Doosan Enerbility

Phase A — Understand the business

Lens 1 · Company Overview

Doosan Enerbility (formerly Doosan Heavy Industries & Construction, and before that state-owned Korea Heavy Industries / "Hanjung", established 1962, listed 2000, rebranded to "Enerbility" = Energy + Sustainability) is South Korea's national heavy-power-equipment champion and EPC contractor, headquartered at the Changwon complex. It is one of a very short list of firms on Earth — GE Vernova, Mitsubishi, Framatome, and it — that can forge and integrate the primary components of a nuclear island (reactor vessel, steam generators, pressurizer) and build H-class gas turbines in-house.

How it makes money. A blend that de-risks project lumpiness:

• New-build equipment & EPC — nuclear islands, gas/steam-turbine combined-cycle plants, desalination, castings & forgings. Project-based, lumpy, backlog-driven.
• Aftermarket / long-term service agreements (LTSAs) — parts, upgrades, inspections on the installed base. Recurring, higher-margin. Example: a ₩480B (~$319M) Long-Term Parts Management deal with KOSPO signed May 2026 on three gas turbines.
• Emerging clean energy — SMRs, hydrogen/ammonia co-firing, fuel cells (via Doosan Fuel Cell), offshore wind.

Three reportable segments (consolidated): Doosan Energy (the core power/nuclear/forging business), Doosan Bobcat (compact construction equipment — the 46%-consolidated cash cow), Doosan Fuel Cell.

Customers: utilities, IPPs, state energy companies (KHNP/KEPCO in Korea, KOSPO), EPC consortium partners, and — the new leg — US "Big Tech" hyperscalers buying gas turbines for datacenter power. Contract structure is milestone/progress-billed EPC (not take-or-pay), which means revenue lags orders by 2–4 years — the backlog is the leading indicator, not the P&L.

Ownership: controlled by Doosan Corporation (the group holdco is the largest shareholder). This is a chaebol — minority-shareholder alignment is a live governance question (see Lens 9/13).

Lens 2 · Supply Chain

Upstream → Doosan → end customer, with named links:

• Upstream inputs: specialty/nuclear-grade steel and forgings ingots, precision-machined components, control systems, and construction subcontractors. Critically, for the nuclear-forging business Doosan is often its own upstream — it operates the large-forging press at Changwon and forges long-lead materials itself, which is the entire point of the moat. It also owns Doosan IMGB (Romania) and historically Doosan Škoda Power (Czech steam turbines, acquired 2009) as in-group capability.
• The company (Changwon): casts/forges the reactor vessel, steam generators, pressurizer, turbine rotors; assembles gas/steam turbines; manages EPC.
• Downstream / end customers (named):
• KHNP (Korea Hydro & Nuclear Power) — domestic reactor fleet (APR1400).
• NuScale Power (US) — Doosan is a strategic partner, equity investor, and the forge for long-lead materials on 12 NuScale Power Modules.
• X-energy, TerraPower — SMR forging partners in the pipeline.
• Rolls-Royce SMR — new SMR component work won Q1 2026.
• Westinghouse — AP1000 component supply under the Korea–US nuclear cooperation framework.
• ČEZ (Czech Republic) — Dukovany Units 7 & 8, via Team Korea/KHNP.
• US hyperscaler (undisclosed) — 7× 380MW-class gas turbines.

Chokepoint / single-source note: the Changwon ultra-large forging press is the chokepoint in the company's favour — there are only a handful of presses globally that can forge Gen-III reactor-vessel-scale components to nuclear code outside China/Russia, and Doosan runs one of them. That is the moat's physical embodiment. The corresponding risk is single-site concentration: a Changwon disruption (labour, seismic, capacity ceiling) is a company-level single point of failure. SMR capacity is being expanded via a dedicated Changwon SMR plant targeted for 2028.

Lens 3 · Competitive Advantages (moats)

The moat is real, physical, and geopolitically scarce. HSBC's word for it: "quasi-monopoly status in nuclear steam supply systems".

1. Nuclear-island forging scale (process + capex + code moat). Forging a reactor pressure vessel to ASME/nuclear code requires a press, decades of qualified process know-how, and regulator-accepted traceability. This is a 20-year, multi-billion-dollar barrier to entry. Bernstein frames Doosan as able to execute Western nuclear at ~$60/MWh, undercutting peers while being non-Chinese/non-Russian — the only politically acceptable large-forge supplier for the US/EU nuclear build-out.
2. Dual nuclear + gas-turbine capability. Localised large gas turbine production in 2019; 17,000+ hours of operational verification since. Very few firms do both nuclear islands and H-class gas turbines — that breadth is a hedge against any single technology's policy cycle.
3. SMR foundry position. By being the forge for NuScale (12 modules), X-energy, TerraPower, and Rolls-Royce SMR, Doosan is positioning to be the arms-dealer to the entire Western SMR industry — it wins regardless of which SMR design ultimately scales. This is the single most valuable strategic option in the story.
4. Installed-base aftermarket — LTSAs on decades of delivered plants = a recurring annuity with switching costs (you service the turbine you built).

Bargaining power: high over customers on nuclear islands (few alternatives, national-security-sensitive, long qualification) — Doosan needs the hyperscaler less than the hyperscaler needs guaranteed 2028–2030 turbine slots. Weaker on commodity EPC (Middle East combined-cycle is competitively bid). The moat is concentrated in the nuclear/large-turbine core, not the whole revenue base.

Lens 4 · Segments

segments.csv is empty — no segment figures exist; all below are/``.

Consolidated FY2025:

• Revenue ₩17.06T (+5.1% YoY); Operating profit ₩762.7B (−25% YoY) — the OP fall is attributed explicitly to Doosan Bobcat's construction-equipment down-cycle dragging the consolidated line.
• Standalone Enerbility (ex-Bobcat) FY2025 revenue guidance ₩10.7T, described as "smoothly achieved," with H2 skewed to higher-margin gas turbine + nuclear.

H1 2025 (consolidated):

• Revenue ₩8.32T (+0.8% YoY); OP ₩413.6B (down YoY); net profit ₩176.7B.
• Q2 alone: revenue ₩4.37T, OP ₩271.1B, net ₩197.8B — a recovery from a Q1 net loss of ₩21.1B.

Segment mix direction: the reported story is Bobcat weakness masking core-business acceleration; the real story is the Doosan Energy segment inflecting up as the order book converts. Bernstein models standalone operating margin rising 3% (2024) → 9% (2027) as mix shifts to nuclear/gas turbine. Cause = margin mix, not volume alone. By-geography and by-EBITDA splits are n/a (would require DART segment notes).

---

Phase B — Measure performance

Lens 5 · Earnings Result (latest print)

Latest fully-reported period is FY2025 (consolidated), with the most granular disclosure at H1/Q2 2025:

• Revenue: FY2025 ₩17.06T (+5.1%); standalone guidance ₩10.7T achieved.
• Operating profit: FY2025 ₩762.7B (−25% YoY) — miss vs. the growth narrative, but the miss is Bobcat, not the core. Q2 2025 OP ₩271.1B vs. ~₩283B estimate — a modest ~4% miss.
• Net income: TTM ₩154.4B; H1 2025 ₩176.7B; the swing from a Q1 loss to Q2 profit is the key quality signal.
• Drivers: gas turbine + nuclear (high margin) accelerating in H2; Bobcat (low margin, cyclical) decelerating.
• Balance-sheet flags: post-2020-bailout deleveraging is the overhang that has lifted — the company survived a near-bankruptcy (see Lens 9). No fresh distress signal in the sourced material; specific net-debt/inventory/receivables figures are n/a (empty financials.csv; would need DART).
• Orders — the number that actually matters: FY2025 new orders ₩14.73T, a record and 2.06× the prior ₩7.13T. Backlog ₩23T end-2025 (up from ~₩16T mid-2025). For a 2–4-year-revenue-lag EPC, this backlog is the forward P&L.
• Market reaction: the stock is up ~339% in the year to May 2026 and +81% market cap YoY to July 2026 — the market has emphatically priced the order inflection, not the trailing OP.

Unusual vs. own history: the record order intake and the return to net profit after a Q1 loss and a 2020 near-death is a genuine regime change — but reported OP still going down YoY on Bobcat is the tell that the consolidated P&L is a misleading lens for this name.

Lens 6 · Earnings Calls (sentiment trend)

transcripts/ is empty — no compiled call history. From public IR/press summaries, the management narrative arc over the last ~4 quarters:

• Consistent, escalating message: "nuclear supercycle + SMR foundry + gas turbine export + datacenter power demand." Each quarter added a concrete proof point — Czech nuclear, first US gas turbine export, Rolls-Royce SMR, record ₩14.7T orders.
• Newly emphasised phrases (2026): "datacenter/AI power demand," "gas turbine export," "SMR mass production" (Changwon 2028), "Team Korea → Europe."
• De-emphasised / dropped: the 2020–2022 language of "financial normalisation," "debt reduction," and "self-rescue" — the balance-sheet-survival framing has been retired in favour of a growth/backlog framing. That shift is itself the sentiment signal: management has moved from defense to offense.
• Tone: confidently guiding a multi-year order ramp (₩108T of segment order targets to 2030) — promotional relative to a company that was on state life-support four years ago. Treat the out-year order targets as aspirational, not contracted (Lens 13).

Caveat: sentiment here is inferred from press/IR summaries, not primary transcripts — a hybrid refresh should ingest the DART/IR call decks to firm this up.

Lens 7 · Comps

Reads:

• Doosan's P/B 5.1x and EV/EBITDA 28.0x (2025) sit below GE Vernova's EV/EBITDA (85.6x) and near Siemens Energy (34.8x) — so on EV/EBITDA Doosan is arguably the cheapest of the Western-aligned power-equipment trio, even as its P/E screens most expensive (because trailing net income is artificially low on Bobcat drag + high D&A + financing). The P/E is the wrong lens; EV/EBITDA and EV/backlog are the right ones.
• 5-yr average ROE: n/a (and would be distorted by the 2020–2022 loss years anyway).
• Bottom line: on a pure-play power/nuclear EV/EBITDA basis Doosan is not obviously more expensive than GEV/Siemens; on trailing P/E it looks absurd. Both are true; the disagreement is the whole point of the comp.

Lens 8 · Stock-Price Catalysts (>5% moves, ~5-yr pattern)

What actually moves this stock:

• 2020 — Bailout / near-bankruptcy (DOWN, existential). ₩3T KDB+KEXIM rescue amid COVID + the Moon administration's nuclear phase-out; forced asset sales + dilutive rights issues. The stock's multi-year base.
• 2022 — Yoon administration pro-nuclear pivot (UP). Policy reversal restored the domestic reactor pipeline; the rebrand to "Enerbility."
• Dec 2024 — Martial-law shock (DOWN, sharp). President Yoon's failed martial-law attempt "pummelled Korean financial markets, particularly nuclear energy stocks," and the plunge killed the Bobcat–Robotics merger as prices fell below buyback floors. Political risk is a named, realised catalyst for this name.
• 2025 — Nuclear supercycle re-rating (UP, violent). Share price tripled in 2025 YTD, outperforming KOSPI by ~209%; +339% in the year to May 2026. Drivers: Czech Dukovany order recognition, first US gas turbine export, HSBC/Bernstein initiations, and the AI-datacenter-power narrative.
• 2026 — Bobcat split-off ex-effects + volatility (choppy). Share prints of ₩139,200 (52-wk high) → ₩128,000 (May) → ₩85,500 (Jul 6) show a sharp pullback off the peak, consistent with profit-taking and structural-reorg noise after the parabolic run.

Pattern: this stock reacts to (1) Korean domestic politics/policy, (2) major nuclear order recognition (Czech, US, SMR), and (3) the AI-power macro narrative — far more than to a given quarter's OP. It is a policy-and-backlog stock, not an earnings-print stock.

---

Phase C — Judge people & books

Lens 9 · Management

CEO/leadership (as of 2025):

• Geewon Park — Chairman & CEO (Internal Director).
• Yeonin Jung — President, COO (Internal Director).
• Sanghyun Park — President, CFO (Internal Director). The chair sits inside the Doosan family / Doosan Corporation control structure — this is a founding-family chaebol, not a professional-manager-run public company. Archetype: controlling-family steward, with all the alignment upside (long horizon, willing to make a 20-year forging bet) and downside (group-level interests can trump minority holders — see Lens 13).

1. Track record — genuinely impressive turnaround. This team took a company that in 2020 needed a ₩3T state bailout to avoid bankruptcy and, by 2025, delivered record ₩14.7T orders and a ₩23T backlog with the balance sheet repaired. Surviving the Moon-era nuclear phase-out, repaying state creditors via asset sales (Infracore, Solus, Doosan Tower) and rights issues, and re-emerging as the KOSPI's leading nuclear play is a real, quantifiable operational feat.
2. Tenure & skin in the game: family-controlled via Doosan Corp; specific insider-ownership % n/a (insider-transactions.csv absent). Alignment is structural (family holdco) not necessarily minority-friendly.
3. Capital allocation — the mixed record. Pro: the forging-capacity and SMR-plant capex (Changwon 2028) is exactly the right bet on the moat; the pivot to gas turbine export is well-timed. Con: the 2024 attempt to hand the cash-cow Bobcat stake to money-losing Doosan Robotics at a ratio widely seen as unfair to Enerbility/Bobcat minorities is a textbook chaebol value-transfer red flag. It was scrapped twice under shareholder + regulatory pressure and finally killed by the Dec-2024 market crash — meaning it was governance-blocked, not voluntarily abandoned.
4. Red flags: the Bobcat/Robotics saga (related-party value transfer); reliance on state banks historically; group-holdco control over minority interests.
5. Founder vs. professional: controlling-family — implies patient capital for the nuclear supercycle but a standing governance discount that a Korea "value-up" re-rating could narrow (an underrated catalyst) or that another self-serving reorg could re-widen.

Lens 10 · Forensic Red Flags

Accounting-risk map (labelled; no `` — DART filings not on the shelf):

• Revenue recognition (highest-attention item): long-cycle EPC percentage-of-completion accounting is inherently judgement-heavy — cost-to-complete estimates, change orders, and milestone timing let management pull/push revenue and margin across periods. With a ₩23T backlog now converting, POC assumptions are the number to audit. ``
• Consolidation optics: consolidating a 46% Bobcat stake means group revenue/OP swings on a business Enerbility only partly owns — the ₩17T consolidated figure overstates the scale of the actual power business (₩10.7T standalone). Any analysis on consolidated multiples is distorted. ``
• Segment reporting: three segments disclosed, but sub-segment (nuclear vs. gas turbine vs. forging) margins are not broken out publicly — opacity that hides where the real profit is. n/a
• Financing / leverage: post-bailout the balance sheet was rebuilt via dilutive equity issuance — share count 640.5M reflects that history; watch for further raises to fund SMR capex. ``
• SBC / non-GAAP flattering: no evidence sourced of aggressive non-GAAP add-backs (Korean reporting is IFRS-based, less non-GAAP-driven than US). n/a
• Cash-flow vs. earnings divergence, receivables/inventory vs. revenue: the classic EPC risk (unbilled receivables, contract assets ballooning ahead of cash) is the forensic watch-item — but the specific figures are n/a (empty financials.csv; requires DART cash-flow statement).

Regulatory findings (required sub-section) — read from regulatory/regulatory-findings.md:

• SEC (EDGAR LR/AAER): none possible. Doosan Enerbility has no CIK and is not an SEC filer; the automated EFTS search returned 0 findings by construction, not by clean record. ``
• Non-SEC enforcement (web): no material FTC/DOJ/FDA/consent-decree hits surfaced for "Doosan Enerbility" in the sourced searches. The relevant governance controversy is the 2024 Bobcat–Robotics merger, which drew Korean regulatory scrutiny and lawmaker objection (an FSS/fairness-of-swap-ratio issue, not an enforcement action) before being scrapped. Historically, Doosan Heavy drew NGO/ESG criticism for financing coal plants in Indonesia (Vietnam/Indonesia coal EPC) — reputational, not legal.
• Item 3 Legal Proceedings (10-K): n/a — no 10-K exists (Korean filer). The DART equivalent (사업보고서 litigation note) is the source to pull on a hybrid refresh.
• Verdict: No material enforcement finding surfaced via SEC EFTS (structurally empty for a non-filer), web search, or available disclosure as of 2026-07-06. The live governance risk is capital-allocation fairness to minorities, not fraud.

---

Phase D — Project & stress-test

Lens 11 · Forward Projection (EPS, next 3 FYs)

No forecast.ts create in unattended/watchlist mode (per skill). Projection is `` built on sourced anchors; EPS is on a standalone-ish, order-conversion basis — the trailing consolidated EPS is not the right base.

Sourced anchors:

• Current price ₩85,500, market cap ₩55.2T, shares 640.5M, trailing P/E 357.6x → trailing EPS ≈ ₩239. Forward P/E 106.2x → FY-forward EPS ≈ ₩805.
• Sell-side OP roadmap: ₩762.7B (2025A) → ₩1,052B (2026E) → ₩1,518B (2027E) → ₩2,000B (2030E).
• Bernstein: +94% EPS growth 2024–27; 2025 P/E 115x normalising to 27x by 2030; standalone revenue ₩14T by 2030 (8% CAGR); OP margin 3%→9% (2024→2027).
• HSBC: backlog triples to ₩48.4T by 2030 from ₩15.9T (2024).

Three-path EPS sketch (KRW/share; illustrative, ``):

The number that actually decides it isn't EPS — it's order conversion. At ₩23T backlog and ~₩10.7T standalone revenue, the book already covers >2 years; the question is margin realisation (does mix really get to 9%?) and out-year order capture (does the ₩48–108T of targets become contracts?). If both hit, the "27x by 2030" normalisation math works and today's price is defensible; if margin sticks at ~5% or SMR FIDs slip, the ~106x forward P/E has no support.

Tracked-forecast candidate (log on a live/interactive pass, not here): "034020.KS FY2027 consolidated OP ≥ ₩1,400B", p≈0.55, resolves 2028-03-31.

Lens 12 · Bull vs Bear

Bull case. Doosan is the only Western-politically-acceptable, at-scale nuclear-island forge outside China and Russia — and the world has simultaneously (a) rediscovered nuclear as the answer to AI-datacenter baseload, (b) launched an SMR industry that all needs the same forge, and (c) turned to gas turbines as the bridge fuel, which Doosan also builds. It is a pick-and-shovel play on the entire nuclear renaissance that wins across NuScale, X-energy, TerraPower, Rolls-Royce, Westinghouse AP1000, and Korea's own APR1400 — design-agnostic. Record ₩14.7T orders (2.06×) and a ₩23T backlog are the leading indicator turning up hard; margin mix is set to lift OP from ₩0.76T to ₩2T by 2030. A Korea "value-up" governance re-rating is free optionality on top. Earnings surprise vector: a single large US nuclear FID or a hyperscaler multi-turbine frame agreement could step-change the backlog overnight.

Bear case. (1) Valuation is the risk. At ~106x forward P/E the stock has priced years of flawless order conversion; any slip (SMR FIDs are notoriously delayed, US nuclear new-build has a decades-long history of cost overruns and cancellations) de-rates it hard. *(2) The order pipeline is mostly targets, not contracts — ₩48–108T of 2030 order "estimates" are aspirational; the gap between MOU/forging-partner status and a signed EPC is where nuclear dreams die. (3) Governance/political impairment — a chaebol that tried to transfer the Bobcat cash cow to a money-losing affiliate in 2024, in a country where a martial-law episode cratered the stock inside 18 months. (4) Bobcat drag — consolidated OP is falling even as the story soars.

Pre-mortem (18 months out, thesis broke): SMR final-investment-decisions slipped to the 2030s (as SMRs historically do); the US nuclear "renaissance" produced MOUs but few signed reactor EPCs; a Korean political/policy shock (election, another governance reorg) hit sentiment; and the stock de-rated from ~106x toward the ~27x "2030 normalised" multiple early — a >50% draw-down on multiple compression alone, even with the business fine.

Are multiples too high? On trailing/forward P/E, yes, extreme. On EV/EBITDA (28x) vs. GEV (85.6x) / Siemens Energy (34.8x), no — arguably reasonable for the growth. The honest answer: the multiple is only justified by out-year order conversion the market is taking on faith.

Contrarian view (what the market refuses to see): consensus treats Doosan as a nuclear pure-play re-rating and frets about the P/E. The under-appreciated point is that the gas-turbine export leg — first US hyperscaler order, 45 units targeted by 2030 — is a nearer-term, higher-certainty cash engine than the multi-decade SMR dream, and it's the AI-datacenter-power trade with a 2027–2029 delivery clock rather than a 2030s one. The market is paying for SMRs; it may get paid by turbines first.

Lens 13 · Devil's Advocate (short-seller)

Dismantling the bull case:

• What structurally breaks the money machine? Nuclear new-build is the most cancellation-prone capital project in the world (Vogtle, V.C. Summer, Hinkley cost blowouts). Doosan's forward value rests on FIDs that are not in its control — utilities, governments, and hyperscalers decide. A US nuclear-policy reversal or a single high-profile SMR failure freezes the whole pipeline.
• Revenue concentration & the shift: the growth is concentrated in a handful of mega-projects (Czech Dukovany ~₩5–6T, US nuclear, specific SMR designs). Lose Dukovany to EDF, or have NuScale/X-energy stumble, and the backlog narrative loses its marquee names.
• Why the moat may be weaker than bulls think: being the forge for SMRs is lower-margin components work, not owning the reactor IP or the plant economics — Doosan could end up the low-value commodity supplier in a high-value industry, capturing forging fees while designers/utilities capture the upside. "Arms dealer" cuts both ways: arms dealers have pricing power only while capacity is scarce; capacity is being added (incl. by Doosan itself and potential Japanese/Western entrants).
• Most dangerous competitor bulls underestimate: GE Vernova (scale, $163B backlog, its own SMR via BWRX-300, hyperscaler relationships) and Japan's Mitsubishi Heavy (nuclear + turbine, geopolitically acceptable, and not carrying a chaebol governance discount). Also China's forging capacity competing on non-Western projects.
• Worst capital-allocation moves: the 2024 Bobcat→Robotics value-transfer attempt — a live demonstration that management will, given the chance, act for the group over Enerbility's own minority holders.
• What must hold for today's price? OP margin actually reaching 9%; ₩48T+ backlog by 2030 converting; no Korean political shock; SMR FIDs this decade. That's four independent bets, each with real failure probability.
• If growth disappoints 20–30%: on a ~106x forward P/E, a 25% earnings miss + de-rating to even 50x is a >60% price decline. The asymmetry is brutal at this multiple.
• Single scenario that permanently impairs: a US/global step-back from new nuclear (a major accident, a cost-overrun scandal, or policy reversal) that pushes the SMR/large-reactor build-out into the 2030s+ — Doosan reverts to a cyclical Korean EPC + turbine maker worth a fraction of ₩55T. Plausibility: moderate — nuclear's history is a graveyard of "this time it's different" build-outs, but the AI-power demand driver is genuinely new and large.

Lens 14 · Management Questions (ordered by information value)

1. Of the ₩48–108T of stated 2030 order targets, how much is under signed contract vs. MOU/forging-partner status vs. internal estimate — and what is the historical conversion rate from your SMR forging agreements to firm EPC orders?
2. What operating margin do you actually underwrite for the standalone Doosan Energy segment in 2027 and 2030, and what is the nuclear vs. gas-turbine vs. forging margin split within it?
3. On the Bobcat 46% stake: what is the definitive plan now that the Robotics merger failed — hold, sell, dividend, or re-attempt a reorg — and how do you guarantee any structure treats Enerbility minority holders fairly?
4. Which SMR designs (NuScale, X-energy, TerraPower, Rolls-Royce) do you expect to reach FID this decade, and what is your exposure if the answer is "none before 2030"?
5. For the US gas-turbine orders: are these firm frame agreements with delivery slots and cancellation penalties, or LOIs — and what is the pipeline of repeat hyperscaler turbine demand?
6. What is your cash-flow conversion on the ₩23T backlog — how large are contract assets/unbilled receivables, and when does order growth translate to free cash flow?
7. How dependent is the nuclear thesis on the Korea–US (123) nuclear cooperation agreement and Team Korea consortium politics — what happens to Westinghouse/AP1000 component work if that framework shifts?
8. What is the Changwon forging capacity ceiling, and what capex (and dilution risk) is required to build the 2028 SMR plant and the 12-unit/yr gas-turbine and 20-unit/yr SMR capacity?
9. On Czech Dukovany: what is your contracted scope and revenue, and what is your exposure to Team Korea vs. EDF competitive dynamics on future EU reactors?
10. How do you defend forging pricing power as Western/Japanese and your own added capacity comes online — where is the pricing floor?
11. What is the realistic revenue-recognition schedule (percentage-of-completion) for the current backlog by year 2026–2030?
12. What further equity issuance should shareholders expect to fund the SMR/turbine capex build-out?
13. How exposed is the order book to Korean domestic political risk (elections, another martial-law-type shock, KHNP/KEPCO policy)?
14. What is the aftermarket/LTSA revenue and margin trajectory as the installed base grows — how large is the recurring annuity beneath the lumpy EPC?
15. What single event would most change your own capital-allocation plan over the next 3 years, and how are you hedged against it?

---