Himile: A 40%-Margin Tire-Mould Franchise Is Funding Two Adjacencies That Have Not Yet Earned Their Keep

The correct demand function is new specification and model cadence, not global tire-unit volume. EV penetration helps at the margin because new vehicle platforms bring different loads, noise requirements, rolling-resistance targets and homologations. It does not create an automatic mould super-cycle. Continental has explicitly said that most of its existing tire lines can already be fitted to electric vehicles. The EU's regulatory direction is similarly evolutionary rather than a one-time reset: current rules regulate rolling resistance, wet grip and noise, restrictions on weaker performance classes have tightened, and future requirements extend toward abrasion and mileage. Those rules encourage product redesign and continuing specification turnover, but they do not require every EV or regulatory change to use a wholly new dedicated tire family.

The often-repeated market-share claim survives a cautious test, though not with the precision implied by sell-side shorthand. Himile's own global materials say it works with 66 of the world's top 75 tire companies and has more than 30% “international” tire-mould share. Older public reporting has cited a lower figure around 25%. Because the denominator is not standardized (outsourced moulds, all moulds, particular radial-tire categories, or international sales can produce different percentages), I would describe Himile as possessing roughly a quarter to one-third of the relevant global outsourced market, rather than present 33% as an audited fact.

The evidence nevertheless points toward a real moat. The top five customers represented only 34.29% of 2025 revenue and the largest 12.48%, so this is not a one-customer captive supplier masquerading as a champion. The company sells directly, runs mould production and service subsidiaries across the United States, Europe and multiple Asian and Latin American markets, and reports that nearly 80% of machine tools used in its tire-mould operations are self-made. Scale therefore sits alongside equipment know-how and accumulated process engineering. The weakness is contractual: public disclosures do not show long-term take-or-pay contracts or exclusivity. Orders are customized and recognized when delivered. Customers can re-tender and can dual-source. The commercial switching cost is qualification, engineering integration, response time and execution risk, not a legal lock.

My moat verdict is a genuine process-and-scale moat with weak contractual lock-in. The strongest evidence is the combination of global customer breadth and nearly 40% mould gross margin, rather than the headline market-share percentage by itself.

The second growth engine deserves a very different treatment. Himile's large-parts business sells to customers including GE, Mitsubishi, Siemens, Dongfang Electric, Shanghai Electric, CRRC and Harbin Electric, and its casting-and-machining capacity has been expanded to more than 300,000 tonnes of designed capacity, including a 65,000-tonne expansion that entered trial operation in the second half of 2025. This intersects with an exceptionally strong gas-turbine cycle: GE Vernova's current turbine pipeline is sold out well into the end of the decade, while Siemens Energy reported record order intake and a record company backlog in 2026.

The temptation is to capitalize that hot cycle directly into Himile's valuation. I do not. “Large parts” also contains products for other end markets, and Himile does not disclose gas-turbine revenue, gas-turbine gross margin, individual turbine components by customer, or a gas-turbine order backlog. The 2025 margin decline inside the entire large-parts category shows that converting demand into returns is not automatic. I assign no standalone gas-turbine value beyond what is already captured in reported large-parts revenue and my group growth scenarios. This is intentionally more conservative than sell-side sum-of-the-parts approaches.

The third engine, CNC, is more interesting and more dangerous. Revenue advanced from just RMB 88m in 2021 to RMB 146m in 2022, RMB 308m in 2023, RMB 399m in 2024 and approximately RMB 968m in 2025. Himile now markets five-axis vertical and horizontal machining centres, mill-turn systems, laser machine tools and direct-drive rotary tables. July 2026 investor communication indicated that internally produced components include structural parts such as machine beds and cradle-type rotary tables. What public disclosure still does not establish is equally important: it does not name the suppliers or localization ratios for the numerical controller, high-precision spindle, linear guides and bearings. Nor does the annual report disclose a meaningful list of external CNC customers or physical shipped units.

I do not underwrite full-stack CNC import substitution. I underwrite only the disclosed RMB 968m of 2025 sales and Himile's proven ability to make structural and selected functional components. The option becomes more valuable if future filings disclose external customer repeat orders, component localization and stable gross margins.

The balance sheet is an unusually strong counterweight to those uncertainties. At December 2025, equity attributable to shareholders was about RMB 12,027m, cash RMB 1,483m and trading financial assets RMB 450m, against short-term borrowings of roughly RMB 210m. Weighted ROE was 21.80%. A simple 2025 DuPont reconstruction produces approximately 21.6% net margin × 0.84 times asset turnover × 1.21 times financial leverage, almost exactly reproducing reported ROE. The high return therefore comes from margin and asset productivity, not debt.

Cash conversion is the uncomfortable counterpoint. Operating cash flow was RMB 1,034m in 2025 versus RMB 2,393m of attributable profit, only about 43%. Across 2021–2025, cumulative operating cash flow was roughly RMB 4,452m against cumulative net income of about RMB 8,270m, a 54% conversion ratio. In 2025, the reconciliation shows operating receivables consumed approximately RMB 1,704m and inventories another RMB 223m, while operating payables supplied only RMB 120m. This is principally a working-capital problem, not an earnings-quality problem caused by repeated asset gains, but it is still cash shareholders do not have.

Capex makes the issue more acute. Cash spending on fixed and intangible assets rose from roughly RMB 302m in 2023 and RMB 372m in 2024 to RMB 913m in 2025. The resulting simple 2025 free cash flow was only about RMB 121m. Physical depreciation was about RMB 354m, while the pre-expansion capex run rate was generally in the RMB 300–400m area, so a reasonable maintenance-capex estimate is RMB 350–450m and the remaining roughly RMB 460–560m of 2025 capex is best viewed as growth investment. That makes the current headline free-cash-flow yield almost meaningless on its own, but it also means the stock cannot be valued responsibly on accounting earnings alone.

At CNY 52.28, trailing-twelve-month attributable earnings are approximately RMB 2,389m, producing a market P/E near 25.4 times. The trailing cash dividend equivalent to the 2025 RMB 800m distribution is about a 1.32% yield on today's market capitalisation. The share price has also retreated materially from a 52-week high around CNY 70, so the market has already removed part of the diversification premium that existed near the high; on today's earnings, CNY 70 would correspond to roughly 34 times P/E.

The qualitative portrait is a company in transition. The tire-mould business already qualifies as a high-quality industrial franchise. Large parts are large enough to affect group margins but do not yet earn tire-mould economics. CNC has become large enough to matter but remains too young to prove external competitiveness. The stock therefore sits between “quality compounder” and “diversified precision-manufacturing growth story.” The market price is roughly consistent with the latter becoming true. It does not provide a large cushion if the transformation merely produces revenue rather than cash.

Vertical history and financial review

Himile's origin explains why moving from one end market to another is less eccentric than it looks. The entrepreneurial story began around 1994–1995, when Zhang Gongyun and several partners acquired the repair workshop of a failed local textile-machinery operation. Tsinghua University's management-school case describes four founders raising RMB 40,000 while taking on roughly RMB 960,000 of liabilities; the young operation had no defining product and accepted assorted metalworking jobs simply to survive. The legal predecessor dates from 1995. Zhang had previously worked in a state-owned forging-machine-tool factory as engineer, technical manager, production manager and deputy factory director.

The first turn came in 1997. Chinese tire-mould manufacturing still depended heavily on manual or semi-manual processes and imported equipment. Himile developed a dedicated electrical-discharge machine for tire moulds; contemporary retrospective accounts describe its selling price as a fraction of imported equipment and its role in accelerating domestic numerical-control adoption. Through the following years the company added dedicated engraving, turning and other machinery for the same customers.

Then management did something that still defines the company. Himile stopped treating the machine tool as the end product and moved downstream into the product made on the machine. The company's own retrospective dates the strategic shift to tire moulds to 2002; a Xinhua profile describes the decisive operating move in 2003. The two sources are best reconciled as a 2002–2003 transition rather than forcing false precision. Zhang's reasoning, as recorded in the contemporary account, was economic: dedicated machine tools are durable and the customer pool is finite, while tire moulds are recurring production tooling and new specifications continually require new sets. Shanghai Double Coin was cited as an early customer that helped validate the switch.

That decision changed Himile from a small equipment maker into a consumable-tooling specialist. It also created the internal-machine-tool culture visible today. The company learned the tire customer's process from inside the mould shop, then kept designing equipment around its own production bottlenecks. In other words, today's CNC business is partly a return to the original competence, after two decades of using machine tools primarily as an internal manufacturing advantage.

The corporate form was standardized in 2008 when the predecessor limited company was converted into a joint-stock company. The IPO followed on June 28, 2011. Himile issued 50m shares at CNY 24, leaving 200m shares outstanding after the offering. Gross proceeds were RMB 1,200m and net proceeds approximately RMB 1,152m. The issue price represented 25.72 times 2010 earnings on the post-issue share count. Founder Zhang Gongyun held 29.90% immediately after listing. The public-market story was then straightforward: radial-tire moulds and related manufacturing equipment, with capacity expansion funded by the listing.

The next decade can be understood in four economic stages rather than as an event log.

The first stage was survival and process invention, from 1995 through the early 2000s. Management's enduring contribution was not any one product; it was the habit of solving manufacturing constraints by building equipment internally. That is the institutional root of the later machining advantage.

The second stage, roughly 2002 through the IPO, converted that know-how into the recurring tire-mould business. The move downstream enlarged the addressable profit pool and placed Himile directly inside the tire makers' product-development cycle. By the time of listing, it had become China's largest specialist in the category described in the prospectus and had begun selling internationally.

The third stage was global scale and quiet diversification. The history is important because large parts did not suddenly appear with the 2025 gas-turbine enthusiasm. By 2021, large mechanical parts were already RMB 2,441m, or 40.6% of group revenue. In 2022 they reached RMB 3,041m, 45.8%, while moulds were RMB 3,274m, 49.3%. That year CNC was still tiny at RMB 146m. Himile was already economically a two-business company before the latest power-generation cycle; the recent change is that CNC has become a credible third line.

The fourth stage began around 2023 and is still under way: simultaneous reacceleration, heavier capex and a shift in how investors have to value the company. Mould revenue grew from RMB 3,790m in 2023 to RMB 5,509m in 2025. Large parts went from RMB 2,770m to RMB 3,964m. CNC went from RMB 308m to RMB 968m. The company committed fresh casting capacity, overseas mould expansion and machine-tool capacity at the same time, pushing 2025 fixed/intangible-asset purchases to RMB 913m.

The financial history since 2020 captures why Himile first appeared on quality screens.

The 2020–2022 revenue and earnings base comes from the company's later comparative statements; 2023–2025 figures and ROE are from the audited annual reports.

From 2020 through 2025, revenue compounded at approximately 15.9% annually and attributable net income at about 18.9%. That is unusually good for a metalworking manufacturer. More revealing is the change in return structure: ROE moved from the high-teens area earlier in the period to above 20%, while financial leverage remained modest. The 2025 DuPont decomposition, 21.6% margin against roughly 0.84 times sales to average assets and about 1.21 times average assets to average equity, shows that leverage is barely doing any work.

A net-cash invested-capital calculation reaches the same broad conclusion. Removing cash and trading financial assets from capital, adding the small amount of interest-bearing debt, and stripping identifiable financial/investment income from pretax profit yields a 2025 ROIC proxy in the low-to-mid-20% range. This is my calculation rather than a company-reported metric, and segment capital cannot be separated. The safe conclusion is that historical group returns are genuinely high; the unsafe conclusion would be to assume the casting and CNC investments individually earn the same return as tire moulds.

Cash flow is the main blemish on that record.

Across those five years, operating cash flow was approximately RMB 4,452m versus RMB 8,270m of profit. Cash capex absorbed another RMB 2,166m, leaving cumulative simple free cash flow around RMB 2,286m. The cash problem is therefore persistent enough to affect valuation, even though individual years are volatile.

The reconciliation shows the culprit. In 2024 inventory absorbed about RMB 806m and operating receivables about RMB 785m, partially offset by RMB 387m of additional operating payables. In 2025 receivables absorbed a much larger RMB 1,704m, inventory another RMB 223m, and payables contributed only RMB 120m. Rapid sales growth is using balance-sheet capital faster than earnings are arriving in cash.

This distinction is central to earnings quality. There is little evidence that reported profitability depends on large recurring disposal gains or aggressive R&D capitalization: 2025 R&D investment was RMB 658m, 5.94% of revenue, and only RMB 2.6m was capitalized. The earnings are mostly operating earnings. The problem is collection and working-capital intensity.

Balance-sheet risk remains low. At year-end 2025 cash was RMB 1,483m, trading financial assets RMB 450m, attributable equity RMB 12,027m and short-term borrowings only about RMB 210m. Receivables at RMB 3,498m and inventory at RMB 2,693m are much more relevant to shareholders than leverage. By the end of the first quarter of 2026, cash was around RMB 1,531m, while total assets and attributable equity had risen further.

Capital allocation is mostly organic rather than acquisitive. The five-year capex pattern shows management using internally generated funds to expand casting, overseas mould service/production and machine tools; there is no evidence of a large recent equity-funded acquisition programme. The 2025 dividend proposal was RMB 800m, equivalent to 33.43% of attributable earnings, alongside a 4.5-for-10 capital-reserve share increase that took the share count to 1,160m. The prior year's cash distribution was only about RMB 319m because management was funding heavier expansion.

The company also uses broad employee share ownership. The 2025 plan covered roughly 2,100 core employees and the 2023 plan roughly 1,900, against a total year-end 2025 workforce of 18,383. Share-based expense remained small at RMB 9.6m in 2025. This makes the celebrated employee-ownership culture somewhat visible in the numbers, but the company does not disclose employee-turnover data that would let an outside investor prove superior retention. Revenue per employee was roughly RMB 603,000 in 2025; the productivity story is respectable, not enough by itself to establish a cultural moat.

Governance is founder-controlled but operationally institutionalized. Zhang Gongyun, who held 29.90% after the IPO, remained the controller with approximately 30.25% in current ownership data. Day-to-day leadership has moved to long-serving internal managers: chairman Shan Jiqiang joined in 2000 and general manager Cao Aijun joined in 2002 after earlier hands-on machine-tool and tire-mould work. The ownership stability and internal succession reduce key-person risk relative to a founder-centric industrial company.

There are still capital-allocation details to watch. In 2025 the group earned roughly RMB 34m from entrusted lending, and related parties accounted for 12.60% of annual procurement. Neither item is material enough today to alter the thesis, but both deserve more scrutiny than the simple “net-cash compounder” narrative usually gives them. The annual report received a standard audit opinion from Xinyong Zhonghe.

The capital-market history is harder to compress into a single valuation percentile without introducing false precision. The IPO itself was priced at 25.72 times post-issue earnings. Today, the stock is again around 25 times trailing earnings, but comparing the two multiples mechanically would miss the business change: 2011 investors were paying for a focused tire-mould growth company; today's buyer is paying for a high-return mould franchise plus two increasingly capital-intensive adjacencies.

Over the latest year the market briefly placed a much richer price on that diversification. The 52-week high of approximately CNY 70 corresponds to roughly 34 times current trailing earnings; at CNY 52.28 the multiple is about 25.4 times. That retreat can be read as partial removal of the CNC/gas-turbine growth premium. I cannot independently audit a full split-adjusted 10-year daily P/E history from the primary filings, so I do not assign a fabricated “73rd percentile” or similar historical statistic.

Business model, moat, industry and cycle

The segment mix shows how rapidly the investment case has changed.

The company reports tire-equipment revenue slightly above mould revenue because the category also contains other tire equipment. Formal accounting segment reporting is limited; management effectively treats the group as one reporting segment, so CNC gross margin and dedicated segment assets are unavailable.

The tire-mould machine is the cleanest business. Each order is customized around a tire specification. Production is low-batch, engineering-heavy and direct-to-customer. A 2025 mould cost reconstruction shows why the economics differ from a commodity foundry: materials were roughly RMB 1,245m, labour around RMB 1,163m, energy roughly RMB 106m and manufacturing overhead about RMB 792m. Material represented about 38% of mould cost, consistent with management's July investor communication. Labour, machining time, programming and process know-how are therefore at least as important as metal prices.

Large parts have a different cost equation. Management said raw material represents roughly 54% of that business's cost, and the annual-report cost data support the figure. Cast metal, furnaces, machining equipment and capacity utilization matter more. This explains why scaling can initially hurt profitability: a new foundry or machining line brings depreciation and labour before utilization reaches mature levels. Management attributed the 2025 large-parts margin decline partly to project-expansion investment and increased staffing.

CNC is different again. Fixed engineering expense, service capability and working inventory matter heavily. Production value reached approximately RMB 1,227m in 2025, external sales roughly RMB 968m, self-use equipment about RMB 459m and year-end CNC inventory around RMB 262m. Inventory was up 121% year on year. Rapid growth justifies some build, but external sell-through now needs to catch up with the manufacturing ramp.

That mix explains why group operating leverage is not behaving like a simple factory scale story. Tire moulds already operate at high utilization and high margin. Large-parts expansion adds furnace and machining capacity before revenue fully absorbs the cost. CNC is still building engineering, sales and service infrastructure. More group revenue can therefore coincide with lower consolidated margin, exactly what happened in 2025.

The tire-mould moat has four components.

First is process scale. The company says it has more than 30% of the international market and relationships with 66 of the top 75 tire companies. Even allowing for denominator ambiguity, this is substantially larger than a normal local tool shop.

Second is captive manufacturing technology. Himile reports that almost 80% of the machine tools operating inside its tire-mould production are self-developed. A toolmaker that designs the machine and the mould together can attack cycle time, accuracy and cost at both levels of the process. That capability is difficult for a small competitor to duplicate simply by buying another machining centre.

Third is customer-specific engineering and global proximity. Himile maintains mould or service operations in markets including the United States, Europe, Thailand, India, Indonesia, Brazil, Vietnam, Mexico and Cambodia. Because moulds need modification, maintenance and timely launch support, that network reduces the disadvantage of manufacturing from Shandong.

Fourth is financial endurance. A nearly debt-free balance sheet lets Himile carry customer receivables, invest through weak cycles and build capacity ahead of demand. Greatoo's current losses illustrate why this matters in a business where qualification and equipment commitments precede revenue.

The moat stops short of contractual captivity. The top-five customer share of 34.29% and largest-customer share of 12.48% prove diversification, but neither tells us customers are locked in. The annual report does not disclose multi-year minimum purchases or exclusivity. The right characterization is a high-reputation qualified supplier that can be replaced at a cost and risk, not a supplier that cannot be replaced.

Margin data support that middle ground. Mould gross margin of about 40% in both 2024 and 2025 is strong evidence that customers do not auction away all of Himile's productivity advantage. The decline from 43.06% in 2023 says the advantage is shared with customers and can vary with mix, material, FX and pricing.

Geography adds another wrinkle. Group export revenue was RMB 4,661m in 2025, 42.07% of total. Disclosed main-business gross margin was 39.96% on exports versus 27.62% domestically. That 12-point gap cannot be interpreted as “foreign tire customers are more profitable,” because domestic revenue contains nearly all CNC sales and much of large parts while exports are more mould-heavy. The company does not publish tire-mould gross margin by domestic versus overseas customer.

Trade measures consequently matter, but public disclosure has not shown a separately quantified tariff hit large enough to isolate. The overseas production-and-service footprint reduces freight, service and some border risks, though it does not eliminate tariffs on China-origin content. FX is much more visible in the financial statements.

The renminbi has become a real 2026 margin input. Himile carried significant dollar-, euro- and yen-denominated cash and receivables at year-end 2025, with the reported year-end USD/CNY conversion at 7.0288. By August 18, 2026, AsianBondsOnline showed CNY 6.744 per US dollar, about a 4.1% appreciation of the renminbi from year-end on that comparison. A stronger renminbi reduces translated export receipts and can create losses on net foreign-currency monetary assets.

The tire industry's relevant cycle is primarily a product-iteration and capex cycle. Replacement-tire unit demand helps customer health, but mould demand is pulled by new tire patterns, dimensions, compounds and homologations; initial equipment for a new plant also matters. Himile's own language explicitly separates overall tire capacity from specification structure and pattern iteration.

The EV thesis is best described as specification acceleration rather than unit-volume leverage. Michelin's electric-Macan development involved numerous references; Bridgestone's bespoke Macan range came in six dimensions; such proliferation is mould-intensive. But Continental's statement that most existing product lines are already suitable for EVs is the counterexample that kills the simplistic “every EV needs a brand-new tire category” model.

Regulation should keep model iteration alive. EU tire rules already impose minimum performance standards through type approval and consumer labels for rolling resistance, wet grip and noise; the Commission notes further abrasion/microplastic and mileage requirements are coming. A tire maker can meet those requirements through compounds, construction and tread changes. Some changes require new mould geometry; others do not. I therefore include regulatory redesign as a steady demand tailwind, not as an abrupt TAM reset.

Replacement from mould wear is a second recurring demand source, although Himile does not publicly disclose a standardized replacement interval. Its own R&D work on laser cladding specifically seeks to improve corrosion resistance and extend mould life, which indirectly confirms that wear and maintenance are genuine economic variables. Longer-lived moulds are good for customer economics and potentially negative for replacement frequency, another reason not to model mould demand off tire volumes.

The large-parts cycle is currently far hotter. GE Vernova reported exceptional gas-power demand in 2026, with gas turbine production effectively sold out through 2029 and a sharply enlarged gas-power pipeline. Siemens Energy's latest reported quarter produced record order intake and a record overall backlog, with Gas Services guiding double-digit comparable revenue growth. This is exactly the sort of OEM environment that can keep qualified suppliers busy for several years.

Himile is positioned to participate because the company explicitly identifies GE, Mitsubishi and Siemens among its large-parts customers. Yet disclosure does not let an investor isolate gas turbines from wind, steam, industrial machinery and the other applications inside “large parts.” There is no gas-turbine revenue line, no dedicated margin and no public qualification schedule by OEM. That absence is material. A hot end market cannot be given an independent valuation when the supplier's exposure cannot be measured.

Capacity is measurable. Designed casting capacity now exceeds 300,000 tonnes and the approximately 65,000-tonne expansion entered trial production in the second half of 2025. Fixed assets jumped from roughly RMB 2,308m to RMB 3,072m during 2025 and construction in progress from about RMB 80m to RMB 500m. That is the physical evidence behind both the growth opportunity and the 2025 margin pressure.

The CNC cycle is a technology-and-capex cycle, with an import-substitution overlay. Himile's machines cover five-axis machining, mill-turn and laser applications. It has a credible structural advantage in castings, beds, machining, assembly and selected rotary-table components because those capabilities sit inside the group already. The key open question is the high-value component stack. Public disclosure does not identify controller, spindle, guide or bearing suppliers or the domestic-content ratio.

That matters because buying an imported controller and spindle, placing them on an excellent locally made structure and assembling a competitive machine is still a valid business, but it is a different moat from controlling the machine's complete functional stack. FANUC is therefore useful as a technology benchmark, not a valuation peer. Himile must prove external machine reliability, service and repeat purchasing before its CNC economics deserve the same quality multiple as tire moulds.

The diversification argument does have one numerical point in its favour. Group ROE has increased rather than collapsed as large parts and CNC grew. From 2021 to 2025, the group moved from roughly 17% ROE to 21.8% while maintaining low leverage. That means shareholders have not yet suffered the classic conglomerate outcome of obvious return destruction. Segment assets are not disclosed, though, so it remains impossible to prove that the incremental yuan invested in casting and CNC earns the core mould return.

Horizontal competitors and current fundamentals

There is no clean listed peer. Greatoo Intelligent Equipment, 002031, is the closest Chinese public comparator because it sells tire moulds and hydraulic vulcanizing presses alongside robotics and intelligent equipment. Even that comparison quickly becomes lopsided. Greatoo reported only about RMB 788m of 2025 revenue, down 19.78%, and attributable net loss of approximately RMB 232m. Overall gross margin was about 10.37%, while operating cash flow remained positive around RMB 238m. Himile generated more than fourteen times Greatoo's revenue and more than RMB 2,393m of profit.

† Himile's figure is tire-mould gross margin; Greatoo's is company-wide gross margin, so the row is a profitability reference rather than a like-for-like segment comparison. Himile data are from its annual report; Greatoo data from its 2025 annual-report release and financial disclosures.

The gap tells us more about business evolution than about product specifications. Greatoo became a mixed tire-equipment-and-automation company without establishing Himile-like profitability. Himile used internally designed equipment to build a much larger global mould franchise and then diversified from a position of surplus cash and high margins. Greatoo's existence shows that tire makers have alternative equipment suppliers; Greatoo's financial profile shows that competing in the category does not automatically reproduce Himile's economics.

Private international mould makers remain relevant, especially because sophisticated tire companies generally avoid depending completely on one tooling vendor, but transparent current financial data are insufficient for a robust public-peer valuation. Historical industry accounts describe Korean and other specialist suppliers that pre-date Himile's global rise. I therefore use them as evidence that the industry is contestable, not as numeric valuation comparables.

FANUC belongs in the peer set only on one narrow axis: the machine-control and automation benchmark against which a Chinese high-end machine tool must compete. GE Vernova and Siemens Energy belong on another narrow axis: their order books are demand indicators for power-equipment supply chains. Treating any of those companies as a consolidated P/E peer for Himile would create more noise than information.

Himile's ecological niche is thus unusual. In tire moulds it is the global-scale specialist taking work that tire companies choose to outsource. In large parts it is a qualified manufacturing supplier whose economics remain below those of the mould core. In CNC it is the challenger trying to commercialize machinery knowledge accumulated internally. The most plausible future competitor is therefore different in each segment: another tooling specialist in moulds, a large foundry/machining supplier in components, and established Japanese, European or Chinese machine-tool manufacturers in CNC.

The latest operating data sharpen the distinction between narrative and fundamentals. The first-quarter 2026 report, filed April 30, showed revenue of RMB 2,668m, up 17.07%. Attributable net income was RMB 515m, down 0.83%, and operating cash flow RMB 117m, down 28.58%. On the face of it, this looks like an extension of the 2025 “sales up, profit not keeping up” problem.

Q1 2026 was operationally better than the net-profit line looks. Gross margin calculated from reported revenue and cost was about 34.05%, versus approximately 33.81% a year earlier. The big change was finance expense: it swung from roughly negative RMB 29m, a gain, to positive RMB 62m, a cost. The RMB 92m year-on-year deterioration is equivalent to roughly 15% of reported first-quarter pretax profit. Holding the finance result constant, pretax profit would have grown by roughly the mid-teens rather than staying nearly flat.

That does not make FX irrelevant. It proves the opposite. With 42% of 2025 revenue exported and a sizable net foreign-currency monetary position, a renminbi appreciation can materially interrupt earnings conversion even when factory margins are holding. The company's July investor communication said it uses or considers hedging selectively rather than eliminating exposure entirely.

The Q1 balance sheet creates another issue to monitor. Inventory rose to roughly RMB 3,043m from RMB 2,693m at year-end, about 13% in one quarter, while receivables rose more moderately to about RMB 3,637m from RMB 3,498m. A quarter of inventory build ahead of deliveries can be normal for a fast-growing industrial company. Repeated inventory growth ahead of sales would instead signal that CNC and new capacity are producing faster than customers are taking output.

The last fully disclosed segment year, 2025, gives bulls three tangible data points. Mould revenue still grew 18.44% with a stable 39.98% margin. CNC more than doubled. Large-parts capacity was commissioned into a power-equipment upcycle. Those are real fundamentals rather than thematic labels.

The bear data are equally concrete. Large-parts margin fell four points. CNC inventory more than doubled and external customer names remain undisclosed. Group cash conversion deteriorated. Q1 2026 net income did not grow despite 17% sales growth, even if FX explains much of the gap.

The market is therefore trading three intertwined expectations: that tire moulds retain their approximately 40% gross margin; that the current gas-turbine/power-equipment cycle raises utilization and repairs the large-parts margin; and that CNC's 2025 growth represents the beginning of a durable external franchise rather than a burst of internally supported capacity. The first assumption is already well evidenced. The second is plausible but not isolated in disclosure. The third remains early.

The bull case on tire moulds is stronger than a simple one-third-share statistic. Himile's customer base is diversified, its global service network is expensive to reproduce, and its product margin has stayed close to 40%. The bear counterargument is that no disclosed contract prevents global tire makers from dual-sourcing or re-tendering; a moat built on qualification and execution can erode gradually through price, not suddenly through customer loss.

The bull case on EV tires is narrower than common commentary. More EV platforms, wheel sizes and new tire references increase mould-development events. The bear evidence is Continental's observation that existing tire lines already serve many EV applications. The rational forecast is continuing specification proliferation, not an EV-driven step-change in moulds per tire sold.

The bull case on gas turbines is that GE Vernova and Siemens Energy have extraordinary forward visibility. The bear case is that Himile has not disclosed the portion of its RMB 3,964m large-parts business attached to gas turbines, and 2025 gross margin deteriorated while this supposedly attractive cycle was strengthening. Demand visibility at the OEM does not equal supplier profitability.

The bull case on CNC is the growth rate and the internal proving ground: Himile has spent years using its own machines in demanding mould production. The bear case is the lack of evidence about high-value component localization and outside repeat customers. A product that works inside a captive factory is not yet equivalent to a service-intensive merchant machine-tool franchise.

There is no verified H1 2026 earnings figure in this report. The scheduled August 29 interim filing is the next event that can resolve several of these disagreements at once. Any secondary figure purporting to be Himile's first-half result before that filing should be checked against ticker 002595 and the legal company name before use.

Valuation, risks and tracking

The starting valuation is neither obviously cheap nor obviously euphoric.

The price is the August 18 close; TTM profit adjusts 2025 earnings for the first-quarter 2025/2026 replacement, and the dividend yield uses the RMB 800m 2025 distribution over current market capitalisation.

Historical valuation gives only a partial anchor. The IPO multiple was 25.72 times post-issue earnings, close to today's trailing P/E, while the recent CNY 70 high was equivalent to roughly 34 times current TTM earnings. The business at each date was different enough that neither number should be treated as a fixed “normal P/E.” A reliable full-history valuation percentile could not be established from primary disclosure, so I do not use one in the target-price work.

Peer valuation is even less useful. Greatoo is loss-making, so its P/E is not meaningful. FANUC is a machine-tool/control-system reference but its economics cover businesses that are only a small slice of Himile. GE Vernova and Siemens Energy are customers/demand indicators rather than peers. Absolute valuation therefore carries most of the analytical weight.

The cash-flow passthrough test changes the picture. Across 2021–2025, OCF/net income is only about 54%. In 2025 itself it was 43%. The shortfall against full conversion is therefore about 46 points, past the 30-point gap at which I stop treating accounting P/E as the default basis, so the scenarios below are built on owner cash flow instead.

Maintenance capex is not disclosed separately. I estimate RMB 350–450m for 2025, centered near RMB 400m. The basis is physical depreciation of roughly RMB 354m and the RMB 300–400m annual capex run rate immediately before the 2025 casting/machine-tool expansion. On that basis, about RMB 460–560m of the RMB 913m 2025 capex was growth capex.

Using the RMB 400m midpoint, 2025 OCF less estimated maintenance capex was only about RMB 634m. On today's RMB 60,645m market value, that is an owner-cash-flow yield of roughly 1.0%, or approximately 96 times the depressed 2025 owner-cash-flow proxy. The number should not be capitalized literally because 2025 working capital was exceptionally heavy. It does prove that the current price depends on a substantial cash-conversion recovery.

The market is already paying for cash conversion to normalize. My valuation therefore models normalized 2027 owner cash flow rather than pretending either 2025 FCF or 2025 accounting earnings alone is sustainable.

† Approximate, including modest dividends; it is scenario arithmetic rather than a forecast or investment recommendation.

The conservative case assumes that tire-mould economics remain sound, revenue growth settles into high single digits, large-parts margins fail to recover much and cash conversion nearly normalizes only because slower growth releases working-capital pressure. The base case requires roughly 11% revenue CAGR, group margin around 21%, and OCF again matching accounting profit. That is not heroic operational growth, but the cash assumption is demanding given the five-year record. The optimistic case requires CNC and power components to sustain faster growth, group margin to regain 22%, cash conversion above 100% as previous working-capital investment is harvested, and the market to retain a premium multiple. These scenario values are within a research framework, not investment advice.

The base case says something useful about market expectations. At CNY 52.28, the stock is already close to my base value. An investor at today's price is effectively underwriting continuing double-digit sales growth plus a large improvement in cash conversion. A simple reverse dividend model points in the same direction: with a roughly one-third payout and an illustrative 8–9% cost of equity, a 25-times earnings multiple needs persistent mid-to-high-single-digit long-run earnings growth rather than flat earnings. The discount rate is my assumption; the payout and current P/E come from reported figures.

The most fragile base-case assumption is cash conversion. Reducing the 100% OCF/net-income assumption to 70%, while leaving the base revenue, margin and 25-times owner-cash multiple unchanged, cuts the estimated valuation to about CNY 34 per share. That sensitivity is much larger than a one-point change in revenue growth or a normal FX quarter.

The independent margin-of-safety test reaches an even firmer conclusion. Current price is above my CNY 42–46 conservative fair-value range, so there is no discount to conservative value. If earnings remain flat for three years and the valuation multiple is unchanged, the mechanical return is essentially the dividend yield of about 1.3% per year. China's official ChinaBond government curve showed a 10-year sovereign yield of 1.6864% on August 18, 2026. On that flat-earnings test, there is no margin of safety at this buy price.

Margin-of-safety verdict: none.

The permanent-loss risks are concentrated enough to be monitored rather than treated as generic boilerplate.

The first is tire-mould moat erosion, probability medium and impact high. The transmission path would be customer re-tendering or stronger competitors forcing lower mould pricing; mould gross margin would fall, group margin would fall disproportionately because this is the highest-margin major line, and the stock would lose both earnings and its “quality industrial” multiple. The observable threshold is mould gross margin below roughly 37% for two reporting periods, especially if revenue is still rising. The 2023–2025 range of 39.6–43.1% makes that threshold economically meaningful.

The second is poor returns on diversification, probability medium and impact high. Large-parts gross margin fell to 21.58% just as capex and fixed assets accelerated, while CNC inventory rose 121%. If the new casting capacity runs below plan and external CNC sell-through lags production, depreciation, labour and working capital will absorb cash without matching the mould core's returns. Watch large-parts gross margin below 20%, CNC inventory-to-sales above roughly 35%, and capex remaining near or above operating cash flow after the current commissioning period.

The third is persistent cash trapping, probability medium-to-high and impact medium-to-high. Five-year OCF conversion is already only 54%. If receivables continue to grow materially faster than revenue, the company can report attractive earnings while free cash remains too low to fund both expansion and shareholder returns. The valuation would eventually migrate from P/E toward cash yield, where today's price looks much more demanding.

The fourth is FX and overseas-trade exposure, probability medium and impact medium. Forty-two percent of revenue is exported, foreign-currency receivables are substantial, and Q1 2026 showed that a roughly RMB 92m year-on-year finance-expense swing can almost erase the operating earnings growth from a 17% revenue increase. A sustained stronger renminbi or new tariffs would act through realized pricing, translation and customer sourcing decisions. Overseas production mitigates but does not eliminate the exposure.

The fifth is valuation compression, probability medium and impact high. At 25.4 times TTM earnings, even a good company can lose a large part of its market value if earnings stall and investors reclassify the new businesses as ordinary cyclical machinery. A move to 15–18 times earnings with flat or lower profit would produce a share-price outcome far worse than normal industrial volatility.

The positive catalysts are equally concrete. The August 29 interim report can show whether Q1's FX distortion reversed, whether cash flow improved and whether the casting/CNC ramp carried into Q2. A rebound of large-parts gross margin toward the mid-20s would show that the 2025 dilution was commissioning-related. Continued CNC growth accompanied by slower inventory growth would be stronger evidence than revenue alone. Further gas-turbine orders at GE Vernova and Siemens Energy lengthen the potential utilization runway for qualified component suppliers, though Himile still needs to disclose its exposure.

Negative catalysts would be an H1 profit-growth rate again far below revenue growth, mould gross margin moving below its recent range, CNC inventory rising faster than sales, another year of sub-60% cash conversion, or a sustained RMB appreciation that keeps finance expense positive. Those events would attack the exact assumptions supporting today's 25-times multiple.

The tracking dashboard below is designed around those failure points.

The baseline ranges come from Himile's 2023–2025 reported margins, cash-flow history, 2025 CNC inventory and the latest FX level; the next filing date is the published interim-report schedule.

The most useful places to track them are the annual/interim segment tables for margins, cash-flow reconciliation for receivables and inventory, the machine-tool production/inventory section for CNC, and the investor-relations records for capacity commissioning. The first number I would read in the next report after revenue and profit is operating cash flow; the second is large-parts margin; the third is whatever new evidence appears on CNC external customers.

Cross-synthesis conclusion, uncertainties and sources

Looking vertically, Himile has proven one capability across thirty years: it can turn difficult manufacturing problems into internally designed equipment and then scale the resulting process. The first important product was a tire-mould machine tool. Management then recognized that the larger recurring profit pool sat downstream in the mould itself. The company subsequently took that production system worldwide. The record after 2020, roughly 16% revenue CAGR and 19% earnings CAGR with ROE above 20% and no material leverage, is consistent with genuine operating capability rather than a one-cycle commodity windfall.

The crucial historical insight is that Himile's advantage was never “tires” in the abstract. It was the combination of machine design, metalworking, process engineering, low-batch customization and organizational willingness to build the missing tool rather than wait for a supplier. This is why the CNC line has more credibility here than it would at a tire-industry conglomerate that suddenly acquired a machine-tool company. CNC is a commercialization of an old internal skill.

The same history does not automatically validate large parts. Large castings use some of the same machining resources, but their economics are much more material- and capacity-intensive. The 2025 data are the clearest warning: RMB 632m of additional large-parts revenue produced essentially no additional segment gross profit because margin dropped from 25.67% to 21.58%. The business may still prove attractive once the new 65,000-tonne capacity matures, especially given current power-equipment demand, but the burden of proof is now on utilization and return on capital.

Horizontally, the tire-mould franchise looks stronger than the nearest listed competitor by almost every observable economic measure. Greatoo's 2025 revenue was below RMB 800m and it was loss-making; Himile generated more than RMB 11,000m of revenue and RMB 2,393m of attributable profit. The gap is too large to explain with simple product pricing. Himile has become the scaled global process company while Greatoo remains a smaller diversified tire-equipment and automation operator.

This is enough for me to choose between the two interpretations of the one-third-share niche. The evidence favours a genuine moat over the “low-value outsourcing work that customers could reclaim at will” interpretation. Tire makers may re-tender, and there is no contractual lock-in, but a low-value commodity supplier generally does not hold roughly 40% gross margins, serve dozens of top global customers, operate a global maintenance network and build most of its own production equipment. The moat is operational rather than legal.

The market's likely error is subtler. I do not think it is dramatically overestimating the quality of tire moulds. I think investors are in danger of extending the proven mould economics to the two adjacencies before the evidence supports it. Large parts are already much lower margin. CNC has spectacular top-line growth but no disclosed segment margin, component-localization ratio or external customer list. The current stock price can work if those businesses become respectable, cash-generative additions. It becomes expensive quickly if they merely add revenue.

The 2025 revenue/profit crossover is therefore the most useful single fact in the report. Revenue grew faster than at any point in the recent five-year period, yet net margin fell. The decomposition says diversification is the main reason. That is preferable to tire-mould price collapse because it leaves the core franchise intact. It is still an economic cost. Shareholders are paying in capex, working capital and group margin to discover how transferable the machining capability really is.

The next twelve months are about conversion. The questions are whether the large-parts margin begins recovering as new capacity fills, whether CNC external sales absorb the inventory build, whether the first-quarter FX hit normalizes, and whether operating cash flow begins to track profit. H1 2026, scheduled for August 29, is the first major checkpoint.

The three-year question is whether Himile can keep group net margin around 20–22% while CNC becomes material and large parts scale. If it can, a low-double-digit revenue growth rate still supports respectable earnings growth. If group net margin settles in the high teens because the adjacencies structurally earn ordinary industrial returns, today's multiple has little reason to persist.

The five-year question is capital allocation. By then there should be enough data to decide whether management created a precision-manufacturing platform or diluted an exceptional tire-mould franchise. Segment ROIC disclosure would be ideal; absent that, cash conversion, group ROE, capex and segment margin will tell most of the story.

The stock itself is less compelling than the company. At CNY 52.28 it sells for about 25.4 times TTM accounting earnings. That looks acceptable when set against a 20%-plus ROE, a net-cash balance sheet and historical earnings growth. It looks less comfortable after examining cash conversion. The five-year OCF/net-income ratio is only 54%, and 2025 OCF less estimated maintenance capex yields barely 1% on today's market value. My base valuation explicitly assumes this improves toward 100% cash conversion.

That expectation is plausible because the receivable and inventory build accompanies unusually fast expansion, and 2023 already showed a year when OCF roughly matched net income. It is not proven because 2021, 2022, 2024 and 2025 all converted poorly. A valuation that requires the weak historical variable to become strong deserves a discount, even when the underlying company is excellent.

The risk-free comparison reinforces the point. A flat-earnings investor today receives roughly a 1.3% dividend yield, below the 1.6864% 10-year China government bond yield on August 18. Growth can readily make the equity superior, but the current price does not let an investor earn an attractive return while being wrong about growth. That is the essence of an absent margin of safety.

The case improves dramatically around the low CNY 30s. At that price, the investor would be paying well below my conservative valuation while retaining the upside from CNC and power components essentially as low-cost options. The opportunity cost is obvious: Himile may never trade there if cash conversion normalizes and new businesses execute. A disciplined valuation framework has to accept that possibility rather than turn every good company into a buy.

The bull reasons that survive the full analysis are:

• Tire moulds generated RMB 5,509m of 2025 revenue at a 39.98% gross margin, and company disclosures indicate relationships with 66 of the world's top 75 tire companies.
• CNC revenue rose 142.59% to about RMB 968m in 2025, converting an old internal machine-tool capability into a material external business.
• Large-parts capacity is expanding into an unusually strong global gas-turbine cycle, with GE Vernova and Siemens Energy reporting multi-year demand visibility.
• Group ROE reached 21.8% in 2025 with little financial leverage, showing that diversification has not yet destroyed consolidated returns.

The bear reasons are:

• Five-year operating-cash-flow conversion is only about 54%, while 2025 receivables alone consumed roughly RMB 1,704m of operating cash.
• Large-parts gross margin fell from 25.67% to 21.58% in 2025, leaving calculated gross profit almost unchanged despite nearly 19% revenue growth.
• CNC lacks disclosed segment margin, major external customer names and a full functional-component localization map, while year-end inventory rose 121%.
• At CNY 52.28 the stock already trades near my base value and requires cash conversion to normalize; the current dividend yield is below the 10-year CGB yield.

A concrete three-year pre-mortem starts with the tire-mould moat, not a macro recession. Suppose by 2027 aggressive tenders from large tire customers and alternative Asian mould suppliers push tire-mould pricing down by around 10%, taking mould gross margin from about 40% to 33–35%. At the same time the large-parts margin remains below 20% because new casting capacity is underutilized. Group net income could fall toward roughly RMB 2,000m even if revenue does not collapse. A market that then values Himile at 15 times earnings rather than 25 times would support equity value around RMB 30,000m, or about CNY 26 per current share: roughly a 50% loss from the research-base price. The mechanism is simultaneous moat disappointment and multiple compression, not short-term volatility.

A second script is failed diversification. CNC production continues rising through 2027 but external customer repeat orders lag; inventory-to-sales rises above 40%, the company must discount machines and provide heavier service support, while controls and precision functional components remain externally sourced. Large-parts capacity remains tied to a hot but uneven power cycle. Annual capex stays near RMB 1,000m, OCF conversion remains below 60%, and shareholders finally stop valuing the business on accounting P/E. A 4–5% normalized owner-cash yield on much lower cash earnings could again put the stock in the high-CNY-20s or low-CNY-30s. Neither script is my base case; both are credible enough that the current price needs a margin of safety it does not provide.

The research judgment would change positively if three facts appear together: tire-mould gross margin stays above 38–39%; large-parts margin returns toward the mid-20s as the new capacity fills; and trailing OCF/net income moves sustainably above 80%. Evidence that CNC is selling repeatedly to named third-party customers with stable inventory and disclosed domestic-content progress would justify another upward revision.

I would overturn the quality thesis negatively if tire-mould gross margin stays below 37% for two consecutive reporting periods, because that would indicate pressure inside the proven moat rather than mere mix dilution. I would also reconsider the diversification thesis if large-parts margin remains below 20% after the current commissioning phase, or if CNC inventory consistently outgrows sales. And I would reduce fair value sharply if OCF/net income remains below 60% after top-line growth slows, because working capital could no longer be dismissed as the temporary price of expansion.

The final judgment is therefore deliberately narrower than “excellent company” or “expensive stock.” Himile has one business whose quality is already proven and two businesses whose strategic logic is plausible. Its history gives management more credibility than the typical industrial diversification story: machine design and precision machining really are ancestral capabilities here. The large-parts margin and CNC disclosure gaps show that capability transfer is not free.

At the current CNY 52.28, an existing shareholder can rationally hold through the transition because balance-sheet risk is low and the base valuation is close to market price. A new buyer receives little protection against being wrong about working-capital normalization. I would wait for either stronger evidence or a substantially lower price.

【Company-profile scores】

• Fundamental quality: high
• Growth: medium
• Moat: medium
• Financial soundness: strong
• Management credibility: high
• Valuation attractiveness: low
• Risk level: medium
• Suitable investor type: long-term growth

【Investment rating】

• Rating: Hold
• One-line thesis: A durable tire-mould franchise is funding credible adjacent growth, but current price already assumes working-capital cash conversion normalizes.
• Ideal buy price: see the dedicated line below.
• Acceptable hold price: 46–62 CNY
• Clearly overvalued price: 84–92 CNY
• Current-price classification: acceptable hold
• Whether to wait for a better price: yes. I would require CNY 33 or below while tire-mould gross margin remains at least 38% and cash conversion shows no structural deterioration. The opportunity cost is missing the upside if CNC and power components mature without a drawdown.
• Target holding horizon: 3–5 years
• Expected annualized return: conservative about -4%; base about 3%; optimistic about 13%, using a three-year framework and modest dividends.
• Max-loss risk: roughly 45–55% in the pre-mortem case where tire-mould margin falls into the mid-30s, diversification remains low-return and the P/E compresses toward 15 times.
• Reassessment-trigger signals: tire-mould gross margin below 37% for two reports; large-parts gross margin below 20% after commissioning; OCF/net income below 60% after revenue growth normalizes; CNC inventory/annual sales above 35%; or a repeat FX drag large enough to consume more than roughly 10% of pretax earnings.

【Ideal Buy Price】30–33 CNY

Basis: this is at least about 20% below the low end of my CNY 42–46 conservative fair-value range, providing a margin for weak cash conversion while preserving CNC and gas-turbine-adjacent upside.

【Valuation Range】

• current: 52.28 CNY (close as of 2026-08-18)
• bear (conservative · ideal buy zone): [30, 33]
• base (fair · acceptable hold zone): [46, 62]
• bull (optimistic · above the clearly-overvalued line): [84, 92]

Research uncertainties remain material. First, the H1 2026 report was still pending as of the research date; August 29 is the scheduled release, so this report deliberately contains no invented Q2 figure.

Second, CNC disclosure is insufficient for a complete import-substitution audit. Himile identifies structural parts and selected rotary-table capabilities but not controller, spindle, guide and bearing suppliers, external customer concentration or physical third-party unit shipments.

Third, the gas-turbine exposure cannot be isolated from the broader RMB 3,964m large-parts category. The OEM cycle is verifiably strong; Himile's exact gas-turbine revenue, margin and qualification scope are not publicly quantified.

Fourth, the commonly cited global tire-mould market-share figure has a denominator problem. Himile says more than 30% “international” share while older public reporting cited approximately 25%. I therefore use a quarter-to-one-third range qualitatively and assign no valuation value to a precise 33% figure.

Fifth, segment assets and invested capital are not reported, preventing a defensible calculation of tire-mould ROIC versus large-parts ROIC versus CNC ROIC. Group ROE and a net-cash ROIC proxy are strong, but they cannot settle the core capital-allocation question.

The primary source hierarchy for this report is the 2025 annual report filed March 31, 2026; the April 30, 2026 first-quarter report; the 2024, 2023 and 2022 annual reports; the July 2026 investor-relations record; and the 2011 listing announcement. Demand-side checks use European Commission tire regulation materials, tire-maker product releases, current gas-turbine OEM disclosures and Greatoo's 2025 reporting.

Other tickers mentioned

• 002031.SHE: Greatoo Intelligent Equipment, the closest listed Chinese tire-mould and tire-equipment competitor, but far smaller and loss-making in 2025.
• 6954.TSE: FANUC, used as a high-end CNC control and automation technology reference rather than a consolidated valuation peer.
• GEV.US: GE Vernova, a disclosed large-parts customer-chain reference whose gas-turbine backlog provides a direct demand read-through.
• ENR.XETRA: Siemens Energy, another disclosed customer-chain reference and current indicator of the global gas-turbine equipment cycle.

This report is based on public information and does not constitute investment advice. Markets carry risk; invest with caution.