Rating: Neutral (avoid chasing highs) | Target Price: RMB 22.0–33.2 (base-case fair value range, midpoint RMB 27.6) | Current Price: RMB 29.32 (2026-10-29 close) | Margin of Safety: -25.0% (base-case fair value floor of RMB 22.0 vs. current price, no margin of safety) | Time Horizon: 12 months
China Jushi is the global leader in both glass fiber rovings and electronic-grade fabrics, ranking first in scale in both segments. In the first three quarters of 2026, net profit attributable to shareholders (excluding non-recurring items) reached RMB 5.313 billion (+103.4%), with Q3 standalone gross margin at 51.24%, delivering the resonance of historically extreme electronic fabric prices (grade 7628 rising from ~RMB 3.9/meter in September 2025 to RMB 13.3/meter in October 2026) and roving volume growth with price recovery. However, the earnings engine is highly dependent on high electronic fabric prices: industry investment in special electronic glass fiber under construction or planned totals RMB 13.38 billion; Jushi's own electronic fabric capacity plans add another 42%; new capacity from Nitobo/Kingboard/Taiwan Glass will be released in concentration from late 2027 through 2028. Sell-side consensus of ~RMB 11.8 billion net profit for 2027E implies fabric prices will not fall back throughout the year; this report's base case is a decline starting from 2027H2 (full-year realized average price RMB 10–11/meter, net profit RMB 8.0–9.5 billion). After the stock price retraced 60% from the June–July high of RMB 74.12, the current price of RMB 29.32 sits in the upper-middle portion of the base-case fair value range (RMB 22.0–33.2): PB of 3.42x is at the ~87th percentile of the past 5 years; on mid-cycle normalized earnings, ~21–26x PE (10-year median 17.4x), while the 2027E consensus PE has compressed to ~9.9x — the market has partially priced in the fabric price decline. Conclusion: Neutral, avoid chasing highs. 12-month odds distribution: bear RMB 12.4–18.8 (-36% to -58%) / base RMB 22.0–33.2 (-25% to +13%) / bull RMB 36.7–53.1 (+25% to +81%). Two repricing windows merit attention: the stock falling below RMB 22 (valuation risk released, reassess for left-side allocation), or volume ramp-up of certified low-dielectric electronic fabric (structural upward revision trigger).
Opportunity positioning: The global glass fiber leader's earnings are at a peak driven by historically extreme electronic fabric prices, and the 2027 supply clock has started; the current price is ~6% above this report's base-case fair value midpoint, and market consensus (2027E net profit ~RMB 11.8 billion) has not yet been tested by a fabric price decline — the current situation does not constitute an actionable opportunity. Use the November–December electronic fabric price adjustment letters and the January 2027 preliminary earnings announcement as repricing nodes; reassess allocation once the stock falls to the lower edge of the base-case fair value range or sustained high prices are validated. (Evidence strength: Grade B — core financial data have been verified against original financial reports, but fabric price/inventory figures are Zhuochuang and Longzhong data as cited via brokerage weekly reports, and some industry supply data have not been verified at the primary source.)
Recent key developments (by event date): (1) Grade 7628 electronic fabric rose another RMB 1.5/meter in October 2026 to RMB 13–13.3/meter; G75 electronic yarn rose RMB 3,000/ton to ~RMB 26,137/ton — both record highs, with 7628 exceeding its 2021 peak by ~50%; (2) the preliminary Q1–Q3 earnings pre-announcement (10-10) of RMB 5.136–5.393 billion net profit was confirmed at the upper half of the range by the Q3 report at RMB 5.346 billion; (3) second-largest shareholder Zhenshi Group reduced holdings by ~RMB 1.34 billion in mid-September (first reduction in nearly 10 years); (4) the 2027 supply release window is clearly set at 2027Q4–2028Q2 (Jushi's own 570 million meters of new capacity + peers' special fabrics + Nitobo's new Fukushima plant).
Primary stock price drivers (by importance): (1) The price center of 7628/G75 — weekly quotes and copper-clad laminate makers' inventory days (currently ~2 days) are the nearest-term reversal signals; (2) the pace of electronic fabric supply release vs. roving volume and price (roving at RMB 3,749/ton is only at the ~21st historical percentile — an upside option — but industry inventory of 905,000 tons has shifted to restocking); (3) the direction of 2027E earnings expectation revisions and shareholder returns (payout ratio 43.66%; interim dividend of RMB 1.32 billion already implemented).
Expectation gap: The verifiable market anchor is sell-side consensus (n=46, 2026-10-29): 2026E EPS RMB 1.96, 2027E RMB 2.955, median target price RMB 52.6, with 83 Buy ratings / 45 Accumulate / zero Reduce, and consensus upward revisions of 21.2% over the past 90 days. This report's base case for 2027E net profit of RMB 8.0–9.5 billion is ~25% below consensus; the disagreement lies in the 2027 fabric price center. The consensus assumption of no fabric price decline requires both supply delays and undiminished AI demand to hold simultaneously.
Key evidence:
Q3 standalone gross margin of 51.24% (calculated in this report from the Q3 and H1 reports: Q1 39.64%, Q2 44.39%, Q3 51.24%) has surpassed the ~45–49.5% level at the top of the last boom cycle in 2021. The preliminary announcement range (net profit RMB 5.136–5.393 billion) matches the actual Q3 report figure (RMB 5.346 billion, upper half of the range), validating management's forecasting credibility. The full-year range of RMB 7.3–7.9 billion implies Q4 standalone profit of RMB 1.95–2.55 billion: the lower end corresponds to roughly -19% sequentially (replicating the seasonal pattern of concentrated year-end provisions in 2025Q4, when RMB 717 million was -18.6% sequentially), and the upper end roughly +6% sequentially — October spot prices at RMB 13.3/meter, significantly above the Q3 industry average of RMB 10.1/meter, support the earnings center, but FX losses from RMB appreciation (FX-related losses of ~RMB 176 million in the first three quarters, translation adjustment in other comprehensive income of -RMB 192 million) and impairment provisions following receivables expansion are two unpriced disturbance items for Q4; at the 10-23 earnings call, management answered only "not yet determined" regarding November fabric prices.
Key evidence:
The supply-side chain of facts: industry investment in special electronic glass fiber under construction or planned totals RMB 13.38 billion; Kingboard's looms expanding from 3,300 to 5,800 units; Nitobo's new Fukushima T-glass plant starting production in early 2027; Taiwan Glass's TD yarn plant expansion; plus Jushi's own Huai'an Phase II 320 million meters + Tongxiang 250 million meters of new lines (ignition in 2027Q4–2028Q1) — high-end and standard electronic fabric supply will be released in concentration from late 2027. On the demand side, AI server shipment growth is decelerating from +46% in 2024 to a projected +28.3% in 2026E (TrendForce). Mean reversion for 7628 from RMB 13.3/meter is therefore clear in direction; but on timing, the industry has no large-scale new capacity in the first three quarters of 2027 (multiple leading companies confirmed to media in October that industry capacity will only be released in concentration at the end of 2027), Toyota loom lead times of 18–24 months and 2-year customer certification lock in short-term supply, and historically expansions achieve only ~70% of planned output, with second-generation low-dielectric fabric yield bottlenecks at ~50% — reversion is more likely to begin in 2027H2. Consensus of RMB 11.8 billion (implying spot prices holding at ~RMB 13/meter all year) carries >25% downward revision risk, but within 12 months this is more likely to manifest as "stagnation at high levels with gradual expectation revisions" rather than "rapid collapse." A reverse test of this thesis is conducted in parallel: if low-dielectric fabric wins volume orders from leading copper-clad laminate makers, changing the high-end supply structure, this thesis is invalidated.
Key evidence:
The valuation has three coordinates: (1) On mid-cycle normalized net profit of ~RMB 4.5–5.5 billion (3.40 million tons of roving × mid-cycle price RMB 4,200–4,400/ton + 1.4 billion meters of electronic fabric × realized price RMB 6.5–7.5/meter), the current price implies a normalized PE of ~21–26x, above the 10-year median of 17.4x; (2) a three-layer value decomposition (detailed in the valuation chapter): asset value ~RMB 9.5/share, EPV ~RMB 12.3/share — ~58% of the current price consists of growth options, indicating the market is paying a sizable premium for the "AI electronic fabric new normal + low-dielectric second growth curve"; (3) but the fact on the other side is: the stock has been cut in half twice from its peak, and the 2027E consensus PE has compressed to 9.9x — "EPS not yet revised down while the stock has already fallen 60%" shows the market is pricing the fabric price decline through valuation rather than through earnings forecasts. After the two forces offset each other, the current price sits at roughly the 65th percentile of the base-case fair value range — lacking a margin of safety on the upside, but on the downside supported by the world's lowest cost curve, a ~2.9% dividend yield (estimated as 2026E net profit of RMB 7.6 billion × 45% payout), and a balance sheet that remains profitable even at the trough.
Key evidence:
Cost leadership derives from large-furnace economies of scale, self-supply of pyrophyllite and sizing agents, smart manufacturing, and the low-cost Egypt base (whose profitability has already exceeded the domestic operations); weighted ROE for the first three quarters of 2026 was 16.11%. An extreme-trough stress test (fabric price reverting to RMB 4.5/meter + roving at RMB 3,300/ton, i.e., 2024 trough prices applied to today's larger capacity base): trough net profit of ~RMB 2.2–3.0 billion, EBITDA of ~RMB 6.5–7.2 billion, net debt/EBITDA of ~1.2–1.3x; interest-bearing debt maturing within one year of RMB 6.13 billion can be rolled over via trough operating cash flow of ~RMB 4.5–5.5 billion/year and unimpeded bond/bank channels — no liquidity risk exists; the company can traverse the next trough while remaining profitable. Two caveats to retain: the unit-cost gap is on a third-party 2024 estimate basis; and the erosion of new-line investment returns from platinum-rhodium price increases (+17% in total investment for the Chengdu project) is partially offset by the company's existing rhodium inventory and gains of over RMB 4 billion from high-price precious metals sales in 2021–2023 — a double-edged sword.
Key evidence:
Upon objective quantification, this signal is weaker than it appears on the surface: the RMB 1.34 billion reduction represents only 1.1% of market cap, about 7.5% of the single-day turnover on September 18 (RMB 17.75 billion), and is a partial realization of Zhenshi's position built at ~RMB 11.63/share in 2024–25 (unrealized gain of ~3x). Over the same period, controlling shareholder CNBM conducted two rounds of increases totaling RMB 1.125 billion, and Zhenshi itself also increased holdings by RMB 1.10 billion in February–March 2026 — the net direction of industrial capital is not one-sided selling. The pledges are rolling legacy operations since 2025 (after release in August, re-pledged the same day until August 2027), with disclosed use being the group's own production and operations; at an estimated 40–50% financing ratio, the warning line corresponds to a stock price of ~RMB 14–17, leaving ample buffer from the current price. The rise in related-party sales from 11.46% to 17.46% corresponds to procurement expansion after Zhenshi Co.'s listing in January 2026 (procurement from Jushi of RMB 2.112/1.895/2.037/3.3 billion in 2022–2025), governed by an annual quota system subject to shareholders' meeting approval. The real monitoring focus is receivables quality: in a downturn, the lenient terms on the combined RMB 6.3 billion of accounts receivable and receivables financing could convert into concentrated impairment provisions.
| Metric | FY2023 | FY2024 | FY2025 | 2026Q1–Q3 (latest period) |
|---|---|---|---|---|
| Revenue (RMB bn) | 14.876 | 15.856 | 18.881 | 18.078 (+30.0%) |
| Net profit attributable to parent (RMB bn) | 3.044 | 2.445 | 3.285 | 5.346 (+108.2%) |
| Non-GAAP net profit attributable to parent (RMB bn) | 1.898 | 1.788 | 3.482 | 5.313 (+103.4%) |
| Gross margin | 28.0% | 25.0% | 33.1% | 44.9% (Q3 standalone 51.24%) |
| Net margin | 20.5% | 15.4% | 17.4% | 29.6% |
| Operating cash flow (RMB bn) | 0.867 | 2.032 | 4.201 | 3.957 (+85.4%) |
| Free cash flow (RMB bn, OCF−CapEx) | −0.593 | 0.729 | 2.833 | 0.824 (CapEx restart phase) |
| Cash and equivalents (RMB bn, period-end) | 3.126 | — | 3.473 | 3.580 |
| Interest-bearing debt (RMB bn, period-end) | 14.004 | — | 12.220 | 12.129 |
| Debt-to-assets ratio | 42.39% | 41.08% | 40.41% | 41.28% |
| Net debt/EBITDA | — | — | — | ~0.86x (TTM estimate; ~1.2–1.3x under extreme trough scenario) |
| Sales volume | Roving 2.4814 mn tonnes / electronic fabric 836 mn meters | 3.0250 mn tonnes / 875 mn meters | 3.2026 mn tonnes / 1,062 mn meters | 2026H1 roving 1.7004 mn tonnes / electronic fabric 544 mn meters |
Note: FY2024 cash/interest-bearing debt figures not available due to missing full annual report (only summary-level key metrics available); free cash flow is net operating cash flow minus cash outflow for acquisition and construction of long-term assets.
Drivers of metric changes (all items with YoY ≥ ±20% include company explanations): ① Gross margin of fiberglass and products rose 7.87pp to 32.21% in FY2025 — company explains this as "implementing incremental volume, price restoration, price stabilization and price adjustment sales strategies, with volume and price both rising"; a cyclical price increase rather than structural cost reduction (per-tonne cost of RMB 3,883 in 2025 was essentially flat vs 2024); ② FY2025 administrative expenses +116% — prior-year base was low due to RMB 0.300 bn reversal of the non-vested portion of the excess profit-sharing plan, recorded as non-recurring income; ③ FY2025 operating cash flow +106.7% — increased cash collections from sales and maturing commercial bills presented for payment; ④ Cash spent on long-term assets in first three quarters of 2026 rose from 0.599 to 3.133 bn RMB (+423%) — concentrated construction at Chengdu 200k-tonne project (total investment raised to RMB 2.748 bn), Huai'an electronic yarn/wind power, and Tongxiang cold repairs; ⑤ Accounts receivable +68.2% vs start of year — increase in receivables within credit terms (faster than revenue growth of +30.0%; collection structure needs verification); ⑥ Contract liabilities +45.1% in 2026H1 (increase in advance receipts — a leading indicator of improving conditions); ⑦ Higher finance expenses — FX swings turned from gains to losses (RMB appreciation).
Q3 report review: Q3 standalone revenue was RMB 6.919 bn (+44.3%), net profit attributable to parent RMB 2.412 bn (+173.8%); single-quarter earnings already exceeded more than half of any full year 2022, 2023 or 2024 at a comparable pace. The results-boost preannouncement range was realized in the upper half, with no signs of excessive packaging. In terms of earnings mix, the jump in Q3 gross margin to 51.24% was mainly driven by electronic fabric (7628 industry average price in Q3 at RMB 10.1/meter, +143% YoY, +49% QoQ; company's H1 electronic fabric average price RMB 6.02/meter, +49% YoY, with realized prices converging toward spot each quarter under "monthly repricing"), while Q3 roving average price of RMB 3,749/tonne was only +5% YoY — the two product lines are severely out of phase in the cycle (see chapter "Cycle Positioning and Through-Cycle Profitability"). Operating cash flow of RMB 3.957 bn (+85.4%) showed acceptable matching with net profit (OCF/net profit ≈ 0.74), but receivables expansion and the CapEx restart narrowed free cash flow to RMB 0.824 bn. As of 2026-09-30: cash RMB 3.580 bn, interest-bearing debt RMB 12.129 bn (net debt RMB 8.549 bn), debt-to-assets ratio 41.28% — expansion funded mainly by debt (SCI-tech super short-term commercial paper rates have fallen to 1.45–1.70%). Relative to market expectations: sell-side 2026E consensus net profit attributable to parent is ~RMB 7.85 bn; the first three quarters delivered 68%, on track. This report's full-year range of RMB 7.3–7.9 bn is consistent with, slightly below, that consensus.
Business model snapshot: Asset-heavy pool-furnace fiberglass drawing manufacturing (fixed assets at 63.6% of total assets), a single operating segment (fiberglass and products at 99.01% of main revenue), plus a small wind power generation business (0.99%, 70.38% gross margin, tax-exempt period). Revenue has no subscription/recurring concept, but fiberglass end markets (wind power, automotive, electronics, construction) are diversified and the company holds ~25% market share, giving it oligopolistic coordinated price-restoration ability (validated by two rounds of collective price increases in 2024-03 and 2025-03) — pricing power is cyclical and organizational, not structural.
Cash conversion test of earnings: OCF/net profit attributable to parent went 0.28→0.83→1.28 over 2023–2025, and FCF/net profit −0.19→0.30→0.86 — a classic cyclical pattern of bill backlog at the trough and receivables recovery in the upswing; since 2025 the cash content of reported profits has been ample. In 2026Q1–Q3, OCF was RMB 3.957 bn, 0.74x net profit, with FCF narrowing (RMB 0.824 bn) after the CapEx restart. On a recurring basis: 2025 non-GAAP profit of RMB 3.482 bn was actually above the reported RMB 3.285 bn (disposal losses of RMB 0.192 bn depressed reported profit), and non-recurring items in 2026Q1–Q3 were only RMB 0.032 bn — current earnings "quality" is clean. But historically (2022–2023), the company propped up reported profit with disposal gains from selling precious metals (2023 disposal gains RMB 0.998 bn; 2022 disposal gains/losses RMB 2.535 bn); in a downturn, watch for renewed divergence between reported and non-GAAP profit.
Return on capital: FY2025 ROIC ~7.6% (early-cycle recovery); annualized ~15.3% in 2026Q1–Q3 — returns on capital in the upswing have crossed above WACC (9% in this report), but the full-cycle average (2022–2025 ROE of 26.4%/10.8%/8.3%/10.7%) is roughly 10–13%, making it a "high-quality cyclical manufacturer" rather than a compounding machine with sustained >15% returns; the moat lies in the survival right of "still profitable at the trough" thanks to the leftmost position on the cost curve, not in the persistence of excess returns.
CapEx quality: 2022 RMB 5.547 bn (previous expansion peak) → 2023–2025 average RMB 1.3–1.5 bn per year (CapEx/depreciation ~0.55, an underinvestment/digestion phase) → 2026Q1–Q3 RMB 3.133 bn (annualized ~RMB 4.2 bn). The company does not disclose a maintenance/growth split; approximating maintenance capex with the 2023–2025 trough average of RMB 1.3–1.5 bn per year, roughly RMB 2.5–2.8 bn from 2026 would be growth capex, directed at Huai'an electronic yarn/electronic fabric (RMB 4.431 bn), Chengdu 200k tonnes (RMB 2.748 bn), and Tongxiang cold repairs (RMB 0.960 bn) — concentrated commissioning in 2027–2028 (synchronized with the industry supply wave; watch for pro-cyclical investment diluting returns; the Chengdu project's expected total ROI is only 7.06%).
Red flag list: ① Rising related-party sales (Zhenshi-affiliated customers among the top five rose from 11.46% to 17.46%, reflecting expanded purchases after Zhenshi Co.'s listing — institutionally disclosed, but the pace is steep); ② leverage at the major shareholder level (Zhenshi has pledged 51.54% of its holdings); ③ 2026 receivables +68.2%, faster than revenue; ④ the related-party competition resolution commitment has been deferred 4 times since 2017 (extended another 5 years in May 2026). All are "friction items"; no signals negating the business model were found.
Credibility of words vs. deeds: pragmatic. Verifiable commitment-delivery record: ① 2025-11 Zhenshi share purchase plan of RMB 0.55–1.1 bn → completed in 2026-03 at the full RMB 1.100 bn; ② CNBM's purchase of RMB 0.125–0.25 bn → completed at the lower bound of RMB 0.125 bn; ③ 2025 buyback of 0.75–1.00% of shares → completed 0.86% (RMB 0.540 bn), granted to 674 key employees; ④ payout ratio commitment for 2024–2026 of 35%/40%/45% escalating → actual 39.30%/43.66%/interim 45.0%, all met; ⑤ the 2026-10 results preannouncement range was realized in the upper half by the Q3 report. The only major undelivered item: the related-party competition resolution commitment with CNBM's fiberglass business (Taishan Fiberglass) has been suspended since 2017 — 9 years and 4 deferrals — now extended another 5 years via asset restructuring — a state-owned-enterprise integration pacing issue, not an integrity issue, but it must be counted. Annual operating plans are qualitative only (2026: "control capacity, stabilize prices"); the company gives no quantitative guidance, leaving forecasting responsibility to investors.
Shareholder friendliness: shareholder-friendly. Payout ratio rose three years running (36.16%→39.30%→43.66%); the 2026 interim dividend of RMB 1.32 bn (RMB 3.3 per 10 shares) set an interim record and has been paid (record date 10-15); cumulative cash dividends over the past three years ~RMB 3.5 bn; no additional offerings/rights issues/convertibles; incentives use restricted stock (0.79% of shares, RMB 0.334 bn subscription amount) rather than dilutive options. On 2026E net profit of RMB 7.6 bn × 45%, the full-year dividend is ~RMB 0.85/share, a yield of ~2.9% at current prices.
Risk signals: founding family (Zhenshi) first-ever high-price share reduction plus pledge of over half its holdings (see C5); deposits of RMB 0.780 bn plus borrowings of RMB 0.200 bn at the group finance company (normal for central SOE cash pooling, within the RMB 0.9 bn cap); core management stable (the Zhang Yuqiang system), with executives bound via the 2025 restricted stock incentive plan.
The company reports a single operating segment; roving and electronic fabric revenue and gross margins are not separately presented — this section breaks down per the disclosed basis, supplemented by volume and third-party data inference.
| Segment/product line | Revenue share | Gross margin | YoY | Business logic in one line |
|---|---|---|---|---|
| Fiberglass and products (roving + electronic fabric combined) | 99.01% | 32.21% (FY2025, +7.87pp) | +18.51% | Pool-furnace drawing at scale; lowest global cost |
| — of which: roving and products | Not separately disclosed (FY2025 volume 3.2026 mn tonnes, +5.9%) | — | — | Wind/thermoplastic/construction commodity reinforcement materials; price at 21st historical percentile |
| — of which: electronic yarn/electronic fabric | Not separately disclosed (FY2025 volume 1,062 mn meters, +21.4%) | — | — | Upstream of copper-clad laminates; global #1 in standard 7628 fabric; price at historical extreme |
| Other (Huai'an wind power) | 0.99% | 70.38% | +194.61% | Self-generated green power sold externally beyond tax-exempt period; supports decarbonization |
Primary profit engine and source of earnings elasticity: on FY2025 static structure, roving contributes the bulk of profit via ~3.2 mn tonnes × per-tonne net profit (estimated RMB 2.6–3.0 bn), while electronic fabric at 1,062 mn meters contributes ~RMB 1.1–1.4 bn; but in 2026 virtually all incremental elasticity comes from electronic fabric — H1 average price of RMB 6.02/meter (+49%) drove blended gross margin sharply higher each quarter (Q1 39.6% → Q3 51.2%), and at Q4 spot of RMB 13.3/meter, electronic fabric has become the #1 single-quarter profit driver. The inter-segment gross margin gap (roving system ~25–30% vs electronic fabric chain 40%+) exceeds 10pp, stemming from differences in bargaining power rooted in technical barriers (electronic-grade glass formulations, fine yarn drawing, weaving) and supply bottlenecks (Toyota loom delivery lead times) — which is precisely the main attack surface of the 2027 supply wave. Domestic revenue is 66.89%, overseas 33.11% (Egypt/US bases provide tariff hedging via "overseas supply for overseas").
Accounting red flags (pattern/severity/evidence): ① Selling precious metals to boost profit in downturns (med) — 2023 disposal gains of RMB 0.998 bn from partial precious metal sales; 2022 non-current asset disposal gains/losses of RMB 2.535 bn; across the 2021–2024 downturn, platinum-rhodium inventory disposal gains propped up reported profit, with non-GAAP profit persistently below reported; ② Excess profit-sharing plan accrual-reversal swings across periods (med) — 2024 reversal of RMB 0.300 bn booked as non-recurring income depressed the expense base, producing the apparent +116% anomaly in 2025 administrative expenses; ③ Concentrated recognition of disposal losses in 2025 (low) — non-current asset disposal losses in non-operating expenses of RMB 0.211 bn (vs only RMB 4.84 mn prior year) — clearing out old lines in the upswing; direction is conservative but adds cross-period noise; ④ Receivables growing faster than revenue (med) — +68.2% at end-September 2026 vs revenue +30.0%; ⑤ Group finance company deposits and fund lending (low) — RMB 0.780 bn deposits plus RMB 0.200 bn borrowings; normal within the central SOE system.
Cross-period consistency checks: ① Non-recurring items reversed direction twice in three years (+1.147 bn → +0.657 bn → −0.196 bn), with complete line-item disclosure each period and consistent definitions, matching company explanations; ② OCF/net profit of 0.28→0.83→1.28 is self-consistent with the "bill backlog at the trough, improved collections in the upswing" explanation; ③ The V-shaped gross margin path of 28.0%→25.0%→33.1%→44.9% matches the 2400tex and 7628 price trajectories; ④ The capex restart of 5.547 bn → 1.3–1.5 bn → 3.133 bn maps one-to-one to project announcements; ⑤ Related-party sales share rose 11.46%→17.46%; the company did not separately explain it in the annual report (though Zhenshi Co.'s purchase expansion is corroborated by announcements and the prospectus) — flagged as "partially explained." Overall judgment: no traces of systematic accounting manipulation were found; all cross-period anomalies can be explained by the cycle and disclosed items; what requires ongoing vigilance is the recurrence of the "disposal gains as a floor" pattern in downturns.
Current Market Data and Multiple Percentiles: Closing price RMB 29.32 (2026-10-29, completed close), market cap RMB 117.37 billion; PE(TTM) 19.4x (69th percentile over the past 5 years — earnings are at the peak of a cyclical upswing, so the percentile is mechanically distorted and for reference only), PB 3.42x (~87th percentile over 5 years, 81st percentile over 3 years), 2026E PE 14.9–16.1x (consensus vs. this report's range), 2027E consensus PE 9.9x, dividend yield ~2.9%. Liquidity is excellent (average daily turnover ~4–7% of market cap, far above the 0.1% warning threshold); no liquidity discount issue. Peer valuation comparison (comparable companies' ROE for the first three quarters of 2026 vs. growth): Sinoma Science & Technology 8.39%/+29.1%, CPIC Fibre 14.57%/+371%, Changhai Co. 5.36%/+0.6%, Hongho Technology 29.96%/+435.6% (A-share peer PE/PB series not obtained; Hongho Technology's mid-2026 market cap once reached ~RMB 190 billion, implying an annualized PE of ~190x, serving as the valuation thermometer for the AI electronic fabric narrative — not directly comparable with Jushi).
What Expectations Are Already Priced In: Back-solving from a mid-cycle normalized PE (median of 17.4x over the past 10 years), the current price implies the market believes normalized net profit attributable to shareholders is about RMB 6.7 billion — treating roughly 85–90% of peak profits at historically extreme fabric prices (2026E RMB 7.3–7.9 billion) as a sustainable norm; on a consensus basis, the current price = 9.9x 2027E, implying 7628 fabric holds at ~RMB 13/meter all year without falling. The verifiable reality is: mid-cycle normalized net profit attributable of about RMB 4.5–5.5 billion, and a 2024 trough non-recurring-adjusted profit of RMB 1.788 billion. The current price requires that high electronic fabric prices largely persist, while reality is that the supply clock has started and AI shipment growth is decelerating — but the stock's prior 60% decline has already paid part of the price, so this is a "balanced level" rather than "significantly overvalued."
Three-Tier Value (EPV): ① Asset value ~RMB 9.5/share (2026Q3-end BPS of RMB 8.57 + a rough ~15% fixed-asset replacement premium; rising platinum-rhodium bushing replacement costs support the premium); ② EPV ~RMB 12.3/share — normalized net profit attributable of ~RMB 5 billion (roving 3.4 million tons × mid-cycle price + electronic fabric 1.4 billion meters × realized price RMB 6.5–7.5/meter, all at already-commissioned capacity). Bridge: + D&A ~RMB 2 billion − maintenance capex ~RMB 1.7 billion − working capital increment ~RMB 0.2 billion + after-tax interest ~RMB 0.17 billion + minority interests ~RMB 0.2 billion → normalized FCFF ~RMB 5.2 billion (range RMB 4.4–5.4 billion), divided by 9% WACC, less net debt of RMB 8.549 billion; ③ Growth option = current price − EPV ≈ RMB 17.0/share, ~58% of current price — what the market is mostly paying for is the option of "electronic fabric new normal + Huai'an low-Dk RMB 4.431 billion project + roving price restoration," whose realization depends on low-Dk certification volume ramp (management's investor-relations wording is only "samples being delivered in an orderly manner," with no quantitative disclosure).
Three Scenarios and Odds (current price RMB 29.32, all 12-month fair-value ranges):
| Scenario | Probability | Fair Range | 2027E Net Profit Assumption | Exit Multiple | vs. Current Price |
|---|---|---|---|---|---|
| Bear | 25% | RMB 12.4–18.8 | RMB 4.5–5.8 billion (fabric price reverts early to RMB 6–8/meter + roving stuck at 3,700–3,900 + escalating trade barriers) | 11–13x (cyclical downswing, anchored to 10-year median 17.4x at a discount) | -36% to -58% |
| Base | 50% | RMB 22.0–33.2 | RMB 8.0–9.5 billion (fabric price holds at 12–14 in 2027H1, falls to 9–10 in H2, full-year realized average RMB 10–11/meter; roving re-priced to 4,100–4,400; electronic fabric 1.4–1.5 billion meters) | 11–14x (peak earnings year, compromise between 2021 top trading range of 10–15x and 10-year median) | -25% to +13% |
| Bull | 25% | RMB 36.7–53.1 | RMB 10.5–12.5 billion (low-Dk certification ramp + fabric price holds above 13 all year + roving at 4,800–5,200) | 14–17x (structural upward revision narrative, still below the multiple implied by consensus) | +25% to +81% |
Probability-weighted fair value ~RMB 28.9 (expected return ~-1.3%); external reference for the bear scenario: sell-side currently has zero sell ratings and no public bearish target prices; the closest bearish markers are the company's own annual-report warning that "structural overheating of capacity warrants vigilance" and industry observers cited by Cailianshe (new capacity concentrated release in 2027 may cause ordinary fabric oversupply) — both qualitative judgments with no comparable quantitative target prices; this report's bear-case fair value is below all sell-side target prices, with the gap stemming from assumptions about the extent of fabric price reversion (RMB 6–8/meter vs. the sell-side implied RMB 13/meter with no fall). External reference for the bull scenario: Soochow Securities (2026-08-24) with 2027E net profit attributable of RMB 14.36 billion is the most optimistic licensed forecast, above this report's bull-case upper bound; its premise is that the loom bottleneck + slower ramp keep prices high — the bull scenario does not require anything so extreme.
In-House Earnings Forecast vs. Consensus vs. Management Guidance:
| Basis | FY2026E Net Profit Attributable | FY2027E Net Profit Attributable |
|---|---|---|
| This report (base) | RMB 7.3–7.9 billion (non-recurring-adjusted RMB 7.2–7.8 billion) | RMB 8.0–9.5 billion |
| Sell-side consensus (n=46) | ~RMB 7.85 billion | ~RMB 11.8 billion |
| Management guidance | No quantitative guidance (Q3 pre-earnings increase announcement has materialized) | None |
The entire divergence lies in the 2027 fabric price center: consensus implies spot prices at ~RMB 13/meter all year, while this report's base case is a fall to a realized average of RMB 10–11/meter. This is the anchor for the next earnings reconciliation: if 2027H1 fabric prices stay above RMB 12, the risk of downward revision to this report's base case exceeds the risk of upward revision.
Conclusion: Fair. The relationship between world-best quality and price must be stated separately: under the base scenario, the current price sits in the upper-middle of the fair range (~65th percentile), with no margin of safety (base-case floor of RMB 22.0 is -25% vs. current price); but a verdict of extreme overvaluation is equally unsupported — the stock has already pulled back 60%, and the 9.9x 2027E consensus PE and 2.9% dividend yield provide downside cushion. Avoid chasing; wait for two types of repricing: stock price ≤ RMB 22 (percentile risk released) or low-Dk volume ramp confirmed (structural upward revision).
Industry Size. Glass fiber is a reinforcement material and electronic substrate made by melting and drawing raw materials such as pyrophyllite and borocalcite through tank furnaces. China data (China Glass Fiber Industry Association "2025 Annual Development Report"): 2025 total glass fiber yarn output of 8.43 million tons (+11.5%), of which tank-furnace yarn was 8.12 million tons, 96.3%; year-end in-production tank furnace capacity of 8.38 million tons; industry main-business revenue +11.0%, total profit ~RMB 11 billion (+78.4%, recovering from the deep-loss zone of 2022–2024); domestic apparent consumption of 6.585 million tons (+16.7%). Global: 2024 output ~10.5 million tons, China ~72% (including 664,000 tons at Chinese overseas bases, +8.5%); 2025 global market size ~USD 19.1–25.1 billion (aggregator figures from Straits Research/Fortune BI etc., unverified, for cross-reference only). Growth forecast: output CAGR of ~6–9% over the next 3–5 years (wind power, thermoplastics, electronics as three engines); global low-Dk electronic yarn market USD 1.5 billion in 2024 → USD 3.2 billion in 2033 (aggregator figures, unverified).
Industry Chain and Value Distribution. Upstream: pyrophyllite (China's reserves among world's largest, concentrated in Zhejiang/Fujian), borocalcite/quartz sand/soda ash, natural gas and electricity, platinum-rhodium bushings (over 15% of single-line investment; a 120,000-ton roving line requires ~RMB 1.2–1.5 billion). Midstream: glass fiber yarn tank-furnace drawing — a highly concentrated oligopoly segment (domestic CR3 over 60%); the 8–10 year cold-repair cycle determines supply rigidity; electronic fabric weaving is constrained by Toyota looms (JAT910) as the sole supplier with 18–24 month lead times, the physical bottleneck of 2025–2026 excess profits. Downstream: construction/building materials ~34–35%, transportation 29%, electronics/electrical 14%, industrial equipment 12%, energy/environmental (wind power) 10%, with the mix tilting toward wind power and electronics. The company sits at the midstream raw-yarn core segment, vertically integrated into electronic fabric (electronic yarn 270,000 tons + electronic fabric sales of 1.062 billion meters, No. 1 globally): strong bargaining power upstream (self-sufficient in pyrophyllite/sizing), fairly strong downstream (led joint price restoration in 2024-03, and in 2025-03 raised electronic yarn prices by RMB 800/ton collectively with Taishan Fiberglass/Kingboard), export side suppressed by US/EU tariffs. Gross margin mainly accrues to the raw-yarn and high-end fabric segments with scale and formulation advantages — Jushi's 2024 glass fiber business gross margin was 9.1pct higher than the comparable-company average.
Supply-Demand and Competitive Landscape. Demand side: 2025H1 domestic wind power new installations of 51.39GW (+98.9%), blade upsizing driving high-modulus yarn output +65.1%; automotive lightweight thermoplastic yarn +9.5%; electronic yarn +9.6% (AI low-Dk fabric boom); traditional thermoset direct roving/construction demand -9.8%; export volume -3.6% (1.9496 million tons) but value +1.6%. Supply side: global capacity accelerating toward China (~72%); Owens Corning (OC) sold its glass fiber reinforcements business in Feb 2025 for USD 755 million, exiting commodity fiberglass entirely — global competition shifts from "China vs. foreign" to "competition within Chinese players"; domestic 2025 new capacity of 860,000 tons, 2026 a light supply year (new additions of 350,000 tons vs. cold-repair exits of 460,000 tons, net ~-110,000 tons, with 70% of additions from the leader), and of the 1.035 million tons planned as of 2025, actual ignitions were restrained. Concentration: the global Big Six (Jushi/OC/NEG/Taishan/Chongqing Polycomp/JM) hold 70–80%; Jushi's global share ~25% (Sinolinks figures: in-production roving 2.59 million tons + electronic yarn 270,000 tons; the company claims the world's largest capacity scale — both point to the same fact). Domestic capacity shares: Jushi ~35%, Sinoma Science & Technology (Taishan) ~17%, CPIC Fibre ~14%. Entry barriers: tank-furnace technology + scale (single 120,000-ton line ~RMB 1.2–1.5 billion investment, platinum-rhodium bushing损耗, 8–10 year continuous operation), steep cost curve (leader's unit cost 20–30% below new entrants), 12–18 months customer certification, high-end electronic fabric additionally requiring loom and major CCL maker certification of 2 years. Substitution threat limited (carbon fiber several times more expensive, quartz fabric ~40x the price, confined to high-end niches). Price-war risk: 2022–2024 saw "involutionary vicious competition" (association's wording); since 2024, leader-led joint price restoration + industry self-discipline improved the landscape, but vigilance is needed after concentrated release of new electronic fabric capacity in 2027–2029.
AI Electronic Fabric: Quantified Chain of a Structural Demand Inflection (this report's verdict: established, and triple-verified through prices). Driver trajectory: AI server shipments 2024 +46% → 2025E +24.3% → 2026E +28.3% (TrendForce); data center capex USD 376.7 billion in 2025 (+71%) → 2026E USD 608 billion (+60%). Leap in unit consumption: NVIDIA Rubin platform PCB layer count rises to 40–78 layers (traditional servers ~10–20 layers); AI server high-speed CCL demand from 42 million sheets in 2026 → 131 million sheets in 2028 (CAGR ~77%, Goldman Sachs), with high-speed copper-clad laminate area and sheet count per unit multiplying. Penetration ramp: second-generation low-Dk fabric reported at ~RMB 160/meter in April 2026 (doubled since the start of the year); ordinary 7628 fabric from RMB 3.9/meter in Sept 2025 → RMB 13.3/meter in Oct 2026 (+241%, ~50% above the 2021 peak of 8.8) — high-end crowding out loom capacity pushes up ordinary fabric in reverse, with "monthly price adjustments" becoming the norm. The company's share: Jushi is No. 1 globally in ordinary 7628 fabric and a full beneficiary of the across-the-board price rise; but in the highest-end T-glass class, Nittobo holds ~90%, and Jushi's low-Dk fabric is still in certification/sample-delivery (no quantitative investor-relations disclosure) — for this inflection, Jushi's exposure is a "broad price-rise dividend + catch-up option," not a fait accompli high-end monopoly. The reverse risk is equally quantifiable: from end-2027, concentrated release of industry new capacity combined with falling AI shipment growth would drive mean reversion in prices (see Investment Thesis Point 2 and the Cycle Positioning chapter).
Cycle and Regulation. Regulatory: trade barriers are a long-term variable — US 55% special tariff on Chinese glass fiber (stacked with the 12.5% temporary additional tariff on 46 economies from 2026-07-24); EU glass fiber yarn anti-dumping duty raised from 24.8% to 33.2% in 2025-11, glass fiber fabric 99.7% still in effect, plus an 11% anti-dumping duty on Egyptian glass fiber (transshipment channels starting to fail); India USD 274/ton; Brazil's preliminary ruling in Aug 2026 imposed the top-tier provisional anti-dumping duty of USD 456.44/ton on Jushi (final ruling expected early 2027); export tax rebate cut from 13% to 9% from 2024-12 (stock drag ~RMB 230 million/year). The hedge is "domestic supply for domestic, overseas supply for overseas": overseas capacity of 456,000 tons covers ~43% of overseas demand; the Egypt base's profits have surpassed domestic, and the US base benefits from local price rises under tariffs. Domestic policy is neutral-to-positive: glass fiber is designated a strategic new material; capacity self-discipline and high-end policies support the price center.
Peer Operating Comparison (first three quarters of 2026, revenue/growth/ROE; valuation multiple series not obtained):
| Company | Positioning | Revenue (RMB 100M) | Net Profit Attributable Growth | ROE | Key Difference vs. Jushi |
|---|---|---|---|---|---|
| China Jushi | Global No. 1 in both roving + electronic fabric | 180.78 (+30.0%) | +108.2% | 16.11% | Leftmost cost, highest single-business glass fiber purity |
| Sinoma Science & Technology (Taishan Fiberglass) | Global No. 2 roving + blades + lithium membranes | 256.68 (+18.3%) | +29.1% | 8.39% | Diversification dilutes; positioned in specialty electronic fabric ahead of Jushi (19.17 million meters sold in 2025) |
| CPIC Fibre | ~10% global capacity, wind yarn specialist | 79.95 (+24.7%) | +371.2% | 14.57% | High elasticity off low base, weaker stability |
| Changhai Co. | Integrated products niche (~1/6 of Jushi's scale) | 24.27 (+2.9%) | +0.6% | 5.36% | Weakest elasticity in the price-rise cycle, highlighting Jushi's pass-through power |
| Shandong Fiberglass | Second-tier roving | 22.87 (+29.0%) | Turned to profit | 2.01% | Consecutive losses 2024–2025, 66% debt ratio — a cost-curve control sample unable to profit at the same prices |
| Hongho Technology | Pure play on high-end/ultra-thin electronic fabric | 19.75 (+131.6%) | +435.6% | 29.96% | Purest AI fabric elasticity but ~1/10 of Jushi's electronic fabric volume, high leverage |
Company's Industry Position: Absolute leader with rising share (counter-cyclical expansion during the industry loss period of 2022–2024; 2024 roving sales +21.9%); three pillars of the moat: cost (per-ton cost ~RMB 1,017 lower, third-party figures), globalization (six bases + 43% overseas demand locally covered), and structural high-end mix (wind/thermoplastic/electronics over 85%). The current cyclical timing positioning is detailed in a dedicated chapter below.
Cycle template (three roving cycles since 2014): The first two cycles each lasted about 3 years; the third cycle's upswing ran from 2020-03 to 2022-01, with the 2400tex peak at roughly RMB 6,000–6,500/ton (Taishan quote of RMB 6,200/ton in 2021-09), and the downswing from 2022-02 to 2024-02 bottomed at RMB 3,000/ton (cumulative -48.7%); joint price restoration by industry leaders has been underway since 2024-03. Peak-to-trough gross margin spread is roughly 20pct (2021Q2 Jushi ~45–49.5% vs. the 2024Q1 crisis trough). Electronic fabric is an independent sub-cycle: 2021 peak at RMB 8.8/m → 2023 trough at 3.4–3.6 → 4.15 at end of 2025Q3 → all-time high of RMB 13.3/m in 2026-10.
Current position: desynchronized cycles. Roving 2400tex averaged RMB 3,749/ton in 2026Q3 (+5% YoY, -0.8% QoQ), at roughly the 21st percentile of the past decade's RMB 3,000–6,500 range — early-to-mid recovery, with price restoration far from complete; industry inventory of 905,000 tons at end-August (roughly 36–39 days, +41,000 tons QoQ, turning to restocking) is the main blemish. Electronic yarn/electronic fabric sit at the 100th historical percentile (G75 +150% over one year, 7628 +241%) — boom-top characteristics, but the top will persist as long as loom and certification bottlenecks remain unresolved.
Supply response: 2026 net supply is roughly -110,000 tons (cold repairs 46 > new capacity 35); 2027 is the key transition year — if roving new builds return to normal (peak historical years saw 500,000–900,000 tons of new capacity) while demand growth slows to single digits, oversupply could return in 2027H2–2028; on the electronic fabric side, high-end capacity concentrates in 2027 year-end through 2029 (Kingboard looms 3,300→5,800 units, Jushi Huai'an 50,000 tons of yarn + 320 million meters of fabric, Taishan Fiberglass 94 million meters of specialty fabric, Unitika Fukushima in early 2027, Taiwan Glass TD yarn plant), with industry-wide specialty electronic fiberglass investment of RMB 13.38 billion released simultaneously; on an 8–10-year cold-repair cycle, the 2020–2022 capacity peak corresponds to a 2028–2031 cold-repair peak that will partially offset new builds. The main clock for cycle peaking: around 2027Q4.
Earnings through the cycle: Mid-cycle normalized net profit attributable of roughly RMB 4.5–5.5 billion (EPS ~RMB 1.25) — roving 3.4 million tons × mid-cycle price of RMB 4,200–4,400/ton + electronic fabric 1.4 billion meters × realized price of RMB 6.5–7.5/m. Normalized PE uses the 10-year median of 17.4x ±5x → 12.4–22.4x corresponding to RMB 15.5–28.0: the current price of RMB 29.32 is slightly above the upper edge of the +1 standard deviation band, i.e., the market has priced in "mid-cycle earnings + a multiple slightly above the historical median." Trough EPS references 2024 non-recurring net profit of RMB 1.788 billion = RMB 0.447, with a 12–15x trough multiple corresponding to RMB 5.4–6.7 — this is an extreme disaster reference (fabric price and roving both returning to 2024 troughs, with attributable profit of roughly RMB 2.2–3.0 billion under the larger current capacity base using 2024 unit profitability, implying a market cap of RMB 26.4–45.0 billion), not the base case. The current 19.4x PE(TTM)/14.9–16.1x 2026E PE is a hybrid of "electronic fabric peak × roving trough": for electronic fabric it is a low PE at peak (a danger signal), for roving it is at a low level (a safety signal) — mechanically citing either side would be wrong.
Downside stress test (based on risk scenarios, industry price basis): 7628 falling back to RMB 8/m → attributable profit impact of roughly RMB 4.6 billion; falling to RMB 6 → roughly RMB 6.3 billion; roving midpoint from 3,750 to RMB 3,300/ton → roughly RMB 780 million. Compound extreme trough (fabric at RMB 4.5/m + roving at RMB 3,300/ton, i.e., both returning to 2024 prices in 2027–2028): attributable profit of RMB 2.2–3.0 billion, EBITDA of RMB 6.5–7.2 billion, net debt/EBITDA of roughly 1.2–1.3x, trough operating cash flow of roughly RMB 4.5–5.5 billion/year — FCF could be negative for consecutive years when the capex peak (annualized ~RMB 4.2 billion) coincides with an earnings trough, but cash plus rolling financing can see the company through the winter without a liquidity crisis; corresponding market cap reference range of RMB 26.4–45.0 billion (-62% to -78% from current price), assessed probability of roughly 18%.
Management's cycle discipline: neutral to slightly positive. After the 2022 expansion peak, capital expenditure contracted to RMB 1.3–1.5 billion in 2023–2025 (CapEx/depreciation 0.55, digestion-period discipline); most new projects were decided and started during the industry trough of 2025 (counter-cyclical decision-making), but fund deployment and production runs concentrate in 2027–2028 at high prices (pro-cyclical deployment risk); at the industry level, the restraint of cold repairs of 460,000 tons > 350,000 tons of new capacity and the discipline of joint price restoration form the organizational basis for the 2024–2026 price recovery. Verdict: counter-cyclical decisions, pro-cyclical deployment — discipline credits and risk items offset, leaving neutral to slightly positive.
Overall rating: Neutral, confidence 0.50, time horizon 12 months. China Jushi is a quality cyclical leader with dual No.1 positions in cost and scale in the global fiberglass industry; this desynchronized cycle of "AI electronic fabric extreme boom + roving low-level recovery" has delivered a doubling of its 2026 earnings; however, the earnings engine — electronic fabric — sits at the 100th historical price percentile, the supply clock points to end-2027, and the consensus 2027E of RMB 11.8 billion implies no fabric price decline. The current price of RMB 29.32 roughly equals this report's probability-weighted fair value (RMB 28.9): no margin of safety on the upside (PB at the 87th percentile, normalized PE of roughly 21–26x), while the downside is buffered by three layers — a 60% drawdown range, 9.9x 2027E consensus PE, and a 2.9% dividend yield. Strategy: avoid chasing highs; existing holders may hold for the dividend; the left-side attention zone for aggressive investors is below RMB 22 (lower edge of base-case fair value); the right-side trigger is disclosure of volume orders for low-dielectric fabric.
Monitoring points (within 12 months, in chronological order): ① Whether electronic fabric/yarn price adjustment letters in Nov–Dec 2026 and weekly 7628 quotes can hold above RMB 13/m (a break below RMB 12 or two consecutive weeks of QoQ declines = early normalization); ② annual earnings preannouncement before 2027-01-31 (lower bound ≥ RMB 7.5 billion = Q4 ≥ RMB 2.15 billion); ③ weekly average price of roving 2400tex and industry inventory (below RMB 3,600/ton or inventory above 1 million tons = roving engine stalling); ④ 2026 annual report and dividend plan (around 2027-03-20, payout ratio ≥45% or institutionalization of interim dividends); ⑤ Huai'an Phase II / Tongxiang electronic fabric lines and Unitika Fukushima commissioning pace (within 2027, earlier ignition = supply clock moving up); ⑥ low-dielectric fabric certification and volume orders (bull-case trigger); ⑦ further Zhenshi Group share reductions/pledge changes and China National Building Material's offsetting purchases; ⑧ timing and magnitude of the first downward revision of sell-side 2027E consensus within 2027 (confirmation signal of the market converging to this report's base case).
Price anchor for this report: RMB 29.32 (2026-10-29 completed close); unless otherwise noted, financial data are as of 2026-09-30 (Q3 report), and price and industry data as of late October 2026. Report date: 2026-10-29.