Zhejiang Supcon Technology: China's Process Automation Champion & The Industrial AI Pivot

I. Introduction & Episode Roadmap

In 1993, a 30-year-old professor at 浙江大学 Zhejiang University borrowed 200,000 yuan of his own money — without funding from the university — to start a company in a country that had never built its own industrial control system.¹ His name was 褚健 Chu Jian. He had scored at the top of his school in the 1978 college entrance exams, entered Zhejiang University's chemical engineering department at age fifteen, and in 1986 became the first jointly supervised doctoral candidate between Zhejiang University and 京都大学 Kyoto University.¹ What he saw in Japan reshaped his perspective. Kyoto in the mid-1980s was the global benchmark for process automation, and Chu returned convinced that a nation unable to control its own refineries could not truly claim ownership of them.

Colleagues suggested a safer path: operate as a distributor, reselling American and Japanese control systems into China's expanding petrochemical industry to collect a steady margin. Chu declined. As he later recalled in 中国经济周刊 China Economic Weekly, his strategy was straightforward: build a proprietary system or accept total failure.¹

Thirty-three years later, that decision has yielded both market leadership and structural financial pressure. 中控技术股份有限公司 Zhejiang Supcon Technology Co., Ltd. — 中控技术 SUPCON Technology, listed as 688777 on the Shanghai Stock Exchange's 科创板 STAR Market — held 45.1% of China's 集散控制系统 distributed control system (DCS) market in 2025, marking its fifteenth consecutive year in the top position.² In chemicals, its market share reached 68.5%, and in petrochemicals, 59.4%.² By conventional industry metrics, this represents clear dominance. Foreign incumbents including Honeywell, Emerson Electric, Yokogawa Electric, ABB, and Siemens — which once divided China's control-room market — have steadily lost market share in the sectors where they originally established the category.

Yet financial results have diverged sharply from market share. In the first half of 2026, SUPCON reported revenue of RMB 3.635 billion, down 5.1% year over year, while net profit attributable to shareholders fell 56.2% to RMB 155 million.³ This downturn followed a weak 2025, when full-year revenue dropped 11.66% to RMB 8.073 billion and net profit fell 60.48% to RMB 441 million.⁴ Operating cash flow in the first half of 2026 turned negative, reaching negative RMB 715 million.³

A company controlling 45% of its core market has seen roughly two-thirds of its net earnings vanish over eighteen months. This gap between market share and financial performance sits at the center of SUPCON's valuation. Either its competitive moat was overstated, or cyclical macro headwinds are temporarily depressing a structurally sound business. The evidence points to a third explanation: the moat is real, but narrower than headline market share figures suggest.

The company's history also reflects significant governance hurdles. In autumn 2013, as Chu Jian was widely tipped for election to the Chinese Academy of Engineering, prosecutors detained him over the privatization of university-affiliated enterprises. He remained in custody for over three years. In January 2017, the 湖州市中级人民法院 Huzhou Intermediate People's Court convicted him of corruption and the intentional destruction of accounting records, sentencing him to three years and three months in prison plus a RMB 500,000 fine — a sentence he had by then largely served.⁵ SUPCON survived his absence, completed its initial public offering in 2020, and raised approximately USD 565 million through a Global Depository Receipt listing in Switzerland in 2023.⁶ It subsequently acquired a Dutch analyzer manufacturer and appointed 崔山 Cui Shan as operational lead — an executive who spent two decades at Honeywell and Yokogawa, the legacy leaders SUPCON was built to displace.

The business evolution unfolds across four main phases:

Act I (1993–2012) covers the initial domestic milestone: building China's first viable DCS, establishing customer trust in mission-critical process plants, and taking market share from foreign incumbents.

Act II (2013–2017) details the governance test: maintaining execution on large-scale 20,000-loop projects while the founder was in custody and his equity was held via nominees.

Act III (2018–2023) tracks the financial expansion: listing on the STAR Market during a national drive for domestic substitution, capitalizing on a major petrochemical investment cycle, and issuing Swiss depositary receipts to fund international expansion.

Act IV (2024 to present) examines the strategic pivot from selling control hardware to marketing Industrial AI software and services, centered on the Nyx universal control system, the TPT time-series foundation model, and process-plant robotics.

The core investment thesis remains balanced. SUPCON's installed base creates substantial switching costs, as replacing the control system of an operating industrial plant carries high operational risk.

However, switching costs defend an installed base rather than pricing power or revenue growth. They did not prevent gross margin from contracting 2.52 percentage points to 31.26% in 2025,⁷ nor did they prevent chemical and petrochemical revenues from dropping 17.41% and 11.01% respectively that year.⁷ Furthermore, with state-owned enterprise customers enforcing multi-vendor procurement policies, converting hardware market share into high-margin software annuities remains unproven.

Management has tied strategic execution metrics to executive compensation. The analysis that follows evaluates how well empirical performance aligns with management's software and AI commitments.

---

↑ Back to Top

II. Origins: The Academic Entrepreneur & The DCS Breakthrough (1993–2012)

Picture a Chinese refinery control room in 1993. The consoles are Japanese or American. The screens display English or transliterated menus. When something goes wrong at three in the morning — and in a continuous process plant something always goes wrong at three in the morning — the plant manager calls a number in Tokyo or Phoenix, waits for business hours in another time zone, and then waits again for an engineer to fly in. Spare boards are quoted in dollars at multiples of manufacturing cost. The software protocol is proprietary and closed, meaning the plant cannot modify its own control logic without paying the vendor to do it.

That was the market Chu Jian entered, and it explains why he declined the distributor's route. The friction customers experienced was not primarily about price; it was about dependency.

The macro backdrop made that dependency politically uncomfortable as well as commercially frustrating. China in the 1990s was industrializing at an unprecedented pace — refining, petrochemicals, coal chemicals, thermal power, metallurgy, and fertilizers were all scaling simultaneously. Each of these is a continuous process industry, where material flows through the plant without stopping and the business model depends on operating at design capacity for years at a time. Continuous processes also carry high operational risks. A batch plant failure creates a bad batch; a continuous plant failure risks an industrial hazard. China was building thousands of facilities whose core control systems were designed, manufactured, serviced, and understood by vendors headquartered in the United States, Japan, Germany, and Switzerland.

What a DCS actually is. Strip away the acronym and a distributed control system functions as a plant's central nervous system. A large refinery has tens of thousands of measurement points — temperature in a reactor bed, pressure across a compressor, flow through a heat exchanger, the position of a control valve. Each measurement represents a loop: measure, compare to setpoint, adjust. The DCS reads these data points continuously, calculates necessary adjustments, and executes control logic in deterministic real time, twenty-four hours a day for two decades.

Layered alongside it is the 安全仪表系统 safety instrumented system (SIS), a separate set of interlocks designed strictly to execute emergency shutdowns. The SIS operates independently of the DCS on the safety principle that protective systems must not share failure modes with primary control systems.

If a corporate ERP system fails, administrative tasks are delayed. If a DCS fails in a chlor-alkali plant, the risk is catastrophic. That asymmetry dictates industrial procurement — driving stringent qualification processes, conservative reference requirements, twenty-year asset lifecycles, and a multi-year barrier to entry for new vendors.

Building it anyway. The predecessor entity began in 1993 as an automation venture attached to Zhejiang University, funded by Chu's borrowed 200,000 yuan.¹ The same year, at age thirty, he was appointed full professor — then the youngest at Zhejiang University.¹ The engineering objective was to build a domestic DCS that a industrial plant would trust with critical interlocks. Achieving durable market acceptance required roughly a decade of development.¹ In an industry where operational failure carries severe downside risks, commercial adoption relies on extended proof of reliability rather than rapid iteration.

Building a DCS required four synchronized elements: industrial-grade hardware controller cards, input/output modules, redundant power supplies, and communication buses capable of enduring electrical noise, temperature swings, and persistent vibration without silent failures; a deterministic real-time operating layer guaranteeing execution within fixed millisecond windows; a configuration environment for plant engineers to define control logic; and a human-machine interface displaying alarm hierarchies clearly across twelve-hour operator shifts. A new vendor had to validate all four components simultaneously to secure customer confidence.

The counter-position. SUPCON executed a classic strategy of counter-positioning: adopting a business model that incumbents could not replicate without compromising their existing revenue streams.

Foreign incumbents relied on high-margin hardware paired with premium service contracts. SUPCON priced hardware lower and bundled responsive, localized engineering support in Mandarin on site. Matching that service intensity in China would have undermined the global service margins supporting foreign incumbents' premium positioning elsewhere. The incumbents largely chose not to match SUPCON's service model, allowing the domestic newcomer to establish traction.

The beachhead. SUPCON initially targeted underserved segments — small municipal utilities, regional chemical plants, and fertilizer producers — where buyers often faced a choice between low-cost automation or manual operations, and where vendor customization took precedence over brand prestige.

Each successful installation established an operational reference point. In process industries, operational credibility compounds steadily: once plant managers verified multi-year uptime without unplanned trips at peer facilities, the economic rationale shifted from risk avoidance to cost efficiency.

The strategic breakthrough occurred when state-owned energy groups 中国石化 Sinopec and 中国石油 PetroChina qualified SUPCON for major refining and ethylene projects. Vendor qualification functions as the gatekeeper to process industry contracts; unlisted vendors cannot participate in formal tenders. Securing major vendor approvals required documented performance histories, rigorous factory audits, functional safety certifications, and internal operational sponsorship, establishing a structural barrier to entry alongside the long replacement cycle.

This installed base created a secondary locking mechanism. Once qualified, plant engineers trained on SUPCON's configuration tools, operators standardized on its interface screens, and maintenance teams stocked its spare parts. Over a decade, plant human capital became aligned with the vendor's software environment, creating structural switching costs that later supported SUPCON's software and services expansion.

Where the moat came from — and where it didn't. By around 2020, SUPCON had held the top position in China's DCS market for nine consecutive years, with a market share of roughly 27%.¹ This position represented market leadership rather than single-vendor dominance.

In 2022, China's total DCS market reached approximately RMB 12 billion across hardware, software, services, and engineering, with foreign brands retaining nearly half the overall market; SUPCON's chemical-sector share reached 54.8% that year.⁸ Domestic competition remained active, as 和利时 Hollysys led in power generation, rail transit, and nuclear automation, while 上海新华 Shanghai Xinhua Control Technology and other domestic players held established market positions.⁸ SUPCON's growth concentrated initially in chemical and petrochemical sectors before expanding into adjacent industries.

That sector focus reflected a specialized competitive advantage built on domain-specific process knowledge — such as reactor behavior dynamics and chlor-alkali operating protocols — embedded into control strategies and application software. While this domain expertise secured high market share in chemical processing, it also tied the company's core revenue base to cyclical capital expenditure trends in downstream chemical industries.

In 2017, SUPCON launched supOS, an industrial operating system designed to sit above the control layer and integrate plant operational data into application software.¹ The architecture represented the foundation for management's current Industrial AI initiatives, marking the start of an extended effort to convert software capabilities into recurring revenue. However, that product launch occurred during a period of organizational uncertainty, as the company's founder had been placed in custody over three years prior.

↑ Back to Top

III. The Storm: Legal Crisis, Arrest & Governance Reconstruction (2013–2017)

The autumn of 2013 should have been Chu Jian's coronation. He was a Yangtze River Scholar, vice president of Zhejiang University with responsibility for personnel, logistics and — the detail that would matter — the university's affiliated enterprises.¹ His name was in circulation for election to the Chinese Academy of Engineering, the closest thing a Chinese engineer has to a knighthood. Instead, prosecutors took him into custody.

The charges reached back more than a decade, into the tangled and largely undocumented process by which Chinese universities converted lab-attached businesses into private companies during the late 1990s and early 2000s. The 校企改制 restructuring wave created thousands of situations in which state assets, university IP, and personal entrepreneurial risk were braided together with almost no legal template. A professor who commercialised his own research typically did so using university facilities, university-affiliated staff, and in many cases university money, on the implicit understanding that the upside would eventually be shared in a manner nobody had written down. When those ventures became valuable, the absence of documentation stopped being a convenience and became an exposure. Some of those knots were later untied in court, and the untying was rarely gentle.

For SUPCON, the immediate operational question was brutally practical. Industrial control is a business in which customers buy a twenty-year relationship. What does a plant manager at a state-owned refinery do when the founder of his control system vendor is detained by prosecutors with an uncertain future? The rational response is to hedge — award the next project elsewhere, delay an upgrade, or quietly qualify a secondary supplier. That SUPCON's market share continued climbing through this period is compelling evidence of institutional resilience beneath the founder.

When the verdict arrived on January 16, 2017, the Huzhou Intermediate People's Court found that between 1999 and 2002, using his positions at Zhejiang University's industrial automation research centre and affiliated control companies, Chu had embezzled RMB 2,381,803. The court also found that in late 2012 he had directed the destruction of accounting books at several related entities. He received three years and a RMB 400,000 fine for corruption, one year and a RMB 100,000 fine for destroying accounting records, and a combined sentence of three years and three months alongside a RMB 500,000 fine. The court noted that all embezzled funds had been recovered and that his confession and expression of remorse reduced the sentence.⁵ Having been detained since 2013, he was released days after the judgment.

Two things deserve to be stated plainly. The sum involved — under RMB 2.4 million, recovered in full — was small relative to the enterprise Chu built, and the conduct dated to a period of genuine legal ambiguity around university spin-offs. That is the mitigating reading, and it is not unreasonable. But the second count is not ambiguous. Directing the destruction of accounting books in 2012, a decade after the underlying conduct, was a deliberate act aimed at erasing an audit trail. For an investor assessing management quality, a founder conviction for destroying accounting records remains a permanent historical fact, regardless of surrounding context.

The claim this tests. The standard bull framing of founder-led Chinese technology companies assumes that visionary founder control delivers long-horizon strategy and governance stability. SUPCON's own record provides clear disconfirming evidence that extends beyond the conviction itself.

Because Chu could not hold equity openly while serving as a university vice president, his ownership sat with nominees. The nominee arrangement originated in 2012; restoring legal title required a 2017 arbitration that produced a written agreement confirming 28.32% of the company had been held on his behalf.⁹ By the time of the IPO prospectus, he controlled 25.30% — 16.37% directly and 8.93% still through nominee holdings — while holding no board or executive position, serving only as strategic advisor.⁹ His younger brother 褚敏 Chu Min held 4.66% and chaired the board.⁹

Nominee shareholding is an informal structure common in Chinese corporate histories: the registered holder owns the shares legally, the beneficial owner holds them in fact, and the gap between those positions relies on trust or private agreements. When the arrangement falters through death, dispute, or non-acknowledgement, the beneficial owner's remedy is litigation supported by sparse documentation. That SUPCON's unwinding required formal arbitration was a standard outcome for such structures.

The prospectus was explicit about the resulting governance risks. Post-issuance, Chu's stake was expected to dilute to roughly 22.77%. With ownership otherwise dispersed, the filing explicitly warned of the risk that an outside investor could acquire control and destabilise management.⁹

Other disclosures in the prospectus clarified the company's operational and financial condition ahead of listing. Forty-eight patents were co-owned with Zhejiang University; while transfers required consent, the university retained the right to license them independently — including, the company cautioned, to competitors at low or no cost.⁹ Financial records also highlighted balance sheet pressure: in 2019, nearly thirty-five percent (34.87%) of accounts receivable were more than a year overdue, and the asset-liability ratio surpassed 60%, above peer averages. State support provided a significant profit cushion, with direct government subsidies contributing 10.42% of 2019 net profit and value-added tax rebates adding a further 28.93%.⁹ Furthermore, two key shareholders were affiliated directly with major customers: Sinopec Capital held 4.95%, while a nuclear industry fund held 3.00%.⁹

Equity stakes taken by state-owned enterprise customers represent a double-edged dynamic in industrial procurement. While customer equity aligns commercial interests and simplifies vendor qualification, it also creates a related-party structure through which commercial terms can be influenced beyond the visibility of public shareholders.

How to weigh it. This history does not refute claims of operational management strength; performance during the crisis suggests corporate resilience. What it refutes is the assumption that founder control equals governance strength. The professionalisation of SUPCON occurred during Chu's absence rather than his presence. A bench of long-tenured engineer-executives maintained project delivery and customer relationships through three years without their founder, proving institutional depth. Following his release, the structural response was not to reinstate the founder, but to recruit experienced executive leadership from international competitors.

Chu returned in 2017 as strategic advisor rather than chief executive, launching an internal initiative termed the Blazing Fire Plan.¹ He did not reclaim the chief executive post, assume the chairmanship, or consolidate voting control through dual-class share structures. He retained the advisory role while operational leadership transitioned over the next four years to executives recruited from global industry peers.

This structure leaves SUPCON as a founder-associated company where the founder holds roughly a fifth of the equity and sits outside the executive hierarchy, while professional managers run operations. For investors, this configuration reduces key-person reliance compared to traditional founder-dominated peers, while leaving the historical legal record part of the corporate profile.

The organisation Chu returned to required three elements: fresh capital, a normalized capitalization table acceptable to capital market regulators, and a compelling technology growth story. All three materialized over the subsequent three years.

---

↑ Back to Top

IV. The Resurgence: STAR Market IPO & Domestic Market Dominance (2018–2022)

On November 24, 2020, SUPCON's shares began trading on the Shanghai STAR Market at an offer price of RMB 35.73. The company issued 49.13 million shares at a price-to-earnings multiple of 64.04 times, raising net proceeds of roughly RMB 1.637 billion, with 8.08 million shares allocated to strategic investors and a post-issue share count of approximately 491 million.¹⁰

That transaction highlighted two distinct market dynamics. First, net proceeds of RMB 1.64 billion represented a modest capital raise for a business that would generate over RMB 9 billion in annual revenue four years later, serving primarily as working capital and listing currency. Second, a valuation of 64 times earnings reflected the policy environment of the STAR Market, which was created to fund 自主可控 autonomous and controllable industrial technology. As a domestic vendor in a sector long dominated by foreign imports, SUPCON aligned directly with Beijing's 国产替代 domestic substitution policy initiative.

The strategic timing coincided with shifting policy priorities. By 2020, foreign technological dependencies in critical infrastructure had become a central policy concern. For industrial process operators, reliance on foreign control software presented operational and strategic risks. Procurement officers at major state-owned enterprises increasingly prioritized domestic vendors, enabling SUPCON to translate its long-established technical record into market share gains.

Listing also provided structural corporate advantages. Publicly traded equity allowed the company to offer equity compensation to engineering talent, facilitate potential acquisitions, and present a strengthened balance sheet when negotiating multi-decade vendor commitments with industrial clients.

The share-gain machine. Examining market trajectory illustrates the scale of this shift. Around its listing, SUPCON held approximately 27% of the domestic DCS market.¹ By 2025, its DCS market share reached 45.1%, while its safety instrumented system (SIS) share reached 31.4%, securing four consecutive years in the top position for that category.² Gaining nearly eighteen percentage points of market share in an industry characterized by fifteen-to-twenty-year equipment lifecycles required capturing a dominant portion of new capital construction, particularly during China's rapid petrochemical expansion between 2018 and 2023.

This growth dynamic operated as a double-edged mechanism. While an installed customer base protects existing facility contracts, revenue growth depended heavily on greenfield plant construction. Consequently, any slowdown in new facility builds would exert a disproportionate drag on expansion.

Revenue in this era. Financial results tracked this market expansion. Annual revenue increased from RMB 1.71 billion in 2017 to RMB 2.53 billion in 2019,¹ eventually reaching RMB 9.139 billion by 2024¹¹ — representing a compound annual growth rate of roughly 29% over seven years. This performance matched the broader 30% compound annual growth in revenue and net profit cited in market commentary for the 2017 to 2023 period.¹²

The 5T framework. During this expansion, management framed its technology strategy around five integrated domains: AT (automation technology), IT (information technology), PT (process technology), OT (operation technology), and ET (equipment technology), collectively termed 5T. The framework posited that defensible industrial software requires combining control engineering with deep domain-specific process knowledge. Under this view, pure software companies lack operational process context, while equipment vendors lack unified data integration layers.

In practice, executing the 5T strategy meant transitioning industrial clients from basic control loops toward higher-margin software solutions, including manufacturing execution systems, digital twin simulations, and energy optimization tools. However, selling operational software required clients to alter organizational workflows rather than simply replace hardware.

Evaluating this framework against financial results shows a more gradual transition. Between the introduction of the framework and 2025, standalone software remained below 10% of main business revenue,⁷ while overall group gross margin contracted rather than expanded.⁷ While the 5T model outlined a strategic roadmap, initial reporting indicated that core earnings remained tied to traditional hardware and integrated engineering delivery.

Reading the era correctly. For investors, SUPCON's financial expansion between 2018 and 2022 reflected both competitive execution and a favorable macro capital expenditure cycle in China's chemical sector. Corporate disclosures combined software delivered within integrated solution contracts alongside standalone licenses, making structural software adoption hard to isolate from overall project volume.

For instance, the company reported RMB 2.653 billion in industrial software revenue in 2024, representing 20.7% year-over-year growth.¹¹ However, because this total bundled software embedded within broader hardware and engineering projects, standalone software products accounted for less than 10% of main business revenue in 2025.⁷

By 2023, as domestic construction activity normalized, management sought new growth avenues across two primary fronts: international market expansion and industrial artificial intelligence.

↑ Back to Top

V. Core Business Deep-Dive & Financial Economics

Inside a 10-million-tonne refinery, the distributed control system is physically understated — a few racks of circuit cards in an air-conditioned room, optical fiber, and a wall of monitor screens. It typically represents roughly 1% of a plant's total capital cost, yet it dictates the operational stability of the remaining 99%. That structural asymmetry — minimal upfront equipment cost combined with existential operational consequence — defines both the strengths and the limitations of the process control business.

The shape of the revenue. In 2025, SUPCON's core segment — industrial automation and smart manufacturing solutions — generated RMB 4.527 billion, dropping 19.77% year over year in a decline far steeper than the 11.66% drop at group level.⁷⁴ This primary division accounted for roughly 58% of main business revenue. Meanwhile, instrumentation contributed nearly 18%, the S2B service and industrial e-commerce platform accounted for roughly 11%, standalone industrial software remained under 8%, and operations and maintenance services represented about 4%.⁷ The revenue breakdown confirms that SUPCON remains fundamentally a control-systems provider with attached adjacencies, rather than a high-margin software enterprise with a hardware legacy.

Margin trends reinforce this structural reality. Group gross margin contracted 2.52 percentage points to 31.26% in 2025, and fell an additional 0.41 percentage points year over year to 31.67% in the first half of 2026.⁷³ Had high-margin software expanded as a proportion of total sales, gross margins would have expanded; instead, they compressed.

What each piece actually does. The solutions segment comprises DCS and SIS hardware alongside installation and commissioning engineering — a project-based business recognized over multi-month acceptance cycles, with gross margins in the mid-thirty percent range that fluctuate based on tender pricing pressure. Instrumentation covers field-level hardware, including pressure and temperature transmitters, flowmeters, control valves, and online analyzers added through the Hobré acquisition. While instrumentation carries lower gross margins and faces direct price competition, field devices wear out and require periodic replacement, generating a steady, high-frequency revenue stream from the installed base.

The supply-to-business platform, marketed as PlantMate, combines industrial e-commerce for spare parts, subscription services, and an effort to capture maintenance spending previously directed to local third-party suppliers. In 2024, the platform reported 622 SaaS member clients,¹¹ a modest figure relative to a customer base exceeding tens of thousands, indicating that the digital platform remains in an early stage of adoption. Operations and maintenance services represent the closest equivalent to a recurring annuity in SUPCON's portfolio, with revenue contracted against the installed base rather than driven by new capital expenditure.

This segment breakdown explains the company's recent earnings volatility. Approximately three-quarters of total revenue — solutions and instrumentation — depends directly on greenfield industrial construction. In contrast, roughly one-sixth — the S2B platform and maintenance services — depends on ongoing plant operations. The sharp earnings decline occurred because new construction spending contracted.

Switching costs: the real moat, precisely bounded. The primary economic moat protecting SUPCON's core business is high switching costs. Continuous process facilities operate around the clock for years between scheduled turnarounds. Replacing an active control system requires a complete facility shutdown, re-engineering thousands of control loops, re-certifying safety interlocks with regulatory authorities, retraining plant operators on a new software interface, and absorbing operational restart risks. Unplanned downtime and lost output can exceed tens of millions of renminbi before accounting for new hardware costs. Consequently, equipment lifecycles typically span fifteen to twenty years.

With more than 39,000 process-industry customers and over 100,000 deployed control systems at year-end 2025, SUPCON commands an exceptionally sticky installed base.²¹³

Now the falsification. The conventional assumption is that high switching costs confer pricing power and insulate a vendor from industry downturns. SUPCON's financial results between 2024 and 2026 challenge that assumption.

As capital expenditure in Chinese refining and chemical sectors contracted, SUPCON's full-year 2025 revenue fell 11.66%, while net profit dropped 60.48% to RMB 441 million.⁴ Revenue from chemical clients dropped 17.41%, and petrochemical revenue declined 11.01%.⁷ The contraction persisted into 2026: first-quarter revenue declined 6.29% year over year to RMB 1.506 billion, with net profit falling 37.81% to RMB 74.68 million;¹⁴ in the first half of 2026, revenue fell 5.10% year over year and net profit dropped 56.20%, with non-recurring-adjusted profit falling 64.47%.³ Return on equity in the first half stood at 1.57%, down 1.92 percentage points.³

High switching costs protect the existing installed base — customers rarely replace a functioning control system. However, switching costs provide little protection during capital expenditure downturns. Greenfield projects require competitive bidding where incumbent advantages are limited and pricing is aggressive. Brownfield upgrades are discretionary and often postponed when customer margins narrow. Meanwhile, spare parts and service volumes track industrial utilization rates, which also declined. A moat based on retention protects market share, but cannot prevent revenue contraction when customer capital deployment halts.

In its 2025 annual report, management cited slowing macroeconomic growth and weak downstream demand, alongside lower returns on bank wealth-management products and foreign exchange losses, as primary drivers of the profit decline.⁴ While demand contraction represents the principal driver, the impact of non-operating factors warrants scrutiny. When reported net income fluctuates based on treasury yields and foreign exchange movements, headline figures reflect financial management alongside core operations. This pattern aligns with historical disclosures: in 2019, government subsidies and value-added tax rebates combined to account for nearly 40% of net profit.⁹ Consequently, evaluating operating profit provides a clearer view of underlying business health than headline net income.

Buyer power is the other half of the story. Evaluating the competitive structure reveals significant buyer concentration. SUPCON's client base includes major state-owned enterprises such as Sinopec, PetroChina, and China Energy Investment, alongside chemicals major Wanhua Chemical and private refining groups Hengli Petrochemical and Rongsheng Petrochemical. State-owned buyers systematically enforce multi-vendor qualification policies to maintain bargaining power and limit supplier pricing leverage. Consequently, holding a 68.5% market share in the chemical sector does not grant pricing power; rather, it prompts large customers to manage supplier concentration risks.²

This procurement dynamic qualifies the broader domestic substitution thesis. While localization policies enabled SUPCON to gain market share against foreign incumbents, policy directives also promote competition among domestic suppliers. Having displaced international vendors across core chemical segments, SUPCON primarily competes against domestic peers within a contracting domestic capital expenditure market.

Furthermore, state-owned enterprise buyers increasingly utilize centralized e-procurement platforms featuring standardized framework agreements and scheduled re-bidding cycles. This administrative structure enforces price transparency and margin discipline, limiting an incumbent's ability to extract pricing premiums.

Other competitive dynamics present mixed conditions. Threat of substitution remains low, as deterministic real-time control of hazardous processes cannot be executed on general-purpose IT infrastructure, and safety instrumented systems remain regulatory mandates. Barriers to entry remain high due to multi-year vendor qualification cycles. However, supplier bargaining power is rising, while industry rivalry continues to intensify.

Scale economies, honestly assessed. SUPCON's extensive field engineering network across China's major industrial clusters provides a cost structure that foreign competitors cannot easily match. However, network scale economies require high utilization to remain accretive. During sector downturns, a large fixed engineering staff creates negative operating leverage, driving selling, administrative, and general expenses higher as a percentage of declining revenue.

Working capital efficiency also presents ongoing operational demands. At year-end 2025, goods shipped but not yet recognized as revenue reached RMB 2.123 billion, representing 11.28% of total assets,⁷ while first-quarter 2026 operating cash flow reached negative RMB 632 million.⁷ In a project-based business dealing with major industrial clients, shipped goods awaiting customer sign-off represent incurred costs prior to revenue realization. When downstream clients experience margin compression, project acceptance and milestone sign-offs frequently slow.

Disclosures from the 2020 IPO indicating that nearly thirty-five percent of accounts receivable were over a year overdue⁹ demonstrate that extended collection cycles are a structural feature of the business model. While this historical context reduces sudden credit surprise risks, it highlights persistent working capital intensity.

The comparison that clarifies. Comparing SUPCON with global peers — such as Honeywell, Emerson, Yokogawa, ABB, and Siemens — puts its strategic pivot in perspective. Global incumbents achieve higher blended gross margins primarily because their revenue mixes shifted toward software, recurring services, and aftermarket maintenance over multiple decades. These peers underwent similar transitions when hardware margins commoditized, expanding higher up the technology stack over ten-to-fifteen-year timelines. While this precedent demonstrates that software transformation is achievable in industrial control, it also indicates that management's strategic execution timeline is ambitious relative to historical industry benchmarks.

SUPCON remains an established market leader within a cyclical industry, rather than an unconstrained compounder. This structural dependence on domestic chemical capital expenditure explains why, starting in 2023, management initiated strategic expansion into international markets and industrial artificial intelligence.

↑ Back to Top

VI. Current Leadership, M&A & Global Expansion: The 崔山 Cui Shan Era (2023–Present)

SUPCON's operational leadership reflects an intentional hire from its legacy international competitors. 崔山 Cui Shan — born in 1971, a Singapore national trained in chemical process automation at the National University of Singapore — began his career in 1998 as a senior advanced process control engineer at Honeywell. He subsequently spent eleven years at Yokogawa Electric in Asia, moved to Yokogawa International, and served as vice president and then executive vice president of Yokogawa China until 2018. He joined SUPCON in December 2018 as a director and executive president, and became chairman and president in January 2021.¹⁵

That background carries direct operational significance. Honeywell and Yokogawa established the distributed control system category. Having spent two decades inside both organizations, Cui brought detailed insight into their pricing structures, vendor qualification frameworks, global service operations, and structural constraints. While recruiting competitor executives is common, appointing a veteran of legacy foreign incumbents as chairman signalled a pivot in corporate strategy: moving beyond domestic market share gains to build an international sales and service organization.

The appointment also addressed a structural hurdle for Chinese industrial equipment vendors expanding globally. International expansion typically stalls on execution rather than product specifications — specifically, the challenges of recruiting Western sales engineers, obtaining international functional-safety certifications, operating global service networks across multiple time zones, and assuring international clients of long-term warranty support. Importing executive leadership from Yokogawa offered a path to establish those institutional capabilities within SUPCON, though revenue results remain mixed.

Two characteristics define Cui's management approach. First, corporate strategy has been organized around explicit, published performance metrics. Second, management maintained research and development expenditure during the 2025–2026 earnings downturn rather than curtailing investments to buffer reported net income — a capital allocation choice that aligns with its strategic focus on technology development.

The Swiss listing. On April 17, 2023, SUPCON completed a listing of Global Depository Receipts on the SIX Swiss Exchange under the ticker SUPCON. The company issued 20,958,000 GDRs at USD 26.94 per receipt — with each GDR representing two A-shares — raising gross proceeds of approximately USD 565 million in the third Chinese GDR listing in Switzerland that year.⁶

The listing provided hard-currency capital via the China-Switzerland Stock Connect mechanism. Management allocated USD 290 million of the proceeds specifically to international expansion: USD 100 million for a Malaysian manufacturing subsidiary serving Southeast Asia, USD 100 million for a Singapore operations and R&D centre focused on control automation, industrial software, and artificial intelligence, USD 30 million for a Saudi Arabian subsidiary targeting industrial robotics and IoT assembly, USD 30 million to expand Dutch operations, USD 20 million for a Japanese R&D centre, and USD 10 million for a Central Asian operations centre in Kazakhstan.[^16]

This capital deployment targeted production, software development, and sales hubs near key client regions. However, issuing equity near the peak of the domestic petrochemical capital expenditure cycle created dilution for existing shareholders. Cross-border GDR listings often involve receipts priced at a discount to domestic A-shares, which can convert into Shanghai-listed shares after a lockup period and create arbitrage selling pressure. Consequently, the net benefit of the transaction depends on generating adequate returns on the USD 290 million deployed overseas, making international revenue execution a key test of management's capital allocation.

Judging it by results. Initial performance showed rapid international traction. In 2024, overseas revenue rose 118% year over year to RMB 749 million, lifting international sales to 8.25% of main business revenue. New overseas contracts grew over 35% to RMB 1.355 billion, supported by 21 international subsidiaries across Singapore, Saudi Arabia, Kazakhstan, Canada, Japan, and other regions, along with around five new international service centres.¹¹

That expansion paused in 2025. Overseas revenue fell 13.89% year over year to RMB 645 million, reducing international sales to 8.07% of main business revenue.⁷ Two years after raising USD 565 million with an explicit commitment to overseas expansion, the international share of main business revenue remained essentially flat, while absolute international sales contracted. This decline presents key evidence against the narrative of uninterrupted international growth.

Advocates of the international expansion strategy emphasize that overseas markets in the Middle East and Southeast Asia significantly expand SUPCON's total addressable market. The company has secured critical vendor qualifications required to bid on major international energy projects, entering the approved supplier lists of Saudi Aramco and the Abu Dhabi National Oil Company,⁷ winning an advanced process control project for Indonesia's SAMATOR, securing a pipeline automation contract for Algeria's state oil and gas company, and gaining shortlisting by Mexico's national petroleum company and Cemex.¹³ Entering vendor registries for major state energy firms involves extensive technical audits and represents a necessary prerequisite for project bidding.

However, vendor qualification does not guarantee consistent order flow. The pattern of sharp growth in 2024 followed by contraction in 2025 reflects lumpy, project-based order execution rather than an established international franchise. Demonstrating a sustainable global expansion will require consecutive years of international revenue growth alongside stable gross margins after accounting for the overhead of foreign service centers.

The Hobré acquisition. In 2023, SUPCON acquired 100% of Netherlands-based Hobré Instruments from private equity firm AAC Capital. Founded in 1978 in Purmerend, Hobré manufactures online gas analyzers, sampling systems, and measurement solutions for petrochemical refining, natural gas, biogas, steel, metal refining, and food processing. Financial terms were not disclosed, and EC M&A acted as exclusive financial advisor to the seller.¹⁶

The acquisition provided SUPCON with proprietary online analyzer technologies, a product category that carries higher gross margins than commodity transmitters and offers established customer relationships across Western Europe. However, because SUPCON did not disclose the purchase price or Hobré's standalone revenue and earnings contribution, the financial return on the transaction cannot be independently verified. Cross-border acquisitions of European instrument makers by Chinese industrial companies frequently encounter integration friction and elevated operating overhead. Without detailed segment disclosures, the net financial impact of the Hobré integration remains unverified.

The credibility question. Under Cui Shan, management has structured corporate performance around explicit, quantifiable benchmarks. Evaluating whether those commitments align with operational reality forms the core test for the company's next phase of development.

↑ Back to Top

VII. The Industrial AI Transformation & Speculative Optionality

On June 5, 2024, in Singapore, SUPCON held a global product launch and unveiled TPT — a Time-series Pre-trained Transformer, presented as the first pre-trained model built specifically for the process industry. The pitch, delivered under the banner of pioneering industrial AI frontiers, was that plants would stop maintaining dozens of bespoke models for dozens of applications and instead run what the company described as a shift from "N models plus N applications" to "TPT plus one software" per facility, with early deployments in chlor-alkali plants, thermal power and petrochemicals.¹⁷ At the same event the company introduced a universal control system, later branded Nyx and built on a real-time cloud operating system, marketed as the first cloud-native general control platform for process industries.¹⁷¹⁵

What this actually is, in plain terms. A refinery generates an enormous stream of numbers over time — temperatures, pressures, flows, compositions, vibration signatures — from thousands of sensors, continuously, for decades. A time-series foundation model is trained on that kind of data the way a language model is trained on text: it learns what normal looks like, what usually precedes an upset, and how variables move together. The commercial promise is that instead of an engineer spending six weeks building a custom model to predict when a particular pump will fail, a pre-trained model can be adapted to that pump in an afternoon. Applied across alarm management, predictive maintenance and closed-loop optimisation, the claim is that operating decisions currently made by experienced humans can be made continuously by software.

Nyx is the other half of the architecture and answers a different question. A conventional DCS is a closed, hardware-bound box: control logic runs on dedicated cards, and adding capability means adding hardware. Nyx is presented as a universal control system built on a real-time cloud operating system, with the control functions virtualised so that a plant can add applications the way a phone adds apps rather than the way a refinery adds a rack.¹⁷¹⁵ If it works as described, the strategic implication is significant — it turns the control layer from a one-time hardware sale into a platform that can host software, including SUPCON's own models, for the life of the plant.

The obvious objection is also the industry's oldest one. Cloud-native and real-time deterministic control are, at first glance, in tension: the entire value of a DCS is that it never misses a cycle, and the entire history of cloud computing is about trading determinism for flexibility. SUPCON's answer involves running the real-time layer locally on a purpose-built operating system with cloud characteristics rather than in a remote data centre. Whether process customers accept that distinction is an empirical question that will be settled by operating hours, not by architecture diagrams.

The strategic logic behind the data asset is genuinely strong. SUPCON sits on process data most software companies cannot access — it has cited on the order of 100 exabytes of industrial data across more than fifty sub-industries.¹³ Data of that kind is not scrapeable. There is no public corpus of what a hydrocracker's bed temperatures did in the six hours before an upset. If any category of industrial AI has a defensible data moat, this is it.

A necessary caveat: possessing customer data is not the same as being permitted to train on it. Process operating data is among the most commercially sensitive information a chemical company holds — it encodes yield, catalyst performance and cost position. The extent to which SUPCON has contractual rights to use client data for model training across customers has not been broken out in public disclosure, and it is a material question for anyone underwriting the data-moat argument.

Now the proportionality check. In the first nine months of 2025, TPT generated RMB 154 million of revenue and the company's software annual recurring revenue reached RMB 76.91 million.¹³ Against RMB 5.654 billion of revenue over the same period, TPT was under 3% of the business. TPT2, launched in August 2025 as a time-series mixture-of-experts model driving an industrial agent platform, recorded 518 pre-orders, 112 completed sales and 3,268 online registrations on launch day.¹³ Those are respectable early-adoption numbers for enterprise software. They are not a business.

The robotics line is smaller still. SUPCON took a stake in the 浙江人形机器人创新中心 Zhejiang Humanoid Robot Innovation Center, developed process-industry robot solutions including quadruped and humanoid units, and in September 2025 began marketing a concept it calls industrial embodied intelligence — perception, cognition, decision-making and execution combined.¹³¹⁵⁷ New robot orders in 2024 came to RMB 167 million¹¹ — orders, not recognised revenue, against a group revenue base fifty times larger.

The envisioned uses are, to be fair, real problems rather than invented ones. A large chemical complex requires continuous physical inspection of equipment in zones that are hot, loud, and classified as explosion hazards. People walk those routes today with handheld instruments, and the work is both dangerous and monotonous — precisely the profile where automation has historically won. Valve turning during startup and shutdown, manual sampling of process streams, and thermal inspection of rotating equipment are all candidates. A humanoid form factor is defensible in this specific setting for a reason that does not apply in most robotics markets: process plants were designed around human bodies, with ladders, walkways and hand-operated valves, so a machine shaped like a person can use the existing infrastructure without rebuilding the plant.

That is the strongest version of the argument, and it should be held next to the weakest fact about it. Explosion-proof certification for mobile equipment in hazardous zones is a slow, expensive regulatory process, and a robot that cannot be certified for Zone 1 cannot go where the dangerous work is. Until certified units are operating in classified areas at commercial scale, this remains an R&D line item with a compelling story attached.

The falsification test that matters. The right question is not whether TPT is impressive technology. It is: what is this company's historical rate of converting technical milestones into revenue?

The answer is available and it is sobering. supOS, the industrial operating system that carried a structurally similar promise — a platform layer above control, turning plant data into applications — launched in 2017.¹ Nine years later, standalone industrial software and the S2B service platform together still account for under a fifth of main business revenue,⁷ and software ARR, the purest available measure of a genuine software annuity, was under RMB 80 million as of the third quarter of 2025.¹³ The 622 SaaS member clients on the PlantMate platform reported for 2024¹¹ represent a real community, and a small one relative to more than 39,000 customers.²

Nor is this pattern unique to SUPCON. Every major automation vendor announced an industrial IoT platform between 2014 and 2018, and the great majority of those platforms generated a fraction of the revenue projected for them. The industry-wide base rate for converting industrial platform announcements into material software revenue within five years is poor. That does not mean TPT will follow the same path — foundation models genuinely do change the economics of building process applications — but it does mean the prior should start low and be moved by evidence rather than by architecture.

This company builds architecture well and monetises it slowly. That is not a criticism of the engineering; it is a statement about the customer. A refinery manager's job is measured in unplanned shutdowns avoided. Asking that person to let a transformer model touch the control loop is asking them to accept a novel and unbounded risk in exchange for an efficiency gain they can partly capture with existing tools — advanced process control, after all, has been optimising these units with conventional model-predictive techniques for thirty years. Adoption in this market is gated by trust, and trust is earned in years of incident-free operation, not in launch-day pre-orders.

There is also a specific analytical trap to avoid here. A model that recommends and a model that acts are separated by an enormous commercial gulf. Advisory AI — flagging an anomaly for a human to investigate — is easy to sell, easy to deploy, and worth relatively little, because the customer still pays for the human. Closed-loop AI — software that adjusts the process itself — is worth a great deal and is exactly what plant safety culture is constructed to resist. Management's stated ambition, autonomous plant operation, sits at the second end. Most current deployments, by their nature and by the industry's risk tolerance, sit closer to the first. Investors reading adoption numbers should ask which kind is being counted.

The pace question, and the number that changes it. There is a counter-signal worth taking seriously, and intellectual honesty requires giving it full weight.

In the first quarter of 2026, SUPCON reported industrial AI revenue of RMB 184 million — 12.2% of quarterly revenue, and more in a single quarter than TPT generated across the first three quarters of 2025.¹⁴¹⁸ Deployments were cited at 贵州磷化 Guizhou Phosphate Chemical, 中天合创, Guangxi Huayi, PetroChina and 中国海油 CNOOC, spanning smart factories, green-hydrogen coupling and virtual plant applications.¹⁸ TPT has been extended from chemicals and petrochemicals into roughly thirteen industries including oil and gas, pharmaceuticals and food,⁷ with over 110 projects deployed across refining, coal chemicals, chlor-alkali and hydrogen, and a partnership with 华为 Huawei on integrated offerings intended to lower the adoption barrier.¹²

The pricing model matters as much as the revenue. An annual subscription charged per production unit, typically on three-to-five-year contracts, is exactly the structure that produces a software annuity if adoption holds — it scales with the customer's asset base rather than with project timing, and it renews. If that model sticks across a few hundred units, the revenue quality of this company changes materially.

Note, however, the definitional shift. "Industrial AI revenue" is a broader and newer category than "TPT revenue," and it was not separately classified in the company's financial reports before 2026.¹⁹ A step-change in a metric that has just been redefined and simultaneously written into management's own incentive plan warrants care. The claim is not that the number is wrong. It is that a single quarter of a newly-created disclosure line, in a category management is paid on, is thin evidence for a thesis that requires sustained conversion.

The cost of the bet. R&D spending has not flexed down with earnings. In 2025 the company spent RMB 951 million on research, or 11.79% of revenue; in the first quarter of 2026 it spent RMB 215 million, 14.28% of revenue, with 2,011 R&D staff making up 37.77% of the workforce and a portfolio of 853 patents including 711 invention patents.²

That is a deliberate choice to hold investment steady through a downturn, and it is defensible if the pivot works. It is also the direct arithmetic explanation of why net profit fell so much faster than revenue: fixed research spending against a shrinking top line is operating leverage running in reverse. In the first half of 2026, revenue fell 5.1% and net profit fell 56.2%³ — a ratio that has almost nothing to do with demand and almost everything to do with a cost base built for a larger company.

Investors should be clear about what they are funding. They are paying for an option on autonomous plant operation, and the option premium is at its most expensive precisely when the core business is at its weakest. That is not necessarily the wrong trade. It is, unavoidably, the trade on offer.

---

↑ Back to Top

VIII. The Playbook: Business & Strategy Lessons

Strip SUPCON down to its core transferable lessons, and four primary insights emerge — each generalizing well beyond Chinese process automation.

1. Counter-position on the incumbent's business model, not its product. SUPCON's early market gains did not stem from building a superior DCS hardware platform compared to Honeywell. Instead, they resulted from targeting the most profitable and vulnerable element of the foreign incumbent model: expensive, slow, and contractually rigid engineering service.

By pricing hardware below foreign alternatives and treating ongoing engineering support as a relationship builder rather than a stand-alone profit center, SUPCON placed incumbents in a strategic dilemma. Matching the offer meant cannibalizing high-margin global service revenue earned across all their other operating regions. For emerging-market challengers, the broader lesson is that an incumbent's highest-margin activity often represents its most vulnerable point, because defending it locally is costly while abandoning it globally threatens overall earnings. However, this competitive window eventually closes: once the challenger becomes the dominant incumbent, it inherits the exact same vulnerability. SUPCON's service and spare-parts margins now represent the exact profit pool a hungrier domestic competitor could attack.

2. Switching costs protect the base; they do not protect growth. This is the lesson SUPCON's financial performance between 2024 and 2026 demonstrates most clearly. Twenty-year equipment replacement cycles ensure that an installed customer base remains secure during sector downturns, providing genuine baseline operational stability. Yet almost all top-line expansion in process control depends on greenfield plant construction and discretionary capital upgrades, both of which are highly pro-cyclical.

Equating high switching costs with guaranteed defensive earnings mistakes customer retention for revenue growth. Switching costs provide a call option on a client's future capital spending rather than a floor under current earnings. As management's strategic pivot reflects, an installed base becomes a predictable annuity only when attached to recurring software subscriptions and ongoing services. This structural reality makes software annual recurring revenue a far more meaningful indicator of business health than aggregate market share. Indeed, the headline 45.1% market share figure offers little insight into price realization or underlying margin expansion inside those contract wins.

3. Institutions can outlast founders, and the test is unplanned. Most corporations claim operational depth beyond their visionary founder. SUPCON had that assertion tested involuntarily for three years while Chu Jian was in custody, and the institution endured: project engineering continued, customer relationships held, and domestic market share expanded.

The key takeaway lies in the organization's post-crisis governance structure. Rather than restoring the founder to chief executive control upon his release, SUPCON retained him as a strategic advisor, recruited seasoned professional operators from international competitors, and tied performance to explicit management metrics. Founder-led companies that navigate a governance crisis and transition successfully to professional management remain rare. However, evaluating overall governance strength requires weighing this operational resilience against historical facts, including nominee shareholding arrangements and the founder's criminal conviction for destroying accounting records.

4. Indigenisation is a tailwind with an expiry date. Domestic substitution, or 国产替代, provided a powerful structural tailwind for SUPCON, transforming a competitive battle against better-resourced foreign incumbents into a policy-assisted market share transition.

Yet the logic of industrial localization extends beyond displacing foreign vendors. Its core objective is domestic supply security, which is served by maintaining multiple capable domestic vendors rather than relying on a single supplier. With market share reaching 68.5% in the chemical sector, domestic substitution in primary industries is largely complete. Future competition is predominantly domestic, price-sensitive, and conducted before state-owned enterprise buyers whose procurement policies intentionally limit vendor pricing power. The same policy environment that drove SUPCON's initial expansion now establishes structural boundaries on its domestic margins — a pattern common across Chinese industrial technology, where market share gains frequently reflect a completed one-time transfer rather than a widening competitive moat.

A fifth lesson embedded in the corporate disclosure record highlights the importance of consistent analytical definitions. A company reporting RMB 2.65 billion in industrial software revenue in one presentation while classifying standalone industrial software as under a tenth of main business revenue in another¹¹⁷ is not misrepresenting facts, but its bundling practices obscure the underlying product mix. The introduction of "industrial AI revenue" as a distinct reporting line in the same period it became part of executive incentive targets¹⁹ reflects a similar pattern. Neither practice is disqualifying, but both emphasize the need for investors to focus on the most transparent, recurring metrics available.

This leaves SUPCON at a critical junction: a market leader navigating a contracting domestic capital expenditure cycle, investing heavily in an unproven industrial AI pivot, with two competing views of its long-term trajectory.

↑ Back to Top

IX. Strategic Position, Risk Radar & Bull vs. Bear Stress Test

The risk radar, in order of what actually matters.

The dominant risk is not exotic. Chinese basic chemicals and refining are in an overcapacity-driven capital spending contraction, and SUPCON's revenue is levered to that spending in the most direct way possible. When Sinopec, PetroChina or a private complex defers a new ethylene unit or stretches a turnaround interval, SUPCON's order book absorbs it within quarters. This has already happened and is still happening, as the first-half 2026 results show.³ The mechanism has no clever workaround; it ends when downstream capex cycles turn, and the timing of that is not in management's control.

The depth of this cycle deserves emphasis because it is not an ordinary demand dip. China built basic chemical capacity through the early 2020s at a pace that has left large parts of the industry — polyolefins, aromatics, refining margins broadly — running below the returns that justify new investment. When an industry's marginal producer is losing money, nobody sanctions a new complex, and the existing ones cut discretionary maintenance first. Control systems sit squarely in the discretionary bucket for brownfield work. Recovery therefore requires not a stabilisation in chemical demand but a genuine absorption of overcapacity, which historically takes years rather than quarters.

Second, supply chain and geopolitics. Real-time control boards depend on specialised microcontrollers, FPGAs and the design software used to build them. Export controls aimed at advanced semiconductors and electronic design automation tools create a genuine dependency risk for a company whose product must be certified, deterministic and available in volume for twenty years of spare-parts support.

SUPCON has been redesigning around domestically controllable architectures, which is the correct strategic response and is consistent with the policy environment that helped build the company. But re-qualifying safety-rated hardware is slow and expensive — every functional safety certification has to be redone — and any transition period carries execution risk that lands on exactly the product line the installed base depends on.

Third, the mirror image: Western scrutiny of Chinese industrial software running critical energy infrastructure. SUPCON's capital plan allocates money to European and Gulf expansion.[^16] In the Gulf and Southeast Asia this is largely a commercial contest, and one SUPCON can plausibly win on price and responsiveness. In Western Europe and US-aligned jurisdictions, a Chinese-controlled system sitting in a refinery's control room is a national security conversation before it is a procurement conversation. That materially caps the realistic addressable market for the international strategy, and it is a structural constraint rather than an execution one — no amount of good selling changes it.

There is a cybersecurity dimension to that same point that cuts in both directions. Industrial control systems are among the highest-value targets in any national infrastructure, and the regulatory attention paid to who supplies them has increased everywhere. Domestically this helps SUPCON, because a Chinese refinery increasingly prefers a domestic control vendor for exactly the reasons a European one would prefer a European vendor. Internationally it is the constraint described above. The asymmetry is structural and unlikely to reverse.

Fourth, and most immediate for anyone reading the reports: cash conversion. Negative operating cash flow of RMB 715 million in the first half of 2026, an asset-liability ratio of 44.80% up 2.49 points, and a large shipped-goods balance awaiting acceptance³⁷ are the fingerprints of a project business under stress.

None of these individually signals distress — the balance sheet remains far healthier than at the IPO, when the asset-liability ratio exceeded 60%,⁹ and half-year cash flow in a project business with seasonal collections is a weak signal on its own. But the direction matters. The company returned 49.47% of 2025 net profit as dividends, or 117.17% of net profit once buybacks are included.⁴ That figure deserves a moment: returning more than all of a year's earnings to shareholders while simultaneously funding a capital-intensive pivot and reporting negative operating cash flow is a deliberate choice. It signals confidence in the balance sheet and a desire to support the share price during a drawdown. It also consumes the cushion that funds the option management is asking investors to underwrite. Both things are true at once, and an activist would press hard on the tension between them.

A related governance point an activist would raise: the 2026 equity incentive plan granted 17 million restricted shares — about 2.15% of share capital — to 1,268 employees at RMB 57.98 per share, against buybacks executed during 2024 and 2025 at prices ranging from RMB 34.92 to RMB 51.49.¹⁹ Repurchasing stock and then reissuing it to employees at a price near or above the repurchase range is defensible practice, and it is also a transfer whose economics depend entirely on whether the performance conditions are demanding. Given how far the 2026 revenue trigger sits above the current run rate, the conditions look demanding on their face — which is the more shareholder-friendly reading, provided the targets are not subsequently reset.

The bear case, argued properly. A skeptical investor would frame it like this. SUPCON is a cyclical industrial hardware company that has learned to describe itself in software vocabulary.

Its core Chinese DCS market is approaching practical saturation at 45% share overall and near 70% in its best vertical² — from here, incremental share is expensive and the market itself is shrinking. Two-thirds of earnings have already disappeared, and gross margin has fallen despite an alleged mix shift towards software, which is the opposite of what a genuine mix shift produces. If software were becoming a larger share of a lower-margin base, arithmetic alone would push the blended margin up.

The international strategy consumed USD 290 million of GDR proceeds and produced overseas revenue that declined in 2025.[^16]⁷ R&D at 12 to 14% of revenue is being sustained through the trough on a bet whose historical precedent — supOS in 2017 — took the better part of a decade to become a modest revenue line.

Then there is the target-setting. The equity incentive plan set 2026 revenue targets of RMB 10.5 billion with a RMB 9.5 billion trigger,¹⁹ against 2025 revenue of RMB 8.073 billion⁴ and a first half of 2026 that was down again.³ Hitting even the trigger would require a second-half acceleration with no visible support in the order trends. When management sets targets that the current run rate cannot plausibly reach, either the plan is aspirational theatre or the operating assumptions embedded in it are optimistic — neither reading flatters guidance discipline. And the new "industrial AI" line that carries the equity story was created in the same season it became a compensation metric.¹⁹ A short-seller would put those two facts in the same sentence and let the reader draw the inference.

The bull case, argued properly. The counter-argument is not weak. Start with the asset that cannot be bought: more than 39,000 process-industry customers and over 100,000 deployed control systems,²¹³ each one a physical connection into a plant's operating data and a standing relationship with the people who authorise spending. That is a distribution channel no software startup can replicate and no foreign vendor can match inside China.

If autonomous plant operation becomes real — if models genuinely close the loop on optimisation and maintenance — the company that already owns the control layer captures the value, because the control layer is where decisions get executed. Everyone else in the industrial AI landscape has to persuade a plant to install something new next to a system they already trust. SUPCON has to persuade them to switch on a feature inside it.

The data advantage is not rhetorical, the subscription structure priced per production unit on multi-year contracts is the right commercial shape, and the first quarter of 2026 provided the first hard evidence that conversion may be accelerating rather than merely being promised.¹⁴ Meanwhile, cycles turn. A company holding 45% share when domestic chemical capex recovers will see disproportionate operating leverage on the way up, because the engineering bench and the research spending have already been paid for through the trough. The bull's strongest claim is not that the AI story is proven. It is that an investor is being offered a cyclical recovery with a free option attached — and that the option is being priced by a market that has just watched two years of falling earnings.

Where the frameworks land. In Helmer's terms, SUPCON's clearest power is switching costs, and it is strong but bounded to the installed base — it defends revenue that already exists rather than revenue the company hopes to win.

Scale economies in field service are real domestically and effectively absent internationally, where SUPCON is the sub-scale player rather than the dominant one. Cornered resource has partial support — process domain knowledge plus proprietary operating data — but the historical monetisation rate argues it has functioned as a technical asset more than an economic one. Counter-positioning was decisive in the 1990s and is now largely spent; the incumbents SUPCON counter-positioned against are no longer the marginal competitor in its core market. Network economies and branding do not meaningfully apply in a business where every sale is an engineered project. Process power — accumulated organisational capability in delivering complex projects repeatably — is arguably present and underrated, and it is precisely what would allow the company to execute a hundred simultaneous AI deployments if the demand ever arrives.

On Porter, buyer power remains the binding constraint and rivalry is intensifying as the domestic pot shrinks; substitution and new-entry threats stay genuinely low. The net picture is of a business with one very strong defensive power, one underrated operational power, and several powers that have either expired or never applied.

Against peers, the comparison is instructive. Honeywell, Emerson, Yokogawa, ABB and Siemens all run higher-margin, more software-weighted automation businesses with genuinely global installed bases and decades of accumulated Western customer trust. SUPCON's advantages against them are cost, speed of response and home-market policy; its disadvantages are brand, geopolitics and a much shorter reference history outside China. Domestically, 和利时 Hollysys and others compete hardest in power, rail and nuclear — the verticals SUPCON entered later and where its share is lowest.⁸

The realistic framing is this: SUPCON has already won its home market and is now attempting two considerably harder things simultaneously — going global and going software — while the home market that funds both of them contracts.

The calibrated verdict. Taking each thesis claim in turn and weighing it against the company's own record produces three narrowed conclusions rather than three verdicts.

The switching-cost moat survives the historical test, but in a materially narrower form than it is usually stated. It protects retention, spare parts and service revenue on an installed base of more than 100,000 systems. It has demonstrably failed to protect margins, growth or earnings through a downturn, and two years of evidence say so unambiguously.

The international vector is credentialed but unproven. The 2025 decline in overseas revenue is genuine disconfirming evidence that must sit alongside 2024's triple-digit growth, and the honest reading of the pair is that project lumpiness currently dominates any underlying trend.

The industrial AI vector rests on a real data advantage and a real distribution advantage, weighed against a company track record of slow commercialisation stretching back to supOS in 2017, with one encouraging quarter recorded in a metric that was defined months earlier and written into management's pay.

None of these claims is rejected outright. All of them are narrower than the promotional version. And each has a specific, observable test attached — which is the most useful thing an investor can carry out of this story.

---

↑ Back to Top

X. Epilogue & What to Watch: The 3 Core KPIs

Chu Jian's trajectory—borrowing personal funds to launch a domestic automation venture, enduring three years in custody, and returning to see the company capture nearly half of a control-systems market once dominated by foreign vendors—represents a notable founding arc. But that initial chapter is now complete. The question defining the next decade has little to do with domestic control-system market share, which is near saturation. Instead, it centers on whether Chairman Cui Shan and his executive team can convert a hardware-centric franchise into a high-margin software enterprise while downstream process industries navigate a prolonged capital expenditure downturn.

The empirical evidence available as of September 2026 supports neither immediate triumph nor outright dismissal. It demands patience and focus on three key operating indicators, as short-term headline results primarily reflect cyclical industrial swings rather than long-term strategic shifts.

One: Software annual recurring revenue and industrial AI revenue share. This dual metric addresses the core strategic question. Annual recurring revenue (ARR) is the most transparent indicator of software adoption in SUPCON's disclosures, unaffected by project milestones, installation timing, or hardware bundling. Tracking ARR alongside industrial AI revenue as a percentage of total sales will reveal whether the growth reported in early 2026 can be sustained across full fiscal years. If ARR compounds steadily while industrial AI contributions expand toward management's targets, the software transformation is taking hold. If AI revenue exhibits lumpy project dynamics while ARR stalls, the slow monetization pattern of the supOS platform will repeat.

Two: Overseas revenue growth aligned with consolidated gross margin. Evaluating the international strategy requires monitoring top-line expansion and margin performance together. International revenue growth alone can be bought through aggressive pricing, while gross margins can be preserved by avoiding competitive tenders. A successful global expansion requires overseas revenue to grow consistently over multiple years toward management's international targets while group gross margins stabilize or expand. If international sales rise while overall gross margins continue to shrink, SUPCON is acquiring overseas market share at the expense of profitability.

Three: Gross and operating margin trajectory relative to R&D intensity. SUPCON has sustained research spending between 12% and 14% of revenue throughout the industry downturn, reflecting a deliberate capital allocation toward technology development. The test is whether this spending generates margin expansion when downstream capital deployment recovers. The key signal will be gross margins stabilizing and operating margins inflecting upward while R&D intensity moderates as a percentage of rising revenue. That combination would demonstrate that research investments are producing operational leverage. Conversely, if margins remain compressed during a cyclical recovery, it will indicate that pricing pressures are structural, proving that the competitive moat is narrower than headline market share suggests.

---

References

1. 褚健：一位科学家的创业史 — 中国经济周刊 / 人民网, 2020-12-15 ↩↩↩↩↩↩↩↩↩↩↩↩↩

2. 中控技术一季度工业AI业务加速放量，核心产品市占率连续15年蝉联第一 — 同花顺财经, 2026-04-27 ↩↩↩↩↩↩↩↩↩

3. 中控技术(688777.SH)2026年中报净利润为1.55亿元、较去年同期下降56.20% — 界面新闻, 2026-08-29 ↩↩↩↩↩↩↩↩↩

4. 中控技术：2025年营业收入80.73亿元 — 证券市场周刊, 2026-04-20 ↩↩↩↩↩↩

5. 浙大原副校长褚健一审被判处有期徒刑三年三个月 — 科学网, 2017-01-16 ↩↩

6. Chinese Company Zhejiang Supcon Technology Lists Its GDRs on SIX Swiss Exchange — SIX Group, 2023-04-17 ↩↩

7. 中控技术：流程工业自动化龙头，DCS很稳，工业AI故事正热 — ZAKER新闻, 2026 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

8. 2024年国内主要DCS厂家 — 剑指工控 ↩↩↩

9. 中控技术冲击IPO：实控人褚健仅担任顾问 核心专利存风险 — 澎湃新闻, 2020 ↩↩↩↩↩↩↩↩↩↩

10. 中控技术：11月24日在科创板上市，股票代码688777，发行价格35.73元/股 — 每日经济新闻, 2020-11-22 ↩

11. 中控技术：2024年营收91.39亿元 海外收入同比增长超118% — 中国证券报·中证网, 2025-04-01 ↩↩↩↩↩↩↩

12. 从年增长30%到业绩双降，中控的工业AI转型承压 — 东方财富财富号, 2026-07-13 ↩↩

13. 中控技术披露三季报："ALL in AI"业绩蓄势，工业具身智能重构流程工业 — 证券时报, 2025 ↩↩↩↩↩↩↩↩

14. 中控技术2026年一季度业绩披露：工业AI业务成增长亮点 — 智能制造网, 2026-04 ↩↩↩

15. Zhejiang Supcon Technology Co., Ltd. Investor Relations — SUPCON ↩↩↩↩

16. Hobré Instruments has been sold to SUPCON — EC Mergers & Acquisitions, 2023 ↩

17. Building Industrial Intelligent Engine, Reshaping a New Paradigm of Industrial Applications: SUPCON Unveils Groundbreaking Time-series Pre-trained Transformer (TPT) — PR Newswire APAC, 2024-06-05 ↩↩↩

18. 中控技术股份有限公司公告披露 — 上海证券交易所 ↩↩

19. 中控技术股权激励设置工业AI收入目标 过去财报未对该收入单独分类 — 新浪财经 / 每日经济新闻, 2026-01-25 ↩↩↩↩↩