LONGi Green Energy

Phase A — Understand the business

Company Overview

What it is. LONGi Green Energy Technology Co. (西安隆基绿能, formerly Xi'an LONGi Silicon Materials; "LONGi" honours Jiang Longji, a former president of the founders' university) is the world's largest producer of monocrystalline silicon wafers and one of the top-two solar module makers globally by cumulative shipments. Founded 2000 in Xi'an, Shaanxi; IPO'd on the Shanghai Stock Exchange in 2012. It is the company that dragged the entire PV industry from multicrystalline to mono-PERC in the 2016–2019 window — a genuinely industry-defining bet that made it the cost and technology leader for a decade.

How it makes money. LONGi is a manufacturer selling into a commodity deflation. Four revenue engines, in order of size:

1. Solar modules (the largest line) — finished panels sold to global EPCs, distributors, utilities and installers. FY2025 module shipments 86.58 GW.
2. Monocrystalline wafers — sold externally to other cell/module makers and consumed captively. FY2025 wafer output 111.56 GW, of which 48.57 GW sold externally.
3. Solar cells — mostly captive; 4.31 GW sold externally in FY2025.
4. Adjacencies — BIPV (building-integrated PV, the former Junma/LONGi roofing line), green-hydrogen electrolyzers (LONGi Hydrogen, see Lens 2/3), and a new energy-storage business launched in 2025 ("solar-storage integration"), targeting 6 GWh of storage shipments in 2026.

The contract structure that matters. This is not a take-or-pay, recurring-revenue business — it is spot-priced commodity hardware. LONGi has no pricing power on finished modules; ASPs are set by a >1 TW/yr overbuilt supply chain against ~half that demand. Where it does use long-dated contracts is procurement: multi-year polysilicon supply deals that lock volume, not price ("monthly negotiation, batch purchase") — a structure that protects supply security but leaves LONGi fully exposed to poly price swings (central to the bear case).

One-line model. Buy polysilicon on the spot/monthly market → pull mono ingots → slice wafers → build cells → assemble modules → sell into a global glut at prices that (industry-wide) sit below cash cost. Survival is a function of being the lowest-cost, highest-bankability node in that chain — which LONGi is — plus a differentiated product (back-contact) that can command a price premium the commodity cannot.

Supply Chain — name every node

Upstream → LONGi → end customer, with the actual companies named (this lens fails if generic):

Upstream — polysilicon (LONGi's single biggest input; LONGi does NOT make it).

• Tongwei (通威, 600438.SS) — the world's largest polysilicon producer AND a LONGi competitor in cells/modules. LONGi–Tongwei signed a multi-year polysilicon deal reported at ~RMB 39.1B / ~$5.4B in May 2024; prior contracts 2018 (55,000 MT) and 2022 (203,600 MT). Co-opetition is the defining feature of LONGi's supply chain — its largest supplier is a direct downstream rival.
• GCL (GCL Technology / 协鑫) — granular-silicon (FBR) supply: ~125,000 MT (2024) / ~150,000 MT (2025) / ~150,000 MT (2026); a separate ~91,400 MT long-term pact also reported.
• Daqo New Energy (DQ), Xinte Energy, Asia Silicon, OCI(M) (Malaysia), Jiangsu Zhongneng Silicon — additional poly suppliers named in LONGi disclosures.
• Xinjiang exposure (the compliance chokepoint): LONGi has historically bought poly from producers who source metallurgical silicon from Hoshine Silicon Industry (Xinjiang) — the trigger for the 2021 US Withhold Release Order and module detention (Lens 10) .

Other inputs: solar glass (long-term deals signed to secure supply ); silver paste (a cost line that rose into Q1 2026 and helped widen the loss ); wafer/cell equipment (LONGi builds much of its own line).

Midstream — LONGi's own footprint: integrated wafer→cell→module bases in Shaanxi, Yunnan (hydro-power for low-carbon ingots), Ningxia, Jiangsu, plus overseas cell/module capacity in Malaysia (LONGi Kuching Sdn. Bhd.) and Vietnam (Vina Solar / Vina Cell) — the nodes that put LONGi inside the US AD/CVD net (Lens 10) .

Downstream — end customers: global utilities, IPPs, EPCs, distributors and rooftop installers across China (still the largest single market), Europe, the Middle East (a growth region), Asia-Pacific and Latin America. 2026 guidance is for >50% of module shipments overseas — a deliberate pivot away from the saturated, price-destroyed Chinese domestic market. Chokepoint on the demand side: the US market is largely closed to LONGi via UFLPA + AD/CVD, so First Solar and US-nexus supply own the highest-margin geography.

Competitive Advantages (moats)

Real, durable advantages:

• Scale + cost leadership. #1 cumulative mono-wafer shipments for a decade straight. FY2025 non-silicon wafer processing cost fell ~29% YoY — LONGi is consistently at or near the bottom of the cost curve, which in a below-cash-cost deflation is the moat that decides who is still standing in 2028.
• Bankability / brand. BloombergNEF Tier-1 module maker; in Q1 2026 it became a BNEF Tier-1 energy-storage maker on first entry. In a market where financiers demand 25-year performance warranties, LONGi's balance-sheet survivability is a commercial moat — buyers won't underwrite panels from a maker that might not exist to honour the warranty.
• Technology optionality — the contrarian BC bet. LONGi broke from the industry TOPCon consensus to commit to back-contact (HPBC / HPBC 2.0) cells — the Hi-MO X10 line at 650–660 W mainstream, 670 W peak, 24.8% mass-production efficiency. Independent/field tests show BC out-generating TOPCon by ~1.9–2.45% per kW in desert/shade/soiling conditions and running far cooler. If BC becomes the next mainstream node the way mono-PERC did, LONGi owns the IP and the ramp; if TOPCon commoditises first, LONGi has stranded a contrarian capexCapital expenditureMoney spent on long-lived things — buildings, machines, servers — rather than on running costs. cycle.

Weak or absent moats:

• No polysilicon integration. Unlike Tongwei/GCL/Daqo, LONGi does not own upstream poly — so it captures none of the value when poly prices recover and eats the margin hit as an input-cost taker. This is the structural hole in the moat.
• No pricing power on commodity modules. The moat is relative (lowest cost) not absolute (no economic profit while ASPs sit below cash cost sector-wide).
• Bargaining power is thin both ways — LONGi needs Tongwei's poly as much as Tongwei needs LONGi's offtake; and downstream buyers have a dozen interchangeable Tier-1 suppliers to play against each other.

Segments (revenue / geography)

By product (volume, FY2025): modules 86.58 GW, wafer output 111.56 GW (48.57 GW external), external cells 4.31 GW, BC modules 22.87 GW. Modules are the dominant revenue line; wafers are the volume/cost-leadership anchor; hydrogen + storage are sub-scale but strategic.

The trend that matters — volume up, revenue down. Group revenue fell from RMB ~129B (2023) → 82.6B (2024) → 70.35B (2025) even as module volumes rose (module shipments were ~66 GW in 2023 and 86.58 GW in 2025). That is textbook commodity deflation: ASP fell faster than volume grew. Module ASPs across the industry roughly halved in 2023 and fell a further ~25% in 2024. LONGi is running harder to stay in place.

By geography: China remains the largest single market but is the most price-destroyed; the explicit 2026 strategy is >50% overseas, chasing Europe, the Middle East and emerging markets where ASPs and margins are higher and (crucially) where LONGi is not barred by US trade actions. Segment operating margins are not cleanly disclosed by line; group gross margin was thin-to-negative through the trough — n/a — not separately sourced at the segment level.

---

Phase B — Measure performance

Earnings Result (latest print + trajectory)

FY2025 (full year): revenue RMB 70.35B (−14.8% YoY); net loss attributable to shareholders RMB 6.42B, narrowed by RMB 2.17B from the RMB 8.62B loss of 2024. Operating cash flow turned sharply positive YoY; inventory turnover days fell ~10; selling and admin expenses fell ~30% and ~24%. The narrowing was driven by cost cuts + lower asset-impairment charges, not by a demand or price recovery.

Q1 2026 (the tell): revenue RMB 11.19B (−18% YoY); net loss RMB 1.92B, WIDENED 34% from the RMB 1.44B loss of Q1 2025; EPS −0.25 vs −0.19. Revenue missed consensus by ~16%. This is the single most important data point in the dossier: the FY2025 recovery reversed in Q1 2026 because the anti-involution price rebound lifted polysilicon and silver-paste input costs faster than module ASPs — hammering a maker that buys poly rather than makes it. Q1'26 shipments: wafers 20.49 GW, modules 12.62 GW of which BC 8.34 GW (~66%) — the BC mix target is already being hit.

Multi-year arc: FY2023 net profit ~RMB +10.75B (EPS +1.42) → FY2024 −8.62B (EPS −1.14, first annual loss in 12 years) → FY2025 −6.42B → Q1'26 loss re-widening.

Balance-sheet flags: asset-liability ratio 60.72%, interest-bearing debt ratio 21.45% (H1 2025) — leverage is elevated but not yet distressed for a company this size; LONGi issued China's first private-enterprise tech-innovation green corporate bond, RMB 2.4B, to shore up cash. Recurring asset-impairment losses (cell lines, inventory) have been a repeated drag — a Q1 2025 impairment of ~RMB 426M is illustrative.

Market reaction: the stock sits near RMB 13.13 (2026-07-10), ~80% below the Nov-2021 peak — the market has already priced a long, grinding trough; prints now move the stock on whether losses are narrowing or widening, which is why the Q1'26 re-widening matters.

Earnings Calls (sentiment trend)

The tonal arc:

• FY2024 (early 2025): shock and defence — "first loss in 12 years," heavy impairments, "adverse operating environment," framing survival and cost discipline.
• H1/FY2025: cautious relief — "loss narrowed," "cost reductions and efficiency gains," "positive operating cash flow," leaning on the BC ramp and cost-per-watt story.
• Q1 2026: a subtle pivot away from the near-term P&L to the strategy narrative — the headline is "solar-storage integration strategy accelerates" and "first-time BNEF Tier-1 storage maker," not the widening loss buried inside. When management leads with a new adjacency rather than the core margin, that is itself a sentiment signal.

Recurring phrases (rising): "back-contact / HPBC 2.0," "cost reduction and efficiency enhancement," "solar-storage integration," "overseas / global," "high-quality development." Things they've stopped saying: aggressive capacity-expansion targets and market-share-at-any-cost language that characterised 2021–2022 — the growth story has been replaced by a survival-and-differentiation story. Net read: defensive-turning-strategic, honest about the environment, betting the recovery on BC premium + overseas mix + storage optionality rather than on a market they can't control.

Comps

Peer set: the Chinese solar complex + First Solar as the profitable Western contrast. The entire Chinese group is loss-making, so trailing P/E is n/a for all of them — they trade on price/book, price/sales and shipment-leadership option value, not earnings.

The comp that indicts the group: First Solar earns $14.21 EPS and trades at ~14.5x earnings and ~4.6x sales; the top-four Chinese makers together lost ~RMB 11B ($1.54B) in H1 2025 alone (2.5x the prior-year loss). The market pays a profitability premium (P/S ~4.6x) for un-exposed US CdTe and a survivorship discount (P/S ~1.4x) for the best Chinese volume. LONGi is the highest-quality name inside the discounted group.

Data conflict flagged (do not adopt): some screeners show LONGi P/E ≈ 36 and P/B ≈ 7.1. These are stale (pre-crash) or erroneous — a company posting a net loss cannot have a positive trailing P/E, and P/B ~7 is inconsistent with a 60.7% asset-liability ratio. A rough current P/B is ~1.3–1.5x. Treat the 36x/7.1x figures as artefacts, not signal.

Stock-Price Catalysts (what actually moves 601012)

The 5-year tape is one long structural de-rating punctuated by macro/policy prints:

• Feb–Nov 2021 — the top. All-time high RMB 73.79 (2021-11-01); market capMarket capitalisationThe share price multiplied by the number of shares. What the market says the equity is worth. peaked >RMB 500B (~$78B). Driver: the mono-PERC super-cycle + China carbon-neutrality euphoria.
• 2021 (Oct–Nov) — UFLPA shock (modest). US detained 40.31 MW of LONGi modules (1.6% of annual US exports); a headline risk, but per LONGi "no major impact on operations" — it dented sentiment more than earnings.
• 2022–2023 — poly whipsaw. Polysilicon spiked then crashed (RMB 230k→65k/ton in 2023); LONGi's 2023 investment gains on poll stakes reversed and Q3 2023 profit fell 44% YoY — the first crack.
• 2024 — the loss. Full-year net loss RMB 8.62B, first in 12 years; the stock broke down as the oversupply thesis became undeniable.
• 2025 — false-dawn relief rally. Anti-involution policy hope + poly rebound (+38.9% YTD to Nov 2025) narrowed losses and lifted the group.
• Jan 2026 — the recovery cracks. China's antitrust regulator (SAMR) suspended the ~$7B (RMB 50B) polysilicon capacity-consolidation fund on monopoly grounds, undercutting the OPEC-style supply-discipline thesis — then Q1'26 confirmed the loss re-widening.

Pattern read: 601012 reacts to (1) polysilicon/module price direction, (2) Chinese industrial policy on overcapacity ("anti-involution"), and (3) US trade actions — far more than to any single customer or product launch. It is a macro/policy commodity beta, not an idiosyncratic story stock. That is why the BC differentiation is strategically vital: it is management's attempt to manufacture idiosyncrasy the market will pay for.

---

Phase C — Judge people & books

Management

Founder-controller, mid-crisis handover. Li Zhenguo (李振国) co-founded LONGi in 2000 with college friends and remains the largest individual shareholder at ~14.08% and the designated "actual controller". In a pivotal May 2025 move, Li stepped down as general manager, board member and legal representative, taking the role of President of LONGi's Central Research Institute and CTO — i.e. the founder deliberately moved from running the company to running the back-contact technology bet. Chairman Zhong Baoshen (钟宝申), a co-founder, absorbed the GM and legal-representative roles.

How to read the handover. Two interpretations, both plausible and worth holding simultaneously: (a) conviction — the founder concentrating on the one thing (BC) he believes wins the next cycle, a bull tell of technical seriousness; or (b) accountability optics — a controlling founder stepping back from the P&L after the first loss in 12 years. The reassuring fact is Li did not exit — he retained control and moved to the highest-leverage seat. For a technology-differentiation thesis, a founder-CTO betting his legacy on BC is the alignment you want.

Track record: genuinely elite through 2012–2022 — LONGi created the mono-PERC transition and compounded to a >RMB 500B cap; ROE ran ~20%+ in the good years. Capital allocation is the debatable part: LONGi expanded capacity aggressively into the 2021–2022 boom alongside the whole industry — the collective overbuild that produced the >1 TW glut. LONGi didn't cause the glut alone, but it participated in it; the 2024–2025 impairments are partly the bill for that expansion. On the constructive side, in the downturn management has shown discipline — deep opex cuts (~30%), inventory-day reduction, green-bond funding, and a refusal to chase share into the worst-priced domestic volume.

Skin in the game: high (founder ~14%, insiders ~23% per screener aggregates ). Dividends were effectively suspended in the loss years — appropriate. Red flags: the Tongwei co-opetition (largest supplier is a rival) is a structural governance oddity more than a scandal; no evidence of self-dealing or promotional accounting surfaced in this pass (see Lens 10).

Forensic Red Flags

Accounting-risk map (web-only; no filings on shelf to tie out — flag every figure):

• Impairment cadence. Recurring asset-impairment losses on cell lines and inventory (e.g. ~RMB 426M in Q1 2025). In a deflation this is expected and arguably conservative — the risk is the reverse (under-impairing to flatter the P&L). Watch whether impairment charges keep falling (management's claim) or re-accelerate as older PERC/early-TOPCon lines are stranded by the BC pivot. Stranded-asset risk on non-BC capacity is the key forensic question.
• Receivables / inventory vs revenue. With revenue falling and volumes rising, inventory and receivables risk outrunning sales; H1'25 reported falling inventory days (−10), a good sign, but the base disclosures aren't on the shelf to verify. n/a — not independently tied out.
• Poly-stake investment income. LONGi has historically booked investment gains/losses from minority stakes in polysilicon ventures — a swing item that flattered 2021–2022 and reversed after. Non-operating and volatile; strip it when judging core margins.
• SBC / non-GAAP: A-share reporting is PRC-GAAP; there is no US-style non-GAAP add-back culture to distort here — a modest positive on accounting-quality risk relative to US peers.

Regulatory findings (required sub-section).

• SEC / EDGAR: None possible. Per regulatory/regulatory-findings.md (fetched 2026-07-10), LONGi has no CIK and no SEC filing obligation; SEC EDGAR EFTS (LR + AAER) returned 0 findings. This is an absence-of-search, not an absence-of-risk.
• US Customs / UFLPA (material). In Oct–Nov 2021, CBP detained 40.31 MW of LONGi modules under the Xinjiang forced-labour enforcement regime; LONGi sources poly from producers linked to Hoshine Silicon (Xinjiang), which carries a Withhold Release Order. The UFLPA names polysilicon a priority enforcement sector. Ongoing exposure: any LONGi product entering the US must rebut the forced-labour presumption — a de-facto barrier to the highest-margin market.
• US AD/CVD (material, active). US Commerce issued final affirmative AD/CVD determinations (2025-04-21) and AD/CVD orders (June 2025) on solar cells/modules from Cambodia, Malaysia, Thailand and Vietnam — directly hitting LONGi's Malaysia (Kuching) and Vietnam (Vina Solar/Vina Cell) capacity. A September 2025 CBP EAPA interim-measures notice on PV imports (Cons. 8163) indicates continued anti-circumvention scrutiny. Net: LONGi's Southeast-Asian export route into the US is effectively closed — reinforcing the >50%-overseas-ex-US strategy.
• Other agencies (FTC/DOJ/etc.): no material web hits beyond the trade/forced-labour matters above.
• Verdict: no accounting-fraud findings surfaced; the regulatory risk is trade and forced-labour compliance, not securities enforcement — and it is largely already priced/known (US market treated as closed).

---

Phase D — Project & stress-test

Forward Projection

No our model logged (unattended watchlist run; and the base case is a range, not a committed conviction). EPS is projected qualitatively; every input.

Framework: LONGi's earnings = (module+wafer volume) × (ASP − unit cost). Volume is guided ~100 GW wafers / ~80 GW modules in 2026. The swing factor is entirely spread: module ASP vs. the poly+silver+processing cost stack.

• Bear path (FY2026 loss ~RMB 5–7B, EPS ≈ −0.7 to −0.9): anti-involution discipline stays blocked (SAMR suspension holds), poly grinds sideways-to-up on input costs while module ASPs stay ~$0.09/W below cash cost; Q1'26's re-widening is the run-rate. Losses persist through 2026, breakeven slips to 2027+.
• Base path (FY2026 loss ~RMB 2–4B, narrowing; FY2027 ~breakeven): cost cuts + rising BC mix (>65%, at a real price premium) + >50% overseas partly offset flat ASPs; operating cash flow stays positive; the loss narrows sequentially through H2'26 and approaches breakeven in 2027 as the worst capacity exits.
• Bull path (FY2026 near-breakeven; FY2027 modest profit, EPS ~+0.2–0.5): anti-involution supply discipline is revived in some form, ASPs clear cash cost, BC commands a durable 5–8% generation/price premium, and LONGi's cost lead converts to positive gross margin ahead of peers. Return to the RMB +10B earnings power of 2023 is a multi-year, not a 2026 proposition.

The pivotal variable is not LONGi's execution — it's the industry clearing price, which is a policy/capacity question LONGi cannot control. That is why this is a WATCHING name, not a BUY: the company is doing the right things (cost, BC, overseas, storage) but its P&L is hostage to a sector deflation whose end date is unknowable and, as of Q1'26 + the Jan'26 SAMR block, not yet in sight.

Bull vs Bear

Bull case. LONGi is the last-man-standing trade. A >1 TW industry earning collective losses must consolidate; when it does, the lowest-cost, best-capitalised, most-bankable node captures disproportionate profit on the other side. LONGi is that node — decade-long wafer cost leadership, a differentiated BC product that out-generates TOPCon and can command a premium, a founder-CTO betting his legacy on it, positive operating cash flow even in the trough, and optionality in hydrogen + storage. Buy the survivor at ~1.4x sales and ~80% off the peak, and let mean-reversion in the sector clearing price + BC premium do the work. Secular tailwind: global PV demand still compounds; someone builds the panels, and it won't be the dozens of sub-scale Chinese makers bleeding out.

Bear case. Three ways this permanently impairs:

1. Structurally short its own input. Every anti-involution "recovery" that lifts polysilicon squeezes LONGi's un-integrated spread — the recovery hurts the stock. Q1'26 is the proof, not a blip. Tongwei/GCL, who own the poly, capture the rebound; LONGi pays for it.
2. The clearing price never clears. SAMR blocked the RMB 50B consolidation fund (Jan 2026) on antitrust grounds; local governments protect zombie capacity; 570kt of poly inventory (~300 GW latent supply) caps any rally. Deflation runs years longer than balance sheets can bear, and even the leader dilutes or levers to survive.
3. BC is a stranded contrarian bet. If TOPCon commoditises to cost-parity before BC scales, LONGi has spent a capex cycle differentiating into a premium the market won't pay — and must impair the non-BC lines it's replacing.

Pre-mortem (18 months out, thesis broken): it's early 2028; module ASPs never cleared cash cost because capacity discipline failed; LONGi posted a third straight annual loss, raised equity/converts to shore up the balance sheet (diluting the ~14% founder stake), the BC premium compressed as peers shipped competitive BC, and the US market stayed shut. The "survivor" survived — but as a lower-return, higher-share-count utility of a business, not the RMB +10B compounder of 2023.

Are multiples too high? No — ~1.4x sales / ~1.3–1.5x book on a loss-maker is a survivorship valuation, not an expensive one. The risk is not over-valuation; it's a value trap where the trough lasts longer than the patience (or the balance sheet).

Contrarian view (what the market refuses to see): the consensus frames a poly/module price recovery as unambiguously bullish for LONGi. It is not — because LONGi doesn't own poly, the first leg of any input-cost recovery compresses its margin before pricing power returns downstream. The clean bull trigger is not "prices go up," it's "module ASP rises faster than the poly+silver cost stack for two consecutive quarters" — a spread event, not a price event.

Devil's Advocate (short-seller)

Dismantling the bull case. Where the money is concentrated: LONGi is a leveraged bet (asset-liability 60.7%) on a single macro variable it does not control — the Chinese solar clearing price. The moat is thinner than bulls think: "lowest cost" is a relative rank in a race where everyone is below cash cost; being least-unprofitable is not a moat, it's a slower death unless the sector consolidates — and Beijing just blocked the consolidation mechanism (SAMR, Jan 2026). The most dangerous competitor bulls underestimate is Tongwei, not Jinko: Tongwei owns the polysilicon, so it can survive lower and squeeze LONGi's input margin at will — the supplier is the assassin. The most dangerous product competitor is TOPCon at scale — if BC's generation premium (~2%) doesn't translate into a durable price premium net of BC's higher manufacturing cost and lower early yields, the contrarian bet is a capex sinkhole. Worst capital-allocation history: boom-era capacity expansion that helped build the very glut now destroying returns. What must hold for today's price: that LONGi survives to the other side without a materially dilutive equity raise and that BC becomes mainstream — two live, unproven assumptions. If 2026 volumes disappoint 20–30% (a demand air-pocket from US-tariff spillover + Chinese subsidy rollback), LONGi burns cash, leverages further, and the "survivor" valuation re-rates down toward book. The single scenario that permanently impairs: a multi-year sub-cash-cost deflation that forces a dilutive recap — the founder's 14% halves, the option value transfers to new capital, and equity holders own a diluted commodity utility.

Management Questions (ordered by information value)

1. LONGi is not polysilicon-integrated — as poly prices recover under anti-involution, how do you protect the wafer/module spread, and would you vertically integrate into poly given the Q1'26 margin hit?
2. What is the all-in cash cost per watt for HPBC 2.0 vs. mainstream TOPCon today, and what net price premium is the market actually paying for BC generation after that cost gap?
3. What are the concrete conditions under which you would cut utilisation or idle capacity rather than ship below cash cost to hold share?
4. With SAMR having suspended the RMB 50B consolidation fund, what is your realistic base case for when industry module ASPs clear cash cost — and what does LONGi do if that is 2028, not 2026?
5. How much of current wafer/cell capacity is non-BC and at risk of stranding/impairment as you migrate to HPBC 2.0, and what is the remaining book value exposed?
6. Under what scenario would you need to raise equity or convertibles, and where is your line on protecting existing shareholders from dilutionDilutionIssuing new shares, so each existing share owns a smaller slice of the same company.?
7. Li Zhenguo's move to CTO/Central Research Institute — what decision rights does he now hold over the BC roadmap vs. the board, and how is capital allocated between defending the core and funding BC?
8. Storage (6 GWh target 2026) and hydrogen — are these real profit engines or diversification optics, and what capital are they consuming while the core loses money?
9. What is your US strategy given UFLPA + the 2025 AD/CVD orders effectively close SE-Asian export routes — do you cede the highest-margin market permanently?
10. On the Tongwei co-opetition: how do you manage a supply relationship with your largest polysilicon vendor who is simultaneously a downstream competitor?
11. Overseas is guided >50% of modules — what are the realised gross margins overseas vs. China, and how defensible are they against local-content/tariff regimes (EU, India, US)?
12. What impairment charges should investors expect in FY2026, and what is your framework for timing impairments in a prolonged deflation?
13. What is your covenant headroom at 60.7% asset-liability ratio, and what leverage ceiling will you defend?
14. If BC generation premiums compress as peers ship competitive back-contact, what is Plan B for product differentiation?
15. What is the single operating metric (not price) by which you want to be judged over the next eight quarters?

---