Range Intelligent Computing Technology Co., Ltd.

Money comes from leasing installed data-center cabinet space, billed either through a telecom-operator partner that bills the end customer and settles monthly with the company by cabinet count and agreed price, or directly under a service contract, plus fees for computing capacity supplied for AI model training, inference and industry applications, billed against agreed service terms. Both lines charge for access to physical capacity the company has already built and owns, rather than for a one-time product, and the business has recorded positive net income in every year CompanyGraph can see on file.

CompanyGraph reads its growth as scaling by adding units of physical capacity: it builds new large, self-owned computing-center clusters, in China and, per its own plans, overseas, then ramps each new cluster's installed and paying share toward the high levels it reports at its already-mature centers. Revenue and operating income have both grown alongside this expanding, heavily capitalized asset base; the depreciation charge running low relative to operating income is consistent with a still-young asset base whose depreciation has not yet caught up with capacity already added. Cash generation and a relatively strong cash position alongside this growth suggest the build-out has so far been funded without straining the business, though CompanyGraph does not verify the specific financing mix behind it.

It depends on land allocation, government-granted energy-consumption quotas, and electricity supply near the cities where it builds, all of which it names as constraints rather than things it fully controls. It also depends on large construction contractors and internationally known equipment suppliers to build and equip each computing center, and, for part of its customer base, on telecom-operator partners who provide connectivity and bill the end customer on its behalf.

Its customers are internet companies, cloud vendors, and companies running AI model training, inference and industry applications, all drawing on physical hosting and computing capacity they have not built themselves. A small number of these customers account for most of its revenue, with the largest counterparties disclosed only by their share of revenue, not by name. Telecom operators that resell its hosting conditions as part of their own connectivity services also depend on it as the physical supplier standing behind that offering.

This way of running the business, building large physical capacity itself and earning recurring fees for access to it, is not rare: CompanyGraph maps a real cohort of other companies worldwide that operate the same way. In its own account, the company points to scarce underlying resources such as land, energy quotas and grid access near major cities, its park-scale build approach, long-standing customer relationships, and access to multiple financing channels as what sets it apart. CompanyGraph does not independently verify whether rivals can replicate any of this, so that part is reported here as the company's own view, not a measured finding.

By the company's own account, its larger customers run complex physical deployments inside its data centers, and moving that equipment to a different provider means physical migration and interruption to the customer's own operations, which the company describes as costly. It also reports that its core customer relationships tend to run for many years rather than being renewed or replaced frequently, consistent with switching being something customers do rarely rather than routinely.

The common assumption for a business like this is that its limit comes from customer retention and from recovering the cost of winning each customer over time. The company's own account does not describe its limits that way. The risk it names first is government control over energy-consumption quotas and energy policy, and the other limits it names are the scarcity of land and grid electricity near the cities where demand concentrates, plus the risk that newly built capacity takes time, or fails, to fill with paying customers. So, on its own account, what shapes its scale looks more like how much physical capacity it can get permission and power to build, and how fast that capacity fills, than a retention-driven constraint.

By its own disclosure, a very small number of customers, none named, account for nearly all of a year's revenue, and nearly all of that revenue itself comes from one region of China rather than being spread nationally or internationally. Because capacity is built in advance, as large, self-owned physical sites with fixed construction and equipment costs already committed, the company itself names the risk that a newly built center's installed and paying share comes in below what was expected. Combined, losing any one of these customers, or a disruption concentrated in that region, would touch a large share of revenue sitting on a cost base that stays mostly fixed regardless of how full each center is, since electricity and depreciation together make up most of operating cost.

By its own account, the pressure it names first is government energy-efficiency and energy-consumption policy, since each new computing center needs an allocated energy-consumption quota before it can be built. It names intensifying competition among AI-infrastructure providers next. It also operates under general securities regulation as a listed company, and each project separately needs construction-planning and environmental-assessment approval before it can proceed. Land and electricity access near the major cities where demand concentrates are named as tightening constraints rather than settled inputs.