JCET Group

Phase A — Understand the business

Lens 1 · Company Overview

JCET (Jiangsu Changjiang Electronics Technology / 长电科技) is the largest OSAT — outsourced semiconductor assembly and test — company in mainland China and #3 globally, behind Taiwan's ASE and US-headquartered Amkor. It does not design or fabricate chips; it takes finished wafers from foundries (SMIC, TSMC, GlobalFoundries, UMC) and fabless/IDM customers and performs the back-end: wafer probe, bumping, assembly/packaging, final test, and global drop-shipment — a turnkey "chip in, tested package out" service.

What it actually is: a capital-intensive, contract-manufacturing utility for the semiconductor back-end — historically a commoditized, low-margin business (wire-bond, flip-chip) that is re-rating into a strategic bottleneck as advanced packaging (2.5D/3D, fan-out, chiplet integration) becomes the binding constraint for AI silicon. The bull narrative is that JCET is China's answer to TSMC CoWoS; the reality is a company whose blended net margin is ~6% and whose core (ex-one-off) profit fell in 2025.

Product lines:

• Packaging — wire-bond (legacy, commodity), flip-chip, wafer-level packaging (WLP), fan-out (XDFOI series — its CoWoS analog), SiP (system-in-package) modules, power-device and MEMS/sensor packaging.
• Testing — wafer sort, final test, burn-in, system-level test.
• XDFOI = eXtreme Dimension Fan-Out Interconnect, JCET's high-density fan-out platform consolidating multiple chiplets into one package for AI accelerators / HPC; JCET claims high-volume packaging of 4nm chips via chiplet integration.

End markets (2025 revenue mix):

• Communications ~38% · Consumer electronics ~24% · Computing ~19% · Automotive + Industrial/Medical ~19%.

Customers / suppliers / competitors: see Lens 2. Blue-chip customer relationships (Qualcomm RFFE "Excellent Supplier" award to its SCK Korea unit; SanDisk/NAND via the 2024 SanDisk Semiconductor Shanghai majority-stake acquisition) came largely through the 2015 STATS ChipPAC acquisition, which is the single most important event in the company's history — it converted a domestic wire-bond house into a global top-3 OSAT with fabs in Singapore and Korea and a Western customer book.

Contract structure: OSAT is not take-or-pay or recurring-subscription; it is capacity-utilization-driven, per-unit contract manufacturing with pricing pressure at the commodity end and better economics at the advanced-packaging end. Revenue tracks fab loading and the smartphone/PC/HPC cycle. This is the core economic vulnerability — no contracted floor.

Lens 2 · Supply Chain

Upstream inputs → JCET → end customer. OSAT sits at the last mile of the semiconductor chain — "a wafer cannot ship as a working product without packaging".

Wafers (SMIC, TSMC, GlobalFoundries, UMC, Samsung)
 +
Packaging materials/substrates (Unimicron, Ibiden, AT&S, Shinko — ABF substrates)
Bonding wire / lead frames / EMC mold compound
Test equipment (Advantest, Teradyne — testers; ASM Pacific, BESI — bonders)
 ↓
JCET back-end: probe → bump → assemble/package (XDFOI, flip-chip, SiP, WLP) → final test → drop-ship
 ↓
Fabless / IDM customers (Qualcomm, MediaTek, Broadcom, AMD, SanDisk/WD, HiSilicon, domestic-China designers)
 ↓
OEM/ODM → smartphone / PC / auto / AI-server end product

Named chokepoints / single-source dependencies:

• ABF substrate (Ajinomoto Build-up Film) — the interposer/substrate layer for advanced packaging is a known industry chokepoint dominated by a handful of Japanese/Taiwanese/Austrian suppliers (Ibiden, Unimicron, AT&S, Shinko). Substrate/interposer shortages are explicitly cited as forcing OSATs to raise resilience. JCET is a price-taker on this input.
• Test equipment — Advantest/Teradyne duopoly on high-end testers; concentrated but not JCET-specific.
• Wafer supply — JCET's fortunes are tied to its foundry customers' loading; the historic SMIC relationship (SMIC was a 14.28% shareholder 2017–2024) gave it a captive domestic-foundry pipeline that has now partly unwound with SMIC's exit (Lens 9).

Manufacturing footprint: 8 major production bases across China (Jiangyin HQ, Suzhou, Chuzhou, Shanghai), South Korea (Incheon — the ex-STATS ChipPAC "JSCK"/SCK plant serving Qualcomm/consumer), and Singapore. This tri-country footprint is the geopolitical hedge (Lens 13) — non-China capacity lets Western customers route around China-origin restrictions.

Concentration: JCET does not disclose named customer concentration percentages publicly. International customers = >60% of 2025 revenue, which is unusually high for a Chinese OSAT and reflects the inherited STATS ChipPAC book. This cuts both ways: it's a quality signal and the primary geopolitical attack surface.

Lens 3 · Competitive Advantages (moats)

Where the moat is real:

1. Scale + capital base. OSAT is a fixed-cost, capacity game. JCET's ~$5B revenue and 8-base footprint put it in a 3-player top tier (ASE > Amkor > JCET control ~45–50% of global OSAT revenue ). A new entrant cannot replicate this cheaply.
2. Advanced-packaging IP + qualification. XDFOI fan-out, 2.5D/3D, SiP — and the customer qualification that takes years. Getting designed-in to a Qualcomm RFFE or an AMD chiplet line is a multi-year switching cost. Advanced-packaging revenue hit RMB 27B in 2025 (~70% of total) — the mix has genuinely shifted up-stack.
3. Policy moat (the big one). As mainland China's #1 OSAT, JCET is the designated national champion for back-end localization. State control (China Resources, Lens 9), Big Fund backing, and the localization imperative give it a captive, growing domestic market insulated from foreign competition inside China. This is a policy moat, not a product moat — and it's the source of both the growth and the valuation.

Where the moat is thin:

• Bargaining power is weak on both sides. Upstream, JCET is a price-taker on substrates/equipment (Lens 2). Downstream, its fabless/IDM customers are giants (Qualcomm, AMD, Broadcom) with alternative OSATs (ASE, Amkor, Tongfu) — customers need JCET less than JCET needs them, except where localization mandates lock in Chinese demand.
• Commodity core. ~30% of revenue is still legacy wire-bond/flip-chip where pricing is brutal and differentiation is near-zero.
• Technology follower, not leader. JCET's XDFOI trails TSMC's CoWoS/SoIC in yield, scale, and ecosystem. It wins the China AI-packaging market by default (export controls push Chinese AI-chip designers to domestic OSATs), not on global merit.

Net: the durable moat is scale + policy designation, not a Buffett-style pricing moat. Bargaining power over suppliers and customers is structurally poor. Who needs whom more: for the global book, customers hold the power; for the China book, JCET holds the policy card.

Lens 4 · Segments

JCET reports by end-market, not by clean product-margin segment, and does not publish segment operating income publicly. FY2025 end-market revenue growth:

The trend that matters: growth is concentrated in computing / auto / industrial (all +30–43%), while the large Communications/Consumer base is flat-to-soft. Group revenue only grew +8.1% in 2025 despite those 40%+ segment prints — because the high-growth segments are still a minority of the mix. Advanced packaging = RMB 27B of RMB 38.87B (~70%). The mix shift up-stack is real and is the entire bull thesis; the drag is that it's diluted by a big, slow commodity communications base. Segment operating-income splits are n/a — not disclosed.

---

Phase B — Measure performance

Lens 5 · Earnings Result

FY2025 (full year, reported RMB):

• Revenue RMB 38.87B (~$5.3B ), +8.1% YoY — record high.
• Profit before tax RMB 1.74B, +5.4% YoY.
• Net profit attributable to shareholders RMB 1.57B — −2.75% YoY (down from RMB 1.61B in 2024).
• Net profit ex-non-recurring items: RMB 1.369B, −11.51% YoY. ← This is the headline the "record revenue" framing buries: core operating profitability declined double-digits.
• R&D RMB 2.09B, +21.4% YoY. Fixed-asset investment ~RMB 10B.
• Q4 2025 gross margin 15.28% (+1.94pp YoY), net margin 6.06% (+1.14pp YoY).

Q1 2026 (most recent print):

• Revenue RMB 9.17B, −1.76% YoY (vs RMB 9.34B in Q1 2025).
• Net profit attributable RMB 290M, +42.7% YoY.
• Auto segment +28.8% YoY in the quarter.

The critical read (falsifiable): JCET is NOT a revenue-growth-acceleration story right now — Q1 2026 revenue fell year-on-year. It is a margin/mix-recovery story: profit is rising because the mix is shifting to advanced packaging and mature plants run at high utilization, while top-line is flat-to-down. The bull case ("AI packaging demand explosion") and the reported numbers (declining core profit in 2025, flat revenue in Q1 2026) are in visible tension. Margin recovery is genuine (Q4'25 GM +1.94pp) but off a very low base (~15% GM, ~6% net — utility-grade).

Balance-sheet flags: positive free cash flow for six consecutive years (2019–2024), which is creditable for a capex-heavy OSAT. But FY25 fixed-asset investment ~RMB 10B against ~RMB 1.5B/quarter net capex historically signals a major capex step-up (the AI-packaging build-out) that will pressure FCF going forward. Debt/receivables detail: n/a (no primary filing on the shelf).

Market reaction: stock is up massively over the cycle (52-wk range 32.88 → 111.11 CNY, trading ~90 ) — a >2x move driven by the China-Resources takeover, AI-packaging narrative, and the capex announcements, not by the modest earnings. The market is paying for the story and the policy put, not the P&L.

Lens 6 · Earnings Calls (sentiment trend)

No earnings-call transcripts exist on the shelf (transcripts/ empty) and A-share OSATs do not host English earnings calls the way US names do — so this lens is built from press-release language across the FY24 → Q1 26 arc.

Management's consistent focus (recurring phrases):

• "Record-high revenue" — used every quarter; a tell that management leads with the top-line even when core profit falls (as in FY25).
• "Advanced packaging" / "high-end" / "one-stop platform delivery" — the strategic drumbeat, escalating each quarter.
• "High-performance computing, storage, embodied intelligence" — 2026 language adds embodied intelligence (robotics), a new demand vector vs. 2024's smartphone-centric framing.
• "Product mix optimization" — the Q1 2026 explanation for +42.7% profit on −1.76% revenue. Honest, and consistent with the margin-recovery read.

Tone shift over time: from capacity-recovery / cost-discipline (2024, coming out of the post-STATS-ChipPAC margin trough) → aggressive expansion (2025–2026: RMB 10B fixed-asset investment, the $1.1B Lingang plant, CPO/glass-substrate R&D). The tone has turned unambiguously growth-offensive, backed by state capital. What they've stopped emphasizing: deleveraging / margin repair as the primary story — it's now assumed, and the narrative has moved to landgrab. Sentiment: increasingly confident, state-backed, capacity-first. The risk in that tone is it's a supply build into an uncertain-demand environment where their own revenue just declined YoY.

Lens 7 · Comps

Peer set: the global OSAT top tier + China peers. Multiples are `` with source/date; where I could not source a specific multiple I write n/a rather than fabricate.

The comps verdict — this is the crux of the dossier. JCET trades at ~22x EV/EBITDA and ~79x trailing / ~50x forward P/E on ~6% net margins and structurally similar (or worse) unit economics to Amkor, which trades at ~11x EV/EBITDA / ~39x trailing / ~31x forward. JCET is priced at ~2x Amkor's EV/EBITDA and a large premium to ASE on a business that is lower-margin and lower-return than either Western/Taiwanese peer. Even ASE — a superior franchise (17.7% GM, 11.6% ROE, pays a dividend) — is flagged "significantly overvalued" by aggregators at ~45x. Simply Wall St's own fair-value model pegs JCET intrinsic value at ~CN¥49 vs the ~CN¥90 price — i.e. the model says the stock is ~45–80% above fair value. The peer table says JCET is the most expensive, lowest-quality name in the OSAT tier by valuation-per-unit-of-margin. The multiple is a China-A-share-scarcity + policy-champion + AI-narrative premium, not an earnings-justified one.

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

Events that have moved / should move the stock (mostly ``):

• China Resources takeover (Mar–Nov 2024) — the single biggest structural catalyst. State conglomerate CRC bought 22.53% from the Big Fund + SilTech for ~$1.6B/RMB 11.6B, becoming controlling shareholder; SMIC exited its 12.79%; board fully reconstituted. Shares jumped ~4%+ on the disposal news and the name re-rated on the "state-backed national champion" thesis. This is the policy put that underwrites the multiple.
• AI advanced-packaging narrative (2025–2026) — every XDFOI / chiplet / 4nm-packaging / CPO / glass-substrate announcement feeds the "China's CoWoS" story.
• $1.1B Lingang plant (Jun 2026) — RMB 7.8B AI-packaging fab in Shanghai; framed as securing China's AI capabilities. A capex-as-catalyst event — market rewards the supply build as a demand signal.
• Quarterly prints — reaction is asymmetric to the narrative, not the numbers: FY25 core profit fell −11.5% yet the stock stayed elevated; Q1 2026 revenue fell −1.76% yet profit +42.7% was received positively.
• 52-week range 32.88 → 111.11 CNY — a ~3.4x peak-to-trough spread confirms this is a high-beta, narrative-driven, retail-heavy A-share, not a stable industrial compounder.

Pattern: the market reacts to (1) state/ownership signals, (2) AI-packaging capex/tech announcements, and (3) the localization policy story — far more than to actual revenue or margin. That is the definition of a name where sentiment, not fundamentals, sets the price — dangerous on the way down.

---

Phase C — Judge people & books

Lens 9 · Management

The 2024 regime change is the headline. Following the CRC takeover, the board was fully reconstituted (Nov 2024):

• Chairman: Huaqiang Chuan — chief accountant of China Resources (CRC); prior chief accountant of FAW Group and deputy chief accountant of China National Machinery. A state-SOE finance executive, not a semiconductor operator. The prior chairman Gao Yonggang (ex-SMIC chairman, on JCET's board since 2017) resigned, along with directors Peng Jin, Zhang Chunsheng, and supervisor Wang Xian — a clean sweep of the SMIC-era board.
• CEO: Zheng Li (Li Zheng) — this is the genuine operating strength. 30+ years in ICs: ex-Senior VP at SMIC, ex-Global SVP & President Greater China at NXP, ex-CEO Renesas Greater China. A credible, globally-experienced semiconductor executive running day-to-day.

(1) Track record: Zheng Li brings real IDM/operator pedigree (NXP, Renesas, SMIC) — a strong hire for a company trying to move up-stack and hold Western customers. The board, by contrast, is now a state-control apparatus oriented to policy execution, not shareholder returns.

(2) Tenure & skin in the game: Zheng Li is a relatively recent CEO. Insider ownership is not meaningfully aligned — this is a state-controlled entity (CRC ~22.5% + Big Fund residual 3.5%); the controlling shareholder is the Chinese state, not founders or management with personal equity at risk. insider-transactions.csv = empty; individual insider stakes n/a.

(3) Capital allocation: the defining move is a massive state-directed capex cycle — RMB 10B fixed-asset investment in 2025, the $1.1B Lingang plant, continuous advanced-packaging expansion. This is policy-driven capacity building, which may or may not be ROIC-optimal — the objective function is national self-sufficiency, not per-share value. Historic FCF discipline (6 straight positive years) is now being spent down. ROE is low-teens [est]; the capex step-up will pressure returns near-term.

(4) Red flags: related-party / non-commercial-actor risk is structural. The controlling shareholder is a state conglomerate; the residual Big Fund stake and the SMIC-era relationships mean capital allocation can be steered by industrial policy over minority-shareholder interest. The board is chaired by a career SOE accountant. This is not fraud — it's the agency risk of a state-champion: the company may be run to build China's back-end capacity even if that means over-investing at sub-cost-of-capital returns.

(5) Founder vs professional manager: neither, precisely — it's a state-owned enterprise (SOE) run by a professional-manager CEO under an SOE board. Implication: reliable state funding and policy tailwind (bullish for survival/scale), but minority shareholders are not the priority and the equity is a policy vehicle as much as a profit vehicle.

Lens 10 · Forensic Red Flags

• Revenue quality / "record revenue" framing. The most important flag is a presentation one: management leads every release with "record-high revenue" while core profit (ex-non-recurring) fell −11.51% in FY25 and Q1 2026 revenue declined YoY. The gap between the headline and the core number is a caution flag on how results are framed — though the disclosure of the ex-one-off number is itself a point of honesty.
• Non-recurring items propping net income. FY25 net profit −2.75% but ex-non-recurring −11.51% → ~RMB 200M of the RMB 1.57B (≈13%) is non-operating/one-off (subsidies, disposals, government grants are common for Chinese semiconductor SOEs). Government subsidies are a recurring feature of Chinese chip names and flatter reported earnings — quantum n/a but directionally material.
• Goodwill / intangibles from STATS ChipPAC. The 2015 ~$1.8B acquisition (paid with only ~$260M own capital via a leveraged two-tier structure) loaded the balance sheet with goodwill + acquisition debt and drove "several years of thin margins". Any goodwill-impairment test is a latent risk if the Korea/Singapore units underperform. Carrying value n/a (no filing).
• Capex vs. cash flow divergence. The RMB 10B fixed-asset investment step-up will push FCF negative if demand disappoints — the six-year positive-FCF streak is at risk in the current build cycle.
• Receivables / inventory outrunning revenue: n/a — cannot verify without the balance sheet. This is a real gap: for a name growing capacity into flat revenue, receivables/inventory quality is exactly what a forensic analyst would want, and it is not on the shelf.
• SBC / accounting standard: reports under PRC GAAP / IFRS-aligned Chinese standards, audited by a Chinese firm — a lower-transparency, harder-to-scrutinize regime than US GAAP + PCAOB. Standard caveat for any A-share.

Regulatory findings (required sub-section):

• SEC (EDGAR LR + AAER): None possible. Per regulatory/regulatory-findings.md (fetched 2026-07-06): "JCET Group has no CIK — it is public and not required to file with the SEC. No EDGAR enforcement search is possible." total_sec_findings: 0.
• 10-K Item 3 (Legal Proceedings): n/a — JCET files no 10-K.
• Non-SEC enforcement (web search): The one material export-control adjacency: GlobalFoundries was penalized by the US BIS for shipments to SJS (a sanctioned SMIC-affiliated entity), and JCET/SMIC had divested their stakes in the relevant entity. JCET itself was not the sanctioned party, but the episode illustrates its proximity to US-China export-control friction via the SMIC/Big-Fund orbit. No direct FTC/DOJ/BIS enforcement action naming JCET was surfaced.
• Net: No SEC/AAER findings are possible (no CIK), no direct enforcement action against JCET was found via web search, and no 10-K Item 3 exists to quote — verified via regulatory-findings.md and web search as of 2026-07-06. The material regulatory risk is prospective, not historical: US export controls / Entity List exposure through its China-champion status and SMIC-orbit relationships (see Lens 13).

---

Phase D — Project & stress-test

Lens 11 · Forward Projection

Build from actuals (all RMB; net profit attributable). FY2024 net RMB 1.61B → FY2025 RMB 1.57B (−2.75%) → Q1 2026 run-rate: Q1 net RMB 290M (+42.7% YoY) on RMB 9.17B revenue. Advanced packaging ~70% of mix and rising; margin recovering off a low base; capex ramping hard.

Assumptions (labeled):

• Revenue: flat-to-down near-term (Q1'26 −1.76% YoY) then re-accelerating as the AI-packaging/Lingang capacity ramps and the auto/computing segments (+30–43%) grow into a larger share. Base: FY26 revenue ~RMB 41–43B (+6–10%).
• Margins: net margin recovering from ~4.0% (FY25, 1.57/38.87) toward ~5–6% as advanced-packaging mix lifts and mature plants stay full — the entire earnings-growth lever, since revenue is soft.
• Dilution: minimal near-term; A-share; no major disclosed raise (state funds capex). Share count ~stable.

EPS: shares outstanding n/a — not precisely sourced (~1.45B shares implied from ~CN¥130B mkt cap ÷ ~CN¥90 = ~1.45B ). Base FY26 EPS ≈ RMB 1.8B / 1.45B ≈ ~RMB 1.24 → against ~CN¥90 price = forward P/E ~50–70x on the base case — corroborating the ~50x forward P/E aggregators report. The projection confirms the valuation problem: even the bull path (RMB 3.8B by FY28) leaves the stock at ~34x FY28 earnings three years out.

Per --watchlist rules, I am NOT logging a forecast.ts Brier forecast in the loop (breadth mode). Base call recorded here for the dossier only.

Lens 12 · Bull vs Bear

Bull case. JCET is the only scaled, state-backed vehicle for China's semiconductor back-end localization at the exact moment advanced packaging becomes the AI bottleneck. Three real levers: (1) policy put — China Resources control + Big Fund + the national self-sufficiency mandate mean near-unlimited capital and a captive, growing domestic AI-chip-packaging market that export controls force toward domestic OSATs; (2) mix shift up-stack — advanced packaging already ~70% of revenue and RMB 27B, with computing/auto/industrial segments compounding 30–43%; (3) inherited global franchise — the STATS ChipPAC Korea/Singapore fabs + blue-chip book (Qualcomm, SanDisk) give it non-China capacity and Western revenue no pure-mainland peer has. Margin is recovering (Q4'25 GM +1.94pp; Q1'26 profit +42.7%). If China's domestic AI-accelerator ecosystem scales and routes packaging to JCET, this is a multi-year volume+mix compounder with a sovereign backstop.

Bear case. Three things that could permanently impair or de-rate: (1) valuation mean-reversion — the stock is priced at ~22x EV/EBITDA / ~50x forward P/E for a ~6% net-margin contract manufacturer whose core profit fell 11.5% in 2025 and whose revenue declined in Q1 2026; Simply Wall St fair value is ~CN¥49 vs ~CN¥90. A re-rate to even ASE-like (already-overvalued) multiples is a >30% drawdown; to Amkor-like EV/EBITDA (~11x) is a ~50% drawdown — with no change in the business. (2) Commodity-OSAT economics + Chinese price war — Tongfu, Huatian, and ASE all chasing the same advanced-packaging demand; OSAT has no pricing moat; a domestic capacity glut (everyone building AI-packaging fabs on state money) compresses the very margin recovery the bull case needs. (3) Geopolitics cutting the global book — >60% of revenue is international; tightening US Entity-List / export controls on a China-state-controlled OSAT could sever the Western customer relationships (Qualcomm, AMD, SanDisk) that are its highest-quality revenue, leaving it with the lower-margin captive-China base.

Pre-mortem (18 months out, thesis broke): It's early 2028. The stock is down 45%. What happened: the AI-packaging capex arms race (JCET + Tongfu + Huatian + SMIC-orbit fabs, all state-funded) produced a domestic advanced-packaging glut; utilization and pricing fell; JCET's margin recovery reversed. Simultaneously, US export-control tightening on China-state-controlled semiconductor entities pushed Western fabless customers to shift qualified volume to ASE/Amkor, cutting the international book. Core (ex-subsidy) profit, already declining in 2025, went negative on operating deleverage against the RMB 10B+ capex. The multiple, which was ~50x forward on a policy-narrative, re-rated to ~15x as the growth story broke — a double hit of earnings-down and multiple-down.

Are multiples too high? Yes, unambiguously, on any fundamental basis. ~50x forward / ~22x EV/EBITDA cannot be justified by ~6% net margins, flat-to-down revenue, and declining core profit. The multiple is a China-A-share scarcity + policy-champion + AI-thematic premium that persists only while sentiment and state support hold.

Contrarian view (what the market refuses to see): the market is treating JCET as "China's TSMC/CoWoS at the packaging layer" and pricing it like a proprietary-technology monopoly, when it is actually a low-margin, price-taking contract manufacturer whose only durable edge is a policy designation — and it is building enormous capacity into an environment where its own revenue just declined. The bull thesis is really a bet on Chinese industrial policy, not on JCET's economics. If you believe in the policy put and China's AI-silicon self-sufficiency, the asset is a buy; the stock at 50x is not.

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

Dismantling the bull case:

• What structurally breaks the model: OSAT is a capacity-utilization business with no contracted revenue floor and no pricing power. The moment the smartphone/PC/AI cycle rolls or domestic capacity out-runs demand, utilization falls and this ~6%-net-margin business swings toward breakeven on its fixed-cost base. The company is adding RMB 10B+ of capacity into flat-to-declining revenue — textbook late-cycle over-build.
• Where revenue is concentrated / what happens if it shifts: >60% of revenue is international. JCET is a China-state-controlled entity (China Resources chair = a CRC accountant). One tightening turn on US export controls or an Entity-List designation and the Qualcomm/AMD/SanDisk relationships — its best revenue — are at risk. The concentration risk isn't a single customer; it's the entire Western book sitting on top of a geopolitical fault line.
• Why the moat is weaker than bulls think: the "moat" is policy designation, not product. XDFOI trails TSMC CoWoS. Domestically, Tongfu and Huatian have the same state backing and are building the same advanced-packaging capacity — the policy moat is shared among three national champions, which guarantees a domestic price war, not a monopoly.
• Most dangerous competitor bulls underestimate: Tongfu Microelectronics (002156.SZ) — #4 global OSAT, +5.6% rev, also state-backed, also raising capital (RMB 4.4B) for advanced packaging. In the China market that is JCET's protected turf, Tongfu is the direct threat that erodes the policy premium. Globally, ASE (17.7% GM, real technology lead) can undercut JCET on the international book on quality.
• Worst capital-allocation / governance concerns: the state-champion agency problem — controlled by a conglomerate whose objective is national self-sufficiency, not per-share returns; a board chaired by a career SOE accountant; ~13% of FY25 net income from non-recurring items (likely subsidies); a leveraged 2015 acquisition that carries goodwill risk. Minority shareholders are structurally subordinate to industrial policy.
• What assumptions must hold for today's ~CN¥90 price: (a) advanced-packaging demand explodes AND JCET captures it at rising margins; (b) no domestic price war despite three state-funded champions building simultaneously; (c) the Western book survives export controls; (d) the market keeps paying ~50x forward. All four must hold. Break any one and the stock re-rates.
• Valuation if growth disappoints 20–30%: if FY27 net comes in ~RMB 1.2B (bear) instead of ~RMB 2.2B (base), and the multiple compresses to a still-generous 25x, the implied equity value roughly halves vs today. At Amkor-like ~11x EV/EBITDA the downside is ~50%+.
• Single scenario that permanently impairs: Entity-List designation of JCET (as a China-state-controlled OSAT) severing Western customers, coincident with a domestic advanced-packaging glut compressing the China book — the international-revenue engine and the margin-recovery engine break at once. Plausibility: moderate and rising given the trajectory of US-China semiconductor policy and JCET's explicit "securing China's AI capabilities" framing of its own capex.

Lens 14 · Management Questions (ordered by information value)

1. FY25 net profit ex-non-recurring items fell 11.51% while headline revenue hit a record — what portion of reported net income is government subsidies / grants / disposals, and what is the trajectory of core operating profit ex-subsidy over 2026–2028?
2. With three state-backed Chinese OSATs (JCET, Tongfu, Huatian) all building advanced-packaging capacity simultaneously on state capital, what prevents a domestic price war that compresses the advanced-packaging margins your entire thesis depends on?
3. >60% of revenue is international. What is your contingency if a US Entity-List designation or tightened export control severs the Western customer book (Qualcomm, AMD, SanDisk)? How much of that revenue could shift to your Korea/Singapore fabs vs. be lost outright?
4. Q1 2026 revenue declined 1.76% YoY while profit rose 42.7% on "mix optimization." Is the margin recovery sustainable if top-line stays flat, or is it a one-time mix benefit from running mature plants full?
5. The RMB 10B fixed-asset investment and $1.1B Lingang plant are being built into flat current revenue. What utilization and ASP assumptions underwrite the ROIC on this capex, and at what demand shortfall does it become value-destructive?
6. What are the named anchor customers for the Lingang AI-packaging capacity, and are they contracted or speculative? Is this a build-it-and-they-will-come bet on Chinese AI-accelerator designers?
7. How does XDFOI actually compare to TSMC CoWoS-L / SoIC on yield, interconnect density, and cost-per-package for a comparable AI-accelerator — with numbers, not adjectives?
8. What is the current carrying value of goodwill and acquisition-related intangibles from STATS ChipPAC, and under what Korea/Singapore-unit performance would an impairment be triggered?
9. Post-SMIC-exit, how has your foundry-customer wafer pipeline changed, and does China Resources control bring new customer channels or only capital?
10. What is your exposure to ABF-substrate and high-end interposer supply, and how do you hedge a chokepoint where you are a price-taker?
11. What are your through-cycle target gross and net margins for the advanced-packaging mix at scale, and how far are you from them?
12. How independent is capital allocation from China Resources / industrial-policy direction — can management decline a state-directed capex project that fails your own ROIC hurdle?
13. What is the realistic timeline and yield roadmap for HBM-adjacent and 2.5D/3D packaging at volume for domestic AI accelerators?
14. CPO and glass-substrate R&D were flagged for 2026 — what is the revenue timeline, and are these defensive (keeping up with ASE/TSMC) or genuinely differentiated?
15. What insider/management equity ownership exists, and how are executives incentivized on per-share value and ROIC vs. on revenue scale and policy-mandate delivery?

---