Xiexie Data, which placed a "bold gamble" on computing power rental, experienced severe stock volatility—the share price plummeted from a high of 308 yuan to around 210 yuan, with total market capitalization evaporating nearly 48 billion yuan, before staging a strong rebound of over 12% on June 17 and another gain of over 4% on June 18, bringing total market cap back above 130 billion yuan, with year-to-date gains exceeding 130%.

At the helm of this company is an entrepreneur who emerged from Foxconn's workshops—Geng Kangming. He is transforming Xiexie Data from a consumer electronics manufacturing company focused on USB drives, data storage devices, and cameras into a heavy-asset, high-leverage AI computing power service provider.

Public records show he is from Bozhou, Anhui Province, with the original name Geng Sihua, born in 1974. After graduating from university in 1995, Geng Kangming moved south to Shenzhen and joined Foxconn, serving in various positions including engineering, quality control, procurement, business operations, and manufacturing.

This experience gave Geng Kangming an almost instinctive understanding of manufacturing cost control, supply chain management, and scaled production—precisely the foundational capabilities most critical for Xiexie Data's subsequent transformation into a computing power rental business.

The company's earliest business was unglamorous. Initial operations included original equipment manufacturing for consumer electronics: USB drives, data storage devices, and cameras. Positioned in the middle of the industrial chain for extended periods, the company earned modest profits from the traditional hard-work manufacturing sector.

During that period, as the Internet of Things gained momentum, Geng Kangming swiftly pivoted the company into smart cameras and other IoT endpoint sectors, eventually becoming a supplier to companies like Xiaomi and 360. The company listed on the Growth Enterprise Board in 2020.

This step transformed Xiexie Data from a pure "hardware contract manufacturer" into an intelligent hardware manufacturing company.

Around 2024, as generative AI exploded, high-end computing power became a scarce resource. Xiexie Data, through its Hong Kong subsidiary Aujia Software, obtained credentials as an Nvidia NCP cloud partner, using this qualification to enter the business of high-end AI computing server procurement, cluster construction, and computing power rental services.

According to public reports, relatively few domestic manufacturers have obtained such credentials, and this qualification is regarded as an important gateway for Xiexie Data to enter the AI computing power circle.

The computing power rental business fundamentally involves centralized procurement of expensive AI servers, constructing server clusters, and leasing them to customers with large model training, inference, autonomous driving, and other computing requirements.

For many customers, directly purchasing dozens of high-end servers costs tens of millions of yuan, while leasing allows them to reduce one-time capital expenditures and pay based on actual usage volumes or time periods.

Xiexie Data states that its computing power rental services can provide elastic scalability for large model training or high-frequency inference tasks, helping customers reduce resource idle time.

Financial reports show that in 2025, the company achieved revenue of 12.236 billion yuan, up 65.13% year-over-year; net profit attributable to parent company shareholders reached 1.164 billion yuan, up 68.32% year-over-year. Notably, intelligent computing power products and services contributed 2.761 billion yuan in revenue, surging more than seventeen-fold year-over-year and becoming the core growth engine.

Entering 2026, Xiexie Data's performance continued to explode. Financial reports show that in the first quarter, the company achieved revenue of 6.085 billion yuan, up 192.90% year-over-year; net profit attributable to parent shareholders reached 750 million yuan, up 343.45% year-over-year.

From a financial perspective, Xiexie Data is no longer a traditional consumer electronics company but rather an AI computing power operator in a period of rapid expansion.

Since 2026 began, Xiexie Data has applied to banks and other institutions for up to 20 billion yuan in credit lines and up to 80 billion yuan in financing lease capacity; in 2025, the company had also applied for credit lines not exceeding 51.5 billion yuan.

In March 2026, the company signed a 4.248 billion yuan project with Industrial Bank Financial Leasing and Xingye Financial Leasing, equivalent to a heavy-asset expansion model of "borrowing to purchase servers and paying in installments."

Financial reports show that as of the end of March 2026, Xiexie Data's asset-liability ratio rose to 85.89%, with interest-bearing debt reaching 25.9 billion yuan, a significant increase compared to the same period the previous year, while first-quarter financial expenses reached 228 million yuan.

Geng Kangming states that essentially all orders are signed before each procurement, with most major cooperation agreements being five-year long-term contracts. The contract periods cover the accounting depreciation lifespan of servers, and the company has cash recovery and cost control measures in place.

Geng Kangming compares the current computing power industry to "1990s real estate," believing that over the next five to ten years, investment scale in the computing power sector could potentially exceed that of real estate at its peak.

In his view, AI training, inference, autonomous driving, robotics, and other scenarios will continue to consume computing power, and what Xiexie Data must do is seize market position before the sector fully explodes.

According to industry research, China's potential computing power rental market revenue scale is expected to reach 260 billion yuan in 2026 and maintain robust growth. As large models transition from training to inference and spread from internet companies to traditional industries, corporate demand for elastic computing power, on-demand computing, and professional operations and maintenance will continue to grow.

As domestic and international computing power supply increases, domestic AI chips mature, and intelligent computing centers accelerate construction, the computing power rental industry will eventually transition from "scarcity premium" to "operational efficiency competition." At that point, the determining factor in success will no longer be who can afford to purchase more servers, but rather who has higher-quality customers, more stable operations, lower capital costs, and higher asset turnover rates.

Computing power rental is a typical heavy-asset business. Server procurement requires enormous upfront capital investment, with cash flows recovered gradually through long-term rental contracts. If customer orders remain stable, rental rates remain elevated, and server supply is sufficient, high leverage can amplify shareholder returns; however, once demand falls short of expectations, rental rates decline, GPU supply tightens, or technology obsolescence causes equipment value to plummet, high debt and depreciation pressures can rapidly erode profits.

Xiexie Data's management has recently released information about "ample order backlogs" on interactive platforms, seemingly attempting to stabilize market expectations regarding its delivery capacity and order authenticity.

But markets will not listen to stories alone. As the company's fixed assets, construction in progress, and computing power equipment scale rapidly expand, investors increasingly focus on several hard metrics: have purchased servers been delivered and fully deployed? Have customers actually accepted delivery and begun billing? Can long-term contract prices and payment terms cover financing costs? How will server depreciation and residual values from re-manufacturing be handled?

These factors will determine whether Xiexie Data becomes an "AI-era computing power infrastructure leader" or falls into a "high-leverage asset trap."

This time, he has chosen to bet on a newly explosive sector using extremely high leverage, extremely rapid expansion, and extremely strong market expectations. Should the company succeed, Xiexie Data could become an important foundational infrastructure platform in China's AI computing power service ecosystem; should it fail, high debt burdens, elevated depreciation, and concentrated technological iteration risks will be fully exposed.

Xiexie Data's story is fundamentally a generational "bold gamble" on "computing power becoming a new factor of production."

Geng Kangming is the figure sitting at the center of the table, pushing all his chips "all in." And at this moment, the market remains holding its breath, awaiting the answer.