Rating: Neutral | Target Price: RMB 74–114 (base-case fair value range, midpoint RMB 94) | Current Price: RMB 101.53 (as of 2026-10-30 close) | Margin of Safety: −27.1% | Investment Horizon: 12–18 months
Market cap: RMB 24.051 billion (total shares outstanding: 236.9 million); currency: RMB. All prices and multiples in this report are anchored to the 2026-10-30 closing price.
Bestechnic is China's leading wearable SoC designer for branded products: its TWS controller chips rank fourth globally in shipments (approximately 10% share, per Chaodain智库/Chaodian Intelligence data), its 6nm process node leads domestic peers by one to two generations, and its blended gross margin of 42.27% in the first three quarters of 2026 hit a record high. 2026 is a year of exogenous shocks — memory price hikes, the phase-out of China's consumption subsidy (national subsidy), and a high base following the share gains realized in 2024–25 — with 9M26 revenue down 26.8% and net profit attributable to shareholders down 57%. However, sequential revenue has improved for two consecutive quarters (Q3 +16.7%), and the unusually low Q4 2025 base of only RMB 591 million makes a Q4 year-over-year turnaround within reach. At the current price of RMB 101.53, the stock trades at 78.1x PE(TTM) (distorted by the earnings trough, for reference only) and 40.5x static PE (13th percentile of the past 5 years); the market has already fully priced in the base case of "recovery to 2025 earnings levels in 2027 + roughly 40x," leaving the AI glasses controller optionality (BES6100) nearly free. Three-scenario fair value: bear RMB 32–51 (30%), base RMB 74–114 (50%), bull RMB 152–232 (20%); probability-weighted value is approximately RMB 97.9, or −3.6% versus the current price — fairly priced with no margin of safety, a deep left tail (−59% to the bear-case midpoint) and a wide right tail (+89% to the bull-case midpoint). Take a neutral stance, driven by two falsifiable checkpoints: whether Q4 revenue in the January 2027 preliminary results announcement is ≥ RMB 800 million; and whether the description of BES6100 in the March 2027 annual report upgrades from "sampling" to "design win/mass production." Key constraints: share-loss concerns implied by the stock's decline outpacing the industry; the audio/shooting sub-segment of China's AI glasses market already down 35.4% YoY in 26Q2; the BES6100 timeline having been delayed roughly two quarters versus original guidance; and the share reduction window of financial investor Run Yuan remaining open until 2026-12-18.
What changes over the next 6–18 months: ① Revenue YoY declines narrow quarter by quarter: −32.7% (Q1) → −30.3% (H1) → −20.0% (Q3), with Q2/Q3 quarter-on-quarter growth of +1.95%/+16.66%; the Q4 2025 base of only RMB 591 million sets a low bar for a Q4 YoY turnaround — but its signal value depends on whether the absolute level exceeds RMB 800 million (roughly +35% YoY); ② BES6100 (6nm, hybrid architecture, integrated multi-core ISP/NPU) taped out in 2026Q3 and entered customer sampling; the company targets mass production in 2027, and the leap from "low-power co-processor slot" to "controller slot" is the variable with the greatest valuation elasticity (per-device SoC value of roughly USD 3 → USD 15–20, based on BOM estimates); ③ The AI glasses industry is growing rapidly in aggregate (IDC: global smart glasses shipments projected at 23.687 million units in 2026) but bifurcating structurally — Meta dominates globally, while China's market turned down 8% in 26Q2 (with audio and shooting models down 35.4%, precisely the segment where Bestechnic's solutions sit); ④ On the shareholding front, the remaining reduction windows of Run Yuan I/II run until 2026-12-18, while the share buyback (cap of RMB 100 million, price cap of RMB 150, for incentive purposes) had only executed RMB 22.51 million as of end-September — limited hedging power.
Primary stock price drivers: The slope of quarterly revenue recovery (Q4 turnaround → RMB 3.3–3.6 billion revenue in 2027 rebuilding confidence) is the first anchor; second is the disclosure pace of BES6100 customer design wins/mass production.
Expectation gaps (verifiable): Sell-side consensus for 2026E net profit attributable to shareholders is roughly RMB 530 million (implying Q4 single-quarter profit of RMB 314 million, +240% YoY — clearly not yet revised down enough relative to 9M26 actuals; consensus has been cut −40.6% over the past 90 days) and 2027E of RMB 796 million; this report's base case is 2026E RMB 300–340 million and 2027E RMB 500–600 million, 25–40% below the Street. But the key point: the current price does not require the Street's estimates to be met in order to be self-consistent — this is the core reason the stock is "fairly valued" rather than "undervalued," and excess returns must come from the 20%-probability bull scenario materializing.
Verification catalysts and falsification conditions: Concentrated in the Chapter VI tracking schedule (Q4 preliminary results, IDC quarterly data, BES6100 language in the annual report, share reduction outcome announcements, 2027 national subsidy policy, inflection point in memory contract prices).
Key evidence:
The mechanism of gross margin rising against the cycle is "mix," not "price increases": new categories such as watches and glasses carry higher per-unit value and integration, with stronger certification barriers. This report's caveats: ① the inventory write-down provision ratio fell steadily from 12.67% at end-2023 to 8.98% at end-2025; reverting to the historical midpoint would imply roughly RMB 40 million of additional provisions (about one-fifth of 9M26 net profit), so the "record high" margin partly reflects a thinner balance-sheet buffer; ② peer Actions Semiconductor's gross margin of ~47.5% exceeds Bestechnic's, but its main battleground is Bluetooth speakers/wireless microphones, so the comparison is not directly apples-to-apples. The pricing power conclusion should read as "significantly above white-label, on par with branded platform-type peers."
Key evidence:
This thesis must confront a contradiction: the company's 26.8% decline over the first three quarters is deeper than industry benchmarks (global TWS −0.7%, China −9.5%), and the company's 2025Q4 inflection point predates the steepest phase of memory price hikes — "pure exogenous shock" cannot fully explain it. After the share-gain year of 2024 (+49.9% for the company vs. roughly +3% for the industry), customer destocking and normalization are also drags, and third-party data are lacking to rule out share losses. Quantified recovery path: Q4E revenue of RMB 640–750 million (YoY +8% to +27%), full-year RMB 2.79–2.90 billion; 2027 revenue of RMB 3.3–3.6 billion with net profit of RMB 500–600 million (implying net margin of 13.3%–18.2%, requiring gross margin to hold near 42% and R&D expense ratio to dilute with revenue). Memory cost is a nested variable: NOR's share of cost is not high (per the company), but TrendForce expects 4Q26 contract prices to still rise +10%–20%; the recovery slope remains uncertain until a cost inflection is confirmed.
Key evidence:
The industry-level inflection is solid (2024 +210% → 2025 roughly +350%, with AI glasses already 88% of smart glasses shipments), but two layers of evidence weaken the certainty of revenue realization: first, the growth is concentrated overseas (Meta-led, controllers locked to Qualcomm) and in AR display models (+123%, outside Bestechnic's home turf), while the domestic-camp audio/shooting sub-segment where Bestechnic sits has already turned negative in China; second, overall smart glasses return rates run 30–70% (per Runto Technology) and vivo halted its project, raising doubts about demand quality. Our estimate of Bestechnic's 2026 glasses-related revenue is RMB 150–280 million (5–10% of revenue; not separately disclosed by the company), and teardowns of Rokid's flagship models also show no Bestechnic chips (the company's disclosure of "onboarding" may refer to future models). Conclusion: the inflection is real, positioning is partially confirmed, and revenue realization remains an estimate.
Key evidence:
Three caveats to this optionality: ① the timeline has already slipped once — in Q4 2025 multiple sell-side reports relayed company guidance as "sampling in H1 2026," but tape-out was actually completed in 2026Q3, roughly two quarters late, so confidence in "2027 mass production" warrants a discount; ② the per-unit value jump of USD 3 → 15–20 is a BOM-based estimate (reference: the Oakley Meta HSTN controller at roughly USD 50, ~30% of BOM), not a disclosed figure; ③ moving to controller means head-on competition with Qualcomm's AR1 Gen2 (4nm), and the co-processor slot itself faces substitution risk from Qualcomm's integration push, NXP, and customers' in-house designs. It is the core pillar of the ~61% growth optionality embedded in the current price — and also its most fragile link.
Key evidence:
Pricing self-consistency check: the current price of RMB 101.53 requires only "recovery to 2025 non-GAAP net profit of RMB 519 million in 2027 + 40x" to support roughly RMB 88, and +50x to support roughly RMB 109 — the current price sits within this band, i.e., the market has already paid for the baseline recovery, leaving the AI glasses controller transition as an approximately free option. Using a more conservative 30x on recovered earnings implies fair value of roughly RMB 66 (−35%). The asymmetry of the distribution is the crux of this judgment: probability-weighted fair value is roughly RMB 97.9 (−3.6% vs. the current price), but the left tail extends to −59% at the bear-case midpoint and the right tail to +89% at the bull-case midpoint — "fairly valued" is not the same as "safe." The rating is Neutral rather than Bearish because the quality and the optionality genuinely exist; we are also not Bullish because the current price has already prepaid the base scenario.
| Metric | FY2023 | FY2024 | FY2025 | 9M 2026 (latest period) |
|---|---|---|---|---|
| Operating revenue | 21.76 | 32.63 (+49.9%) | 35.25 (+8.0%) | 21.47 (−26.8%) |
| Net profit attributable to shareholders | 1.24 | 4.60 (+272.5%) | 5.94 (+29.0%) | 2.16 (−57.0%) |
| Non-GAAP net profit attributable to shareholders | 0.29 | 3.95 (+1279%) | 5.19 (+31.4%) | 1.88 (−59.4%) |
| Blended gross margin | 34.2% | 34.71% | 38.69% | 42.27% (+3.76pp) |
| Net margin | 5.68% | 14.11% | 16.85% | 10.04% (Q3 single-quarter 5.58%) |
| Net operating cash flow | 4.70 | 5.80 | 4.47 (−23.0%) | 3.51 (−7.5%) |
| Free cash flow (OCF−CapEx) | 2.76 | 4.09 | 2.90 | 1.97 |
| R&D expense (100% expensed) | 5.49 (FY2024 basis: 6.17) | 6.17 | 6.91 (19.6% of revenue) | 5.52 (25.7% of revenue; Q3 single-quarter +54.3%) |
Latest balance sheet stock (as of 2026-09-30): cash and equivalents of RMB 3.371 billion + trading financial assets of RMB 1.619 billion = cash-like assets of RMB 4.990 billion; interest-bearing debt limited to lease liabilities of RMB 28 million; net cash of roughly RMB 4.963 billion (≈ RMB 20.95/share); debt-to-asset ratio 6.81%; inventory of RMB 906 million (+3.4% vs. beginning of year); accounts receivable of RMB 457 million (+23.2% vs. beginning of year).
Drivers of metric changes (line items with YoY ≥ ±20%): ① 9M26 revenue −26.8%: the company attributes this to "weak consumer electronics demand and memory chip price increases upstream," compounded by the high base after share gains in 2024–25; ② Net profit −57%: revenue decline against rigid expenses (9M26 R&D still +2.4%, Q3 single quarter +54.3%), with negative operating leverage amplifying the drop; ③ Selling expenses +72.3%: the company explains this as increased product promotion spending; ④ Asset impairment losses of roughly −RMB 21.89 million in 9M26 (+83.3%): increased inventory write-down provisions (the company did not separately disclose details); ⑤ FY2024 revenue +49.9% / net profit +272.5%: smartwatch chip revenue of RMB 1.045 billion, +116%, with shipments exceeding 40 million units — the biggest driver; ⑥ FY2025 operating cash flow −23.0%: the company explains that cash paid out grew faster than cash received (not elaborated); ⑦ Other current assets +795%: RMB 200 million of newly added large-denomination certificates of deposit.
Q3 single-quarter revenue was RMB 796 million (−20.0% YoY, +16.7% QoQ, the second consecutive quarter of sequential improvement); net profit attributable to shareholders was RMB 44.43 million (−77.4%) and non-GAAP net profit RMB 37.29 million (−79.2%); gross margin 42.27% (+3.76pp YoY, stable QoQ). The large scissors gap between profit and revenue stems from rigid expenses: Q3 R&D of RMB 223 million (+54.3%, the post-BES6100-tape-out investment peak) + a surge in selling expenses + increased inventory impairment provisions, leaving net margin at only 5.58%. Against sell-side consensus: Q3 net profit fell significantly below the path implied by the Street's 2026E (roughly RMB 530 million) — the Street's model implies Q4 single-quarter profit of RMB 314 million (+240% YoY), which this report deems unattainable; consensus still faces downward revision pressure ahead of the January 2027 preliminary results window. The market reaction was equally split: on 10-20, an edge-AI theme catalyst drove a 20cm limit-up close (RMB 106.32), with single-day turnover surging to RMB 1.3–1.6 billion around the Q3 disclosure, and on 10-30 a single-day drop of −9.3% to RMB 101.53 — the gap between theme fervor and earnings reality is converging. Whether results "beat expectations" must be judged against consensus: Q3 came in below the already-downgraded consensus trajectory, but the stock has partially digested this (down roughly 4.5% from the 10-20 high).
Business model snapshot: Fabless, asset-light SoC design — foundry (TSMC 12nm/6nm) and packaging/testing fully outsourced; R&D staff make up 86.7% of employees; revenue is entirely one-off chip sales (no subscription/recurring revenue); gross margin is the thickest link in the value chain (design 42%+ vs. white-label peers 22–30%). Weak bargaining power versus upstream (TSMC, memory suppliers) and passive exposure to price hikes; downstream stickiness with brand customers based on process-node leadership and certification barriers. Pricing power signal: gross margin rose four years straight, 34.2% → 34.7% → 38.7% → 42.3% (structural), but revenue is highly sensitive to consumer electronics cycles and memory costs — "pricing power" holds only on structure, not on volume.
Cash content of earnings: OCF/net profit attributable to parent 3.80 (FY2023) → 1.26 (FY2024) → 0.75 (FY2025) → 1.63 (9M26); FCF/net profit 2.23 → 0.89 → 0.49 → 0.91. FY2025 broke below the 0.8 warning line (profit +29% while OCF −23%), mainly due to inventory +151 million and reduced payables; the 9M26 recovery relied on trimming procurement payments rather than improved collections. It must be clearly stated: 2025's book profit had weak cash backing, and in 9M26 revenue fell −26.8% while receivables rose +23.2% and inventory grew further to 906 million — working capital moving against revenue. On recurring earnings: FY2025 non-recurring items of 75 million (wealth-management returns 53.4 million + subsidies) accounted for 12.7% of net profit attributable to parent; the gap between recurring (519 million) and reported profit is manageable. But in 2023 recurring profit was only 29 million (non-recurring 77%) — that year's profit was essentially propped up by wealth-management returns and subsidies, so all growth metrics based on 2023 are distorted. No aggressive SBC add-back Non-GAAP treatment (100% expensing of R&D is a plus).
Return on capital: Headline ROIC ~1.7% (2023) → 6.0% (2024) → 7.1% (2025), diluted by roughly 5 billion of net cash; excluding net cash, operating-asset ROIC is roughly 19–20% (NOPAT 506 million / operating assets ~2.6 billion). True operating returns far exceed WACC (~10%); the 2024–25 smartwatch volume ramp reflects genuine improvement in capital returns, not "pseudo-growth."
CapEx quality: CapEx/D&A ~1.5 (FY2025, 156 million/104 million), mainly the R&D center under construction (budget 400 million, 53.5% invested); the company does not disclose a maintenance/growth split — the level of maintenance spending cannot be determined. Under the Fabless model, capacity elasticity is borne by the foundry, so capital-black-hole risk is low.
Red flags & concentration: Top five customers account for 78.40% of revenue (FY2024: 74.22%, concentration rising) and top five suppliers account for 80.26% of purchases (wafers are 65% of cost) — dual concentration of customers and supply is the biggest structural constraint of this business. Effective income tax rate was only 0.21% in FY2025 (15% high-tech rate + extra deductions), so earnings are sensitive to tax incentives; wealth-management returns account for ~9% of net profit.
Words-vs-actions consistency: partially delivered, technically oriented, occasional slippage on new-product timing. Positives: BES2800 (6nm) reached mass production on schedule and fully ramped (FY2024 annual report promise → FY2025 annual report confirming deployment in TWS/smartwatches/glasses/microphones); the 2024 smartwatch chip growth commitment of +116% was delivered; BES6100 taped out and sent samples in 2026Q3 per the company's latest guidance. Negatives: BES6100 sample delivery slipped roughly two quarters versus the company guidance relayed by sell-side in Q4 2025 (H1 2026); operating plans over the years are all qualitative with no quantitative revenue guidance and no falsifiable quantitative commitments — this is both prudence and avoidance.
Shareholder friendliness: neutral to friendly. Cumulative cash dividends over the past three years total RMB 490 million, 124.7% of the three-year average net profit; FY2025 payout ratio 39.76% (RMB 14 per 10 shares + 4 bonus shares). A buyback of RMB 50–100 million (cap at RMB 150/share) for employee stock ownership/equity incentives (not cancellation), proposed by the Chairman about a month after public investor questioning of the share price — the signal matters more than the actual hedge (the cap covers only ~28% of this round's sell-down market value; RMB 22.51 million executed as of end-September). Two consecutive years of 10-for-4 bonus share issues plus rolling vesting across three equity incentive plans have raised total shares outstanding from 120 million to 236.9 million (+97%) — ongoing dilution.
Risk signals: Financial investor Run Yuan I/II has established a continuous sell-down pattern (1.5% Feb–Apr 2026, another ≤1.5% Sep–Dec); Bifu, a related party acting in concert with the controller, also sold ~1.12% before April 2026; on 2026-10-23 the company disclosed a related-party transaction co-investing in a VC fund with a relative of the controller (RMB 50 million for fund units — small amount, compliant procedures); a positive signal: a core technician bought 20,600 shares in February 2026 at RMB 166.31/share.
| Product | Revenue share | Gross margin | YoY | One-line business logic |
|---|---|---|---|---|
| Standard Bluetooth (basic TWS/audio SoC) | 12.2% (RMB 429 million) | 32.14% | −17.98% | Mature category, volume and price both falling (units −14%), heart of the white-label price war |
| Smart Bluetooth (smart earphone/speaker SoC) | 41.0% (RMB 1.444 billion) | 35.37% | −3.56% | The base business, gross margin +3.2pp recovery, main battlefield of BES2800 |
| Others (watch/band, smart hardware, Type-C, incl. AI glasses and wireless mics) | 46.9% (RMB 1.653 billion) | 43.28% | +32.82% | High-margin growth engine, main force of mix upgrade |
By application: Bluetooth audio 53% / watch & band 35% / smart hardware & others 12% (2024: 62%/32%/6%). The profit engine is "Others": gross profit contribution ~RMB 715 million, 52.4% of the RMB 1.364 billion total — a segment with 46.9% of revenue contributes more than half of gross profit; watches + smart hardware, not the "Bluetooth audio" calling card, are the profit base. The spread between the highest- and lowest-margin segments is 11.1pp: watch/glasses SoCs require low-power architecture + display/sensor integration + long certification cycles, giving strong pricing power; standard Bluetooth has been dragged into a price war by white labels. A hidden worry on growth quality: the largest segment, smart Bluetooth, turned negative in 2025 (−3.6%), and the growth baton has been passed entirely to "Others" — if smartwatch penetration slows before glasses ramp, a gap will emerge.
Accounting red flags: ① Inventory impairment provision ratio falling steadily (end-2023 12.67% → end-2024 12.57% → end-2025 8.98% → end-2026H1 7.88%) while inventory balance keeps rising (658 → 724 → 876 → 906 million), turnover days 124 → 148 — provision policy diverging from inventory direction; reverting to the midpoint would require ~RMB 43 million of additional provisions, severity medium (9M26 impairments already +83.3%, direction correcting); ② effective tax rate near zero (FY2025 0.21%); the 15% high-tech status is subject to triennial review and is policy-sensitive, severity low; ③ 2023 non-recurring items were 76.9% of net profit (wealth-management returns + subsidies propping up profit), distorting growth off a low base, severity low; ④ related-party co-investment in a private fund (RMB 50 million), severity low; ⑤ receivables +23.2% against revenue decline — track aging and credit policy, severity low.
Cross-period consistency: ① OCF/net profit 3.80 → 1.26 → 0.75 stepping down for three straight years without systematic explanation ("cash paid operating increase exceeded cash received" is too vague) — inconsistent; ② rising inventory coexisting with 2026 revenue −26.8%; the prior explanation "rising with business scale expansion" contradicts falling volumes and rising inventory — inconsistent; ③ contract liabilities 2.26 million → 20.96 million → 9.42 million → 5.53 million (2023–2026Q3), spiking in 2024 then falling for two straight periods; advance-payment leading indicator weakening, though single-period explanations are self-consistent — consistent but negatively directional; ④ R&D spending first compressed (26H1 −16.5%) then jumped (Q3 +54.3%), matching the BES6100 tape-out timeline; the company's explanations are consistent — consistent. Overall: no evidence of malicious financial engineering, but both "inconsistencies" point to genuine weakening in working capital and demand — inventory drawdown and receivables aging need verification in the annual report.
Current market data & percentiles: Closing price RMB 101.53 (2026-10-30), market cap RMB 24.051 billion. PE(TTM) 78.1x (1-yr 97%/3-yr 48%/5-yr 51% percentile) — denominator collapse; percentiles for reference only. Meaningful anchors: static PE 40.5x (5-yr 13% percentile, cheap vs. own history), PB 3.41x (5-yr 50% percentile), PS(TTM) 8.8x, EV/Sales(FY2026E) 6.7x. Growth basis: forward PE (street 2027E) 30.2x / own 2027E basis 40–48x; PEG 0.61–0.97. Liquidity is ample (average daily turnover ~RMB 785 million, ~3.3% of market cap), but event-day volatility is high.
Peer comparison:
| Company | PE(TTM) | PS(TTM) | 2026 9M revenue growth | Gross margin | Key difference |
|---|---|---|---|---|---|
| Bestechnic (Hengxuan) | 78.1 (distorted) / static 40.5 | 8.8 | −26.8% | 42.3% | Branded wearables main-SoC leader, 6nm |
| Amlogic | 24.2 / static 40.8 | 4.0 | +40.7% (net profit +85.4%) | — | Proven platform SoC, profit base 6x larger |
| Actions Technology | 29.6 | 6.1 | +30.6% | 47.5% | Speaker/mic niches, 1/4 the scale |
| Bluetrum | 9.1 (tainted by fair-value gains; ~55 on recurring basis) | 5.6 | +11.2% | 22.1% | King of white-label, one price tier and one process node lower |
Market-implied expectations: Current RMB 24.051 billion market cap = net cash ~RMB 4.96 billion + zero-growth business value ~RMB 4.4 billion + growth option ~RMB 14.65 billion. Back-solving three questions: ① Recovery basis — "2027 net profit back to 2025's recurring 519 million × 40x" supports ~RMB 88, upper band 109 at 50x; current price sits within the band: the current price demands "recovery," not "re-acceleration"; ② Perpetuity basis — implied perpetual growth of normalized earnings ~7.7%, on the full side; ③ Glasses increment — 2026 glasses revenue only 5–10%; the current price does not depend on 2026 glasses volume, but the option layer requires all three of: 2028 global AI glasses 35–45 million units + Bestechnic share ~40% + per-unit value rising to USD 8–20 (BES6100 as main SoC). In one line: the current price requires the company to "return to 2025 levels + open the ceiling via AI glasses main-SoC adoption on schedule," while reality is revenue still −26.8% and main-SoC adoption still in sampling.
Three-layer value (EPV): Asset value (net-cash floor) ~RMB 20.95/share; zero-growth EPV ~RMB 39.55/share (FY2025 recurring profit of 519 million normalized at a sustainable 15% tax rate to 441 million ≈ RMB 1.86/share sustainable cash flow; Fabless maintenance capex negligible; discount rate 10%; sensitivity range RMB 37.9–41.6); 61.0% of the current price (~RMB 61.98/share) is the growth option — this name has hit a structural demand inflection, so EPV is a floor, not a verdict; whether to pay for growth is decided by scenario math: the 20% bull scenario is precisely the path through which this premium is realized, while the 30% bear scenario would compress price back between EPV and asset value.
Three scenarios & odds (12–18 months):
| Scenario | Probability | Fair range | Midpoint | vs. current | Key driver |
|---|---|---|---|---|---|
| Bear | 30% | RMB 32–51 | 41.5 | −59.1% | AI glasses falsified (China audio-camera segment −35.4% continues) + BES6100 delayed to 2028 + memory price hikes throughout → 2027E net profit only 250–300 million × 30–40x |
| Base | 50% | RMB 74–114 | 94 | −7.4% (range −27%~+12%) | Cost easing + base effect digested → 2027E revenue 3.3–3.6 billion, net profit 500–600 million × 35–45x (anchors: Amlogic static 40.8x, Actions TTM 29.6x, SW comps 44x, own static percentile 13% — all independent of current price) |
| Bull | 20% | RMB 152–232 | 192 | +89.1% | BES6100 main-SoC success: 2028 global AI glasses 40–45 million units × Bestechnic blended share 40–45% × USD 10–20 per unit → glasses revenue 1.2–2.4 billion, 2028E net profit 800 million–1 billion × 45–55x |
Probability-weighted fair value ~RMB 97.85 (bear 41.5×30% + base 94×50% + bull 192×20%), −3.6% vs. current price: odds roughly symmetric, slightly rich; current price sits just above the median of the distribution (~55–60% percentile). Ramp math and scenario self-consistency: the bull chain "units (IDC/Wellsenn: 2028 global 40–45 million) × share 40–45% (inferred from existing product lineup) × USD 10–20 per unit (BOM inference)" yields glasses revenue of RMB 1.2–2.4 billion, consistent with 2028E net profit of 800 million–1 billion and a 45–55x exit multiple (anchor: lower end of Bestechnic's own 2024–25 range of 40–78x + high-growth SoC ramp comparables) → RMB 152–232; the two methods agree. The base chain: 2027E net profit 500–600 million × 35–45x = RMB 74–114, fully consistent with this report's target range.
Own forecasts vs. third-party consensus (reconciliation anchor):
| Basis | FY2026E | FY2027E |
|---|---|---|
| This report | Revenue 2.79–2.90 billion; net profit 300–340 million (recurring 240–270 million) | Revenue 3.3–3.6 billion; net profit 500–600 million |
| Sell-side consensus (n=10) | Net profit ~530 million (implies Q4 standalone of 314 million, +240%, unachievable) | Net profit 796 million, median target price RMB 140.9 |
| Management | No quantitative guidance; framed as "steady QoQ growth in peak season," BES6100 "mass production next year" | — |
This report is 25–40% below street, with the entire gap in 2027 glasses profit realization (this report: 50–100 million; street-implied: 200 million+): main-SoC adoption requires complete-device design cycles; displacing Qualcomm in year one is too optimistic.
Conclusion: Undervalued / Fair / Overvalued — Fair. Assess quality and price separately: quality is above average (42% gross margin, RMB 5 billion net cash, true operating returns ~20%, no malicious financial engineering), but the current price has already prepaid the base recovery scenario, with margin of safety −27.1% (base-case fair floor RMB 74 vs. current price). Rationale for "fair" rather than "overvalued": static percentile 13%, and the growth option has a verifiable realization path and timeline. Rationale for not judging "undervalued": probability-weighted fair value is below the current price, and the left tail of −59% cannot be ruled out by existing evidence.
Industry definition and market size: BES's true competitive segment is "smart audio/video / low-power wearable main control SoC" — single-chip main control/coprocessor solutions for battery-powered terminals such as TWS earphones, smartwatches/bands, AI smart glasses, and wireless microphones. It excludes phone SoCs and cloud compute chips, and only overlaps with white-label Bluetooth audio SoCs at the low end. Size by layer: ① TWS earphone SoC: global TWS shipments up ~3% in 2025 (Counterpoint), entering a mature plateau; the five domestic audio SoC vendors shipped a combined 6.253 billion units in 2023–2025; the global TWS main-control SoC market is roughly RMB 5–8 billion (2025, estimated in this report from BES's RMB 1.873 billion Bluetooth audio revenue and ASP of ~RMB 10/unit, to be verified); ② Smartwatch/band main-control SoC: global smartwatch shipments of 120 million units in 9M 2025 (+7.3%, IDC); China wrist-worn devices 73.9 million units for the full year (+20.8%); BES's watch/band chip revenue in 2025 was RMB 1.234 billion; ③ AI glasses SoC: global smart glasses shipments of 14.52 million units in 2025 (+42.5%), 2026E 23.69 million units (IDC); main-control SoC accounts for ~30% of total BOM (Oakley Meta HSTN teardown, BOM USD 165.47); based on 2026E tiered solutions at USD 10–50/unit, the SoC market is estimated at ~USD 150–750 million for 2026E, expanding to USD 0.9–4.5 billion by 2030E (Wellsenn XR: 90 million pairs; estimate, to be verified); macro view: Frost & Sullivan forecasts global AI edge hardware at RMB 321.9 billion in 2025 → RMB 1.22 trillion by 2029 (~40% CAGR). Growth engines are clearly tiered: TWS low single digits, watches mid single digits, AI glasses doubling-level growth in 2024–2026 (2024 +210% → 2025 ~+350% → 2026E +63%, Counterpoint/Wellsenn/IDC).
Quantified chain of the structural demand inflection (AI glasses): driver trajectory (shipments 1.55 million in 2024 → 7 million in 2025 → 15–23.69 million in 2026E → 40 million+ in 2029E) × jump in per-unit content (~RMB 10 SoC per traditional TWS unit vs USD 10–60 dual-chip solutions for AI glasses, a 5–10× uplift per unit) × penetration ramp (AI glasses as % of smart glasses shipments 46% (2024H1) → 88% (2025H2); smart glasses included in China's national subsidy program in 2026 at 15%, capped at RMB 500 per item) × the Company's share (domestic camp model inventory: Xiaomi AI Glasses, Alibaba Quark S1/N1, Li Auto Livis powered by BES2800; Thunderbird and Rokid added in 2026H1; no authoritative third-party share statistics; 2026 glasses-related revenue estimated at RMB 150–280 million). Two-way validation: upside — if BES6100 succeeds in main-control adoption, per-unit value migrates toward tens of dollars, with revenue elasticity approaching the RMB 1 billion level in 2027–2028 (estimate); downside — China smart glasses −8% in 26Q2 (audio & camera categories −35.4%), return rates of 30–70%, Meta's main control locked to Qualcomm — counter-evidence has already emerged in both the domestic segment and demand quality. Conclusion: the inflection is real (aggregate data is solid), but BES's realization elasticity depends on main-control adoption, not on the coprocessor position already in hand.
Value chain and value distribution: IP licensing (ARM et al.) → foundry (TSMC) → memory (2026 AI servers crowding out capacity, DRAM/NAND prices surging, which design houses passively absorb) → packaging & testing (JCET et al.) → Fabless design (the Company's position, the thickest link in the chain) → ODM/complete units → brand customers (Samsung, Xiaomi, OPPO, vivo, Huawei, Google, ByteDance, Alibaba, Xiaotiancai, Li Auto, Thunderbird, Rokid). BES has weak bargaining power upstream (small scale; foundry and memory are both seller's markets), while downstream it gains stickiness via its 6nm process lead and long certification cycles; overseas revenue accounts for 55.6%, with USD settlement creating FX exposure (26H1 FX loss of RMB 67.63 million, ~40% of net profit attributable to shareholders for the period).
Supply-demand and competitive structure: clearly tiered — premium brand main control = three-way race among Qualcomm (AR1 Gen2 already at 4nm) + BES (6nm) + Apple in-house; white label = Jieli (No.1 globally by unit sales, ASP ~RMB 1.2, 2025H1 gross margin −7.6pp to 30.3%) + Bluetrum (22nm, 22% gross margin); other AI glasses main-control candidates include UNISOC W517 and Amlogic. CR3 by unit shipments ~50%+ (Chaodain Intelligence, mixed 2022/2025 basis, unit sales only). Four entry barriers: process-generation gap (6nm vs 22nm, 1–2 generations ahead of domestic peers), multi-technology-stack integration (low-power architecture × BT/Wi-Fi × DSP/NPU/ISP), brand certification cycles (years), and capex for advanced-node tape-outs (BES's annual R&D of RMB 691 million, 19.6% of revenue). Price wars concentrate in the white-label tier; the brand market is mildly tiered — BES's 42% gross margin validates differentiation, but after BES6100 main-control adoption it will clash head-on with Qualcomm. Substitution threats: terminal in-house design (Apple already self-supplies AirPods; Xiaomi/Samsung have histories of in-house phone SoCs), OWS diverting TWS demand, single-chip displacing dual-chip solutions (BES is both a substitution target and a substitution initiator).
Cycle and regulation: lifecycle tiers — TWS mature stage, smartwatches early maturity, AI glasses at the introduction-to-growth inflection (TrendForce calls 2026 the volume inflection year). Business cycle: consumer electronics in a downturn destocking phase in 2026 (memory super-cycle pushing up BOM + fading national subsidies; 26Q3 global phone/PC shipments ~−20%); leading indicators: memory contract price increases (26Q3 narrowed to +13~18%, first time below 20% in over a year), BES's quarterly revenue QoQ (improving for two consecutive quarters), AI glasses new product launches and sales, and terminal brand inventories. Regulation: ① US export controls — BES is not on the Entity List, consumer-grade low-power SoCs are not restricted categories, TSMC 6nm foundry operations are normal, but the restoration of the BIS "50% rule" effective 2026-11-09 and tightening scrutiny of advanced-node foundry constitute a tail risk (this report assigns ~12% probability); ② national subsidy phase-out is the biggest demand headwind for 2026 (the Company's first attribution in its Q3 report), while smart glasses' inclusion in the 2026 national subsidy is a structural tailwind; 2027 policy to be revealed at end-December; ③ the Big Fund Phase III supports edge AI chip R&D.
Peer benchmarking (operational): Bluetrum — 2025 revenue RMB 1.842 billion (+1.2%), gross margin 22.1%, non-GAAP net profit RMB 233 million (−4.5%); the white-label king is penetrating the brand tier with 22nm, one price band and one process generation behind BES; Actions Semi — 9M 2025 revenue RMB 722 million (+54.7%), gross margin 47.5%, R&D ratio 33%, focused on speakers/microphones, AI glasses chips still at early 22nm; Jieli — 2024 revenue RMB 3.120 billion (+6.5%), 2025H1 −4.6%, gross margin 30.3%, industry leader by unit sales and the white-label price setter; Qualcomm — FY2025 revenue USD 44.3 billion (+13.7%), gross margin ~54.6%, incumbent AI glasses main-control supplier (Snapdragon AR1 across Meta's lineup), the benchmark and biggest threat to BES's main-control push (~70× size gap, 3–4nm process gap).
Company positioning in the industry: absolute domestic leader and global Tier-1 in brand wearable main-control SoCs — No.1 among domestic vendors in Bluetooth audio SoC revenue (RMB 1.873 billion in 2025, ASP ~RMB 10/unit, ~10× the white-label level), No.4 globally in TWS main-control shipments (~10% share, Chaodain Intelligence 2025 unit-sales basis), first 6nm solution in Samsung Galaxy Buds3 Pro continuously displacing Qualcomm, watch chip shipments exceeding 40 million units in 2024 with share rising rapidly, and AI glasses positioned at the coprocessor slot in mainstream domestic models. Moat sources: process-generation gap + multi-technology-stack integration + brand certification cycles, with 42%+ gross margin validating pricing power. The essence of its positioning is "niche leader-turned-main-control challenger": TWS has proven the ability to displace Qualcomm, and the AI glasses main-control slot is the decisive battle over whether it can leap from "earphone chip vendor" to "smart terminal platform chip vendor." Share-basis note: unit-shipment share (Chaodain Intelligence) and revenue share (BES No.1 domestically) are different bases; the former includes the massive low-ASP white-label volumes, and the two are not interchangeable.
Overall rating: Neutral, confidence 0.55, time horizon 12–18 months. BES is an above-average-quality platform SoC prospect: structurally favorable gross margin environment, net cash of RMB 5 billion, real operating returns ~20%, and BES6100 opening up the main-control imagination; but the current price of RMB 101.53 has already prepaid the base recovery scenario (probability-weighted fair value RMB 97.85, −3.6% vs current price, safety margin −27.1%), and left-tail risks are not ruled out — China AI glasses segment turning negative, BES6100's prior delay, and Run Yuan's ongoing sell-down not yet exhausted. Strategy implications: holders can stay invested and adjust based on verification points; non-holders should wait for two types of opportunities — a pullback near the lower bound of the base range (~RMB 74–85) to provide safety margin, or targeted disclosures on BES6100 that raise the bull-case probability before stepping up; chasing the theme at elevated heat is not advisable.
Key risks (ranked by impact): ① collapse of the valuation anchor if the AI glasses narrative is falsified (PS 5–6× or normalized earnings 25–30× implying ~RMB 48.6–75.2, −26%~−52% vs current price); ② BES6100 missing the 2027 main-control adoption window (Qualcomm integration / terminal in-house design); ③ memory price increases persisting through 2027H1 and a second demand trough (double hit to gross margin and recovery slope); ④ single-customer attrition under 78.4% top-five customer concentration; ⑤ Run Yuan's consecutive sell-down pattern (watch for a new plan after December 18); ⑥ working capital divergence of receivables +23.2% / inventory of RMB 906 million and inventory top-ups; ⑦ geopolitical tail from advanced-node foundry scrutiny (restoration of BIS rules effective 2026-11-09 as the watch point).
Monitoring schedule (next 6–12 months): early December 2026 — IDC 26Q3 global/China smart glasses data (global ≥+40% and China stabilizing as confirmation signals); 2026-12-18 — Run Yuan sell-down window expiry and results announcement; late December 2026 — whether the 2027 national subsidy retains smart glasses; January 2027 — earnings pre-announcement window to validate Q4 revenue (≥RMB 800 million confirms the inflection; <RMB 800 million delays the recovery narrative); early January 2027 — TrendForce 4Q26 memory contract prices (increase at zero or negative as the cost inflection point); late March 2027 — 2026 annual report (upgraded BES6100 language + revenue mix by product + inventory/receivables validation); 2027H1 — teardown of the first BES6100 launch device for confirmation. Price triggers: a close below RMB 74 (lower bound of base fair value) triggers re-examination of the entry thesis; reaching RMB 114 (upper bound of base) implies the base scenario materializing — assess realization/exit.
Data and sources: financial data from BES Technology's 2023–2025 annual reports and 2026 Q1/interim/Q3 reports (cninfo.com.cn); industry data from IDC, Counterpoint, Omdia, TrendForce, Wellsenn XR, and Chaodain Intelligence (as relayed by Cailian Press, Yicai, Gelonghui, and other media); sell-side consensus is the aggregate median of 10 reports over the past 6 months (as of 2026-10-30); valuation multiples and percentiles are computed in this report using the 2026-10-30 closing price and the latest financials; the three scenarios, EPV, and probability-weighted fair value are this report's own calculations. This report does not constitute investment advice.