’s own large gas turbine entered commercial operation at Gimpo in July 2023, and the filing counts 24 units contracted since — 1 in 2023, 4 in 2024, 7 in 2025 and 12 in the first half of 2026 alone (5 domestic, 7 overseas) — for total supply capacity above 10.7GW, including a US project of 4.6GW. In offshore wind it has confirmed supply of 1,203MW of its own 8MW and 10MW turbines across seven Korean projects, plus manufacturing participation in two Siemens Energy projects. Those are the filing’s own totals, printed as totals; we have not re-summed the individual projects behind them.

Read the table honestly and the “global leader” framing narrows to something more specific and more useful: Doosan is a monopoly supplier inside one national programme, and a credible but small entrant in two global equipment markets. 100% at home and 3.0% abroad in nuclear is that sentence in two numbers.

🤝 Who Actually Signs Doosan Enerbility’s Nuclear Contracts?

Korea Hydro & Nuclear Power signs every one of them — all six nuclear contracts in Doosan’s disclosed book, worth $8.33bn, name KHNP as the counterparty, including both Czech Dukovany awards.

Korean listed companies must disclose material single supply contracts individually, and then report progress on each one in every periodic filing. The half-year report carries that progress schedule with a base date of 30 June 2026. It lists 22 contracts, of which 20 disclose a counterparty and a value and two are withheld under a confidentiality reservation. Here is the nuclear half of it.

Source: DART half-year report, rcpNo 20260814003782, section XI.1.(나), “Progress on single supply-contract disclosures,” base date 30 June 2026. Counterparty is the filing’s 계약상대방 field, printed as 한국수력원자력(주) on all six rows. The “project location” column is ours, not the filing’s — the progress schedule carries no location field, so we have taken each project’s country from the contract name and, where it is not in the name, from the individual single supply-contract disclosure filed when the deal was signed. That limitation is the subject of the section below. Won contract values before conversion: ₩4,928,968m, ₩711,106m, ₩2,338,118m, ₩1,091,844m, ₩532,040m, ₩1,595,395m.

Our arithmetic, computed separately from the filing: those six rows sum to ₩11,197,471m, which at ₩1,345.06/$1 is $8.325bn. The filing does not print a nuclear subtotal; the grouping and the sum are ours.

Widen the lens to all 20 disclosed contracts and the concentration gets sharper rather than softer.

Grouping and totals computed by us from the same 20-row schedule (rcpNo 20260814003782, section XI.1.(나), base date 30 June 2026). The disclosed book totals ₩21,247,113m = $15.796bn; two further contracts — both described as “Doosan large gas turbine package supply” and disclosed on 17 December 2025 and 6 March 2026 — have counterparty and value withheld for commercial confidentiality, so they are excluded from every percentage in this table. They are not, however, unmeasurable, and the section below floors them. Listing status was checked against DART’s official corporate-code master file, which assigns KHNP corp code 00382001, KOMIPO 00396518 and Korea Southern Power 00382834 with no stock code in any case.

Add the first four rows together and Korean state-owned entities account for 12 of the 20 contracts that carry a disclosed value, and $10.774bn (the four rounded rows above add to $10.773bn; the exact figure converts ₩14,491,476m at ₩1,345.06) of the $15.796bn — 68.20% of that portion by value. The grouping and the sum are ours; the filing prints no such subtotal. And 68.20% is a figure with an explicit scope: it excludes the two contracts whose value is withheld. The next section is about those two, because they can be floored, and because they change this number.

Two more contracts have been signed since the reporting date

The schedule above closes on 30 June 2026. Doosan has disclosed two further supply contracts since, and both extend the pattern rather than break it. On 21 August 2026 it signed the Misfah Independent Power Project in Oman with Jabel Power SAOC; the disclosure states the contract value in its own currencies as US$284,250,000 plus €323,346,000, at its own stated rates of ₩1,402.50 and ₩1,637.49 as of the signing date. On 1 September 2026 it signed a turnkey supply contract for the Hadong combined-cycle plant in South Gyeongsang province, worth $495.0M, with Korea Southern Power — another wholly owned KEPCO generation subsidiary with no stock code.

Sources: DART single supply-contract disclosures rcpNo 20260821800022 (Misfah IPP) and rcpNo 20260901800387 (Hadong CCGT, contract value ₩665,850,000,000, construction 2 January 2027 to 31 December 2029). We quote the Misfah amounts in the currencies and at the rate the filing itself prints rather than re-converting the won figure, because re-converting a disclosed foreign-currency contract at a different rate manufactures a number that exists nowhere.

Two contracts are American, ongoing, and withheld — and the floor under them is disclosed

Both of the withheld contracts are gas-turbine packages for projects in the United States, and although the values are confidential, each one is disclosed as being worth at least 2.5% of revenue — a floor of $301.7M apiece.

This required correcting our own method, so it is worth saying how. The periodic report’s contract-progress schedule — the source for the table above — prints contract name, counterparty, dates and value, but it does not print the project’s location. That field exists only in the individual single supply contract disclosure filed when each deal is signed. We therefore re-read the location field in all 35 such filings by Doosan Enerbility between 1 January 2022 and 8 September 2026. Exactly two carry the region 미국 — United States — and they are the same two whose value is withheld.

One nearby case deserves flagging before a reader finds it and assumes we missed it. Doosan built a 200MW combined-cycle plant with 25MW of battery storage on Guam, and the region field of that disclosure reads 괌 — Guam — not 미국, so it is not one of the two above. Guam is a United States territory, and the same half-year report lists the project company, Doosan Ukudu Power, LLC, in its consolidated subsidiary schedule under the country 미국. The filing therefore classifies the same project as American in one field and not in another. The judgement that Guam counts as American is ours, not the filing’s — and it changes no arithmetic here: that contract’s period ended on 31 October 2024, it does not appear in the 30 June 2026 progress schedule, and it is therefore in none of the percentages in this report.

Source: DART single supply-contract disclosure rcpNo 20220531800931, an amendment filed 31 May 2022 to the original 21 December 2020 disclosure: contract name “Guam Ukudu Power Plant”; region 괌 Dededo, Piti; counterparty Guam Ukudu Power LLC; contract value ₩623,988,800,000, which the disclosure itself states as US$571,000,000 at its own stated rate of ₩1,092.80/$1 on the 18 December 2020 order date. We use the disclosure’s own dollar figure and do not re-convert the won.

Sources: DART single supply-contract disclosures rcpNo 20251217800425 and rcpNo 20260306801165. Each states, verbatim, that “the contract value in item 2 is 2.5% or more of recent revenue and is therefore subject to mandatory disclosure,” and each prints that recent revenue as ₩16,233,054,771,559, which the same note identifies as the FY2024 consolidated figure. Both also state that value and counterparty will be re-disclosed once the confidentiality reason lapses.

Our arithmetic on those two disclosed constants, computed separately and written out so it can be checked: ₩16,233,054,771,559 × 0.025 = ₩405,826,369,289, and ₩405,826,369,289 ÷ 1,345.06 = $301.7M for each contract, $603.4M for the pair. Both operands are printed in the disclosures linked above.

Read that as a floor, not an estimate. The disclosure establishes a minimum and says nothing about how far above it the true value sits.

Three things follow, and the first is a correction of our own framing.

Doosan does have American work, and it is bigger than the American nuclear work ever was. Each of these two contracts is worth at least $301.7M — 1.47 times the entire $204.9M Vogtle contract that took fourteen years and finished in 2022. But they are gas-turbine packages, not reactors, which is exactly what the rest of this company’s disclosure says: the two live commercial mentions of the United States anywhere in the half-year report are both about gas turbines, and the filing lists a US project of 4.6GW in its turbine supply record.

The concentration figures need their scope stated. Korean state-owned entities are 68.20% of the $15.796bn that carries a disclosed value. Add the two withheld contracts at their floor and the book becomes at least $16.400bn, on which the state share is 65.69% and KHNP’s is 50.76%. Because any larger withheld value only dilutes them further, those are ceilings: on the full book, Korean state entities are no more than 65.69%, and the United States is at least 3.68%.

And there is a date on it. The reservations expire on 30 September 2028 and 30 November 2029, and both filings commit to re-disclosing value and counterparty when the confidentiality reason lapses. This is one of the rare cases where a gap in the record has a published closing date.

One counterparty on that list is not Korean, and it is finished

The word “Westinghouse” appears exactly once in the half-year report, and it appears as a customer name. Doosan supplied AP1000 equipment for Vogtle Units 3 and 4 under a contract with Westinghouse Electric Company, LLC, signed on 8 May 2008, scheduled to 31 December 2022, with a base contract value of $204.9M and cost-input progress of 99.94% at the reporting date.

It is a useful row for calibration. It establishes that Doosan has manufactured to a non-Korean reactor programme for a non-Korean customer, and it establishes the scale at which it did so: $204.9M over fourteen years, against a Czech NSSS contract of $3.664bn — a ratio of about one to eighteen. On the filing’s own schedule it is finished.

Who owns Doosan’s biggest customer

KHNP is not a private company keeping to itself. It files annual reports with DART like any listed issuer — its FY2025 report is rcpNo 20260331003643 — and that report’s largest-shareholder schedule reads, in full: Korea Electric Power Corporation, 100.00%, 242,442,838 shares. KEPCO’s own FY2025 report (rcpNo 20260317000686) shows the Korea Development Bank at 32.90% and the Government of the Republic of Korea at 18.20% — 51.10% combined.

So Doosan’s largest customer relationship terminates in a government balance sheet: reactor equipment goes to KHNP, KHNP is wholly owned by KEPCO, and KEPCO is majority owned by the Korean state. For a shareholder that is a credit and cyclicality fact more than a governance one — the buyer of half this company’s contract book does not go bankrupt, and it also does not order on a commercial timetable. KHNP itself has no shares to buy; the nearest listed thing to it is its parent, KEPCO, in Seoul as 015760 and in New York as an ADR.

One more counterparty fact from the same filing, and it is worth reading carefully. The consolidated segment note discloses a single customer accounting for more than 10% of group revenue: $1.129bn in H1 2026 (up from $779.9M in H1 2025), or 16.90% of the $6.681bn of consolidated six-month revenue. The filing anonymises that customer as “Group A” and does not name it. We are not going to guess.

🔬 What Does Doosan Actually Own of NuScale and X-energy?

0.69% of NuScale and 0.30% of X-energy — combined carrying value $40.4M, which is 0.10% of Doosan’s market capitalisation — and the filing classifies both as “simple investment,” the Korean disclosure category that explicitly excludes participation in management.

Doosan’s SMR relationships are usually described in English as strategic partnerships. The Korean filing describes them with a specific, checkable vocabulary, and the numbers are small.

Source: DART half-year report, rcpNo 20260814003782, detailed schedule 3, “investments in other corporations” (타법인 출자현황). Won amounts before conversion — NuScale: carrying value ₩39,690m → ₩30,181m, fair-value loss ₩9,509m, investee total assets ₩2,026,813m, investee net loss ₩510,529m. X-energy: carrying value ₩8,173m → ₩24,155m, fair-value gain ₩15,982m, investee total assets ₩1,843,652m, investee net loss ₩246,429m. The investee figures are the “most recent financial year” columns the schedule prints for each holding; the filing does not state which year-end that is for each investee.

Three observations, each of which is invisible in an English summary.

The X-energy stake was diluted, not sold. Doosan still holds exactly 852,992 shares, but its percentage fell from 0.42% to 0.30% over six months. The share count did not change; the denominator did. Someone else was issued stock.

The two positions moved in opposite directions in the same six months. Doosan booked a $7.1M fair-value loss on NuScale and an $11.9M gain on X-energy. Anyone treating “Doosan’s SMR investments” as one thing is averaging away the only interesting part.

The scale comparison is uncomfortable but fair to state. The combined net losses those two investees reported in their latest financial years total $562.8M. Doosan’s own consolidated operating profit for the whole of FY2025 was $567.0M. These are different bases — one is a pair of investee net losses, the other a full-year consolidated operating result — and we are not netting them. The point is only that the businesses Doosan hopes to manufacture for are, at present, losing money at roughly the rate Doosan earns it.

What the filing does say about the commercial relationship is careful and worth quoting rather than paraphrasing. In the business-overview section it states that Doosan, “following material fabrication for the NuScale SMR, is conducting sales activity to win equipment-manufacturing orders, and is continuing discussions with multiple other SMR designers on equipment supply.” That is a pipeline described in the present continuous. It is not a backlog. And in the R&D schedule, the NuScale work appears as a project titled “optimisation of the automated welding process,” classified under “improvement,” which is the filing’s lowest of three R&D tiers.

📈 Is the Business Getting Better or Worse?

It depends entirely on which set of accounts you read: consolidated earnings per share rose from ₩97 to ₩248 in the first half of 2026, while the parent company’s standalone earnings per share fell from ₩312 to ₩55 over exactly the same six months.

Both figures are printed in the same document, four pages apart. Neither is wrong. They describe different companies.

Every column is the six-month cumulative (누적) column, not the three-month column. Source: DART half-year report, rcpNo 20260814003782, statements 2-2 (consolidated income statement) and 4-2 (separate income statement). Won figures — consolidated: revenue ₩8,985,938m / ₩8,317,595m; operating profit ₩547,774m / ₩413,581m; net income ₩286,594m / ₩176,652m, of which owners ₩158,377m / ₩62,067m and non-controlling interests ₩128,217m / ₩114,585m. Separate: revenue ₩3,760,610m / ₩3,518,028m; operating profit ₩199,640m / ₩191,673m; profit before income tax ₩51,658m / ₩251,297m; income tax expense ₩16,192m / ₩51,628m; net income ₩35,466m / ₩199,669m. EPS figures are the filing’s own. Operating margins computed by us from the rows above them.

The consolidated line looks like an inflection: operating profit up 32.4% on revenue up 8.0%, and the operating margin recovering from 4.97% to 6.10% after three years of compression that we documented in our earlier valuation report on this company. Profit attributable to Doosan’s own shareholders more than doubled, to $117.7M.

The standalone line says something quieter and less comfortable. On its own accounts — the actual power-equipment company, without Bobcat, without Fuel Cell — Doosan’s operating margin fell, from 5.45% to 5.31%, and net income fell 82.2%, to $26.4M. The gap between operating profit of $148.4M and net income of $26.4M is almost entirely below the operating line: standalone finance costs of $640.5M against finance income of $585.0M, plus $54.5M of net other non-operating losses, which leaves pre-tax profit of $38.4M; a $12.0M tax charge — an effective rate of 31.3% — takes it to $26.4M. On a project business with decade-long contracts in five currencies, that line is volatile by construction — but it is the line that determines what the parent actually keeps.

The order book is the part that is unambiguously improving

Contracted revenue not yet recognised reached $21.72bn at 30 June 2026, up 57.4% year on year, with the Enerbility segment alone up 69.5%.

Source: DART half-year report, rcpNo 20260814003782, consolidated note 20, performance-obligation schedule, base date 30 June 2026 with the prior-year comparative at 30 June 2025. Won figures — opening ₩22,910,598m / ₩1,880,732m / ₩228,657m / ₩25,019,987m; additions ₩8,071,328m / ₩95,519m / ₩45,511m / ₩8,212,358m; recognised ₩3,941,764m / ₩46,013m / ₩25,348m / ₩4,013,125m; closing ₩27,040,162m / ₩1,930,238m / ₩248,820m / ₩29,219,220m; prior year ₩15,957,392m / ₩1,827,780m / ₩777,232m / ₩18,562,404m. Column-order check: the schedule’s headers and its “recognised as revenue” row were cross-checked against the separate disaggregation table, which prints Enerbility ₩3,941,764m, Fuel Cell ₩46,013m and Bobcat ₩25,348m of over-time revenue — matching this row in that order. The bridge reconciles exactly in each column and in the total.

Two details in that table deserve to survive the summary. The Enerbility segment added $6.00bn of new contracted revenue in six months — roughly twice what it recognised. And Bobcat’s obligation book fell 68.0%, from $577.8M to $185.0M, which is the sort of thing that never appears in a headline reading “record backlog.”

The flagship contract inside that book is barely started. The consolidated contract schedule in the half-year report (rcpNo 20260814003782) puts Dukovany 5&6 NSSS at 6.43% cost-input progress at 30 June 2026, against a contract that runs to 18 April 2038. On the turbine-generator award the same filing shows cumulative supply of ₩8,832m against a contract of ₩711,106m, and states plainly why, in a table headed “reasons for non-receipt of payment”: “to be received in stages according to the client’s permitting schedule.”

Our conversion of those two figures at ₩1,345.06/$1: $6.6M delivered against a $528.7M contract, or 1.24%.

💰 What Do You Actually Pay for Doosan Enerbility Today?

₩87,900 a share — about $65 — for a $41.86bn company whose profit attributable to its own shareholders over the last twelve months was $134.6M, which is 311 times earnings.

Price, share count and market capitalisation from the Korea Exchange settled close of 7 September 2026 (market cap ₩56,305,324,733,400). Treasury and free-float share counts from DART rcpNo 20260814003782, section I.4. Earnings, equity and backlog from the same filing.

All ratios computed by us, on the KRX close cited above. The trailing-twelve-month owners’ profit of $134.6M (₩181,072m) is derived as FY2025 owners’ profit ₩84,762m, less H1 2025 owners’ profit ₩62,067m, plus H1 2026 owners’ profit ₩158,377m; it is not a figure any filing prints. Subsidiary stake values are Doosan’s disclosed holdings — 48.23% of Doosan Bobcat and 30.33% of Doosan Fuel Cell — applied to those companies’ KRX market capitalisations on the same date (₩6,106,031,340,500 and ₩3,268,136,927,400).

A P/E of 311× is not a valuation input and we are not offering it as one. It is a description. What the market is actually pricing is the backlog, at 1.93×, and the question that decides whether that is cheap is the margin at which $21.72bn of contracted revenue converts — on a business whose standalone operating margin has been 5.31% and 5.45% in the last two half-years.

The other number in that table worth sitting with is the last one. Doosan’s stakes in its two listed subsidiaries are worth $2.926bn — just 6.99% of its own market capitalisation. The market is not paying for the Bobcat stake. It is paying, almost entirely, for the power and nuclear business.

How Doosan is priced against its own customer

Market capitalisations from the Korea Exchange settled closes of 7 September 2026: KEPCO ₩21,024,323,521,750; KEPCO E&C ₩4,674,306,000,000; Doosan Enerbility ₩56,305,324,733,400; KOGAS ₩3,240,186,300,000. Listing status verified against DART’s official corporate-code master file: KHNP (00382001), KEPCO Nuclear Fuel (00160515), KOMIPO (00396518) and KOSPO (00382834) each appear with a corporate code and no stock code. Ownership percentages from the FY2025 largest-shareholder schedules filed by KEPCO (rcpNo 20260317000686), KHNP (20260331003643), KOMIPO and KOSPO.

The equipment supplier is worth 2.68 times its customer’s own parent. Doosan Enerbility’s $41.86bn against KEPCO’s $15.63bn is the cleanest single statement of how the market has chosen to express the Korean nuclear trade: not through the utility that owns the reactors and signs the exports, but through the company that forges the pressure vessels. Whether that is the right expression is a judgement. That it is the expression is a fact you can read off two closing prices.

🏷️ Which Doosan Ticker Are You Actually Buying?

034020 is the operating company; 000150 is the holding company that owns 30.39% of it; 241560 is the Bobcat business already consolidated inside 034020; and the Czech turbine subsidiary trades separately in Prague.

Ownership percentages and the withdrawn Doosan Robotics reorganisation from DART half-year report rcpNo 20260814003782 (consolidated subsidiary schedule; shareholder schedule at 30 June 2026; corporate-history table, which lists both the merger decision and its withdrawal). Market capitalisations from the Korea Exchange, 7 September 2026 settled closes. The Prague listing is stated in the filing’s name-change table; we have not sourced a Prague market capitalisation and do not print one.

On the ownership figure, a small decode that matters if you are reconciling two sources. Doosan Corporation’s 194,690,723 shares are 30.39% of the 640,561,146 shares in issue, but 30.46% of the 639,132,818 shares excluding treasury. The half-year report prints both numbers, in different sections, without reconciling them. They are the same holding on two denominators.

For how a foreign investor actually opens access to KRX-listed shares, see our practical guide to buying Korean stocks; for the small number of Korean names with proper US listings, our ADR directory is the place to start — KEPCO is on it, and Doosan Enerbility is not.

🌏 Is Foreign Money Buying or Selling Doosan Enerbility?

Selling, and consistently so across every window we can measure — foreign investors were net sellers of Doosan Enerbility (034020) of $68.6M over five sessions, $223.4M over twenty and $249.7M over fifty-six, while domestic institutions bought and retail investors also sold.

Method: investor flows arrive as share counts, so every currency figure here is derived. We multiply each session’s net share count by that session’s settled close and sum — an approximation, not execution prices — then convert at ₩1,345.06/$1, the rate our price history records for 7 September 2026. Share counts are Korea Investment & Securities Open API investor-flow data held in our own database; the daily closes they are multiplied by are KRX settled closes; the 56-session window is the full span of investor-flow data we hold for this ticker, beginning 18 June 2026. Price changes are from the settled close on the first session of each window to the settled close on 7 September 2026: ₩81,400 → ₩87,900, ₩79,900 → ₩87,900 and ₩99,600 → ₩87,900. Foreign net share counts: −1,172,302 (5 sessions), −3,867,401 (20), −3,934,237 (56).

Three windows, one sign on the flows — and two signs on the price. That combination is worth stating plainly rather than smoothing over: foreigners have been net sellers of Doosan Enerbility over every span we can measure, while the shares rose 7.99% over five sessions and 10.01% over twenty but fell 11.75% over fifty-six. Domestic institutions have been on the other side of the foreign selling in every window. Our guide to reading Korean flow data explains why that split matters more than the headline direction, and the weekly numbers live in our Foreign Flow Watch series — where Doosan Enerbility turned up as the sharpest counter-example in the week to 28 August: foreigners were net sellers and the stock still posted the largest market-capitalisation gain on the KOSPI.

We are not claiming the selling caused anything. It plainly did not: the shares rose in two of the three windows while foreigners sold in all three.

And a fourth window, because the year has been violent

The three flow windows above all sit inside a much larger round trip. On settled closes from the Korea Exchange, Doosan Enerbility peaked at ₩136,400 on 7 May 2026 and bottomed at ₩60,200 on 30 July 2026, before closing at ₩87,900 on 7 September 2026.

The company’s own half-year report confirms the top from the other side, printing a May monthly high of ₩136,400 against a June high of ₩106,900. (Monthly high/low table: DART rcpNo 20260814003782, section VII.4.)

Our arithmetic on those three closes: a fall of 55.9% from peak to trough in under three months, a recovery of 46.0% off the low, and a level still 35.6% below the May peak. Extremes are drawn from settled daily closes between 31 March 2026 — the earliest date in our price history for this ticker — and 7 September 2026, so an earlier peak in the first quarter would not be captured.

Anyone quoting a “year-to-date” move on this stock without a date is quoting an accident. Depending on where you start, the same twelve months contain a 56% collapse and a 46% rally.

🏛️ Who Controls It, and What Do Minority Holders Get?

Doosan Corporation controls it with 30.39%, and minority holders currently get no dividend, no Value-Up plan, no share cancellation, and one buyback whose stated purpose in the filing is employee compensation rather than shareholder return.

Source: DART half-year report, rcpNo 20260814003782, section VII.3, base date 30 June 2026. “Small shareholder” is the filing’s own definition: a holder of less than 1% of shares in issue.

The shareholder register moved sharply in six months, and in two directions at once.

At 30 June 2026 the company had 1,732,023 small shareholders holding 342,137,373 shares, or 53.41% of the shares in issue. (Source: DART half-year report, rcpNo 20260814003782, section VII.3.)

At 31 December 2025 it had 1,117,843 small shareholders holding 405,658,523 shares, or 63.32%. (Source: DART FY2025 annual report, rcpNo 20260320001246, same schedule, base date 31 December 2025.)

Our arithmetic across those two filings: the number of small-shareholder accounts rose 54.9% while their aggregate holding fell by 63,521,150 shares and their share of the company fell 9.91 percentage points. Six hundred thousand new small shareholders arrived while small shareholders in aggregate sold. Note the filing’s definition — a “small shareholder” is any holder of less than 1% — so the two counts are not a fixed group measured twice.

The buyback that is not a shareholder return

Between 2 and 15 April 2026 Doosan Enerbility bought 1,332,350 of its own shares for ₩130,342,615,600 — $96.9M — executing 99.99% of the ₩130,342,656,000 it had announced. In English that reads as a buyback. The filing’s own treasury-holdings table states the acquisition purpose as 임직원 주식보상 — employee share compensation — with an AGM approval date of 31 March 2026 and a disposal deadline (보유처분기한) of March 2027. One detail for anyone checking the Korean: the same row of that table also has a separate column headed 소각기한 — cancellation deadline — which reads “2027”. We read it as a deadline field rather than a commitment, because: the acquisition purpose in the row beside it is employee share compensation, and the filing states elsewhere in the same section that no treasury shares were cancelled in the period. A separate line in the same section reads, in full: “There were no treasury-share cancellations during the reporting period.”

Shares bought to be handed to staff and re-enter the float within a year do not reduce the share count. They are compensation expense, funded on market.

Our arithmetic: ₩130,342,615,600 ÷ 1,332,350 shares implies an average purchase price of about ₩97,830, which is 11.3% above the 7 September 2026 close of ₩87,900.

What we searched for and did not find

Three absences here are load-bearing, so here is exactly how we tested them. We queried DART’s disclosure index for every filing by Doosan Enerbility (corporate code 00159616) between 1 January 2022 and 8 September 2026 — 779 filings — and matched on report-type name, which is how Korean disclosure categories are labelled:

• 기업가치제고계획 (Corporate Value-Up plan): 0 filings. Also 0 for 밸류업. Doosan Enerbility has not published a Value-Up plan, in a market where the programme has become the standard vehicle for shareholder-return commitments.
• 자기주식소각 (treasury-share cancellation): 0 filings. Consistent with the half-year report’s own statement above.
• Dividend disclosures: 0 filings whose report-type name contains 배당. Consistent with the FY2025 dividend schedule, which shows a dash in every cash-dividend row.

The last governance item is one English coverage never mentions. Doosan Enerbility’s borrowings are secured on the shares of its own two listed subsidiaries. The parent has pledged 28,700,000 Doosan Bobcat shares and 22,200,000 Doosan Fuel Cell shares across six separate agreements, five of which carry a collateral-ratio test obliging the company to post more shares or repay early; the overseas bond instead carries a reference-price top-up with no early-repayment limb:

Source: DART half-year report, rcpNo 20260814003782, note 27 “commitments and contingencies”, parent-company paragraphs. Facility sizes are the filing’s own; the share totals of 28,700,000 and 22,200,000 are our addition of the rows above. We have deliberately not translated these into a percentage of either subsidiary, because the filing does not state one.

The rating and control covenants are narrower than they first look, and worth stating precisely. Exactly one facility — a ₩49,900m short-term loan from China Everbright Bank — carries a covenant requiring Doosan Corporation to remain the largest shareholder. A credit-rating floor of BBB− is the common covenant, appearing as a maintenance condition on four separate borrowings. Two of the share-secured loans carry early-repayment triggers keyed to the same ratings instead: the ₩280,000m Yuanta loan if the senior unsecured rating falls to BBB− or below at one or more of Korea’s three rating agencies, and the ₩150,000m Enerbility Hana 3rd loan if it falls to BB+ or below at two or more. (Source: same filing and note. “Enerbility Hana 3rd Co. Ltd” is our rendering of 에너빌리티하나제삼차 유한회사, a securitisation vehicle and a 유한회사 rather than a Co. Ltd; the other five lenders are given by their own English names.)

🏭 What Is Doosan Actually Building for SMRs?

A dedicated SMR plant costing $599.8M that, at 30 June 2026, was still in detailed design and had not been built.

Doosan disclosed the investment on 17 December 2025 as “construction of a dedicated SMR plant, optimisation of existing plants, and establishment of innovative manufacturing facilities,” at about ₩806.8bn. The half-year report’s schedule of progress on previously disclosed material matters gives the status in six words: “basic design complete, detailed design under way.”

Behind it sits a three-year capital plan for the Enerbility segment of $1.401bn — $524.2M in 2026, $586.5M in 2027 and $290.5M in 2028 — of which $890.8M is plant new-build and refurbishment tagged in the filing as “SMR, GT and wind equipment investment.” Actual segment capital spending in the first half of 2026 was $111.5M, against $171.0M for the whole of FY2025.

Source: DART half-year report, rcpNo 20260814003782, section II.3 “new facility construction and purchase plans” and section XI.1.(가). Won figures: planned total ₩1,884,725m (₩705,085m / ₩788,922m / ₩390,718m by year), of which plant new-build and refurbishment ₩1,198,198m; investment in progress ₩149,986m in H1 2026 and ₩229,974m in FY2025. Conversion at ₩1,345.06/$1 by us.

One more piece of the manufacturing story that is only in the Korean text. The filing’s schedule of material business agreements lists the technology Doosan licenses in: spent-fuel cask technology from NAC International to 2035, steam-turbine technology from its own Czech subsidiary to 2030, boiler technology from Altrad Babcock in perpetuity, and post-combustion carbon capture from HTC Purenergy to 2033. The gas-turbine entry is the interesting one: Doosan licensed gas-turbine technology from Mitsubishi-Hitachi Power Systems in 2007, and the footnote records that it returned the manufacturing and sales licence after 20 December 2017, retaining only the service licence to 2033. That is the localisation story stated in the driest possible way — Doosan handed back the Japanese manufacturing rights and built its own model. The filing counts 24 gas turbines contracted in total, twelve of them in the first half of 2026 alone (five domestic, seven overseas).

📚 Lingo Check

🎯 Why It Matters for K-Export Stars

Every number in this report was available to anyone on 14 August 2026, filed in Korean, in one document, for free. The revenue mix that makes Bobcat the largest segment is in a table headed 매출실적. The counterparty on every nuclear contract is printed in a column headed 계약상대방. The classification of the NuScale and X-energy stakes as 단순투자 is a single word in a schedule of investments in other corporations. None of it is hidden. It is simply in a language most of the people trading this stock do not read, in a document long enough that even those who do tend to read the summary.

That is the whole of our thesis. The wires will tell you Doosan Enerbility won a Czech contract, and they will be right. Only the filing will tell you who signed it, whether you can own them, and what fraction of the company you are buying actually makes reactors.

Disclaimer: This report is for informational and educational purposes only and does not constitute investment advice. All figures are point-in-time as noted and go stale; market data is as of the 7 September 2026 settled close and filing data as of the 30 June 2026 reporting date. Check the live quote and the current filings before acting. Investors should conduct their own research before making financial decisions.