Goldwind

Phase A — Understand the business

Lens 1 · Company Overview

Goldwind is China's largest wind-turbine manufacturer and the world's #1 onshore OEM by installed volume — 29.3 GW installed globally in 2025, holding the top global spot for the fourth consecutive year and #1 in China for fifteen straight years. Founded in 1998 by Wu Gang out of the state's "863 Program," it is a partially state-linked champion built on permanent-magnet direct-drive (PMDD) technology licensed from Germany's Vensys in 2008 (Goldwind later acquired Vensys). Every Goldwind turbine is a PMDD machine — no gearbox — which is the company's core technical identity.

The business is four segments, and the interesting story is that the smallest ones make the money:

The contract structure matters: turbine sales are lumpy, competitively-bid, project-linked equipment revenue with thin margins and long receivables. The wind-farm segment is a build-operate-rotate model — Goldwind develops its own wind farms (3.2 GW in the self-development backlog), earns 43%-margin generation income, then periodically sells matured assets to recycle capital. The service segment is the genuinely attractive recurring-revenue leg: 50,315 MW under O&M management at year-end 2025, a growing installed-base annuity. Main customers are Chinese state power generators (State Power Investment, Huaneng, Datang, etc.) plus a fast-growing overseas book; main suppliers are the rare-earth-magnet, casting, bearing and blade supply chain (Lens 2). Main competitors: domestically Envision, Mingyang, Windey, Sany Renewable; globally Vestas and Nordex.

Plain-terms verdict: this is a commoditized heavy-equipment manufacturer wrapped around a small, high-quality energy-asset and services business. The market narrative ("wind leader, +223%") is written by the 79% of revenue that earns 8.9%; the actual quality sits in the 20% that nobody talks about.

Lens 2 · Supply Chain

Upstream → Goldwind → end customer, named:

Upstream inputs (the chokepoints):

• Rare-earth permanent magnets (NdFeB) — the defining input for PMDD generators. Neodymium and dysprosium prices have swung 30–50% YoY since 2022. China controls the NdFeB magnet chain, which is a strategic advantage for Goldwind (domestic sourcing) but a cost-volatility and geopolitical-export-control exposure for the same reason. Goldwind has historically held a stake in Jinli Permanent Magnet — a magnet supplier — which it has been selling down (see Lens 9, an earnings-quality tell).
• High-precision large bearings — a genuine global bottleneck; lead times of 12–18 months for certain configurations. Main-bearing capacity is a structural constraint for the whole industry.
• Castings (hub, bedplate) — steel-price-linked; concentrated in a handful of Chinese foundries.
• Blades — Goldwind sources long carbon/glass blades; the 100m+ blades for its 10–14 MW Ultra series push material and logistics limits.
• Pitch & yaw drives — the wind industry is >85% import-dependent from Europe (Germany, Denmark, Spain) and a few Chinese players for high-spec drives. Goldwind produces a meaningful share in-house (like Vestas/Siemens Gamesa), which insulates it partially.

Midstream: Goldwind assembles at plants in Urumqi (Xinjiang), across China, and now its first overseas factory in Brazil (opened Aug 2024).

Downstream (end buyers): Chinese SOE utilities dominate; overseas the customer set spans 47 countries across six continents as of Sept 2025 — with South America (>2.3 GW installed), Asia ex-China (~2.5 GW), and Australia (~2.0 GW) the largest overseas footholds.

Single-source / concentration flags: (1) rare-earth magnet chain concentration is both moat and risk; (2) the Xinjiang manufacturing base is a supply-chain liability in Western markets — Goldwind signed an agreement with the sanctioned XPCC in Dec 2020 and has faced Uyghur-forced-labor allegations (it denies them). This is a hard cap on the US market under UFLPA and a live reputational risk for the overseas expansion (Lens 13).

Lens 3 · Competitive Advantages (moats)

Real moats:

1. Scale & cost leadership. 165 GW cumulative installed base across 42 countries; 15 years of #1 China share (20–25% each year). China's onshore LCOE has fallen to ~USD 33/MWh vs a ~USD 40/MWh global average — Goldwind is the volume anchor of the cheapest wind supply chain on earth. Manufacturing scale + domestic supply chain = a structural cost advantage no Western OEM can match.
2. PMDD technology depth + installed base → service annuity. 20+ years of direct-drive engineering and 50,315 MW under management. Switching an installed fleet's O&M provider is costly; this is the closest thing to a switching-cost moat Goldwind owns.
3. In-house drivetrain + magnet integration partially insulates it from the bearing/drive bottlenecks that constrain rivals.
4. State proximity — access to China Development Bank financing (a historic $6bn low-interest loan) and a domestic policy tailwind.

Where the moat is thin:

• The core product is a commodity. Six Chinese OEMs took the top six global spots in 2025 — Goldwind competes with Envision, Mingyang, Windey and Sany on price in a domestic market that just went through a 62% price collapse (RMB 3,800→1,400/kW, 2019→2024). An 8.9% manufacturing margin is the evidence: there is no pricing power in the hardware.
• Bargaining power is weak on both sides — SOE utility customers are large and concentrated; the rare-earth supply chain is a strategic chokepoint. Goldwind sits in the low-power middle of its own value chain.

Net: the moat is cost + scale + service annuity, not product differentiation. It protects share, not margin.

Lens 4 · Segments

Segment revenue and margin (FY2025, ``) are in the Lens 1 table. The trend that matters:

• WTG manufacturing is accelerating in volume, not margin. External sales capacity hit 26,626 MW, +65.9% YoY, but the segment still only earns 8.9%. Product mix is shifting hard to large turbines: 6–10 MW machines were 70.7% of capacity (18,818 MW), 10 MW+ another 17.6% (4,682 MW), sub-6 MW down to 11.7%. 89% of the backlog is 6 MW+. Bigger turbines = better LCOE and (eventually) better unit economics, but the 2024–25 price war compressed the benefit.
• Wind Farm Investment (43.2% margin) is the margin engine and is being actively rotated — Goldwind builds, operates (self-run utilization 2,290 hrs vs 1,979 national average ), and sells. This is capital-recycling, not a growth annuity.
• Service (20.4% margin) is the quiet compounder — 50,315 MW under management, the highest-quality recurring leg, growing with every turbine shipped.
• Geography: the standout is overseas — international sales rose to 29.4% of revenue in 1H25, up from 21.2% in 2024, with a stated target of one-third of sales from overseas within five years. Overseas order backlog was 7.36 GW at mid-2025.

Segment read: the growth is volume-led and margin-dilutive at the top line; the earnings quality is carried by two small segments (wind farm + service) that together are ~20% of revenue but a disproportionate share of profit.

---

Phase B — Measure performance

Lens 5 · Earnings Result (FY2025)

The headline is a genuine recovery — and the cash-flow footnote is the whole story.

• Revenue: RMB 72.78bn, +28.7% YoY (from RMB 56.52bn). Cross-checked against a second FY2025 read: "RMB 73bn, +28.7%".
• Net profit attributable to owners: RMB 2,774M, +49.1% YoY — more than the +45.8% earnings growth flagged elsewhere; either way, profit roughly doubled off a depressed 2024 base.
• Comprehensive profit margin: 14.18%, +38bps YoY.
• Weighted-average ROE: 7.08%, +217bps YoY — recovering but still thin (see Lens 7 comps: Vestas 23%, Nordex 28%).
• External sales capacity: 26,626 MW, +65.9% — volume drove the top line, not price (though the Dec-2025 bid price had rebounded to RMB 1,622/kW from the ~1,400 trough ).
• Order backlog: 53.7 GW total (50.5 GW external — 11.0 GW awarded bids + 39.5 GW signed — plus 3.2 GW self-development). ~2x annual shipments = strong revenue visibility.

The balance-sheet / cash-flow flags (this is the important part):

• 1H25 operating cash flow was NEGATIVE RMB 2,949M despite record revenue. Growth is being funded by working capital, not generated by it.
• Trade receivables at ~173 days; inventory ~130 days — Chinese-industrial working-capital bloat; revenue is booked well ahead of cash.
• Credit-impairment loss +70.2% YoY in the reporting period, "mainly due to trade receivables" — the receivables aren't just slow, they're getting riskier.
• Interest-bearing debt RMB 48.8bn (~41% of total liabilities); asset-liability ratio ~73%. De-levering slowly (debt −RMB5bn YoY) but still a leveraged balance sheet for a cyclical.
• Cash ~RMB 11.3bn at mid-2025 (~5.9% of assets).

Market reaction / what's priced in: the H-shares are up ~223% over the trailing year — massively outperforming the HK market (+29%) and the HK electrical sector (+90%). The recovery is already in the tape.

Unusual vs its own history: the +65.9% volume surge and the near-doubling of profit are both anomalous — this is a cyclical snapping back from the 2024 price-war trough, not a new steady state. The negative operating cash flow during the best revenue year is the anomaly that should worry a bull.

Lens 6 · Earnings Calls (sentiment trend)

No research-layer transcripts exist; drawn from web coverage of the Q3-2025 (Oct 27 2025) and FY2025 calls, led by Chairman Wu Gang, President Ma, and CFO Wang Hongyan.

• Q3 2025: record revenue, EPS beat, backlog 52.5 GW, interest-bearing debt down, ROE up — but management explicitly flagged that "comprehensive profit margin decreased… the primary driver was a shift in revenue mix toward lower-margin turbine manufacturing". Management is telling you the mix is diluting margin.
• Recurring themes management is pushing: (1) overseas expansion — the "one-third of sales overseas in five years" ambition is now the headline strategic message; (2) larger turbines / Ultra series (GWH204-10MW, GWH266-14MW launched Oct 2025) for better LCOE; (3) cash-flow discipline — the forthcoming 15th Five-Year strategy reportedly centers on overseas markets and improving cash-flow management, a tacit admission the cash conversion is the weak point.
• Tone shift: from defensive (surviving the price war) in 2024 to cautiously offensive (globalize, premiumize) in 2025 — consistent with a management team that thinks the domestic trough has passed.

Lens 7 · Comps

Peer table — wind-turbine OEMs. Multiples are `` with source/date or n/a. Nothing here is fabricated.

`` reconciliation: Goldwind's EV ~$20bn on EV/Sales 1.7x implies ~$11.8bn revenue ≈ RMB 84bn — slightly above the RMB 72.8bn FY2025 actual, i.e. the multiple is on a forward revenue base (consensus FY2026 rev ~RMB 84.6bn).

What the comps say: Goldwind trades cheaper on forward P/E (~12x) than Vestas (~23x) or Nordex (~21x) — but that discount is earned, not free. Its ROE (7%) is a third of Vestas' (23%) or Nordex's (28%). You are paying a low multiple for a low-return, low-margin, working-capital-heavy business in a market with China-discount and forced-labor overhangs. On EV/EBITDA (15.1x) it is actually more expensive than Nordex (12x) and only modestly below Vestas — so the "cheap Chinese leader" framing is a P/E artifact of higher leverage, not a genuine EBITDA-level bargain. Mingyang and Envision multiples: not sourced — I will not fabricate them.

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

Mostly ``, directional (no research-layer tape):

• The +223% trailing-year H-share rally is the dominant recent move, driven by two linked catalysts: (1) the Oct-2024 "self-discipline" anti-involution pact among 12 Chinese OEMs (incl. Goldwind) that ended the price war and rebounded bid prices; and (2) the resulting 2025 earnings recovery (profit +49%).
• Document 136 (Feb 2025) — the NDRC/NEA shift from feed-in tariffs to market-based renewable pricing — is a macro catalyst cutting both ways: bullish for near-term installation rush, structurally uncertain for wind-farm generation economics (Lens 13).
• Record China wind installs — 120 GW grid-connected in 2025 (China = 74% of the 169 GW global market) — the volume tide that lifted the whole sector.
• Historical: Xinjiang/UFLPA headlines (2021) capped the US-market thesis; the 2015 "world #1" milestone and 2019 Vestas overtaking were prior sentiment inflection points.

Pattern the market reacts to: for Goldwind, the stock is a China-policy + industry-pricing beta play far more than an earnings-precision name. It moves on (a) domestic wind-installation policy, (b) turbine-price direction (involution vs pact), and (c) the overseas-share story. It does not trade on cash-flow quality — which is exactly why the working-capital risk is under-priced.

---

Phase C — Judge people & books

Lens 9 · Management

• Wu Gang (Chairman & CEO / founder). Former hydroelectric-engineering professor; joined Xinjiang Wind Energy in 1987, founded Goldwind 1998 with state backing; CCP member, former NPC delegate. Track record: built the #1 China turbine maker and (twice) the #1 global OEM from a state pilot program — a genuinely impressive multi-decade operating record. Archetype: founder-operator with deep state ties — which at this stage means patient, policy-aligned, and unlikely to run the balance sheet for short-term shareholder returns.
• Skin in the game / tenure: founder-led for 25+ years; leadership continuity is high (Wu Gang + President Ma + CFO Wang Hongyan present the results). Specific insider-ownership % not sourced — no insider-transactions.csv on the shelf.
• Capital allocation — the mixed record:
• Good: the build-operate-rotate wind-farm model recycles capital into a 43%-margin asset base; de-levering RMB5bn of debt YoY; growing the service annuity.
• Flag: heavy working-capital absorption (negative operating CF) and a leveraged balance sheet suggest capital is being consumed by growth-at-any-margin in the hardware business.
• Earnings-quality flag (concrete): in 2025 Goldwind's subsidiary sold 28.89M shares of Jinli Permanent Magnet for ~RMB 193M of gains — equal to ~10.53% of 2024 attributable net profit. Booking asset-disposal gains that are ~a tenth of prior-year profit flatters reported earnings and should be stripped out to judge underlying operating quality.
• Red flags: (1) the Xinjiang/XPCC association (Lens 13) — a governance/reputational risk unusual among global peers; (2) reliance on non-operating gains and capital recycling to support the P&L; (3) SOE-adjacent governance means minority H-share holders are not the primary constituency.

Net: a capable, durable founder running a national champion for share and strategic goals as much as for per-share returns. Trust the operating competence; discount the reported-earnings quality.

Lens 10 · Forensic Red Flags

Forensic lens. This is where Goldwind is genuinely concerning — the accounting isn't alleged-fraudulent, but the quality of earnings is low and the cash conversion is poor.

• Cash flow vs earnings — the core divergence. Net profit ~doubled to RMB 2.77bn, yet 1H25 operating cash flow was NEGATIVE RMB 2.95bn. Full-year OCF not cleanly sourced, but a business printing record profit while burning operating cash is the single biggest forensic flag.
• Receivables outrunning revenue. ~173 days trade receivables, with credit-impairment losses +70% YoY driven by receivables. Revenue recognized on turbine sales to SOE customers is being collected slowly and increasingly written down — classic sign that reported revenue quality > cash quality.
• Inventory ~130 days — elevated, consistent with building ahead of the backlog but also a channel/working-capital risk if installs slip.
• Non-operating gains flattering the P&L. The Jinli share-sale gains (~10.5% of prior-year net profit) mean a meaningful slice of "profit growth" is portfolio disposal, not operations. Segment reporting is clear (a positive), but the group net-income line is being helped by below-the-operating-line items.
• Leverage & related financing. RMB 48.8bn interest-bearing debt at ~73% asset-liability ratio; historic reliance on state-bank financing (CDB). Goldwind Capital / finance arm adds financial-services assets to the balance sheet that complicate a clean industrial read.
• SBC / non-GAAP: not a material distortion vector for a Chinese A/H industrial (unlike US tech) — not sourced as a flag.

Regulatory findings (required sub-section):

• SEC (EDGAR LR + AAER): none possible. Goldwind has no CIK and files nothing with the SEC; the research-layer regulatory/regulatory-findings.md (fetched 2026-07-07) confirms total_sec_findings: 0 and notes no EDGAR search is applicable.
• Non-SEC / trade enforcement: the material finding is US trade-compliance exposure, not a securities-fraud action. Goldwind's ties to the sanctioned Xinjiang Production and Construction Corps (XPCC) — an agreement signed Dec 2020, months after the July-2020 US Treasury sanction — and Uyghur-forced-labor allegations from the Tech Transparency Project make it a UFLPA-detention risk for US-bound goods. Goldwind categorically denies forced-labor use. No monetary FTC/DOJ fine sourced; the risk is import-ban / market-access, not a settlement.
• Item 3 (Legal Proceedings) from a 10-K: n/a — Goldwind files no 10-K. Chinese/HK annual-report litigation disclosure not machine-read on the shelf.
• Summary: No securities-enforcement findings (no US regulator has jurisdiction). The live regulatory risk is US/EU trade policy (UFLPA, potential tariffs/anti-dumping on Chinese turbines) rather than accounting enforcement — verified via SEC EDGAR EFTS (nil, no CIK) and web search as of 2026-07-07.

---

Phase D — Project & stress-test

Lens 11 · Forward Projection

Built bottom-up from FY2025 actuals + consensus, into base/bull/bear. Fiscal year = calendar year; FY2025 is the latest actual. Output ``, inputs labeled.

FY2025 actuals anchor: revenue RMB 72.8bn; net income RMB 2.77bn; ~4.23bn shares (H+A) ``; comprehensive margin 14.18%.

Consensus reference: FY2026 revenue ~RMB 84.6bn, net profit ~RMB 3.95bn (~14% earnings CAGR); another read had 2026 EPS ~$0.16, 2027 ~$0.18 (≈RMB 1.1–1.3/sh ``).

Key judgment: the base case is a mid-single-digit-RMB-billion earner growing ~14% — respectable but already the consensus the +223% rally priced in. The dispersion is wide because so much depends on (a) whether the anti-involution price discipline holds and (b) whether overseas margins are genuinely better than the 8.9% domestic hardware margin (unproven at scale).

Brier forecast: skipped in unattended --watchlist mode per SKILL rules (no forecast.ts create). If committed later, the scoreable base call would be: "2208.HK FY2026 net profit ≥ RMB 3.9bn," p≈0.55, resolves 2027-03-31.

Lens 12 · Bull vs Bear

Bull case. Goldwind is the volume anchor of the cheapest, fastest-scaling wind supply chain on earth, at the exact moment global wind installs hit a record 169 GW (+38%) and China alone added 120 GW. It is #1 globally four years running, has a 53.7 GW backlog (~2x shipments), a growing 50 GW service annuity, and a credible overseas-expansion optionality (21%→29% of sales in a year, targeting one-third). The anti-involution pact has stopped the margin bleed and bid prices are rising. It trades at ~12x forward earnings — a fraction of Vestas' 23x. If mix-shift to 10MW+ and overseas both lift the WTG margin even 200–300bps, the earnings base re-rates and the low multiple compounds with the growth.

Bear case (permanent-impairment risks).

1. The hardware is a structural commodity. Six Chinese OEMs at the top of the global table means perpetual price competition; 8.9% is the good-year margin. There may simply be no durable pricing power, ever — the pact is a cartel that history says frays.
2. Cash never shows up. Negative operating cash flow in the record year, 173-day receivables, rising impairments — a business that grows by financing its SOE customers' turbines. A slowdown turns the working-capital build into a writedown cycle.
3. Policy dependence. Document 136's removal of feed-in tariffs shifts wind-farm income to merchant/CfD pricing (provinces underwriting only 40–80%) — directly threatening the 43%-margin wind-farm segment that carries the earnings quality.

Pre-mortem (18 months out, thesis broken): it's early 2028. The self-discipline pact frayed as a new entrant grabbed share; domestic installs cooled after the tariff transition; a couple of large SOE receivables were impaired and a matured wind-farm sale printed a loss instead of a gain. Reported EPS fell despite "record" revenue because the non-operating gains that padded 2025 didn't recur. The stock gave back a chunk of the +223% as the market re-discovered that a 7% ROE doesn't deserve a growth multiple.

Are multiples too high? On P/E, no (12x is cheap). On EV/EBITDA (15x) and on a cash-flow basis, arguably yes — you're capitalizing low-quality, low-cash earnings. The multiple is cheap for a reason.

Contrarian view (what the market refuses to see): everyone is trading Goldwind as "the cheap #1 wind leader riding a record install cycle." What they're under-weighting is that the reported profit is being carried by two small non-commodity segments (wind-farm + service) and by portfolio disposal gains — and Document 136 is aimed squarely at the wind-farm segment's margin. The commodity hardware that the bull thesis celebrates is the part that can't earn its cost of capital.

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

Skeptical short dismantling the bull.

• What structurally breaks the model? The anti-involution pact is a voluntary cartel with no enforcement — the moment one signatory (or a hungry #4/#5 like Windey/Sany) defects for share, the 8.9% margin is back to 5%. China's turbine market has demonstrated it will race to the bottom (−62% price in five years).
• Revenue concentration: heavily domestic SOE utilities buying on price; the receivables prove the customer holds the power. The overseas book is small and the highest-margin overseas markets (US, much of EU) are structurally closed or hostile to a Xinjiang-based OEM under UFLPA and prospective anti-dumping duties.
• Why the moat is weaker than bulls think: it's a cost/scale moat on a commodity, not a product moat. Five other Chinese OEMs have the same cost base. There is no switching cost on the turbine sale (only on the service tail).
• Most dangerous competitor bulls underestimate: Mingyang (601615). It's the domestic offshore leader (~31–41% China offshore share), built the world's first 20 MW turbine, has a 46.9 GW backlog and 2025 revenue growing ~58% — it is out-innovating Goldwind on the highest-value (offshore, mega-turbine) frontier while Goldwind's mix is still onshore-heavy. Envision is #2 globally by volume and privately funded to compete on price.
• Worst capital-allocation / accounting moves: funding SOE customers via receivables; booking Jinli disposal gains worth ~10.5% of prior-year profit to help the P&L; a leveraged, finance-arm-complicated balance sheet.
• What must hold for today's price: (1) the price pact holds and bid prices keep rising; (2) overseas scales to one-third of sales at better margins than domestic; (3) Document 136 doesn't gut wind-farm economics; (4) receivables don't turn into a writedown cycle. That's four things, and at least two are outside management's control.
• If growth disappoints 20–30%: the base-case RMB3.9bn net profit becomes ~RMB2.7–3.1bn, ROE stays ~5–6%, and a 15x EV/EBITDA on a de-rating cyclical with poor cash conversion compresses fast — the +223% rally is the fuel for the drawdown.
• Single scenario that permanently impairs: a hard UFLPA/anti-dumping wall goes up across the US and EU simultaneously and a domestic price-war reignites — capping the high-margin overseas escape route while the domestic core re-commoditizes. Plausibility: moderate. Trade barriers on Chinese cleantech are trending up, not down.

Lens 14 · Management Questions (ordered by information value)

1. Operating cash flow was negative RMB2.9bn in 1H25 on record revenue. What is the concrete plan and timeline to convert reported profit into operating cash, and what receivables-days target are you managing to?
2. The self-discipline pact is voluntary. What happens to your turbine margin if a competitor defects on price, and what is your actual floor margin in that scenario?
3. Overseas is targeted at one-third of sales. Is the margin on overseas turbine sales structurally higher than the ~8.9% domestic hardware margin — and by how much, net of the Brazil factory and localization costs?
4. Document 136 removes feed-in tariffs. What is the revenue and margin impact on the 43%-margin wind-farm segment under a 40–80% provincial underwriting range, and how does it change your build-operate-rotate economics?
5. How much of FY2025 net-profit growth was operating versus non-operating gains (e.g. the Jinli Permanent Magnet disposals), and how should we think about the underlying run-rate?
6. What is your exposure, by revenue and by manufacturing footprint, to markets that apply or may apply UFLPA-style import restrictions or anti-dumping duties on Chinese turbines?
7. Mingyang leads offshore and has a 20 MW machine. What is your offshore and mega-turbine roadmap, and are you ceding the highest-value frontier?
8. Rare-earth magnet prices swing 30–50% a year and China may tighten magnet exports. How hedged is your PMDD cost base, and does export control on magnets help or hurt you net?
9. What is the credit quality of your trade receivables by customer type, and what impairment coverage are you carrying given the +70% YoY jump?
10. What is your through-cycle target ROE, and how do you close the gap to Vestas/Nordex at 23–28%?
11. How large can the service segment (50 GW under management, 20% margin) get as a share of profit, and is it your intended path to margin quality?
12. What is your capital-return policy (dividend/buyback) versus continued reinvestment, and how do minority H-share holders' interests weigh against strategic/state objectives?
13. How much of the 53.7 GW backlog carries firm pricing versus repriceable terms, and what's the cancellation/renegotiation history?
14. What does the 15th Five-Year strategy assume for domestic install volumes post-tariff-transition — flat, growing, or declining from the 2025 record?
15. Under what conditions would you slow volume growth to protect margin and cash, rather than defend share?

---