Range Intelligent Computing Technology Group: The Story of China's AI Data Center Titan

I. Introduction & Episode Roadmap

On September 13, 2022, a founder in her early sixties stood on the bell-ringing platform of the Shenzhen Stock Exchange. The ticker on the wall read 300442—a code that, for the preceding seven years, had belonged to a modestly profitable manufacturer of aseptic milk-carton filling machines. That morning the ticker symbol remained unchanged while the operating entity behind it was completely transformed. 普丽盛 PuLiSheng—Shanghai Precise Packaging—ceased operations as a standalone packaging company, replaced by 润泽智算科技集团股份有限公司 Range Intelligent Computing Technology Group, one of China's largest wholesale data center operators. The founder holding the mallet, 周超男 Zhou Chaonan, had just executed a rare transaction: a complete backdoor listing on China's ChiNext growth board under the country's registration-based system.¹

This landmark listing marked an unusual origin story for a primary physical foundation of China's artificial-intelligence infrastructure expansion. While typical AI infrastructure narratives center on semiconductor designs, foundation models, or research laboratories, Range's foundation rested on land acquisition—specifically, several square kilometers in 廊坊 Langfang, an industrial city in Hebei province located roughly 50 kilometers southeast of Beijing. The site was close enough to capture spillover data demand from the capital, yet far enough to access lower-cost land and municipal electricity allocations without strict urban power rationing.

Zhou's core thesis, initiated around 2009, depended on a simple premise: China's expanding internet sector would require vast host facilities, while primary hubs like Beijing and Shanghai would eventually restrict new builds due to land, power, and carbon constraints. By acquiring land rights and securing power capacity early, Range built at a scale few competitors underwrote, initially operating as a capital-intensive real-estate vendor before the surge in generative AI reshaped industry demand.

Range operates across two primary business segments. The legacy Internet Data Center (IDC) division provides wholesale colocation, supplying the building, power, cooling, and operational infrastructure while telecom carriers and hyperscalers lease cabinet space. The newer Artificial Intelligence Data Center (AIDC) segment uses what management calls a "self-invested, self-built, self-owned, self-operated" model. In this higher-tier framework, Range purchases servers and networking hardware directly to rent out computing clusters rather than bare server racks.² In 2025, AIDC generated RMB 2.51 billion in revenue, accounting for 44.24% of total sales—a business line that was negligible three years prior.³

On the surface, 2025 financial results appeared strong, with revenue rising 29.99% to RMB 5.674 billion and net profit attributable to shareholders surging 182.07% to RMB 5.050 billion.² However, core operations accounted for little of that earnings increase. Excluding non-recurring items, underlying net profit was RMB 1.901 billion, up 6.87% year-over-year.² The remaining RMB 3.15 billion gap resulted largely from selling a mature data center asset into a publicly listed infrastructure real estate investment trust (C-REIT). Whether this transaction represents a repeatable capital-recycling model or a temporary earnings enhancement is a key question for the company's long-term valuation.

Additional operational challenges persist. Range missed the profit targets set during its backdoor listing, requiring controlling shareholders to return and cancel 87 million shares as compensation.⁴ Concurrently, the company is expanding internationally, winning a land tender from the Hong Kong government and committing to a development project in Indonesia. It also carries significant debt to support its target of 6 gigawatts in operating capacity, while its revenue remains concentrated among a small group of large domestic buyers.

This analysis examines Range's evolution across several key areas: its initial land and power strategy; the wholesale carrier model that drove early capacity utilization; the structure and consequences of its backdoor listing; its liquid-cooling technological shift for AI workloads; its 2025 financial structure and C-REIT monetization pipeline; a competitive assessment using Helmer's 7 Powers and Porter's Five Forces; an evaluation of management execution against public targets; and the core operational metrics determining its future trajectory.

The company's arc began when its founder left a state-sector career at age forty to launch an independent venture.

---

↑ Back to Top

II. The Langfang Vision & Founding Origins (2007–2012)

Zhou Chaonan was born in 1960 in 衡阳 Hengyang, Hunan. Her father named her 超男—literally "surpassing men"—in the spirit of the era's slogan that women hold up half the sky.⁵ She has often described herself as embodying the regional Hunanese temperament: resilient, patient, and persistent through adversity.

In 2000, at age 40, she left a secure state-enterprise career to launch 天童通信网络有限公司 Tiantong Communications Network in Beijing.⁵ The business focused on civil telecommunications infrastructure, specifically constructing underground conduits, utility trenches, and shared pipelines to carry fiber-optic cables beneath municipal streets. Over the following decade, the company laid more than 2,000 kilometers of conduit across twelve regions, saving state telecom carriers an estimated RMB 200 million by consolidating physical infrastructure into shared corridors.⁵

That initial venture established three operational insights that would guide her subsequent strategy. First, in Chinese telecommunications, the primary bottleneck was rarely server or networking technology, but physical right-of-way, municipal permits, and land corridors. Second, state-owned telecom carriers operated less as typical buyers than as institutional gatekeepers with whom infrastructure providers had to deeply integrate. Third, shared, multi-tenant physical assets generated superior unit economics compared to fragmented single-user builds, allowing the shared builder to capture structural operating margins.

Buying dirt before the demand arrived

In 2009, Zhou established 润泽科技发展有限公司 Range Technology Development, becoming one of the earliest domestic developers to plan data centers at an industrial scale.⁵ Selecting a site in the Langfang Economic and Technological Development Zone, she announced the development of the 润泽国际信息港 Range International Information Harbor—designed from inception as a multi-building data center campus rather than an isolated standalone facility.⁵

This campus model was highly contrarian in 2009. Domestic cloud computing was in its infancy; 阿里云 Alibaba Cloud had been launched only that year. China's third-party colocation market was fragmented, dominated by regional vendors leasing individual floors in city office towers to supply enterprise IT departments. These legacy facilities required minimal upfront capital and generated quick, modest cash flows, but lacked scalability.

Zhou took the opposite approach: acquiring land titles outright, funding heavy civil engineering, and securing high-voltage grid capacity years ahead of committed tenant demand. The strategy targeted a customer profile that had not yet fully materialized in China: hyperscale internet platforms seeking turnkey, megawatt-scale facilities.

Why Langfang, and why not Beijing

The economic logic behind selecting Langfang relies on fundamental physical constraints. Data centers effectively convert bulk electric power into thermal output and rentable space. They demand extensive real estate, heavy electrical grid capacity, and continuous industrial cooling, while generating few direct local jobs per square meter. Consequently, tier-one municipal authorities naturally restrict large-scale data center construction. Beijing's municipal land is prioritized for commercial offices and housing, its power grid serves residential and high-value industrial consumers, and municipal energy-consumption caps tightened throughout the 2010s.

However, core computing demand from digital platforms, financial institutions, and government ministries remained anchored in Beijing. Because fiber-optic signal propagation latency is driven by physical distance, locating a facility 50 kilometers away adds approximately one millisecond of round-trip network latency—an imperceptible difference for cloud storage, batch analytics, AI training, and general enterprise workloads.

This geographic positioning created a structural arbitrage: hosting high-density power loads in Hebei while serving Beijing-based end users. Langfang fell directly within the 京津冀 Beijing-Tianjin-Hebei coordinated development corridor, a national regional integration policy that incentivized Hebei municipalities to absorb non-core industrial functions from Beijing. A tax-paying, land-intensive infrastructure park aligned directly with municipal targets, giving Range access to power and land allocations that were increasingly rationed in the capital.

The patience problem

The primary vulnerability of this land-first strategy was capital intensity. Developing a data center campus requires substantial upfront expenditures—land premiums, dedicated power substations, building shells, industrial chillers, backup generators, and uninterrupted power supply (UPS) systems—years before generating tenant revenue. Zhou financed the initial Information Harbor buildout during an era when domestic internet traffic was expanding rapidly, but wholesale long-term leasing by single hyperscale tenants had not yet become industry standard.

By the time wholesale colocation gained widespread market adoption, the Langfang park had expanded to twelve data center buildings housing more than 40,000 standard server racks, establishing it as the largest single third-party data center campus in China.⁵ Completing that initial phase required roughly fourteen years of continuous construction and capital deployment following the original land acquisition—highlighting the multi-year development timelines inherent to large-scale wholesale infrastructure.

For investors, Range's founding phase illustrates the fundamental asset dynamic underpinning its valuation: its initial competitive moat was built on early-stage optionality across secured land and power capacity in strategic corridors. Because such land and power reserves depreciate as existing capacity is filled, long-term growth requires continuously replicating this site-acquisition strategy in new regional hubs at higher prevailing land and energy costs.

While Range successfully secured physical capacity in Langfang, it faced an immediate operational challenge: filling a multi-building campus without an established enterprise sales force.

↑ Back to Top

III. The Wholesale Model & Grid Access Infrastructure (2013–2019)

Data center business models generally fall into two main categories: retail colocation, which functions like a hotel with many small tenants, active sales teams, higher churn, and continuous occupancy management; and wholesale infrastructure, which operates like a cargo port with major counterparties, multi-year contracts, and minimal marketing overhead. Range selected the wholesale port model.

The primary distribution engine relied on state telecom carriers. Under what the company calls its cooperative operating model, Range contracts with 中国电信 China Telecom, 中国联通 China Unicom, or 中国移动 China Mobile. Range supplies the climate-controlled, uninterrupted-power facility environment 24 hours a day, 365 days a year, while the carrier provides bandwidth and telecommunications services, billing the end customer directly.² In practice, Range's principal contracting counterparties have been the Beijing branch of China Telecom and the Hebei branch of China Unicom, with ultimate end users including major Chinese internet platforms.⁶ Alongside this carrier-led channel, Range also operates a direct-signing track, contracting directly with large enterprise end users.²

Why renting through a state monopoly is smarter than it sounds

Ceding the end-customer relationship to state-owned carriers might initially appear to be a strategic compromise, as it sacrifices direct brand visibility, pricing transparency, and upselling opportunities. However, an analysis of the operating cost structure reveals the strategic rationale.

A retail colocation operator must maintain a nationwide sales force, solution architects, marketing budgets, and churn-management operations—creating a permanent overhead drag equal to several percentage points of revenue. Range avoided these expenses. State carriers already maintained deep enterprise relationships, dedicated account managers, credit assessment systems, and billing infrastructure. The carriers required physical capacity faster than they could build it, particularly in locations with constrained energy quotas, while Range required distribution.

This partnership structure also altered the credit profile of Range's revenue. A wholesale master contract with a provincial branch of China Telecom effectively functions as a sovereign-backed receivable. For a company financing multi-hundred-million-renminbi data center builds through bank debt, holding state-backed tenant credit significantly lowers borrowing costs.

The trade-off remains straightforward: Range does not own the ultimate customer relationship on carrier channels, and lease renewals require negotiating with sophisticated counterparties that understand the underlying facility operating costs. The model trades gross margin for volume and rapid capacity absorption, leaving pricing power with carrier buyers during periods of market oversupply.

Power as the real product

While data center operators are often viewed as real estate providers selling floor space, their core product is energized capacity.

The binding physical constraint in wholesale colocation is rarely square footage; it is grid power delivery and the acquisition of local energy-consumption quotas (能耗指标). In China's regulatory framework, these quotas are administratively allocated, strictly capped to support national 双碳 dual-carbon objectives, and heavily restricted near tier-one cities. Securing approved, unconsumed energy quotas in prime industrial corridors provides an asset reserve analogous to a mining concession.

Range's quota disclosures function as a key leading operating metric. In 2025, the company reported that its energy-consumption indicators grew 30% year-over-year, framing this quota expansion as the foundation for future capacity growth.² In wholesale infrastructure, approved power allocations serve as an early operational signal, with energized infrastructure following two to three years later and revenue generation occurring upon tenant deployment.

Range also constructed its own dedicated high-voltage substations to feed its campuses rather than relying solely on shared municipal utility lines. This direct grid infrastructure improves operational reliability while reducing power costs, as bulk industrial tariffs improve at higher transmission voltages. Additionally, Range secured direct power-purchase agreements with energy generators and participates in national wholesale electricity markets alongside its long-term local grid distribution contracts.² Because electricity represents the single largest operational cost in legacy IDC colocation, energy procurement functions as a primary competitive driver.

From one campus to a national grid

During the late 2010s, Range's focus shifted from filling its initial Langfang campus to scaling its model nationally. China's policy landscape provided the blueprint through the 东数西算 Eastern Data, Western Computing initiative—a national plan directing data center construction into designated regional clusters to balance energy use and shift non-real-time computing to renewable-rich western regions.

Range expanded its multi-building campus model across these designated hubs. By 2025, the company had established seven artificial intelligence infrastructure clusters across six core domestic regions: Beijing-Tianjin-Hebei at Langfang, the Yangtze River Delta at 平湖 Pinghu, the Greater Bay Area at 佛山 Foshan and 惠州 Huizhou, the Chengdu-Chongqing economic zone at 重庆 Chongqing, the northwest at 兰州 Lanzhou, and Hainan at 儋州 Danzhou—representing a combined planned scale of approximately 5.4 gigawatts.²

This footprint followed a deliberate geographical pattern: securing sites adjacent to major demand centers inside policy-approved corridors prior to land price inflation. The Danzhou facility in Hainan served a distinct role as an international computing gateway targeting Southeast Asian connectivity, anticipating Range's subsequent overseas expansion.

By 2020, Range had established a highly structured business model characterized by long-term carrier contracts, sovereign-grade customer credit, secured energy quotas, and multi-year construction visibility. However, remaining a private enterprise constrained its capital structure. As Range held tens of billions of renminbi in physical assets and planned further gigawatt-scale developments, the gap between its expansion capital requirements and available debt financing made access to public capital markets an operational necessity.

↑ Back to Top

IV. Capital Structure, M&A & The ChiNext Backdoor Merger (2020–2022)

Every capital-intensive infrastructure company eventually arrives at the same crossroads: its operational assets are proven, its development pipeline is expansive, and its balance sheet is out of room. Range reached this threshold in 2020.

The conventional path forward was a domestic initial public offering, but that route introduced significant timing and regulatory uncertainty. China's A-share listing regime was in the middle of a major transition—the registration-based system had been launched on the STAR Market in 2019 and expanded to ChiNext in 2020. Approval queues remained long, and regulatory evaluation of a heavily leveraged data center operator on an exchange favoring light-asset, high-growth technology firms was untested. Data center operators occupied an awkward middle ground in Chinese financial regulation: too asset-heavy to fit the traditional technology profile, yet too specialized to be classified as real estate.

The alternative was securing a listed shell entity.

Anatomy of the PuLiSheng restructuring

Shanghai Precise Packaging, known domestically as PuLiSheng, was founded in 2007 and listed on ChiNext in 2015, manufacturing liquid-food packaging machinery and paper-aluminum composite aseptic packaging materials.¹ By the late 2010s, it had become a mature business with eroding profitability—making it an ideal shell asset: clean, listed, and held by owners willing to exit.

The resulting deal structure combined a major asset swap with new share issuance. PuLiSheng's entire operating business, excluding its equity stake in Italian machinery subsidiary COMAN, was valued as the outgoing asset at approximately RMB 602 million. Range Technology's 100% equity was appraised as the incoming asset at RMB 14.268 billion. The RMB 13.666 billion valuation difference was settled by issuing approximately 720 million new shares to Range's fourteen shareholders at RMB 18.97 per share, accompanied by a private placement raising up to RMB 4.7 billion from up to 35 institutional investors.⁷

Regulators approved the transaction on May 27, 2022, marking the first backdoor restructuring and listing cleared on ChiNext following the October 2019 regulatory revision allowing such transactions on the board.⁷ Trading under the new entity began on September 13, 2022, under the Range name, with 京津冀润泽（廊坊）数字信息有限公司 Jing-Jin-Ji Range (Langfang) Digital Information holding 71.27% of the listed company and Zhou Chaonan serving as the controlling person.¹

For a company facing capital constraints, the transaction's value proposition was compelling. Combining an asset swap with a targeted private placement provided public equity, fresh cash, and an acquisition currency in a single process on a predictable timeline, avoiding the delays of an IPO queue. Crucially, it established a public valuation marker of RMB 14.268 billion for an asset base that had previously been appraised primarily by commercial lenders.

Did the valuation hold up?

Evaluating whether the RMB 14.268 billion valuation represented a fair market price or a founder-favorable appraisal requires context. Chinese backdoor restructuring valuations are established by administrative appraisal rather than an open order-book auction, meaning the price is negotiated between related parties and validated by an independent appraiser—a structurally less dynamic price-discovery mechanism than a traditional book-built IPO.

Comparing Range against peer benchmarks offers additional perspective. Its primary domestic peers included listed wholesale operator 万国数据 GDS Holdings, retail-wholesale operator 世纪互联 VNET Group, Chindata Group (which Bain Capital took private in 2023), and A-share peers 宝信软件 Baosight Software and 光环新网 Sinnet. Against these comparables, Range's listing valuation implied a lower multiple of trailing earnings than offshore-listed peers, which were valued on enterprise value to EBITDA and pipeline power capacity rather than net income. However, unlike GDS or VNET at the time, Range consistently generated positive net income and operating cash flow.

The key distinction between Range and its peer group was structural rather than financial. Major competitors expanded substantially through leaseholds and build-to-suit agreements, incurring long-term lease obligations. Range pursued an asset-ownership model, purchasing land titles outright, constructing facility shells, and retaining underlying real estate. While this asset-heavy strategy yielded a larger balance sheet, lower reported returns during early construction phases, and a slower initial ramp, it ensured that long-term terminal value accrued directly to equity holders. This ownership foundation also laid the groundwork for the company's subsequent C-REIT monetization strategy. Comparing valuation multiples across operators without accounting for underlying asset ownership obscures fundamental business model differences.

The earn-out, and what actually happened

Backdoor restructurings of this nature in China require a binding profit commitment (业绩承诺) from selling shareholders, backed by share-forfeiture mechanisms if targets are missed. Range's controlling shareholders committed to achieving net profit after non-recurring items of at least RMB 612 million in 2021, RMB 1.098 billion in 2022, RMB 1.794 billion in 2023, and RMB 2.095 billion in 2024—a cumulative target of RMB 5.599 billion over four years.⁷

Performance exceeded targets in the early phase of the agreement. Range reported RMB 711 million in net profit in 2021 and RMB 1.153 billion in 2022, exceeding its two-year cumulative target by RMB 154 million, or a 109% fulfillment rate. However, growth slowed in subsequent years. Net profit reached RMB 1.703 billion in 2023 against the RMB 1.794 billion target, with 2024 trailing projections further. Over the full four-year commitment period, cumulative net profit reached RMB 5.239 billion against the promised RMB 5.599 billion—achieving a 93.56% completion rate and leaving a shortfall of approximately RMB 360 million.⁸

The contractual compensation mechanism functioned as designed. Four selling entities—Jing-Jin-Ji Range, 北京天星汇市政工程有限公司 Beijing Tianxinghui Municipal Engineering, 润徽（廊坊）企业管理合伙企业 Runhui (Langfang) Enterprise Management Partnership, and 共青城润翔投资合伙企业 Gongqingcheng Runxiang Investment Partnership—surrendered 87,138,076 shares (representing 5.06% of total equity) for a nominal consideration of RMB 1.00. The Shenzhen branch of China Securities Depository and Clearing cancelled these shares on July 10, 2025, reducing total share count from 1,721,481,649 to 1,634,343,573, while cash dividends paid on those shares during the commitment period were returned to the company. This cancellation mechanically adjusted restated 2024 earnings per share upward from RMB 1.04 to RMB 1.10.⁴

Market observers view this outcome through two distinct lenses. The constructive interpretation highlights that the contractual governance mechanism operated effectively, fully compensating public shareholders while holding controlling owners accountable for missed projections. The critical perspective notes that management set aggressive four-year earnings targets while raising public capital and ultimately missed them, providing a key benchmark for evaluating future company forecasts.

The profit shortfall was not caused by a decline in underlying colocation demand, but rather by the execution of a capital-intensive strategic transition.

↑ Back to Top

V. The AI Pivot & High-Density Liquid Cooling Architecture (2023–2024)

The physical demands of artificial intelligence workloads forced a fundamental transformation in data center design.

Traditional cloud server racks draw roughly 6 to 8 kilowatts of power. For decades, operators dissipated this thermal load using conventional air cooling, blowing chilled air across server racks. Air cooling remains effective up to approximately 15 kilowatts per rack. Beyond that threshold, air cannot remove thermal energy quickly enough to prevent processors from throttling performance to avoid heat damage.

High-density AI training racks packed with accelerators generate 40 to 50 kilowatts or more—a fivefold to eightfold increase in thermal output within the same physical footprint. To manage this load, the industry shifted to liquid cooling. Cold-plate liquid systems circulate fluid through metal plates attached directly to processors, transferring heat straight from the silicon into a water loop. Because water carries nearly 3,500 times more thermal energy per unit volume than air, liquid cooling provides the heat dissipation required for modern AI chips.

For infrastructure operators, adapting to high-density computing required extensive structural retooling. Facilities built for legacy air cooling feature inadequate floor-load ratings, piping, power distribution, and chilled-water systems. Operators could not simply slot high-density AI clusters into legacy halls without executing comprehensive structural overhauls.

Building it before the demand was proven

Range committed early to this architectural shift. In 2023, the company completed and operationalized what it described as China's first fully liquid-cooled intelligent computing center—a dedicated facility rather than a pilot project or retrofitted hall.³ Constructing a greenfield liquid-cooled campus represented a major engineering and financial commitment at a time when domestic demand for generative AI workloads remained largely unproven.

Concurrently, Range expanded its position within the computing value chain. Under its AIDC framework, the company moved beyond supplying physical shells and power capacity. It began procuring high-performance servers and networking hardware directly, assembling them into clusters to sell training and inference capacity to AI developers and enterprise clients.² This shift fundamentally transformed Range's asset profile: alongside long-life physical real estate, the balance sheet absorbed shorter-duration IT hardware subject to rapid depreciation and technological obsolescence.

Operational efficiency in data centers is measured by Power Usage Effectiveness (PUE)—the ratio of total energy consumed by a facility to the energy delivered to IT hardware. A lower PUE indicates less energy wasted on cooling and power distribution. With Chinese regulatory authorities enforcing strict PUE caps on new builds and power representing the primary variable expense in colocation, liquid cooling provided both regulatory compliance and a structural cost advantage. To support these operations, Range aligned its facilities with international management frameworks—including ITIL, Uptime Institute, and IFMA—and reported employing 81 Uptime Accredited Operations Specialists as of 2025, satisfying technical audit standards required by hyperscale clients.³

Who is actually buying

Identifying Range's ultimate end users requires distinguishing between indirect wholesale channels and customer disclosures. Because master lease contracts are executed through state telecom carriers, end users are separated from direct billing relationships. Media reports have identified key tenants in Range's wholesale facilities as major technology platforms, including 字节跳动 ByteDance, 华为 Huawei, and 京东 JD.com.⁶ In its 2025 financial disclosures, management categorized its client base without naming specific entities, citing coverage across China's top three internet platforms, major cloud service providers, and leading AI developers, alongside framework agreements with all three national telecom carriers.² Because revenue breakdowns by end user are omitted from public summaries, specific tenant exposure figures in market analysis remain estimates.

However, disclosures indicate a shift in customer concentration. Management reported that reliance on its largest single customer declined significantly compared to pre-listing levels, broadening its revenue base across multiple platforms.² While reduced concentration mitigates exposure to any single customer's capital expenditure cycle, detailed percentage breakdowns by tenant remain unverified in summary filings.

The bill came due in 2024

The financial impact of this transition became evident in 2024. Full-year revenue reached RMB 4.365 billion, up just 0.32% year-over-year, while net profit grew 1.62% to RMB 1.790 billion—missing consensus sell-side estimates of RMB 2.1 billion to RMB 2.2 billion.⁹ The slowdown was concentrated in the fourth quarter, when revenue dropped 45.64% year-over-year to RMB 908 million and net profit declined 56.89% to RMB 277 million, interrupting the momentum of the preceding three quarters.⁹

Segment performance reflected the ongoing infrastructure retooling. Legacy IDC revenue fell 7.56% to RMB 2.914 billion, and segment gross margin contracted by 6.93 percentage points to 47.53%, driven by taking server racks offline for liquid-cooling upgrades. While AIDC revenue expanded 21.03% to RMB 1.451 billion, the gain was insufficient to offset legacy downtime. Concurrently, capital expenditures rose roughly 20% to support customized liquid-cooling builds. This increased depreciation and amortization expenses by RMB 158 million, while research and development spending climbed 48.63%, and finance costs nearly doubled to RMB 247 million as debt funded ongoing construction.⁹

This divergence illustrates the classic capital mismatch inherent to infrastructure transitions: upfront capital commitments and debt service immediately elevate operating costs, while monetization lags by several quarters. Investor reaction was swift: Range's shares fell up to 17% intraday before closing down 14.16% following the release of the 2024 financial results. Management subsequently noted that market expectations had remained anchored to the profit targets set during the backdoor listing rather than reflecting the near-term realities of facility retooling.⁸⁹

In hindsight, the 2023–2024 performance reflected a strategic transition accompanied by mismanaged market expectations. Range allocated substantial capital to convert operational facilities into high-density AIDC clusters ahead of anticipated AI demand, but failed to adequately adjust public earnings guidance during the buildout phase. Whether this capital deployment would yield target returns depended on capacity absorption in 2025.

↑ Back to Top

VI. Financial Anatomy & Capital Recycling via C-REITs (2025–Today)

On August 8, 2025, a new security began trading on the Shenzhen Stock Exchange under ticker 180901: the 南方润泽科技数据中心REIT Southern Range Technology Data Center REIT, one of China's first public data center REITs, raising RMB 4.5 billion by issuing one billion units at RMB 4.500 each.²¹⁰ The underlying asset comprised a single facility—the A-18 data center in Range's 国际信息云聚核港 International Information Cloud Hub project—along with its ancillary infrastructure.²

This public listing marked a pivotal financial milestone for Range, directly shaping its reported 2025 financial performance.

Reading the 2025 income statement properly

Evaluating Range's 2025 results requires distinguishing recurring operational growth from one-time asset sales.

Revenue rose 29.99% year-over-year to RMB 5.674 billion, reversing the previous year's stagnant top-line growth. Segment performance explains the recovery: AIDC revenue expanded by roughly 73% to RMB 2.51 billion, contributing 44.24% of total sales at an above-average gross margin of 48.50%. Meanwhile, legacy IDC revenue rebounded 8.6% to RMB 3.164 billion.³ This dual expansion reflected upgraded liquid-cooled capacity returning online alongside new high-density deployments entering service. Operating cash flow grew 57.17% to RMB 3.292 billion, demonstrating improved cash conversion as higher-margin computing capacity was absorbed.²

In contrast, headline net profit figures warrant careful parsing. Net profit attributable to shareholders surged 182.07% to RMB 5.050 billion, but net profit excluding non-recurring items grew by a far more modest 6.87% to RMB 1.901 billion.² The RMB 3.15 billion variance stemmed primarily from a single quarter: third-quarter net profit reached RMB 3.822 billion compared to non-recurring-adjusted profit of RMB 475 million.² The C-REIT transaction closed during that third quarter, meaning the earnings surge reflected the non-cash accounting gain realized by transferring a mature asset into the listed fund.

While the RMB 4.5 billion in REIT proceeds generated substantial liquidity, the transaction represents a balance-sheet asset revaluation rather than ongoing operational income. Valuing Range on headline 2025 earnings or its restated 41.80% return on equity relies on a non-recurring event.² Because depreciation expenses on newly delivered capacity absorbed top-line gains, core operating profit grew by under 7% despite a 30% expansion in total revenue.

What the REIT actually does

China's public infrastructure C-REIT framework, introduced in 2021, allows operators to transfer stabilized, cash-generating assets to listed funds while retaining operational mandates and equity stakes. For Range, this mechanism offers a structural capital management tool.

Management frames this strategy as a transition from heavy-asset ownership toward asset-light operational scaling, creating a capital loop of fundraising, development, asset management, and exit.² In practice, Range develops data centers using debt and equity, secures long-term tenant commitments, sells stabilized facilities to yield-seeking public investors, collects ongoing asset management fees, and redeploys the capital into new construction. Executed repeatedly, this model converts balance-sheet capacity constraints into execution-driven growth.

Range has positioned its asset recycling platform as an early mover advantage under tight regulatory quotas. On February 13, 2026, the board approved an expanded REIT issuance backed by the A-7 and A-8 data centers, which shareholders ratified on March 3, 2026. However, as noted in the annual report, final execution remains subject to regulatory review by the National Development and Reform Commission and the China Securities Regulatory Commission.² Regulatory approval timelines represent the key operational variable, as capital recycling depends entirely on government clearance speeds.

Asset recycling also introduces accounting complexities. Transaction gains depend on the spread between an asset's book value and its appraised transfer value—a calculation reliant on valuation assumptions. While auditor 容诚会计师事务所 Rongcheng CPAs issued an unqualified audit opinion on the 2025 financial statements, analysts must track whether asset sales become an indispensable earnings buffer over full market cycles.²

The balance sheet behind the growth

Range's asset base expanded 36.36% during 2025 to RMB 42.694 billion, while shareholder equity rose 37.03% to RMB 13.656 billion.² The balance sheet leverage ratio remained stable at 62.84%, compared to 63.11% in 2024, demonstrating that REIT proceeds enabled substantial asset growth without elevating leverage.² Debt coverage metrics strengthened markedly: EBITDA to total debt rose from 14.36% to 27.63%, while interest coverage expanded from 3.86 times to 9.10 times.²

The company also benefited from low domestic borrowing rates. Range issued RMB 1 billion in technology-innovation commercial paper in August 2025 at an interest rate of 1.71%, followed by an additional RMB 1 billion tranche in January 2026 at 1.65%.² Access to sub-2% debt financing to develop infrastructure assets yields strong interest rate spreads, though profitability remains sensitive to potential shifts in China's monetary policy environment. For 2026, management outlined a four-part funding plan combining equity financing, public REIT recycling, interbank debt issuance, and credit facilities.¹¹

Capital allocation remains focused on reinvestment. Management proposed a 2025 dividend payout of RMB 2.458 per ten shares—a conservative distribution relative to reported net income—confirming that cash flows are being directed into project development rather than shareholder distributions.²

Going offshore

International expansion represents Range's principal strategic initiative in 2026, accelerating faster than initially signaled in past corporate outlooks.

On March 2, 2026, Hong Kong's Innovation, Technology and Industry Bureau awarded Hong Kong Range Intelligent Computing Technology Limited the land tender for the Sha Ling data park in the Northern Metropolis. Range secured a 50-year lease over a 110,000-square-meter site with 250,000 square meters of gross floor area for a land premium of HK$581 million, with 88% of the area designated for high-end data center development.¹² The tender evaluated proposals using a dual-envelope framework weighting technical criteria at 70% and financial bids at 30%, indicating the award was secured on technical specs rather than raw land pricing.¹² The planned 240-megawatt facility entails HK$23.8 billion in cumulative investment over three years, with operations targeted within 42 months and a projected compute capacity of 180 EFLOPS by 2032—equivalent to roughly 36 times Hong Kong's total computing capacity at the time of the award.¹²

In parallel, Range is developing a Southeast Asian computing hub in Indonesia. The planned campus in Nongsa on 巴淡岛 Batam island envisions 360 megawatts across two phases—160 megawatts in phase one and 200 megawatts in phase two—representing a US$500 million total investment with initial delivery expected by late 2026.¹¹ Characterizing Southeast Asian hyperscale demand as a supply-constrained seller's market, management has framed its expansion as exporting both operational talent and capital, deploying over 300 technical personnel to support construction and facilities management.¹¹

Together, the Hong Kong and Indonesia projects add 600 megawatts of planned international capacity alongside 5.4 gigawatts of domestic developments. Across nine regional clusters, Range's total global footprint targets 6 gigawatts of planned capacity against approximately 750 megawatts currently in operation.² This eightfold gap between operating capacity and long-term targets highlights that Range's valuation remains anchored to project execution and capacity conversion.

The 2025 delivery footprint offers benchmark evidence of construction execution. Range operationalized roughly 220 megawatts of new capacity during the year—including a single 100-megawatt hyperscale facility—representing a single-year delivery volume exceeding 40% of its total operational builds over the preceding sixteen years combined.²

While operational momentum has accelerated, converting planned capacity into long-term contract returns will determine whether Range can maintain its growth profile as broader industry capacity expands.

---

↑ Back to Top

VII. Competitive Moats: Hamilton Helmer's 7 Powers & Porter's 5 Forces

If an investor set out with RMB 20 billion to replicate Range Intelligent Computing, the primary barrier would not be building shells or cooling infrastructure. Concrete, steel, and cold-plate liquid loops are commercially available commodities. Nor would the bottleneck be customer access, as carriers and hyperscalers readily evaluate new operators with available capacity. The true impediment arises at the site acquisition stage: securing a contiguous, buildable, grid-connected land parcel within a policy-designated hub with hundreds of megawatts of approved energy quotas attached. That is where capital encounters administrative limits.

The 7 Powers, honestly graded

Cornered Resource — the strongest claim, and a depleting one. Range's primary structural advantage rests on a stock of land titles and energy-consumption approvals accumulated prior to regulatory tightening in locations that subsequently faced strict capacity caps. Energy quotas cannot be purchased once a jurisdiction ceases issuing them. Evidence that this power remains active includes a 30% expansion in energy-consumption indicators during 2025 and winning a competitively tendered Hong Kong site on technical merit.²¹² However, the advantage is inherently depleting: reserves are consumed as facilities are built, requiring each new cluster to be assembled at prevailing land prices in regions where Range lacks established municipal ties. This represents a resource that must be continuously re-acquired rather than a self-compounding intellectual property asset.

Scale Economies — real at the campus level, weaker at the corporate level. Developing twelve data center buildings within a single campus delivers tangible cost efficiencies per server cabinet compared to operating twelve dispersed facilities. A centralized campus shares high-voltage substations, security infrastructure, industrial water facilities, operating teams, and equipment procurement leverage. Fixed-cost leverage contributed directly to operating cash flows in 2025. However, these efficiencies reflect campus-density economics rather than national corporate scale; a competitor operating a single hyperscale hub captures comparable unit benefits, which explains why major operators like GDS and VNET, along with telecom carriers, focus heavily on campus-style developments.

Counter-Positioning — a deliberate cost choice rather than a trap. The carrier-cooperative distribution model allows Range to maintain a lean commercial team while filling megawatt-scale facilities, creating a cost structure that legacy retail colocation operators with large sales organizations cannot easily adopt. Yet this does not constitute strict counter-positioning under Helmer's framework, as competing wholesale entrants face no structural barrier to adopting the same carrier-led channel. Range holds an operating cost advantage rather than an asymmetric business-model trap over incumbents.

Switching Costs — moderate, and often overstated. Relocating an active enterprise workload involves re-routing fiber connectivity, reconfiguring network architecture, and re-certifying compliance, creating meaningful friction that supports contract renewals. Nevertheless, these switching costs do not grant unconstrained pricing power. Contracts expire periodically, wholesale buyers maintain detailed insight into facility operating costs, and hyperscale tenants operating across multiple third-party vendors routinely direct new capacity deployments to competing sites.

Process Power — the plausible emerging claim. Operationalizing a single 100-megawatt intelligent computing facility and bringing a greenfield, full-building liquid-cooled data center into service ahead of domestic peers reflects specialized organizational capabilities that are difficult to replicate quickly.²³ While it remains early to categorize this execution speed as a durable process power, single-year delivery velocity provides the primary evidence supporting the claim.

Branding and Network Economies — essentially absent. Wholesale colocation buyers procure capacity based on unit power costs, power usage effectiveness (PUE), grid reliability, and facility uptime. Consumer brand equity provides no pricing premium, and network effects are absent: adding a new tenant to a facility does not increase the platform value for existing occupants.

Five Forces, war-gamed

Buyers: high power, structurally. Range's revenue remains concentrated among a small group of major counterparties—specifically provincial branches of state telecom carriers and large internet platforms—that possess greater financial scale and build portions of their own infrastructure. At contract renewal, this scale asymmetry limits Range's pricing leverage. Management's reported reduction in single-customer concentration mitigates this vulnerability but does not eliminate structural buyer power.²

Suppliers: moderate to high, and rising. Electrical utility grids establish power tariffs through regulated pricing mechanisms, leaving limited scope for bilateral negotiation despite direct power-purchase agreements. A more acute constraint exists in upstream hardware: global supply limits and export controls affect the procurement of advanced AI accelerators. This bottleneck impacts Range more directly under its AIDC model—where the company purchases servers and computing hardware—than under pure colocation, where hardware sourcing remains the tenant's responsibility. Additionally, specialized vendors supplying high-density power distribution and cooling equipment retain pricing leverage during industry-wide build cycles.

New entrants: low for equivalent scale, but not zero. Administrative energy quotas, strict PUE caps, and multi-year municipal permitting requirements protect incumbent operators from rapid competitive entry. However, these barriers do not stop state-backed infrastructure entities. Municipalities across China have sponsored localized computing centers, creating a wave of subsidized, regional supply that risks depressing local market pricing even when facility utilization remains low.

Substitutes: low. AI model training, inference processing, and digital workloads require physical silicon housed in climate-controlled facilities supplied with reliable electric power. Although chip-level energy efficiency continues to improve, aggregate computing demand has consistently outpaced hardware efficiency gains. China's total intelligent computing capacity reached 1,590 EFLOPS by the end of 2025—nearly doubling within twelve months—while daily model token generation reached approximately 140 trillion by March 2026.³ Substitute risks to data center infrastructure as an asset class remain negligible; the operational risk is confined to facility location and energy efficiency.

Rivalry: high, and bifurcating. Competition among wholesale operators remains intense. Competitor GDS reported 2025 net revenue of RMB 11.43 billion, up 10.8% year-over-year, with an adjusted EBITDA margin of 47.3% and its first annual net profit since listing, supported by asset monetization through asset-backed securities and REIT structures.[^13] This highlights that capital-recycling models are widely deployed across the sector rather than proprietary to Range. Concurrently, Baosight Software benefits from industrial parent backing, Sinnet holds key positions near Beijing, and state telecom carriers maintain substantial internal data center portfolios while serving as Range's primary distribution channel.

The data center market is undergoing a structural split. Standard air-cooled rack capacity in secondary locations faces pricing pressure and oversupply risks. Conversely, high-density, liquid-cooled capacity adjacent to tier-one demand hubs has experienced tightening conditions, with management reporting that AI data center pricing expanded in 2026 while traditional IDC pricing remained under competitive pressure.¹¹

Range's long-term competitive position depends on its ability to continually secure scarce land and power assets while maintaining execution speed, as its structural protection rests on a single depleting asset moat paired with replicable operating efficiencies.

↑ Back to Top

VIII. Management Credibility & Investor Stress Test

For a Chinese founder-led company, Range's ownership structure is remarkably straightforward. Controlling entity Jing-Jin-Ji Range (Langfang) Digital Information holds 59.18% of the listed company—representing 967,135,107 shares, of which roughly 54.6 million were pledged as of the 2025 annual report. The disclosed actual controllers are Zhou Chaonan alongside 张娴 Zhang Xian and 李莹 Li Ying, with Zhou holding 75% of the controlling vehicle. An additional 0.27% stake is held by Beijing Tianxinghui, a vehicle tied to Zhou's original telecom-conduit venture. The three controllers also maintain small direct equity stakes through the company's management incentive scheme.²

A 59% controlling block acts as a double-edged sword. On one hand, it enables long-horizon capital allocation, shielding management from quarterly earnings pressure and forcing the controlling shareholders to absorb the earn-out shortfall directly. On the other hand, minority investors lack effective mechanisms to challenge corporate strategy. Furthermore, the pledged portion of the controlling stake requires ongoing monitoring, as elevated pledge levels can amplify financial risk during sharp market drawdowns.

Judging management by what it promised versus what it delivered

The most objective test of management credibility lies in evaluating execution against its self-imposed targets—a record that reveals distinct strengths and weaknesses.

On the earn-out commitment made during its backdoor listing, management set ambitious four-year profit targets. It surpassed expectations in the first two years before missing projections in the final two, ultimately achieving 93.56% of the cumulative four-year target. Crucially, the controlling shareholders honored the contractual compensation mechanism in full and on schedule. This outcome is neither a failure nor a triumph; rather, it demonstrates that while management's long-term financial forecasts require an analytical discount, its regulatory and contractual compliance is dependable.

Regarding public disclosures, the company's explanation for the 2024 earnings miss was somewhat deflecting. Management noted that sell-side consensus had been anchored to legacy restructuring commitments—a factually accurate observation that nevertheless overlooked the reality that management itself had authored those targets.⁸ However, the operational explanations were concrete and verifiable: capital expenditures escalated to meet clients' customized liquid-cooling requirements, depreciation charges preceded revenue recognition, and legacy server racks were temporarily taken offline for conversion.⁹ This provided a transparent accounting of operational delays rarely seen in peer disclosures.

On narrative consistency, corporate execution has aligned with stated strategy. The AIDC pivot was introduced in 2023 as a deliberate long-term transition. By 2025, management maintained a consistent narrative, framing the pivot as a three-year deployment that built advantages in supply-chain diversity, cluster-networking technology, and client trust—with AIDC revenue, gross margins, and operating cash flows matching or exceeding legacy IDC operations.² The core strategy did not shift unpredictably, though its capital intensity required more time to materialize than originally guided.

The equity incentive structure warrants careful scrutiny. In July 2023, Range granted 22.88 million restricted shares—representing 1.33% of total equity—at RMB 14.77 per share to 99 key personnel, including six senior executives. Vesting is contingent on achieving net profit growth targets relative to a 2024 baseline of at least 50% in 2025, 75% in 2026, and 100% in 2027.¹³ While these represent demanding performance thresholds, the metrics are tied to aggregate net profit—a benchmark that non-operating REIT asset sales can satisfy without corresponding operational growth. Investors must monitor whether future incentive vesting reflects underlying operational performance or capital-recycling gains.

What an activist would attack

Quality of earnings. The most obvious vulnerability is that Range's reported 2025 net profit growth was driven primarily by non-recurring asset sales, while core operating profit expanded by less than 7%. An activist investor would argue that the company trades on non-repeatable earnings, pressing for disclosures that explicitly decouple recurring operational cash flow from REIT transaction gains, alongside detailed guidance on future asset monetization pipelines. The board's decision to expand the REIT framework with two additional data center buildings makes this earnings distinction even more critical.²

Related-party asset transfers. Selling stabilized assets from a controlled operating entity into a public fund managed by the operator is an established regulatory structure. However, it inherently involves asset appraisals between affiliated parties. Appropriate governance oversight requires strict scrutiny of valuation methodologies, retained equity stakes, ongoing management fees, and whether transferred assets represent peak portfolio quality rather than average operational yield.

Capital intensity versus financial disclosure. Management envisions scaling global capacity to roughly 6 gigawatts against approximately 750 megawatts currently in operation, supported by an extensive domestic pipeline comprising dozens of facilities and hundreds of thousands of server cabinets.²¹³ With balance sheet leverage hovering near 63% and a high proportion of assets under construction, equity returns remain sensitive to project delivery schedules. Institutional investors require more granular capacity tracking than high-level megawatt totals—specifically requesting campus-by-campus breakdowns of delivered, contracted, and utilized capacity, alongside lease expiration schedules.

Customer concentration transparency. Although management reports reduced reliance on key clients, detailed customer-level revenue breakdowns remain omitted from annual report summaries.² Given that renewal pricing with hyperscale buyers serves as the primary driver of long-term operating margins, the absence of explicit customer concentration data remains an ongoing governance limitation.

Geographic and strategic expansion. Over an eighteen-month period, Range committed to two major international developments, a multi-billion-dollar Hong Kong data park project, a proposed convertible bond issuance to acquire a 42.56% stake in 广东润徽 Guangdong Runhui to expand its Greater Bay Area footprint, and a comprehensive executive management reorganization completed in April 2026.¹¹ While each initiative carries a logical rationale, they collectively represent a company executing multiple complex projects across unfamiliar regulatory jurisdictions while relying on debt financing shortly after missing a four-year profit target. This concentration of concurrent initiatives presents a clear operational execution risk.

Myth versus reality

Comparing market consensus against disclosed operating data highlights three key realities:

Myth: 2025 performance proved that the AI pivot is generating extraordinary profits. Reality: The operational transition is succeeding—evidenced by 30% top-line revenue growth, AIDC generating 44% of total sales at premium gross margins, and operating cash flow rising 57%. However, the 182% headline profit surge was driven by a single asset disposal rather than core operational expansion.²³ Core operational growth remains solid, but it does not match the headline figure.

Myth: Range's land and power reserves form an unassailable competitive moat. Reality: Secured land and energy quotas represent a real strategic advantage, but they are depleting assets that must be continuously re-acquired at market rates in new geographical markets against established incumbents. Capacity awards in Hong Kong and Indonesia were secured through competitive tendering rather than inherited cost advantages.

Myth: Management consistently fulfills its financial forecasts. Reality: Management achieved two out of four annual commitment targets during its restructuring period, surrendering and canceling 87 million shares to compensate public shareholders for the shortfall.⁴ While the company demonstrates rigorous contractual compliance, its multi-year financial projections justify an analytical discount.

The broader analytical takeaway is that Range operates as a competent, ambitious, majority-controlled infrastructure provider that has executed its operational strategy effectively while over-promising on medium-term financial guidance. For investors, evaluating company performance requires prioritizing delivered megawatt capacity and recurring cash flow over long-term strategic projections.

↑ Back to Top

IX. Current Risk Radar & Key Investment Drivers

Stripping away corporate narrative leaves a focused set of operational indicators. While infrastructure analysis often tracks dozens of metrics, evaluating Range's terminal value depends on monitoring a critical subset of operational and financial performance measures.

The three metrics that matter

First: delivered capacity, in megawatts. The critical benchmark is delivered and energized power, rather than planned or signed capacity. This metric converts land rights and energy quotas into revenue, determining whether the company's target of roughly six gigawatts translates into operational revenue or remains a theoretical expansion target. The primary operational baseline compares the 2025 delivery of approximately 220 megawatts against an existing operating base of roughly 750 megawatts, alongside a 2026 pipeline featuring Langfang B, Pinghu D, Foshan phase one, and Huizhou phase one domestically, in addition to the first phase of the Batam project in Indonesia at approximately 160 megawatts.²¹¹ A widening divergence between growing planned capacity and slowing physical delivery serves as an early indicator of execution friction.

Second: AIDC revenue share and its gross margin, evaluated together. Revenue contribution alone provides an incomplete picture of segment health. The Artificial Intelligence Data Center segment generated 44.24% of total revenue in 2025 at a 48.50% gross margin, outperforming the company's overall average.³ Management expects the AIDC gross margin to settle at around 45% in 2026, noting that the segment carries heavier hardware investments and a larger depreciation burden, whereas the legacy colocation cost structure is dominated by electricity expenses.¹¹ This distinction highlights that AIDC is not merely a higher-margin extension of colocation; it is a more capital-intensive model with shorter hardware asset lifecycles. A scenario where revenue share expands while gross margin contracts would indicate that top-line growth is being driven by hardware capital expenditure without corresponding pricing leverage.

Third: mature-campus utilization and Power Usage Effectiveness (PUE). Rack occupancy across stabilized facilities provides the most direct measurement of tenant demand absorption against new supply. Combined with PUE—which governs regulatory compliance and direct power overhead—these metrics reflect facility operating efficiency. Performance in the first quarter of 2026 highlighted this dynamic: while revenue rose 53.55% year-over-year to RMB 1.84 billion and net profit grew 35.35% to RMB 582 million, gross margin contracted by 2.93 percentage points to 46.80%.¹¹¹⁴ Net income growing at a slower pace than revenue alongside margin compression illustrates the initial cost pressure associated with scaling new capacity.

The risk radar

Hardware and accelerator supply constraints. Component access represents the most distinctive operational vulnerability in the AIDC framework. Under a pure colocation model, a tenant's inability to source advanced chips functioned as an indirect demand lag. Under the AIDC model, where Range directly purchases servers and leases computing clusters, supply chain bottlenecks in high-performance silicon directly constrain revenue growth. Export restrictions on advanced accelerators and the commercial maturity of domestic alternative chips jointly dictate deployment timelines and pricing power. Shifts in trade policy or domestic semiconductor yields consequently impact segment performance within quarters.

Power tariffs and energy regulation. Electricity represents the primary operating expense in legacy colocation, with utility pricing subject to state regulation rather than pure market dynamics. Participating in direct power trading and wholesale electricity markets offers partial cost insulation.² At the same time, tightening municipal PUE regulations present a dual impact: while stricter energy caps increase entry barriers for prospective competitors, they simultaneously require ongoing capital expenditures to upgrade existing legacy facilities.

Depreciation drag from new builds. Facility expansion introduces immediate accounting pressure, as demonstrated during the 2024 earnings miss and the margin compression in early 2026. Newly delivered megawatt capacity begins depreciating immediately upon completion, whereas tenant revenue ramps up over subsequent quarters. During periods of accelerated facility delivery, this timing mismatch depresses operating margins before revenue fully materializes, creating a temporary headwind during heavy construction phases.

Financing conditions and capital costs. Range's expansion model relies on securing sub-2% debt financing alongside recycling equity by transferring stabilized facilities into public REITs at favorable valuations.² Changes in domestic interest rate environments or diminished liquidity in the C-REIT market could reduce asset transaction valuations and delay regulatory approvals, narrowing the capital spread that supports project reinvestment. The pending regulatory approval for the company's proposed REIT expansion underscores this structural dependency.²

Regional oversupply in standard capacity. Secondary markets face oversupply risks driven by state-subsidized localized computing centers built ahead of firm tenant demand. Although Range's primary campus locations and focus on high-density liquid cooling mitigate this exposure, its legacy colocation segment remains vulnerable to broader pricing pressure—a key consideration given that traditional colocation still accounted for 55.76% of total revenue in 2025.³

International execution and cross-border risks. Operating projects in Hong Kong and Indonesia exposes Range to unfamiliar operational environments, including foreign construction oversight, currency fluctuations, cross-border capital regulations, and partner dependencies. Furthermore, initial favorable pricing dynamics in Indonesia may evolve before the project's multi-year development cycle completes.¹¹¹² With operational delivery in Hong Kong projected several years out, international projects require sustained capital commitments prior to generating cash flows.

While no single risk factor undermines Range's operational thesis, collectively they describe an enterprise that is operationally, financially, and administratively tied to a core market condition: that demand for artificial intelligence computing across China and regional hubs continues to outpace the licensed supply of high-density infrastructure.

↑ Back to Top

X. Epilogue & Playbook Lessons

Technology cycles frequently reveal unexpected operational bottlenecks. In cloud computing, market attention initially centered on software capabilities before capital scale emerged as the primary constraint. In artificial intelligence, the bottleneck shifted from model architecture to semiconductor availability, and ultimately to physical power grid connections and municipal energy quotas.

Range's evolution illustrates this shift, yielding three core takeaways for digital infrastructure investors.

Land and power can create more durable moats than software. Software advantages can erode quickly with talent mobility, whereas an energy quota in a restricted jurisdiction remains intact until consumed. By securing land rights and power approvals ahead of demand, Range accumulated physical capacity before energy rationing tightened across major domestic hubs. The strategic insight for regulated infrastructure is that binding constraints are often administrative rather than technological, making early acquisition essential. However, an advantage based on pre-purchased land and power reserves is inherently depleting; maintaining growth forces the operator to continuously secure new capacity in locations like Foshan, Danzhou, Sha Ling, and Batam at prevailing market prices.

Partnering with state telecom giants prioritizes capital efficiency over sales control. By routing facility distribution through China's three state carriers, Range absorbed megawatt-scale capacity while maintaining lower commercial overhead than retail colocation operators. This channel also backed its receivables with sovereign-grade counterparty credit. The strategic trade-off is structural: ceding direct tenant relationships limits pricing power upon contract renewal. For long-life infrastructure operators facing capital constraints, leveraging third-party distribution offers a high-return path for capacity utilization, provided management accepts reduced control over end-user pricing.

Asset recycling shifts growth constraints from balance-sheet leverage to operational execution. Utilizing China's C-REIT framework allowed Range to expand its total asset base by more than one-third in 2025 without increasing financial leverage.² This monetization loop enables operators to recycle capital faster than traditional debt-financed development. However, asset recycling requires analytical distinction: while cash proceeds provide real liquidity, transaction gains represent non-recurring accounting items rather than sustainable operational profit. Peer operator GDS achieving its first annual net profit in 2025 through similar asset monetization demonstrates how widely capital-recycling tools are being deployed across the industry.[^13]

Where this leaves the company

Seventeen years after securing its initial land in Langfang, Range operates as a primary domestic data center provider expanding internationally into Hong Kong and Southeast Asia. Its original Hebei campus has grown into seven domestic regional hubs and two overseas developments, with a planned global footprint of roughly 6 gigawatts against approximately 750 megawatts of operational capacity.² Public equity markets have priced this expansion trajectory with substantial volatility: in early August 2026, the company's stock traded around RMB 68.80 per share—yielding a market capitalization near RMB 113 billion—within a 52-week range spanning from roughly RMB 45 to RMB 111.¹⁵

The investment thesis for Range depends on three primary factors: whether regional demand for artificial intelligence computing continues to outstrip licensed infrastructure supply; whether Range can convert its land, power quotas, and liquid-cooling delivery speed faster than competitors; and whether its asset-recycling model continues to generate liquidity without diluting shareholders or increasing balance-sheet leverage. Investors can evaluate these factors through quarterly trends in delivered megawatt capacity, AIDC operating margins, and facility utilization rates.

Conversely, key downside risks center on buyer concentration and structural pricing power, the depleting nature of pre-acquired land reserves, depreciation pressure from rapid capacity delivery, potential hardware supply constraints in the AIDC segment, and management's execution track record following the profit shortfall during its restructuring earn-out period.

Both perspectives rely on the same underlying operational realities. For investors, evaluating Range requires focusing on concrete metrics—delivered capacity, recurring operating cash flow, and asset transaction terms—rather than long-term strategic projections.

---

↑ Back to Top

XI. Outro & Further References

The next chapters of Range's story will be written in regulatory filings rather than corporate announcements.

In the near term, three operational benchmarks warrant close observation: the 2026 interim financial results, which will clarify whether early-year margin compression reflected a temporary construction ramp or broader pricing pressure; the regulatory approval process for expanding the C-REIT platform with the A-7 and A-8 facilities, testing whether asset recycling operates as a repeatable platform; and the delivery of the initial 160-megawatt phase in Batam by late 2026, which will provide the first empirical evidence of international execution.²¹¹

On future earnings calls and investor presentations, several operational questions remain central: how management frames AIDC pricing relative to its roughly 45% gross-margin target; whether disclosures provide granular utilization and tenant concentration metrics rather than qualitative summaries; how chip procurement is managed during the computing expansion; and whether corporate strategy translates into a predictable recycling cadence rather than opportunistic transactions.

Over a longer horizon, the HK$23.8 billion Hong Kong Sha Ling development represents the largest single commitment in the company's history. Because the facility will not generate revenue for several years, it serves as the primary test of management's capital discipline. Successfully funding a multi-year overseas build while expanding domestic capacity and maintaining target leverage ratios will demonstrate whether Range can execute its global expansion without overextending its balance sheet.

References

1. 获资本加持步上发展快车道 润泽科技借壳普丽盛登陆创业板 — Jiemian News, 2022-09-13 ↩↩↩

2. 润泽智算科技集团股份有限公司 2025年年度报告摘要（公告编号：2026-027） — Shenzhen Stock Exchange disclosure, 2026-04 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

3. 润泽科技年报：2025年营收56.74亿元 净利润50.50亿元同比增182% — Xinhua, 2026-04-10 ↩↩↩↩↩↩↩↩↩

4. 润泽科技：关于业绩承诺补偿股份回购注销完成的公告 — 同花顺财经, 2025-07-11 ↩↩↩

5. 周超男：享受奋斗的人生 — 长江商学院 DBA Voice ↩↩↩↩↩↩

6. 润泽科技研究报告：字节跳动影子股 — 搜狐财经 ↩↩

7. 创业板首单借壳过会！润泽科技将借壳普丽盛上市，承诺四年累计净利达56亿元 — Jiemian News, 2022 ↩↩↩

8. 业绩不及预期股价大跌 润泽科技回应：重组上市做出的业绩承诺未完成 — 财联社, 2025-04-24 ↩↩↩

9. 财说｜年报公布次日股价暴跌14.16%，数据中心龙头润泽科技发生了什么？ — Jiemian News, 2025 ↩↩↩↩↩

10. 润泽科技：2025年净利预增幅最高近200%，AIDC提速与REITs成重要增长引擎 — 证券时报, 2026 ↩

11. 润泽科技：目前东南亚区域超大规模数据中心市场为卖方市场，近期机柜价格持续攀升（投资者关系活动记录） — 东方财富财富号, 2026-04-14 ↩↩↩↩↩↩↩↩↩↩

12. 香港算力将增36倍！内地企业5.81亿港元中标最大智算中心 — 21世纪经济报道, 2026-03-03 ↩↩↩↩↩

13. 润泽科技拟推限制性股票激励计划 2025至2027三年业绩锁定高增长 — 证券时报, 2023-07-28 ↩↩

14. 润泽科技Q1营收增53.55%，多算力项目将于2026年陆续交付 — 经济观察网, 2026-04-30 ↩

15. 300442:CH Range Intelligent Computing Technology Group Co Ltd — Bloomberg Quote ↩