Hua Hong Semiconductor

Phase A — Understand the business

Lens 1 · Company Overview

Hua Hong is a pure-play specialty wafer foundry — it manufactures chips to other companies' designs and gets paid per wafer, plus qualification/engineering fees. It does not chase the leading edge (no 5nm/3nm logic, no AI-GPU fabrication). Its franchise is mature and specialty process nodes (0.35µm down to 40nm, with a 28nm ramp underway), where reliability, analog performance, and local Chinese supply matter more than transistor density.

Five specialty process platforms (the company's own taxonomy):

1. Embedded non-volatile memory (eNVM) — smart cards, MCUs, secure elements (~31% of FY24 revenue).
2. Discrete / power devices — IGBTs, super-junction MOSFETs, the automotive/industrial power franchise (~26% FY24; ~15% of revenue was automotive as of 2022).
3. Analog & power management (incl. BCD) — PMICs, the platform most levered to AI-server power delivery (~22% FY24).
4. Logic & RF — connectivity, ultra-low-power specialty logic; 40nm ULP entered mass production in 2025 (~14% FY24).
5. Standalone non-volatile memory — NOR flash, EEPROM (~6.5% FY24, but the fastest grower, +44.2% in FY25).

Wafer mix (FY24): 8-inch 50% / 12-inch 50% by revenue. By Q1 2026 the 12-inch share had climbed to 62.7% of revenue — the ramp is visibly shifting the mix toward the newer, higher-value 12-inch lines.

Customers: fabless designers and IDMs — automotive, industrial, energy, connectivity, consumer. The strategic pitch is "China for China": be the domestic manufacturing option for Chinese and multinational customers who need to de-risk supply out of Taiwan (a theme the new Intel-veteran president has made explicit). STMicroelectronics is a named 12-inch technology/capacity partner in Wuxi.

Contract structure: short-cycle pay-per-wafer, not take-or-pay and not recurring/subscription. Revenue is a function of (wafers shipped × ASP × utilization). That makes the model highly cyclical and ASP-sensitive — the single most important variable in the entire thesis (see Lens 5, 11, 13).

Scale/rank: #2 Chinese foundry (behind SMIC) and roughly 7th-largest foundry globally by revenue.

Lens 2 · Supply Chain

Hua Hong sits in the middle of the chain — it converts equipment + materials into wafers for fabless/IDM customers. Named stakeholders:

Upstream — equipment (the chokepoint):

• Lithography: ASML DUV immersion tools are the binding input. Hua Hong runs mature nodes reachable with DUV + multipatterning — it does not need EUV, which is precisely why US/Dutch controls have (so far) not decapitated it the way they did SMIC's leading edge. China was ASML's largest market at ~33% of 2025 revenue, and ASML sold ~70% of its DUVi systems into China in 2024. Chokepoint risk is the future tightening (MATCH Act — see Lens 10), not today's toolset.
• Deposition/etch/metrology: Applied Materials, Lam, KLA — Western-dominated, increasingly substituted by domestic tools (AMEC/Naura) as localization advances.
• Materials: silicon wafers (Shin-Etsu, SUMCO, and domestic Chinese suppliers), photoresists, specialty gases.

Midstream — Hua Hong's fabs:

• Legacy 8-inch fabs in Shanghai (Fabs 1–3, the HHNEC heritage).
• 12-inch Fab 7 (Wuxi, first line, up to ~28nm).
• 12-inch Fab 9 (Wuxi second line, the growth engine — 65/55/40nm, ramping to ~83K wpm by YE2026).
• HLMC (Huali) fabs in Shanghai — absorbed Dec 2025, adding ~38K wpm of 65/55/40nm 12-inch capacity.

Downstream: wafers ship to OSAT/packaging then to end-OEMs across automotive (the strategic prize), industrial, consumer electronics, smart-card/security, and — increasingly — AI-server power management (PMICs feeding the >$700B/yr hyperscaler capex wave indirectly).

Single-source dependency: ASML DUV lithography is the one genuinely irreplaceable input, and it is exactly the node the US is trying to close. Everything else has a domestic substitution path underway.

Lens 3 · Competitive Advantages (moats)

Hua Hong's moat is real but shallow and geopolitically contingent — it is not a Buffett moat, it is a policy moat.

• Specialty process depth (moderate durable moat): two decades of qualified eNVM, BCD, and power-discrete recipes. Process qualification + automotive reliability certification (AEC-Q100) creates switching costs — a customer who qualifies a power IC at Hua Hong does not re-spin lightly. This is the closest thing to a genuine moat.
• "China for China" localization (policy moat, not economic moat): as US-China decoupling forces Chinese fabless (and multinationals selling into China) to dual-source domestically, Hua Hong is one of only two credible domestic options at scale. This is a real demand tailwind but it is a moat granted by Beijing and Washington, not earned by the company — it can be widened (more sanctions on Taiwan supply) or eroded (SMIC/Nexchip out-expanding it) by forces outside management's control.
• State-underwritten capital (double-edged): access to the Big Fund, Shanghai state co-investment, and cheap debt lets Hua Hong build $6.7B fabs it could never finance on its own ~11-13% gross margins. This is a moat against private competitors but it also subsidizes the industry-wide overcapacity that crushes its own ASPs — the state funds both Hua Hong and its rivals.
• Bargaining power — WEAK: Hua Hong is a price-taker on both sides. Upstream, it needs ASML more than ASML needs it. Downstream, in a commoditized mature-node market it competes on price with SMIC, VIS, UMC, PSMC, Nexchip, GlobalFoundries, and Tower. The only reason ASPs are rising in 2026 is an industry-wide supply/demand tightening (AI power demand + capacity discipline), not Hua Hong's pricing power.

Verdict on moat: specialty qualification = a moderate switching-cost moat; localization = a strong but exogenous demand tailwind; pricing power = essentially none. Rivals matter enormously.

Lens 4 · Segments

By process platform (share of revenue and growth) — all ``, no segments.csv on disk:

Trend (accelerating): the mix is tilting toward the highest-growth, higher-value platforms — standalone memory (+44%) and analog/BCD (+41%) — which are the two most exposed to (a) the NOR-flash cycle and (b) AI-infrastructure power management. Q1 2026 confirmed the acceleration: MCU, standalone flash, and BCD were the top growth drivers.

By wafer diameter: 12-inch rose from 50% (FY24) → 62.7% (Q1 2026) of revenue — the Fab 9 ramp is doing exactly what it was built to do, shifting revenue to the newer lines.

By geography: not cleanly disclosed. Company describes itself as a China-based platform with global reach; one source puts Asia + North America at ~75% of revenue. The strategic direction is explicitly more China-domestic ("China for China"). No hard geographic segment table is publicly broken out — flagged as an open item.

---

Phase B — Measure performance

Lens 5 · Earnings Result (Q1 2026, reported ~May 2026)

The print:

• Revenue: $660.9M, +22.2% YoY, +0.2% QoQ — a record.
• Gross margin: 13.0%, +3.8pp YoY (from ~9.2%).
• Net profit attributable to shareholders: $20.9M, +458.1% YoY, +19.9% QoQ.
• Capacity utilization: "high across all platforms" (FY25 ran at 106.1% — above 100% on both 8- and 12-inch, among the highest in the industry).
• 12-inch revenue share: 62.7% (up from ~50% a year earlier).

Drivers: MCU, standalone flash, and BCD led; growth came from both wafer volume and ASP — the first time in the recent cycle pricing turned into a tailwind rather than a headwind.

Guidance (the bullish part):

• Q2 2026: revenue $690–700M (~+5% QoQ), gross margin 14–16%.
• Full-year ASP: +10–15%, with some platforms +20–25%. This is the crux of the bull case — management is guiding to a genuine pricing cycle.

Context — the recovery arc:

• FY2024 was the trough: revenue $2.00B, −12.3% YoY, earnings −79% — the mature-node price war bit hard.
• FY2025 recovered: revenue $2,402.1M, +19.9%, GM 11.8% (+1.6pp), shareholder net profit $54.9M.
• Q4 2025: revenue $659.9M (record), +22.4% YoY, GM 13.0%, shareholder NI $17.5M.

The balance-sheet / structural flag (do not skip): FY25 shows a positive $54.9M net profit attributable to shareholders but a consolidated net LOSS of −$110.8M (which itself narrowed 21.1% YoY). The gap is minority interest — the Wuxi JV and HLMC assets are majority-owned but heavily co-funded by the Big Fund and Wuxi state entities, so a large slice of the losses (from depreciation on the new fabs) sits below the line with the minorities. The consolidated enterprise is still losing money on a fully-loaded basis; the shareholder line is flattered by who absorbs the fab depreciation. This is the single most important number in the dossier and it is invisible if you only read the headline EPS.

Margin reality check: at 13–16% gross margin, Hua Hong earns roughly half the gross margin of UMC (~29%) or GlobalFoundries, and well below SMIC (~22%). It is the low-margin end of an already-cyclical business.

Lens 6 · Earnings Calls (sentiment trend)

No transcripts on disk; sourced from web summaries of the last several quarters.

• FY24 calls (trough): defensive tone — surviving the price war, defending utilization, deferring the margin question.
• Q3 2025: cautious optimism on the 12-inch ramp; "signals optimism for 3Q on 12-inch ramp-up".
• Q4 2025 / FY25: confident — "record revenue," specialty platforms "flourishing," explicit pride in +44% standalone memory and +41% analog growth.
• Q1 2026 (most bullish in the cycle): management raised guidance, called out ASP as a tailwind for the first time, and guided full-year pricing +10–15% (some platforms +25%). The stock jumped ~12% on the beat-and-raise.

Tone shift: a clean arc from survival (2024) → ramp execution (mid-2025) → pricing power returning (2026). What they stopped saying: the defensive overcapacity/price-war language of 2024. What they started saying: ASP increases, "China for China" localization, and specialty-platform leadership. New president Bai Peng (ex-Intel) is pushing a sharper localization narrative. Caveat: this is management's framing on the way up in a cyclical — exactly when tone is least reliable as a forward signal.

Lens 7 · Comps

Mature-node / specialty foundry peer set. Multiples are `` with source/date, or n/a. Do not read these as precise same-day marks — they are drawn from multiple aggregators across early-mid 2026 and mix trailing/forward conventions; treat as order-of-magnitude.

Read: the whole mature-node group is trading at cyclical-peak-anticipation multiples (Tower ~50x EV/EBITDA, GFS ~20x) because the market is pricing an ASP rebound. Hua Hong is the worst-quality name in the group on margin (11–16% GM vs peers 22–29%) yet trades at a rich P/B and a P/E that is barely meaningful because shareholder earnings are tiny and the consolidated entity loses money. On any normalized-earnings basis it is expensive; the bull must argue that H-share/A-share scarcity + policy + a steep ASP ramp justify it. The A-share (688347) trades at a large premium to the H-share (1347) — a persistent China-listing quirk, not a fundamental. Note the aggregator disconnect: one screen shows a 22-analyst "Buy" consensus with a HK$113 target — i.e. ~41% below the ~HK$193 spot — a rare case where the consensus rating and consensus price target flatly contradict each other, which tells you the sell-side itself is torn.

Lens 8 · Stock-Price Catalysts (moves that mattered)

52-week range: HK$33.00 → HK$201.60; spot ~HK$192.80 (June 2026). That is a ~6x move off the low — an extraordinary re-rating that dwarfs the fundamental improvement (revenue +20%, still sub-16% GM). The tape is being driven by narrative and flows, not earnings.

Pattern of >5% movers over the cycle:

• Aug 2023: STAR Market (A-share) IPO — China's biggest IPO of the year; raised capital for the Wuxi build.
• 2024: grind lower — mature-node price war, revenue −12%, earnings −79%. The stock bottomed near HK$33.
• H2 2025: the turn — "panic over mature-node capacity crunch drives orders to Chinese foundries"; SMIC/VIS/Nexchip price hikes signalled the cycle inflecting.
• Late 2025: HLMC absorption announced (Dec 31, 2025) — consolidation/scale narrative.
• Q1 2026 (May 2026): ~+12% on the beat-and-raise (ASP guide +10–15%).
• 2025–26 broadly: swept up in the China AI / self-sufficiency / "buy domestic semis" thematic trade (the same flow lifting SMIC, Cambricon, etc.).

What the market actually reacts to for this name: (1) mature-node pricing signals (any SMIC/VIS/TrendForce price-hike headline), (2) China policy / self-sufficiency flows (thematic, not company-specific), (3) capacity/consolidation news (Wuxi, HLMC), and — much less — the actual margin/EPS print. This is a policy-and-flow stock wearing a foundry costume. That cuts both ways: the same thematic flow that took it 6x can reverse violently on a China-tech risk-off.

---

Phase C — Judge people & books

Lens 9 · Management

Bai Peng (Peng Bai) — President (from Jan 1, 2025) and Chairman (from Oct 31, 2025). The pivotal recent change: Hua Hong recruited a 62-year-old Intel veteran (30+ years in semiconductor manufacturing, ex-Intel VP, ex-CEO of Rong Semiconductor) and, within 10 months, handed him both the president and chairman roles. Predecessor Junjun Tang moved from president to chairman/executive director at end-2024, then Bai consolidated the chair.

Read on the operators:

1. Track record: Bai brings genuine leading-edge manufacturing pedigree (Intel) into a company that has deliberately avoided the leading edge — the bet is that Intel-grade process/yield discipline can be applied to make Hua Hong the best-run mature-node fab, and to sharpen the "China for China" localization playbook. Credible hire; unproven at this company.
2. Tenure & skin in the game: Bai is on a 3-year service contract — this is a professional-manager, state-appointed structure, not a founder-owner with meaningful equity. Insider ownership by management is immaterial; the real "owners" are the Shanghai state and the Big Fund (see Lens 13). No insider-transactions.csv on disk; no evidence of large personal management stakes.
3. Capital allocation — the flag: management does not really control capital allocation; Beijing does. The decision to plow $6.7B into Wuxi and to absorb HLMC via share issuance are state-industrial-policy decisions executed through Hua Hong, funded by the Big Fund + Shanghai + Wuxi. ROE/ROIC is structurally depressed — the company reinvests enormous state capital at sub-cost-of-capital returns (consolidated net loss) because the objective function is national capacity, not shareholder IRR.
4. Red flags: heavy related-party dynamics — the controlling shareholder (Hua Hong Group), the Big Fund, and Shanghai Alliance Investment sit on both sides of the JV and the HLMC deal (they were the counterparties who received the 191M new shares at RMB43.34). This is not fraud, but minority shareholders are structurally junior to the state agenda.
5. Archetype: classic state-owned-enterprise professional management — competent operators executing an industrial-policy mandate, with a marquee Western-trained technologist bolted on for credibility and yield. Implication: expect capacity and localization to be prioritized over margins and per-share returns at every fork.

Lens 10 · Forensic Red Flags

No financials.csv/filings on disk — this is a web-only forensic read, so it is directional, not line-item verified. That caveat is itself a flag: as a Chinese SOE-foundry, disclosure granularity is lower and auditability (from a Western minority-holder seat) is weaker than for an EDGAR filer.

Accounting / structural risks:

• Minority-interest masking (the big one): the $165M gap between +$54.9M shareholder profit and −$110.8M consolidated net loss (FY25) means the headline EPS materially overstates the economic profitability of the consolidated enterprise. Any bull citing "profitable and growing" off the attributable line is being misled by the JV structure.
• Depreciation timing: a company mid-way through a multi-fab ramp ($6.7B Wuxi + HLMC) has a wall of future depreciation hitting as the new lines are placed in service. Today's 13–16% gross margin is on a partly-depreciated base; as Fab 9 and HLMC fully load, D&A steps up and can compress reported margins even if ASPs hold. Watch the gross-margin bridge vs. capacity additions carefully.
• Capex vs. cash flow: FY26 capex guided ~$1.62B against ~$2.4–2.7B revenue and thin margins → deeply FCF-negative on a consolidated basis; the gap is plugged by state equity + debt ($2.68B of the Wuxi JV alone is debt-financed). This is a capital-consumption story, not a cash-generation one.
• Related-party asset injection: the HLMC absorption (RMB8.27B, ~97.5% via new-share issuance to Hua Hong Group / Big Fund / state funds) is a related-party transaction that dilutes existing minorities and whose "fair value" is set by affiliated counterparties, not an arm's-length market.
• R&D intensity: CNY 1.994B (~$288M) FY25 R&D, +21% YoY, ~11.5% of revenue — genuinely high for a mature-node foundry; this is a real cash cost supporting the specialty roadmap, not a flatter.
• SBC: the new-share issuance for HLMC is dilution, but not classic SBC-flattered non-GAAP — Hua Hong reports on IFRS/HK-GAAP, not a heavy non-GAAP adjustment culture. Lower risk here than a US growth name.

Regulatory findings (required sub-section):

• SEC (EDGAR): None possible — Hua Hong has no CIK and does not file with the SEC. regulatory/regulatory-findings.md confirms total_sec_findings: 0 and notes the no-CIK limitation.
• US export-control exposure (the material item): Hua Hong is specifically named in the proposed MATCH Act (Multilateral Alignment of Technology Controls on Hardware) alongside SMIC, Huawei, CXMT, and YMTC — legislation that would impose Entity-List-style restrictions on exports/services/technical support. Separately, Tom's Hardware reported the US "stopped exports of tools" to Hua Hong/Huali in the context of a reported 7nm effort in Shanghai. This is the dominant regulatory risk — not accounting enforcement, but the risk of losing access to ASML DUV and Western equipment.
• Non-SEC (FTC/DOJ/FDA/etc.): web search for "Hua Hong Semiconductor" (FTC OR DOJ OR settlement OR fine OR penalty) enforcement surfaced no material US civil/criminal enforcement action against the company itself (as distinct from export-control listing risk). No consent decrees or fines found.
• Item 3 / Legal Proceedings: n/a — no 10-K on disk (foreign filer). Not verifiable from the research layer.
• Net: no accounting-fraud or enforcement findings surfaced. The regulatory risk that matters is US/allied export controls, which is a forward policy risk on inputs, not a past enforcement finding. Verified via SEC EDGAR EFTS (no CIK → n/a), web search, and industry press as of 2026-07-06.

---

Phase D — Project & stress-test

Lens 11 · Forward Projection

Hua Hong reports EPS in cents; the shareholder earnings base is small and volatile, so I project revenue and gross-margin dollars (the honest drivers) and derive an attributable-earnings range. All `` with arithmetic; anchored on FY25 actuals + Q1/Q2-2026 guidance. No forecast.ts create in watchlist mode.

Anchors: FY25 revenue $2,402.1M, GM 11.8%, shareholder NI $54.9M. Q1-26 $660.9M/13.0%GM; Q2-26 guide $690–700M/14–16%GM. FY26 ASP guide +10–15% (some +25%); capacity ramping to ~83K wpm (Fab 9) + ~38K wpm (HLMC).

FY2026 (base):

• Revenue: ~$2.8B.
• GM: ~14–15%.
• Attributable NI: ~$90–120M.
• Bull: revenue ~$2.95B, GM ~16%, NI ~$150M if ASP hits the top of the +15–25% guide and utilization stays >100%.
• Bear: revenue ~$2.6B, GM ~12%, NI ~$40M (or a fresh consolidated loss deepening) if China overcapacity re-collapses pricing in H2.

FY2027:

• Revenue: ~$3.2–3.5B.
• GM: ~15–18%.
• Attributable NI: ~$130–200M base, but this is where the depreciation wall is most dangerous — margins could stall.

FY2028:

• Revenue: ~$3.6–4.0B.
• GM: ~16–19%, NI ~$180–280M base.

Sensitivity that dominates everything: at ~13–16% gross margin, Hua Hong's earnings are a thin residual — a 3-point ASP swing (well within the guided +10 to +25% range or a downside price war) moves gross profit by ~$80–120M, i.e. more than the entire attributable net income. EPS is essentially levered ×several to ASP. This is not a compounder you can DCF with confidence; it is a cyclical option on mature-node pricing. Any "fair value $120" from an aggregator is a single-scenario point estimate on a distribution this wide — treat with skepticism.

Lens 12 · Bull vs Bear

Bull case. China's mature-node cycle has inflected: after the 2024 price war, capacity discipline plus surging AI-infrastructure demand for power ICs has flipped the market to shortage, and foundries are pushing price increases — SMIC +10% BCD, VIS +4–8%, and Hua Hong guiding +10–25% ASP for 2026. Hua Hong is a triple-levered play on that turn: (1) ASP up, (2) volume up as Fab 9 (~83K wpm) and HLMC (~38K wpm) ramp, (3) mix up as 12-inch and the +44%-growth memory / +41%-growth analog platforms scale. Layer on the structural localization tailwind — every escalation of US-China chip tension forces more domestic dual-sourcing straight into Hua Hong's lap — plus a marquee ex-Intel operator and a state that will fund the build no matter what. Utilization >100%, revenue records every quarter, guidance being raised. If margins normalize toward even 20% (still below UMC) on a $3.5B+ revenue base, the earnings power is multiples of today's.

Bear case (permanent-impairment risks).

1. Margin is structurally broken and the depreciation wall is coming. 11–16% gross margin is half the peer group, and the consolidated entity loses money (−$110.8M FY25) once you stop letting the minorities eat the fab losses. As $6.7B of Wuxi + HLMC assets get placed in service, D&A steps up — margins can stall or fall even with higher ASPs. The "profit" is a JV-accounting artifact.
2. The bull case IS the bear case — Chinese overcapacity. The exact same state-funded expansion that gives Hua Hong volume is being replicated at SMIC, Nexchip, PSMC, and every other Chinese fab, all funded by the same Big Fund. ~70% of global 12-inch mature-node expansion is Chinese. The 2024 price war showed what happens when that capacity outruns demand; the 2026 pricing rebound is cyclical and could reverse the moment AI-power demand cools or the new capacity floods in.
3. Export-control decapitation risk. The MATCH Act and reported tool-export stoppages threaten Hua Hong's access to ASML DUV and Western equipment. Mature nodes have survived so far, but a genuine DUV cutoff would cap the entire growth roadmap.
4. Valuation. A ~6x stock (HK$33→193) on ~+20% revenue and sub-16% margins prices a normalization that is far from certain; one consensus screen implies ~41% downside to a HK$113 target.

Pre-mortem (18 months out, thesis broke): it's early 2028. AI-server power-IC demand normalized, and the wave of Chinese 12-inch capacity (Hua Hong's own Fab 9 Phase II + SMIC + Nexchip) all came online into a softer market. Mature-node ASPs rolled over −15%. Hua Hong's just-placed-in-service Wuxi/HLMC depreciation hit the P&L at exactly the wrong time. Gross margin fell back to ~10%, the consolidated loss widened, and the thematic "China semis" flow that took the stock to 6x reversed on a broader China-tech risk-off. The A-share premium compressed. The stock is back to a fraction of the peak — and it never actually earned its cost of capital in a single year of the cycle.

Contrarian view (what the market refuses to see): the market is treating Hua Hong as a growth compounder riding an AI-adjacent secular ramp, when it is really a capital-intensive, price-taking, state-directed commodity fab whose "profit" depends on who absorbs the depreciation. The bull narrative ("China for China," AI power ICs, ASP up) is all true and fully consistent with the company never generating a positive consolidated return through the cycle — because the objective function isn't shareholder IRR, it's Chinese self-sufficiency. The scarce H-share float + thematic flow can keep the price up regardless of the fundamentals for a long time. This is a momentum/policy vehicle, not an owner's compounder — and it should be traded as one.

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

What structurally breaks the money machine: Hua Hong has no pricing power of its own — its ASP is set by the industry-wide China mature-node supply/demand balance, which the Chinese state is actively over-supplying. It is a price-taker funding a capacity arms race whose losers are pre-selected by whoever has the deepest state pocket. Its "earnings" only exist because minority co-investors (the Big Fund, Wuxi state) absorb the fab losses below the attributable line.

Where revenue is concentrated / what shifts it: concentration is in cyclical commodity mature-node wafers — consumer, industrial, automotive power/analog. The +44% memory and +41% analog growth are riding a cyclical up-leg (NOR flash pricing + AI-power PMIC scarcity), both of which mean-revert. If AI-infra power demand normalizes or the NOR cycle turns, the fastest-growing segments become the fastest-declining ones.

Why the moat is weaker than bulls think: the "specialty qualification" moat is real but narrow; the "localization" moat is granted by geopolitics, not earned; and there is zero moat against the Chinese state funding an identical competitor next door. SMIC is bigger, better-margined, and equally state-backed; Nexchip is scaling 55/40nm to 100K wpm; PSMC and VIS compete directly. Hua Hong is the #2, lower-margin player in a subsidized commodity war.

Most dangerous competitor bulls underestimate: SMIC — larger, higher gross margin (~22% vs Hua Hong's ~13%), the same state backing, and moving faster on both capacity and (reportedly) advanced nodes. In any real price war, SMIC's superior margin structure lets it undercut Hua Hong and survive longer. Also Nexchip (SMIC-adjacent) flooding 55/40nm.

Worst capital-allocation / governance moves: the related-party HLMC absorption (RMB8.27B, ~97.5% via new shares issued to affiliated state funds at a price those affiliates helped set) dilutes minorities to execute a state consolidation agenda. The whole capital-allocation posture — plow state money into sub-cost-of-capital fabs — is value-destructive from a minority-shareholder lens even if it's rational national policy.

Assumptions that must hold for today's price: (1) mature-node ASPs keep rising through 2026–27; (2) Chinese overcapacity does NOT re-collapse pricing; (3) no ASML/DUV export cutoff; (4) the new Wuxi/HLMC depreciation gets absorbed without crushing margin; (5) the thematic "China semis" flow persists. Every one of these is exogenous to management.

Growth disappoints 20–30%: if FY27 revenue lands ~$2.6B instead of ~$3.4B (an ASP roll-over + demand softening), the thin-margin math means attributable earnings collapse toward zero or negative and the consolidated loss widens — and a stock priced for compounding de-rates hard. On sub-16% margins there is no earnings cushion.

Single scenario that permanently impairs: a hard ASML/Western-equipment export cutoff (MATCH Act enacted with teeth) that starves the fabs of tools and spares — this caps capacity, blocks node migration, and strands the $6.7B Wuxi build. Plausibility: moderate and rising — Hua Hong is already named in the legislation and reportedly hit with tool-export stoppages.

Lens 14 · Management Questions (15, ordered by information value)

1. On a fully consolidated basis you lost $110.8M in FY25 — walk us through the exact bridge from the +$54.9M attributable profit to that loss, and tell us the fiscal year in which the consolidated enterprise turns cash-flow and net-income positive. (This is the whole thesis.)
2. As Fab 9 Phase I/II and the HLMC assets are placed in service, how much incremental annual depreciation hits the P&L in FY26/27/28, and at what ASP level does gross margin still expand net of that step-up?
3. Your FY26 ASP guide is +10–25%. What share of that is durable (structural localization/mix) vs. cyclical (current mature-node scarcity), and what is your ASP assumption if AI-power-IC demand normalizes in H2?
4. What is your explicit contingency if the MATCH Act is enacted or ASML DUV access is cut — which capacity/node roadmap milestones become unachievable, and what is the domestic-tool substitution timeline?
5. Chinese state-funded mature-node capacity (yours + SMIC + Nexchip + others) is ~70% of global 12-inch expansion. What prevents a repeat of the 2024 price war in 2027 when this capacity fully loads?
6. Return on invested capital has been below your cost of capital through the cycle. What ROIC do you target for the Wuxi and HLMC investments, and by when?
7. The HLMC acquisition issued 191M new shares to affiliated state funds. How was the RMB8.27B valuation set at arm's length, and what did independent minority holders get to protect them from dilution?
8. How much of your revenue is China-domestic vs. export today, and where do you expect that mix in three years under "China for China"?
9. UMC and GlobalFoundries earn ~2x your gross margin on similar nodes. What is the specific path to closing that gap — pricing, mix, utilization, or cost — and the target margin?
10. Which of your five platforms has genuine switching-cost lock-in (qualified designs that won't re-spin) vs. commodity capacity a customer could move to SMIC tomorrow?
11. What is your automotive revenue share now, and your qualified-design pipeline for AEC-Q100 power/analog — the stickiest part of the book?
12. How do you think about the A-share (688347) premium to the H-share (1347) and what, if anything, do you do for H-share holders given that gap?
13. What are your utilization assumptions for FY26/27 — you ran >100% in FY25; what is the plan (and margin impact) if new capacity dilutes utilization to the 80s?
14. President Bai brings Intel manufacturing pedigree — what specific yield/cost programs has he initiated, and what quantified margin benefit do you expect and by when?
15. What is your net-debt trajectory given ~$1.6B/yr capex against thin cash flow, and at what point does the balance sheet constrain the expansion?

---