Rating: Neutral | Target Price: RMB 77–92 (base-case fair value range) | Current Price: RMB 91.5 (close on 2026-10-23) | Margin of Safety: approx. -16% | Time Horizon: 12 months | Market Cap: approx. RMB 38.27 billion (418.3 million total shares)
ASR Microelectronics is the No.1 domestic cellular baseband chip vendor by shipments (37.8% global share of cellular connectivity chip shipments in 2025, per Frost & Sullivan), and is now at the critical validation stage of "market-share realization → profitability realization": 2026H1 revenue of RMB 2.451 billion (+29.15%), first-ever half-year reported profit (net profit attributable to parent of RMB 84 million), with ASIC custom chip orders on hand exceeding RMB 1.5 billion forming the second growth curve. But earnings quality remains unproven — RMB 143 million of non-recurring gains underpin the headline turnaround, while the company still lost RMB 59 million on an adjusted (recurring) basis, with operating cash outflow of RMB 377 million. The current price of RMB 91.5 implies PS(2025) of about 10.0x, in line with the median of profitable design peers (about 10.5x) despite vastly weaker earnings content; reverse DCF shows the market has already priced in the success path of "2026–30 revenue CAGR of 19–21% with net margin climbing to 10–12%." Base-case fair value is RMB 77–92; the current price sits at the upper end of the range with no margin of safety. The sell-side consensus median target of RMB 152 (18 Buy, 0 Sell) effectively prices this report's bull case. Key constraints: the pace of converting ASIC orders into revenue (Q2 custom revenue already fell 28.7% QoQ), the potential reignition of the baseband price war, and dilution from a discounted H-share offering. Conclusion: Neutral — the inflection in fundamentals is real in direction, but the price already pays full value; await the two validation points of the 2026-10-31 Q3 report and Q4 delivery acceptance; the risk/reward only becomes attractive on a pullback to RMB 77–85.
The most likely changes over the next 12 months: ① Recurring net profit converging from -RMB 59 million (2026H1) toward breakeven or turning positive — annualized revenue of about RMB 4.9 billion vs. a breakeven revenue level of about RMB 5.03 billion (at 29.43% gross margin), with extremely thin operating leverage; direction depends on price increases taking hold and the pace of ASIC revenue recognition. ② ASIC custom business moving from "order narrative" to "delivery and acceptance" phase (the 3D DRAM cloud-computing project begins phased delivery and acceptance from 2026Q4). ③ The A+H dual-listing platform either materializing or lapsing (the HK application is valid until 2027-03-11).
The primary share price driver is the pace of ASIC custom revenue recognition and the recurring profit turning positive (both the core engine and the most fragile link); next is overall gross margin (falsification threshold: 27%); then share supply (exit expectations for Alibaba's remaining 10.58% stake + H-share offering pricing). The verifiable expectation gap: market consensus FY2027E net profit of about RMB 320 million (EPS RMB 0.77, n=6) and median target of RMB 152, versus this report's base-case 2027E net profit of RMB 300–450 million (midpoint RMB 380 million), implying fair value of only RMB 77–92 — the disagreement is essentially about exit multiples and the slope of net-margin ramp, not the direction of orders. Validation catalysts: Q3 report (2026-10-31), Q4 delivery acceptance, H-share hearing and pricing; falsification triggers: Q4 acceptance deferred, gross margin falling below 27%, deeply discounted H-share issuance. Opportunity type: a validation-driven opportunity, not an expectation-gap arbitrage (evidence grade B).
The structural fragility of the headline turnaround is concentrated in a single quarter: 2026Q1 net profit attributable to parent was -RMB 28 million, Q2 +RMB 112 million, with nearly all non-recurring gains realized in Q2. The divergence between profit and cash continues to widen — while revenue grew +29.15%, inventory rose +26.5% (RMB 1.704→2.155 billion), receivables +47.6%, and prepayments +73.8%; every RMB 1 of incremental revenue requires roughly RMB 1 of working capital. 2025H2 operating cash outflow was about RMB 785 million, so there is no seasonal pattern of "natural recovery in H2"; however, 2026Q2 outflow narrowed to about RMB 79 million from Q1 (-RMB 299 million), a roughly 74% improvement — if Q3 continues to narrow, the "cash-conversion inflection" would be confirmed ahead of this thesis. This report's reverse test is the Q3 report: quarterly recurring profit turning positive and narrowing operating cash outflow must appear simultaneously.
The order evidence needs to be parsed: the A-share semi-annual report only discloses "over RMB 1.5 billion" and structural percentages; the widely cited "+416.5% YoY" comes from a footnote in the H-share prospectus (implying a 2025-06-30 base of about RMB 290 million), and its original wording is "order volume on hand," with the value basis unverified — citations should be down-weighted. Three weak links in the conversion chain: ① NRE revenue is recognized in one lump upon "delivery and passing functional acceptance," so management can influence the recognition year through delivery scheduling; the company's IR has itself admitted "relevant delivery amounts cannot be accurately estimated at present"; ② Turnkey volume production is priced on a cost-plus basis, capping gross margin after conversion; ③ training-grade large-compute projects require ex-ante export-control compliance screening, shrinking the addressable order pool. The 3D DRAM delivery acceptance from 2026Q4 is the nearest hard test — if Q3 custom revenue rebounds QoQ (>RMB 134 million) and contract liabilities keep growing, the evidence chain will strengthen materially; if acceptance slips to 2027H1, full-year recurring profitability most likely fails.
The gap between share and profitability stems from "trading price for volume": discounting in the low-price Cat.1bis segment to drive shipments; the 2023 price war caused gross margin to plunge 12.78pp in a single year. Two quality caveats on this round of gross margin recovery: first, the company's IR attributed price increases to "rising supply-chain costs upstream and downstream" — whether cost-push price increases can persist in a fully competitive market is questionable; second, 2026Q1 gross margin was about 30% while H1 fell back to 29.43%, implying a QoQ decline in Q2 — 29.43% may be the annual peak rather than an extrapolable center. On the competitive timeline: HiSilicon returned to the NB-IoT market in 2024, and the company itself acknowledged in early 2026 that "the cellular IoT chip market has been fully competitive since H2 2025" — if HiSilicon expands from NB-IoT into Cat.1/Cat.4 (ASR's shipment heartland), price pressure could emerge before 2027. Warning line: overall gross margin falling below 27% (recurring breakeven requires roughly 28%+ gross margin).
Reverse DCF shows: the current price requires 2030E net profit of about RMB 1.24 billion (exit PE 45x, 10% discount rate), corresponding to 2030E revenue of about RMB 10.4 billion at a 12% net margin — i.e., a 2026–30 revenue CAGR of about 19–21% with net margin climbing from about 0% to 10–12% and operating cash flow turning positive, all simultaneously. Note: a 45x exit PE is a lenient assumption; the more conservative the exit multiple, the higher the implied growth requirement — 19–21% is the lower bound of the required growth rate. Industry math (cellular connectivity chip shipment CAGR of 18.4%; ASIC custom market CAGR of 53%) can support that implied growth, but execution allows zero errors. See the valuation section for the three-layer value decomposition — EPV at zero growth of about RMB 11.6/share plus book value per share of RMB 13.97 forms the floor; 87% of the current price is the option premium paid for ASIC/5G/edge-AI growth.
This report's view on the shareholder register is flatter than consensus: Alibaba's sell-down plan has been terminated, the transferred shares are locked up for about 6 months, the unlock queue is empty, and there is zero share pledging — there is currently no definitive sell pressure in transit. The true overhang is the future exit of the remaining 10.58% stake (about RMB 4.0 billion market value) and the controlling shareholder's annual 2% sell-down quota (Dai Baojia's direct holdings were unlocked in July 2026). The H-share offering is the bigger unknown: if priced at a 30%+ discount with 10% dilution, the A/H spread would reinforce the "A-share valuation bubble" narrative; if the offering proceeds smoothly with the discount narrowing to single digits (several large A+H offerings in 2025–26 carried only single-digit discounts), the completed raise would instead validate the asset's scarcity. The milestone resolves by 2027-03-11.
| Metric | 2023FY | 2024FY | 2025FY | 2026H1 (latest) |
|---|---|---|---|---|
| Revenue (RMB bn) | 2.600 | 3.386 | 3.817 | 2.451 |
| YoY | +21.48% | +30.23% | +12.73% | +29.15% |
| Net profit attributable to parent (RMB bn) | -0.506 | -0.693 | -0.390 | +0.084 (first half-year profit) |
| Recurring net profit attributable to parent (RMB bn) | approx. -0.658 (est.) | -0.707 | -0.579 | -0.059 |
| Gross margin | 24.35% | 23.19% | 24.95% | 29.43% |
| Net margin | -19.5% | -20.5% | -10.2% | +3.4% |
| Operating cash flow (RMB bn) | -0.678 | -0.404 | -0.432 | -0.377 |
| Free cash flow (RMB bn, OCF-CapEx) | — | -0.658 | -0.543 | -0.435 |
| Cash + cash-like assets (RMB bn, period-end) | — | — | — | 2.915 (as of 2026-06-30) |
| Interest-bearing debt (RMB bn, period-end) | — | 0 | 0.350 | 0.619 (as of 2026-06-30) |
| Debt-to-asset ratio (period-end) | — | 13.08% | 21.81% | 25.02% |
| R&D expense (RMB bn) | 1.116 (42.9% of revenue) | 1.242 (36.7%) | 1.299 (34.0%) | 0.663 (27.1%, -0.77% YoY) |
Note: 2023 recurring loss is an estimate of "net profit attributable to parent minus non-recurring items of RMB 152 million"; free cash flow = net operating cash flow - cash spent on long-term asset purchases; the company capitalizes no R&D and pays no dividends (parent-company undistributed profit of -RMB 5.108 billion, structurally undistributable); net debt/EBITDA not applicable (negative EBITDA).
Reasons for Metric Changes (items with YoY ≥ ±20%, all per company disclosures): ① Chip custom revenue +155.47% in 2026H1 — the company explains this as "order scale continuing to expand, with projects progressively entering delivery and acceptance" (it had strategically tilted ASIC resources toward in-house chips over the prior two years, creating a -35.30% low base in 2025); ② Contract liabilities +101.39% — increased advance payments for custom R&D services, a leading indicator of the order backlog; ③ Accounts receivable +47.63% — "increased sales revenue and corresponding customer receivables" (growth about 1.6x that of revenue; the company has not disclosed any aging deterioration); ④ Inventory +26.5% — "stocking up in line with sales expansion," with an inventory write-down provision balance of RMB 140 million (about 8.2% of inventory); ⑤ Other income (mainly government grants) -91.65% — fading grants are a drag during the turnaround period; ⑥ Gross margin +4.72pp — "improved product shipment mix and scale effects," plus differentiated price increases amid rising supply-chain costs.
Revenue of RMB 2.451 billion set a half-year record (cellular baseband RMB 2.022 billion, +23%; custom RMB 322 million, +155%), with gross margin of 29.43% (+4.72pp YoY) the highest since the 2023 price war. A three-factor decomposition of the headline turnaround: gross margin improvement contributed about RMB 116 million of incremental gross profit, lower share-based compensation expense (after-tax impact reduced by about RMB 46 million YoY), and non-recurring gains of RMB 143 million (fair-value changes of RMB 72 million plus investment income, etc.); the YoY reduction of about RMB 91 million in government grants was the main drag. The deviation from the 2026-07-23 profit preannouncement (revenue ~RMB 2.45 billion, net profit ~RMB 85 million, recurring ~-RMB 58 million) was under 1% — management's results communication has historically been precise. Market reaction: after the semi-annual report, sell-side estimates were widely revised up (78.9% of samples raised in the past 90 days; median target RMB 152), but the stock has pulled back about 27% from its July high of RMB 125.68 to RMB 91.5 — the market is oscillating between the "order narrative" and "realization evidence"; the Q3 report (2026-10-31) is the next adjudication point.
Business model: Fabless chip design (asset-light) — in-house chip sales account for 93.78% of revenue (volume-driven, low margin); ASIC custom is recognized as one-off NRE plus follow-on production orders (project-based); IP licensing is high-margin but has shrunk to 0.53%. Revenue is non-recurring (no subscriptions); pricing power is weak but improving: forced to cut prices in the 2023 price war, in 2026 it achieved differentiated price increases for the first time, leveraging cost inflation and mix upgrade.
Cash content of earnings (core test): Book profit cannot yet be converted into freely disposable cash — 2026H1 net profit attributable to parent +RMB 84 million vs. operating cash flow -RMB 377 million (OCF/net profit of -4.48x); FCF persistently negative (2024 -RMB 658 million, 2025 -RMB 543 million, 2026H1 -RMB 435 million). The cause is not capex (CapEx/D&A only 0.42; the main investment is expensed R&D of RMB 1.299 billion/year) but working capital: every RMB 1 of incremental revenue absorbs about RMB 1 (inventory + receivables + prepayments) — the "growth = bleeding" structure has not yet reversed. Recurring earnings test: the gap between 2026H1 reported and recurring net profit was RMB 143 million (≥±15%), dominated by fair-value changes and investment income — core earnings (recurring) still a loss of RMB 59 million, though the trend has narrowed for three and a half years (-707→-579→-59 million on a semi-annualized basis). The company reports no Non-GAAP adjustments of its own, but the disclosure of "net profit of RMB 123 million excluding share-based compensation" hints that the fading of share-based payments (RMB 180→138 million, ~RMB 39 million in H1) flatters the loss-narrowing slope across periods.
Return on capital: With consecutive years of losses, ROIC is negative and meaningless; 2025 weighted ROE was -6.99%, and 2026H1 annualized about 3% — below the cost of equity of about 10.5%; the growth-quality gate (ROIC vs. WACC) has not yet been passed. Capex quality: As a Fabless, the company has no fabs; maintenance and growth investments are mixed within expensed R&D and cannot be separated (zero capitalization). Moat and red flags: Moat = self-developed baseband IP and protocol stack covering all standards from 2G to 5G (one of very few globally), certified by four of the world's top five carriers; Red flags = top-5 customer concentration of 82.55% (largest at 37.57%) and trending higher, largest foundry supplier accounting for 47.13% of purchases (two-way concentration), and RMB 2.155 billion of inventory requiring recurring write-downs (consuming about RMB 40–90 million of profit annually).
Say-do consistency: Pragmatic. Three cross-checks: ① The 2023 annual report promised mass production of ASR1903 (RedCap) in 2024 — mass production was achieved in 2024Q4 and certified by China Mobile/China Unicom; ② The 2026H1 earnings preannouncement (revenue RMB 2.45 bn / net profit attributable RMB 85 mn / non-GAAP net loss RMB 58 mn) deviated from actuals by <1%; ③ The 2025 annual report product roadmap (second-generation 4G octa-core SoC in volume production within 2026, 5G octa-core SoC in production by end-2026, ASR1906 Design-in in 2026Q4) stood, as of 2026H1, at the "customer introduction / field testing" stage — in the latter part of the timeline but not yet falsified. The track record of delivering on guidance is an important calibration input for this report's C2 probability judgment.
Shareholder friendliness: Neutral-to-cautious. Positives: in 2024, repurchased and cancelled shares with RMB 815 mn of excess proceeds; precise and prudent performance communication. Negatives: no dividends (structurally unable, not unwilling); the H-share issuance is proceeding in the fifth year of trading below issue price, and amid concentrated protests from minority shareholders the company made no hedging commitments; the ASIC order backlog figure was declined in July citing confidentiality, but was disclosed in late August alongside the H-share filing — the disclosure standard is coupled to the financing timetable, warranting a discount.
Risk signals: Actual controller Dai Baojia and concert parties hold only 21.93% (22.32% voting rights) vs ~20.4% for the Pudong SASAC system — a thin control structure, though both the state-owned entity and Alibaba have written commitments not to seek control; the actual controller is 70 years old with unclear succession plans; Alibaba, after cumulatively reducing and cashing out ~RMB 1.76 bn since 2025, has terminated its reduction plan (still holding 10.58%); the related-party transactions where "core customers are simultaneously suppliers" have drawn media attention and may become a focus of H-share regulatory queries.
| Segment | 2025 Revenue / Share | Gross Margin | YoY | Business Logic in One Line |
|---|---|---|---|---|
| Chip products (cellular baseband as the absolute core, 93.81% of chip revenue) | RMB 3.579 bn / 93.78% | 24.04% (+3.72pp) | +18.77% | Cat.1/Cat.4 IoT base volume; 5G/smart SoC ramp; trading low margin for share |
| Chip customization (ASIC) | RMB 217 mn / 5.69% (rising to ~13% in 2026H1) | 33.10% (-7.73pp) | -35.30% (2026H1 +155.47%) | NRE recognized one-off + volume follow-on orders; backlog RMB 1.5 bn+, delivery cycle ~1.5 years |
| Semiconductor IP licensing | RMB 20 mn / 0.53% | 99.02% | -42.95% | High-margin licensing; company proactively shrinking to focus on in-house R&D |
The profit engine is the chip products segment (93.78% of revenue × 24.04% gross margin — gross profit ~RMB 860 mn, the overwhelming majority); ASIC carries a higher gross margin (33.10%) but a small base, and with its 2026H1 revenue share rising to ~13%, it is the main marginal driver of gross-margin mix improvement. The gross-margin spread (IP licensing 99% vs chip products 24%, a 75pp gap) reflects entirely different businesses: licensing is the rental/sale of existing IP, while chips are a volume business bearing manufacturing costs plus price wars — two business models within a single income statement that should not be valued on a uniform multiple. The ASIC segment's gross margin swung at 29.41% → ~40.8% → 33.10% over 2023–2025, essentially driven by shifts in the NRE vs volume-revenue mix.
Accounting red flags: ① Non-recurring gains propping up the turnaround (High) — 2026H1 fair-value changes plus investment income of RMB 143 mn, still loss-making excluding non-recurring items; fair-value gains fluctuate with markets and are unsustainable. ② Receivables growth significantly above revenue growth (Medium) — +47.6% vs +29.1%; the company offered only the vague explanation of "increased sales," with no aging quantification. ③ Elevated inventory + normalized write-downs (Medium) — balance of RMB 2.155 bn, inventory volume +55.76% in 2025 (sales volume +34.36%), annual provisions of RMB 39/88/59 mn in 2023–2025. ④ Share-based payment timing distortion (Low) — swings of RMB 11 mn → 180 mn → 138 mn → 39 mn in H1 amplified the 2024 loss and flatter the loss-reduction trajectory of 2025–26. ⑤ One-off recognition of NRE revenue at acceptance (Low) — management can influence the recognition year via delivery scheduling.
Cross-period consistency: ① Gross margin trajectory of 24.35% → 23.19% → 24.95% → 29.43% is consistent with management's period-by-period explanations (price war → mix improvement); ② OCF negative for three and a half consecutive years with no narrowing of the gap in the turnaround period — consistent with the explanation of "cash paid for purchases growing faster than cash collected from sales" (not an accounting-caliber manipulation but genuine working-capital expansion); ③ The boom-bust of customization revenue at RMB 226 → 340 → 217 → 322 mn is consistent with the NRE recognition policy; ④ Working capital growing faster than revenue (inventory/receivables/prepayments) — unexplained by the company, an ongoing tracking item of this report; ⑤ Non-recurring gains/losses tilt reported profit direction every year (+152/+14/+188/+143 mn) — composition is transparent, not concealed, but a reminder to investors: always read ASR's income statement via the non-GAAP (ex-non-recurring) column.
Current market data (2026-10-23 close): share price RMB 91.5, market cap RMB 38.27 bn, net cash ~RMB 2.42 bn (EV ~RMB 35.85 bn). PB 6.55x (1/3/5-year percentiles of 34%/66%/73%); PS(TTM) 8.76x, PS(2025) 10.0x (market cap at the 81st percentile of the past 5 years); PE(TTM) is distorted in the early turnaround period (barred as a pricing signal); forward PE (FY2027E consensus) ~119x, PEG ~1.9 — base effects in the early turnaround phase, for cross-validation only. This stock's valuation anchors are PS/EV-Sales and scenario-based earnings ramp; PE percentiles are for reference only.
What the market has already priced in: The current price requires ~RMB 1.24 bn of 2030E net profit attributable (45x exit PE, 10% discount rate) — i.e., "revenue CAGR of 19–21% + net margin 0% → 10–12% + cash-flow turnaround" all holding simultaneously; the reality is that non-GAAP profit has just reached breakeven (annualized revenue RMB 4.9 bn vs breakeven revenue ~RMB 5.03 bn), cash flow still bleeding, but order evidence (backlog RMB 1.5 bn+, contract liabilities +101%) is strengthening. Industry headroom (cellular chip shipment CAGR 18.4%, ASIC customization CAGR 53%) can support the implied growth — neither undervalued nor a bubble; this is "paying full price for the path to success."
Three-layer value (EPV): Asset value (book value per share) RMB 13.97; zero-growth EPV ~RMB 11.6/share (normalized FCFF RMB 0.61/share ÷ 10.5% WACC + net cash RMB 5.78; normalized basis = 2026H1 annualized revenue RMB 4.9 bn, gross margin 29.43%, maintenance R&D at 70% of RMB 929 mn, tax rate 15%; range RMB 3.9–14.7 — the floor assumes full expensing of R&D). The growth option = current price 91.5 − EPV 11.6 ≈ RMB 79.9/share, ~87% of the current price. A note on bridging calibers: EPV measures cash-flow value under "current run-rate, zero growth, maintenance R&D fully expensed," while the bear case (RMB 38–49) measures the value the market would pay on a revenue multiple (peer floor of 3.5–4.5x PS) under "growth decelerating but revenue scale maintained at the RMB 5 bn level" — different calibers, not contradictory; the substance of the difference is that even if growth is falsified, the market will still pay a premium on revenue scale, but the company's own zero-growth cash-flow value is far below that. Whether growth should be paid for depends on the scenario ramp:
Three scenarios and odds (exit multiple anchors: base case uses 2028E revenue RMB 7.8 bn × PS 5–6x ≈ half of the peer median 10.5x, the discount reflecting gross margin of 32% vs peers' 38–46% and the ROE ramp period; not anchored to the company's current 8.76x TTM PS. Cross-check: 2028E net profit RMB 700 mn × PE 55–66x ≈ peer PE(TTM) median of 53.7x — the two methods converge):
| Scenario | Probability | Fair Range | vs Current Price | Decisive Factors |
|---|---|---|---|---|
| Bear | 30% | RMB 38–49 | -58% to -46% | ASIC acceptance failure/deferral + price war reignition: 2027 revenue ~RMB 5 bn, GM 27%, exit at PS 3.5–4.5x (5–6折 on Allwinner's trailing 10.5x), plus escalation-of-controls tail risk (~-70%) |
| Base | 50% | RMB 77–92 | -16% to +0.5% | ASIC delivered on schedule and converting to volume: 2026E RMB 5.2 bn → 2028E RMB 7.8 bn (2025–28 CAGR 26.9%), GM 32%, 2028 net profit ~RMB 700 mn |
| Bull | 20% | RMB 127–157 | +39% to +71% | Continued large AI ASIC orders + 3D DRAM ramp + smooth H-share listing: 2028E revenue RMB 9.9 bn (CAGR 37.4%), net margin 13.6% (benchmarked to Amlogic), exit at PS 6.5–8x |
Probability-weighted fair value ~RMB 84, about -8.5% vs the current price — odds skewed to the downside; the current price sits at the top of the base-case range. Sum-of-the-parts (SOTP) sensitivity: baseband + others, 2027E revenue ~RMB 5 bn × 4–5x PS = RMB 20–25 bn; ASIC RMB 1.2–1.5 bn × 5–15x PS = RMB 6–22.5 bn; total ~RMB 26–47.5 bn (RMB 62–114/share) — the range spans both sides of the base case, indicating that ASIC segment pricing is the decisive variable for the whole. External coordinates: the most conservative sell-side target of RMB 125.1 (Guohai, 9.3x 2026E PS) still sits above this report's base-case ceiling — under a coverage structure of 18 Buy ratings with zero Holds or Sells, no genuine bear-case anchor exists; this report's bear case is constructed on independent stress assumptions, not the sell-side floor; the most aggressive credible target is RMB 175.72 (Guotai Haitong, segment PS, 2028 revenue RMB 8.9 bn / net profit RMB 780 mn — though its international subsidiary is an H-share sponsor, warranting a discount for independence).
Own earnings forecast (reconciliation anchor): 2026E revenue RMB 5.0–5.3 bn (+31% to +39%; H2 requires RMB 2.55–2.85 bn), net profit attributable RMB 150–220 mn (including non-recurring RMB 140–200 mn), non-GAAP -RMB 100 mn to 0; 2027E revenue RMB 6.2–6.7 bn, net profit attributable RMB 300–450 mn (midpoint ~18% above consensus of RMB 320 mn; the divergence lies in the gross-margin slope and ASIC volume-ramp pace). Downside risks: ASIC subsidiary headcount expansion plus H-share issuance costs will raise the expense base, making breakeven revenue higher than the static estimate of RMB 5.03 bn. Management gave no quantified full-year guidance (only qualitative "optimism for H2").
Conclusion: Fair — a genuine inflection in fundamentals, but full price already paid. Target price RMB 77–92 (base case), safety margin ~-16% (base-case fair-value floor of RMB 77 vs current price of 91.5); fundamentals (share leadership + full-standard IP) and price (87% growth option) should be judged separately: a good company, an expensive ticket — the odds only turn positive on a pullback to the RMB 77–85 range.
Industry headroom: ASR's main track is cellular IoT connectivity chips. Frost & Sullivan (cited in ASR's H-share prospectus): global cellular connectivity chip shipments in 2025 of 634.5 mn 4G units + 63.3 mn 5G units, projected at 914.4 mn units in 2026 → 1,797.4 mn units in 2030, CAGR 18.4% (2030: 4G 1,262 mn units, 5G 535.5 mn units); cellular IoT module shipments in 2025 of 544 mn units (+23%, IoT Analytics). In adjacent edge-AI chips, the global market in 2026 is projected at USD 68.8 bn (via Gartner); the ASIC customization chip market shows a 2023–28 CAGR of 53%, reaching USD 55.4 bn in 2028 (Marvell basis, cited via Soochow Securities) — this is the industry beta underlying ASR's second growth curve. Demand inflection chain (partially quantified): ① RedCap penetration — only 0.4% of module shipments in 2025; Counterpoint projects 18% by 2030 (generational replacement + rising value per unit: RedCap terminals command a 20–30% premium over Cat.4 with 2–3x performance; US carrier AT&T halted new 4G certifications from 2025, forcing migration). But ASR's share of the RedCap segment rests only on company targets (the "over one million units shipped in 2026" figure is yet to be verified) and faces competition from HiSilicon's first-mover position — the inflection chain is unproven at the company level, hence this report does not switch its primary valuation anchor to a growth basis. ② ASIC — hard company-level data (backlog RMB 1.5 bn+, contract liabilities +101%) is partially confirmed.
Value chain and value distribution: EDA/IP licensing → wafer foundry (ASR mainly uses leading overseas foundries plus SMIC mature nodes; the largest supplier accounts for 45.5%–52.4% of purchases) → packaging & testing → fabless design (where ASR sits) → module makers (Quectel with 41% global share in 2026Q2, Fibocom, Lierda, etc.) → end devices (smart metering/POS/trackers/gateways/wearables/automotive/industrial). Gross-profit retention: design (24–50%) > module makers (low single digits to 15%); but within design, dispersion is extreme — high-end platform players (Rockchip 45.8%) >> low-price volume players (ASR 24.95%). Upstream: foundry concentration, with SMIC raising prices on part of its capacity ~10% in December 2025, 93.7% utilization in 2026Q2, and memory price hikes squeezing BOM — fabless firms have weak bargaining power and are in a rising-cost cycle. Downstream: chip supply CR5 = 92% vs fragmented module/end-device customers, giving ASR category-level pricing power (its first differentiated price increase landed in 2026).
Supply-demand and competitive structure: Chipset CR5 ≈ 92% (ASR/Asr Micro/Chipwing/UNISOC/Qualcomm, IoT Analytics 2025); Chinese vendors hold ~85% of chipsets and ~90% of modules — a highly concentrated structure but with extreme internal price competition: Asr Micro is attacking with the market's lowest-price Cat.1bis (2026Q2 shipments +20%, expanding overseas as Qualcomm's exclusive agreement expires); UNISOC's revenue is roughly 4–5x ASR's (RMB 14.5 bn in 2024) but its IoT baseband share declined in 2025 and is being eroded by ASR; the biggest structural variable is HiSilicon's return (already back in NB-IoT in 2024, early leader in RedCap). Entry barriers are extremely high: full-standard baseband IP + in-house protocol stack (one of very few globally), carrier certification cycles, ~RMB 1.3 bn in annual R&D, and software stacks for fragmented scenarios. Substitution threats: WiFi/BT displacing cellular in fixed scenarios, NB-IoT contraction in China, and 4G→5G as an intra-industry generational shift.
Cycles and regulation: Dual cycles overlap — the industry life cycle is in the late growth stage (cellular IoT connection growth of 13.3% is the lowest since 2020, volume up price down); the semiconductor cycle is in an upswing (structural tightness in mature nodes, order visibility to 2027). Regulation is broadly favorable: China's mandatory GNSS tracking rules for e-bikes (annual sales of 60–65 mn units, directly creating Cat.1bis increments), wearables included in national consumption subsidies, and Xinchuang domestic substitution; tail risk: US BIS advanced-node export controls — ASR's existing IoT business, mostly on domestic mature nodes, is largely unaffected, but ~80% of the ASIC backlog is on 6nm/4nm advanced nodes; the company has front-loaded compliance screening for training-grade projects, and the controls' boundary effectively caps the order book it can take on.
Peer comparison (2025 annual report / 2026-10-23 close basis):
| Company | 2025 Revenue | Growth | Gross Margin | ROE | PS(2025) | Key Difference vs ASR |
|---|---|---|---|---|---|---|
| ASR Microelectronics | RMB 3.817 bn | +12.7% | 24.95% | -6.99% (loss) | 10.0x | Only player with full-standard baseband + ASIC customization dual engine; no profit |
| Rockchip | RMB 4.402 bn | +40.4% | 41.95% | 25.79% | 20.6x | Edge-AI application processor leader; benchmark for profit delivery |
| Amlogic | RMB 6.793 bn | +14.6% | 37.97% | N/A | 6.3x | Large-customer platform model; net margin ~15% (bull-case reference) |
| Espressif | RMB 2.565 bn | +27.8% | ~40% | 11.84% (first three quarters of 2026) | 14.1x | WiFi MCU ecosystem; non-cellular reference |
| Bestechnic | RMB 3.525 bn | +8.0% | 38.6% | 8.81% | 8.0x | Wearable audio SoC; -26.8% in first three quarters of 2026 (a cautionary mirror of key-customer risk) |
| Allwinner | RMB 2.838 bn | +24.0% | 43.12% (2026H1) | N/A | 10.5x | The most comparable "profitable version" of ASR (2025 net profit RMB 262 mn) |
| UNISOC (unlisted) | ~RMB 14.5 bn (2024) | +11% | N/A | Slim profit | ~3.5x in primary market | One of the two domestic baseband champions; IoT share being eroded by ASR |
Company positioning: ASR has completed the leap from niche challenger to global leader in specific segments — No.1 globally in cellular connectivity chip shipments (37.8% on a shipment basis), and first in shipments across multiple sub-segments including Cat.1/Cat.4/cellular gateways/smart wearables baseband; share trending upward (in 2025, Qualcomm/UNISOC declined while ASR rose against the tide); moat = full-standard IP legacy (from the 2017 acquisition of Marvell's mobile IoT business) + carrier certification + platform flywheel. Shortcomings are equally clear: with profitability at the bottom of peers under a price-for-share strategy, the company stands at the threshold from "share delivery → profit delivery."
The essence of chip design pricing is the product cycle portfolio: tape-out → certification → design-in → volume ramp, with each product generation determining 2-3 years of revenue elasticity. ASR's current matrix:
| Product Platform | Process/Positioning | Commercialization Status (as of 2026H1/latest IR) |
|---|---|---|
| Cat.1/Cat.4 baseband (ASR1603/1606, etc.) | Mature process | Mature mass production, core business (2025 chip sales of 351 million unit equivalents, +34.4%) |
| 5G RedCap: ASR1903 | Release 17 | Mass production since 2024Q4, passed China Mobile/China Unicom certification; ASR1906 in engineering validation, customer design-in in 2026Q4 |
| RedCap+Android: ASR8603 | World's first such combination | Small-batch shipments in 2026Q1, ramping toward large-scale commercial deployment |
| 5G eMBB platform | 5G data-type | Over 500,000 units shipped in 2026Q1, approximately 1.5 million units cumulative |
| 4G smart SoC: ASR8601/8661 | Quad-core/mass production | Quad-core cumulative shipments exceeded 5 million units; 8661 in volume production ramp |
| 2nd-gen 4G octa-core ASR8861 | 6nm, 20 TOPS edge AI | In joint customer debugging; company guides gradual mass production in 2026Q4 |
| 5G octa-core smart SoC | High-end smart | Field testing, targeting mass production by end of 2026 (per 2025 annual report plan) |
| ASIC customization (incl. 3D DRAM cloud computing) | 6nm/4nm accounting for 80% | Backlog of RMB 1.5 billion+; 3D DRAM project delivery and acceptance starting 2026Q4 |
Two structural observations: ① High new-product density but high-end node mass production timing generally falls in the "late stage" — the three key 2026 catalysts (ASR8861 / 5G octa-core SoC / ASR1906) are all concentrated in Q4, overlapping with ASIC delivery acceptance; Q4 is the concentrated realization window for the company's product cycle, as well as an amplifier of volatility; ② The ASP midpoint of the baseband core business is low (2025 chip revenue of RMB 3.579 billion / 351 million units ≈ RMB 10 per unit equivalent); revenue elasticity mainly depends on smart SoCs (ASP at roughly USD 8-15 level) and the upward shift toward ASIC — product structure upgrade is the only path to lift gross margin from 24% to 30%+, with validation lagging by 2-4 quarters.
Overall rating: Neutral (confidence 0.55). ASR's investment story = "world's No.1 shipment share × unproven profitability × share price already priced on a success path." The ASIC order evidence is real (backlog RMB 1.5 billion+, contract liabilities doubled) but the conversion chain has three weak points (advance payments covering only 30%, Q2 sequential decline, acceptance timing admitted by the company as unpredictable); the baseband core business ranks first in volume but with thin margins, with the HiSilicon variable possibly arriving early; the H-share offering is a double-edged sword. At the current price of RMB 91.5, the stock sits at the upper edge of the base-case fair value (RMB 77-92), with odds skewed downward (probability-weighted roughly -8.5%). Strategy: do not chase; accumulate in batches within the RMB 77-85 range; existing holders should use the RMB 92 upper edge and 27% gross margin as trigger lines for trimming/re-rating.
Key Risk and Catalyst Alerts: Upside — Q3 report shows sequential recovery in customization revenue + narrowing of non-GAAP losses, Q4 delivery acceptance initiation, smooth H-share pricing, ASR8861 mass production, gross margin surpassing 31%; downside — acceptance deferral, gross margin falling below 27%, H-share discount of 30%+ with dilution of 10%+, export control escalation affecting 6nm/4nm tape-outs, pre-disclosure of a new Alibaba stake reduction plan. Ongoing monitoring items: inventory aging and write-down provisions, receivables aging (unexplained working capital expansion), proportion of non-recurring gains/losses, drift in top-five customer concentration.
Data sources: ASR Microelectronics 2023/2024/2025 annual reports, 2026 Q1 report, semi-annual report and earnings pre-announcements (cninfo), H-share prospectus application version (HKEXnews, 2026-09-11), SSE e-interaction company replies (2026-07~09), public reports from Securities Times / Cailian Press / National Business Daily / Eastmoney, industry data from IoT Analytics / Counterpoint / Frost & Sullivan / ABI Research. Valuation calculations are based on a frozen anchor at the 2026-10-23 closing price of RMB 91.5.