Zhejiang Hechuan

Phase A — Understand the business

Company Overview

Zhejiang Hechuan Technology is a Chinese industrial-automation "motion-control" component maker — think servo drives, servo motors, PLCs, HMIs, encoders and low-voltage inverters — that has, since 2024, bolted a humanoid-robot component story onto the top of that base business.

• Founded Nov 22, 2011 in Quzhou/Longyou, Zhejiang (Longyou Industrial Park); IPO'd on the STAR Market 28 Apr 2022, the first STAR-listed company out of Zhejiang's 26 mountainous counties.
• Core product is the servo system (drive + motor + encoder), sold into OEM machine-builders. Adjacent lines: PLC/PAC/HMI/IO controllers, VFD inverters, vision systems, linear modules, and PV/energy-storage inverters.
• Go-to-market = price penetration. Hechuan built share as the cheap domestic-substitution challenger, undercutting Yaskawa/Panasonic/Delta and chasing Inovance. Its own IPO materials claimed "#2 domestic servo brand, ~3.4% share (2022)" — a claim that needs heavy caveat (see Lens 3): it is a distant second within the domestic-brand cohort, not #2 of the whole market.
• End-markets (the tell for everything that follows): photovoltaics, lithium-battery equipment, 3C electronics, laser, machine tools, textiles, packaging, logistics — heavily weighted to PV + lithium capex, both of which entered brutal overcapacity busts in 2024.
• Contract structure: short-cycle OEM component sales (not take-or-pay, not recurring SaaS). Revenue is project/capex-driven and cyclical, with stretched receivables to capital-goods customers — the opposite of a sticky recurring model (see Lens 10).

What it actually is: a small ($0.9B-cap) sub-scale servo/motion-control challenger whose legacy engine is a commoditizing, price-war-hit automation-component business, and whose equity narrative is an unproven bet that its motor/encoder/actuator know-how converts into a humanoid-robot supply-chain seat.

Supply Chain

Map: rare-earth & silicon → Hechuan (vertically integrated) → capex-goods OEMs / (aspirationally) humanoid OEMs.

Upstream inputs (named where sourced):

• Power semiconductors / MCUs / DSPs — IGBT & control chips for drives. Hechuan is unusually vertically integrated here: it designs its own control chips via subsidiary 杭州禾芯 (Hangzhou Hexin) and does chip→PCBA→assembly in-house. It still buys foundry capacity and power devices externally — a sanctions/import chokepoint for higher-end nodes.
• NdFeB rare-earth magnets — the raw input for servo & hollow-cup/frameless torque motors. Specific magnet suppliers n/a, but this ties Hechuan's motor cost to the Chinese rare-earth complex.
• Encoders — self-designed and self-made (magnetic encoders were founder Wang's original technical contribution — see Lens 9). This is a genuine vertical-integration edge vs assemblers.
• Precision mechanics — die-casting → machining in-house; planetary roller screws (行星滚柱丝杠) increasingly self-made for humanoid linear actuators.

The company: three closed-loop chains — (1) electronics: chip design→PCBA→test; (2) precision: die-cast→machining; (3) motor: encoder→stator/rotor→motor assembly. This vertical integration is the single most defensible thing about the business.

Downstream customers (named): Foxconn/FII (工业富联), CATL (宁德时代), SF Express (顺丰), LONGi (隆基), S.C./Jiejia (捷佳伟创, PV equipment), Wuxi Lead/Advanced (先导智能, lithium+PV equipment), Lens Technology (蓝思科技), and robot-maker EFORT (埃夫特). Prospective humanoid channel: Huawei's Embodied-Intelligence ecosystem (MOU Nov 2024) and rumored (unconfirmed) links to Tesla Optimus / Zhiyuan(AgiBot) / Unitree.

Chokepoints / single-source risks:

1. Demand concentration in PV + lithium capex — when those two capex cycles rolled over in 2024, Hechuan's revenue fell 27% and receivables ballooned. This is the dominant risk in the whole chain.
2. Higher-end power/control silicon still partly external → export-control exposure.
3. The humanoid downstream is aspirational — orders exist but "规模不大" (small scale) as of Q1 2026. No confirmed design-win at a scaled OEM. Names or it didn't happen — and for humanoid, it hasn't happened at scale yet.

Competitive Advantages (moats)

Honest verdict: a narrow, shallow moat in a commoditizing category, plus one genuine capability (vertical integration) that the humanoid story tries to lever.

• Brand/perceived value: Hechuan is the cheap challenger, not the quality benchmark. Domestic servo brands already sit 20–30% below Yaskawa/Siemens on price; Hechuan discounts within the domestic tier to take share. That is the opposite of pricing power — and it is exactly what detonated gross margin in 2024–25.
• Scale: Sub-scale. 2024 revenue ¥811M vs Inovance ¥46.3B. In servo, Inovance holds ~28% share; Hechuan ~3–4%. The "servo duo (伺服双雄) 汇川 vs 禾川" framing flatters Hechuan — a 2023-H1 industry piece titled it "still vastly disparate (实力依旧悬殊)". Being the distant #2 domestic brand is a marketing line, not a moat.
• Switching costs: Modest. Servo drives are designed into a machine, so there is some stickiness per-OEM, but multi-sourcing is normal and the price-war dynamic shows customers switch on cost.
• Real edge = vertical integration. Self-designed chips (Hangzhou Hexin), self-made encoders, self-made motors and (now) roller screws. This lets Hechuan (a) protect BOM cost and (b) plausibly claim a humanoid-component seat because a hollow-cup motor + frameless torque motor + encoder + roller-screw actuator is exactly the joint content a humanoid needs, and Hechuan makes all four in-house. 90% of its humanoid core parts are self-produced. This is the one durable, non-obvious asset.
• Bargaining power: Weak both ways — it needs its PV/lithium customers more than they need it (hence the 200-day receivables), and it competes for the same rare-earth/silicon inputs as far larger buyers.

Moat rating: LOW on the legacy servo business; UNPROVEN OPTION on the humanoid actuator stack. The bull case is entirely that (5) becomes a moat if humanoids scale and Hechuan's full-stack in-house content wins sockets. Today it is capability, not moat.

Segments

No our figures on the shelf — segment splits are ``, unaudited.

• 9M-2025 mix: Industrial-control products ~92–93% of revenue; machine tools ~6% (¥32.3M); the residual is other/ODM. (One scrape gave "industrial control ¥471M = 92.54%" against a ¥733M base — internally inconsistent, so treat the share as directional, not the absolute.)
• By geography: predominantly domestic China; overseas is small and growing off a low base. Precise split n/a.
• Trend & cause: the H1-2025 rebound (+5.3%) and 9M (+13.8%) were driven by lithium-battery, laser, packaging and ODM orders recovering, offsetting still-soft photovoltaics — management's own attribution. So the segment story is a rotation away from PV (the 2024 wrecking-ball) toward a broader OEM base, with the humanoid line a rounding error in reported revenue but the entire driver of the multiple. The direction is decelerating losses / re-accelerating top-line, cause = PV base effect lapping + product-mix broadening.

---

Phase B — Measure performance

Earnings Result (latest print + trajectory)

The tape in one table (RMB, attributable net unless noted):

Read: the arc is peak (2023) → collapse (2024, PV bust) → topline recovery but still loss-making (2025) → near-breakeven inflection (Q1-2026).

• Vs "consensus": no clean sell-side consensus sourced; the relevant comparison is vs its own cohort — Hechuan's Q3-2025 net profit ranked LAST (30th of 30) in the automation-equipment group, and a separate risk screen put it last of 91 on financial health with 8 flagged risks. It is the worst-financials name in its own sector.
• Margin: gross margin ~23% and falling on price competition; net margin negative. The Q1-2026 near-breakeven + positive operating cash flow is the first genuinely encouraging print in two years.
• Balance-sheet flags: total assets ¥1.98B (−8.2% in 2024), equity ¥1.32B (−15.5%). Accounts receivable ~¥553M at H1-2025 — see Lens 10; this is the scariest line.
• Guidance/tone: management frames 2026 as the humanoid "mass-production year (量产元年)"; reports orders +135% YoY and parent-level monthly profitability in early 2026. Tone has flipped from defensive (2024) to offensive (2026). Whether that converts is the whole question.

Earnings Calls / Investor Communications (sentiment trend)

Chinese STAR issuers don't hold US-style calls; the analogue is the 投资者关系活动记录表 (investor-relations activity records) and 业绩说明会 (results briefings). No transcripts on the shelf; drawn from IR-record filings.

• 2024 tone: defensive/explanatory — PV overcapacity, price competition, impairment, "awaiting recovery (静待复苏)".
• 2025 tone: pivot — heavy emphasis on the humanoid full-chain layout ("核心部件+组件+整机本体"), Huawei-ecosystem entry, YOLO(游龙)01 prototype, dexterous hand + hollow-cup motor sampling; the IR narrative increasingly leads with robotics and treats the industrial base as the recovering-cash-cow.
• 2026 tone: confident — "financials bright + track positioning (财报亮眼+赛道卡位)", orders +135%, breakeven-in-sight.
• Phrases they added: "humanoid mass-production year," "core-parts self-sufficiency 90%," "planetary roller screw," "drive-control integration." Phrases they stopped leaning on: the pure servo-market-share ranking claim. Net sentiment trend: sharply more promotional over 24 months — which, given the still-negative earnings, is itself a yellow flag (Lens 13).

Comps

Peer set = Chinese motion-control / servo / robot names. Multiples ``, snapshot ~mid-2026; loss-makers have no meaningful P/E.

The provenance-critical takeaway: Hechuan trades at ~6–7x sales — a premium to the profitable, 28%-share leader Inovance (3.9x) and roughly in line with profitable Leadshine (7.1x) and loss-making-but-far-larger Estun. For a sub-scale, worst-in-cohort-financials, loss-making servo maker, ~6–7x sales is not an automation multiple — it is a humanoid-optionality multiple. You are paying Leadshine's profitable multiple for a company still losing money. A broker (Guotai Junan/国泰海通) carried a "strong buy," ¥90 target in Jan-2025 — ~2x the current price and, on these numbers, aspirational sell-side hype rather than a grounded DCF.

Stock-Price Catalysts (what moves it)

The tape reveals a name that trades on humanoid-robot narrative, not fundamentals:

• 22 Aug 2024 — "another Tesla-Optimus rumored partner! Servo stock Hechuan takes off (起飞)" — the stock jumped on an unconfirmed Optimus-supplier rumor. Pure theme.
• Aug 2024 — set up JV Zhejiang Hechuan Humanoid Robot Co.; completed YOLO01 prototype.
• 15 Nov 2024 — Huawei Embodied-Intelligence Center MOU (one of 16 firms) — sector-wide humanoid rally participant.
• Dec 2024–Jan 2025 — sell-side "forward-positioned in humanoid" notes; ¥90 strong-buy.
• 2026 — "mass-production year" narrative; record market cap (市值创新高) on humanoid concept despite still-negative FY2025 earnings; orders +135%.
• Overhangs that moved it down: the terminated convertible-bond issuance (Sept 2024); lockup-share unlocks (Oct 2025); and insider reduction plans (GM + a VP filed 减持).

Pattern: >5% moves cluster on humanoid headlines and sector beta, not on earnings (earnings have been uniformly poor while the stock re-rated). This is a momentum/theme vehicle — which cuts both ways for risk.

---

Phase C — Judge people & books

Management

• Wang Xiangbin (王项彬) — founder, chairman, actual controller. Born Jul 1974; mechanical-design degree, Zhejiang Univ. of Technology (1997); prior stints at HK Hongtu, Taiwan Acer group, Zhongshan Anke Xun; founded Hechuan 2011 under the "Zhejiang entrepreneurs return" initiative. Genuine technical founder — personally credited with the servo system + magnetic encoder development that "filled domestic gaps." Grew sales ~69x over 7 years, R&D +200% (2016–18).
• Track record: built a real, vertically-integrated component company from zero to a STAR IPO — that is a genuine operating accomplishment. But he has not yet demonstrated he can run it profitably at scale through a down-cycle — 2024–25 is the first real stress test and the P&L failed it.
• Tenure & skin in the game: controls ~19–26% (varies by date/vehicle) directly + via 禾川投资 (Hechuan Investment). High founder alignment. But: GM Xu Xiaojie (徐晓杰) and director/VP Xiang Henghui (项亨会) filed share-reduction (减持) plans — insiders trimming while loss-making is a confidence yellow flag (common on STAR post-lockup, but noted).
• Key execs: Xu Xiaojie (GM, ex-Aokang finance); Xiang Henghui (VP, high-school education, ex-Foshan Gongtai); core tech — Yan Pengfei (servo/PLC), Li Bo (control R&D), Zhang Yu (chip/software via Hangzhou Hexin & Dalian Chuanpu).
• Capital allocation: aggressive reinvestment — new "digital factory" + Hangzhou research institute (whose depreciation now drags earnings), self-chip and humanoid build-out. ROE/ROIC currently negative. Tried and failed to raise a convertible bond (terminated Sept 2024), then leaned on external debt (financial expense +281%) — i.e., funding an aggressive build-out into a downturn with debt after equity-linked funding fell through. That is the riskiest capital-allocation posture in the file.
• Archetype: founder-engineer / promoter, not professional-manager caretaker. Upside: real R&D conviction and speed into humanoid. Downside: promotional cadence has outrun the numbers, and the balance sheet is being pushed.

Forensic Red Flags

Forensic lens — every figure ``, unaudited, no filings on shelf to cross-check. Treat as flags to verify against the primary 年报 PDF, not proven fraud.

1. Accounts receivable is the headline risk. ~¥553M AR at H1-2025 against ~¥1.0B annualized revenue ≈ ~55% of sales, ~200-day DSO. Selling capital goods to cash-stretched PV/lithium customers on extended terms is a textbook setup for credit-impairment charges — which management already flagged as a loss driver. Watch the AR aging + bad-debt provision in every print.
2. Inventory obsolescence. Repeated write-downs of aged electronic raw materials bought in prior years and consumed slowly amid product iteration — an explicit 2024 and H1-2025 loss cause. Signals over-buying into a demand peak, then eating the markdown.
3. Revenue-up-profit-down divergence. 9M-2025 revenue +13.8% but loss widened — "增收不增利." Cash from goods sold << reported revenue historically (the AR gap). Operating cash flow only turned positive in Q1-2026 — before that, earnings quality was poor (profits, when they existed, were not cash).
4. R&D capitalization / depreciation drag. New digital-factory + research-institute capex is now depreciating into the P&L; STAR issuers frequently capitalize development spend — verify the capitalized-vs-expensed R&D split in the 年报 (not sourced here). If a chunk of "R&D" sits on the balance sheet, reported losses understate the cash burnCash burnHow much more cash goes out than comes in, per period. The clock on a company with no profits..
5. Debt-funded build-out after failed CB. Financial expense +281% on new external borrowing + hedging derivatives. Rising interest burden on a loss-making base.
6. Promotional gap. IR narrative pivoted hard to humanoid while earnings stayed negative and orders remain "small scale" — the classic theme-stock tell (Lens 13).

Regulatory findings (required sub-section). Per regulatory/regulatory-findings.md (generated 2026-07-10): Hechuan has no SEC CIK — 0 EDGAR findings (LR + AAER), and no EDGAR search is possible. Non-SEC / China (CSRC, SSE) web search for enforcement, consent decrees, fines or penalties returned no material regulatory or legal actions against the company as of 2026-07-10. The only regulatory event found is the SSE-terminated convertible-bond review (Sept 2024) — a withdrawn financing, not an enforcement action. Item-3-equivalent (litigation) disclosure could not be read (no filing on shelf) — n/a from primary. Net: no enforcement red flags surfaced; the risks here are accounting-quality (receivables, inventory, R&D cap), not known misconduct.

---

Phase D — Project & stress-test

Forward Projection (FY2026 / FY2027 / FY2028)

No our model logged (unattended/watchlist rule). All outputs `` with arithmetic; ~151M shares; RMB.

Base inputs: FY2025 revenue ~¥1.0B, net −¥130M; Q1-2026 already +18.7% and near-breakeven with positive operating cash flow; gross margin recovering off ~23% as PV base-effect laps and mix broadens; humanoid immaterial to revenue through 2026, small from 2027.

• BASE:
• FY2026 rev ¥1.2B (+20%), GM→26%, opex leverage → net ~+¥15M → EPS ~¥0.10.
• FY2027 rev ¥1.5B (+25%, first real humanoid + servo cycle up), net margin ~6% → net ~¥90M → EPS ~¥0.60.
• FY2028 rev ¥1.85B (+23%), net margin ~9% → net ~¥165M → EPS ~¥1.10.
• → At ¥43, that's ~72x FY2027 / ~39x FY2028 base EPS. Priced for the bull path.
• BULL: humanoid actuator sockets convert (Huawei-ecosystem/domestic OEM design-wins), servo up-cycle + share gains; FY2028 rev ~¥2.6B, net margin ~12% → net ~¥310M → EPS ~¥2.05. ~21x FY2028 — the only path that justifies today's multiple.
• BEAR: price war persists, PV/lithium receivables impair, humanoid stays a rounding error; loss-making through 2027, dilutiveDilutionIssuing new shares, so each existing share owns a smaller slice of the same company. raise to fund burn → EPS negative, book value erodes.

The base call as a scoreable line (not logged): "688320 returns to full-year GAAP net profit (attributable) in FY2026," p≈0.55 — genuinely coin-flip, hinging on receivables not blowing up and the Q1 breakeven holding.

Bull vs Bear

Bull case. Hechuan is a rare full-stack, in-house humanoid-actuator supplier — hollow-cup + frameless torque motors, self-made encoders, planetary roller screws, dexterous hands and drive-control integration, 90% self-produced — sitting inside the Huawei embodied-AI ecosystem just as China's humanoid industry hits its "mass-production year" with component orders reportedly booked into 2027. The industrial base is inflecting (Q1-2026: +18.7%, −79% narrower loss, cash-flow positive), the worst of the PV bust is lapping, and the founder is a genuine motor/encoder technologist. If even a modest share of humanoid joint content converts, a ¥1B-revenue company re-rates on a multi-billion TAM — and the vertical integration that crushed margins in the price war becomes the cost moat that wins sockets. Contrarian read the market is pricing.

Bear case (2–3 permanent-impairment risks).

1. The receivables are a live grenade. ~200-day DSO to distressed PV/lithium capex buyers → a wave of bad-debt provisions could wipe multiple years of thin projected profit and force a dilutive raise.
2. The legacy business has no pricing power. Structural servo price deflation vs Inovance (10x its size) and a field of domestic discounters means gross margin may not durably recover — the 2024–25 margin loss could be secular, not cyclical.
3. The humanoid seat is unproven. Orders are "small scale," the marquee links (Tesla/Zhiyuan) are rumors, and Hechuan competes for the same joint content against Inovance, Leadshine, and dedicated motor specialists with more scale. The multiple assumes a win that hasn't been demonstrated.

Pre-mortem (18 months out, thesis broke): Q1-2026 breakeven proved to be a seasonal head-fake; a big PV customer defaulted, forcing a ¥100M+ impairment; humanoid orders stayed sub-¥50M; the company did a discounted equity raise to plug the balance sheet; the humanoid multiple compressed toward Inovance's 4x sales → the stock halved even though "nothing changed operationally."

Are multiples too high? Yes, on any near-term fundamental basis. ~6–7x sales / ~5x book for a loss-maker is a pure option premium on humanoid. The contrarian thing the market is refusing to see: that a cost-leadership servo challenger and a high-margin, moated humanoid-joint supplier are contradictory identities — the same vertical integration is being valued as a cost weapon (to survive the price war) and a margin moat (to justify the multiple) at once. It can be one or the other; the bull case needs it to flip from the first to the second, and that transition is unproven.

Devil's Advocate (short-seller)

Dismantling the bull:

• Revenue concentration: legacy is ~92% industrial control, historically PV/lithium-heavy — the exact end-markets in overcapacity. Shift the mix and you still have a cyclical, price-taking component vendor.
• The moat is weaker than bulls think: "90% self-made humanoid parts" describes capability, not design-wins. Anyone can build a hollow-cup motor to sample; whose humanoid is it shipping in, at what volume, at what margin? Answer today: effectively none, at scale.
• Most dangerous competitor bulls underestimate: Inovance itself — 28% servo share, profitable, ¥46B revenue, and also pushing into robot joints with vastly more R&D budget, plus dedicated motor houses. Hechuan is the sub-scale price-taker in that fight.
• Worst capital-allocation moves: debt-funding an aggressive build-out into a downturn after the convertible bond was pulled; insiders (GM + VP) filing to sell while the company loses money and the stock rides a theme.
• What must hold for today's price: durable GM recovery and a real, scaled humanoid design-win and receivables not impairing — three independent things, each uncertain.
• If growth disappoints 20–30%: with negative-to-thin margins, a revenue miss flips the model back to losses and likely triggers a raise → the valuation has no floor except book (~¥1.3B, i.e., ~¥8–9/share of tangible equity vs ¥43 price).
• Single permanent-impairment scenario, plausibility: a major PV/lithium customer default cascade → large impairment + covenant/liquidity stress + dilutive rescue. Plausibility: moderate-and-rising given the AR profile. This is the short's cleanest angle.

Management Questions (ordered by information value)

1. Of FY2025 revenue, how much came from humanoid-robot parts — in RMB — and what is the signed backlog by named customer for 2026–27? (The entire multiple rests on this one answer.)
2. What is the current accounts-receivable aging, and what bad-debt provision have you taken (and modeled) for PV/lithium customers on extended terms?
3. Is the Q1-2026 return to breakeven seasonal, or do you expect full-year GAAP net profit in FY2026 — and on what gross-margin assumption?
4. Name the humanoid OEMs you have shipped qualified parts to (not MOUs) and the stage (sample / A-sample / production PO).
5. How do you win humanoid-joint sockets against Inovance given its scale and R&D budget — what specifically can't they replicate?
6. After the terminated convertible bond, what is the funding plan for the humanoid build-out, and will you need an equity raise in the next 24 months?
7. What is the capitalized vs expensed R&D split, and what is true cash burn vs reported net loss?
8. Is the ~23% gross-margin trough structural or cyclical — where does servo GM settle once PV normalizes?
9. What share of BOM (esp. power semiconductors) is import-dependent, and what is your export-control exposure?
10. Why are the GM and a VP reducing shares now, and what is the board's view of that signal?
11. What return (revenue/margin) has the digital factory + Hangzhou institute capex generated vs the depreciation it added?
12. What is your planetary-roller-screw yield/cost position vs specialists — is it competitive or a science project?
13. How concentrated is revenue in your top-5 customers, and how has that shifted 2023→2025?
14. What is the rare-earth-magnet cost pass-through mechanism if NdFeB prices spike?
15. Three years out, are you a cost-leading servo challenger or a high-margin humanoid-joint supplier — and how can you be both?

---