Computing power leasing Q3 reports show a tale of two extremes: leading firms' profits multiply severalfold, while small and mid-sized players see revenue growth without profit growth

Main body of the article

Since 2026, the computing power leasing business has been red-hot, with large orders appearing frequently and rental prices rising.

How did listed companies engaged in the computing power leasing business perform in the first three quarters? With the close of the third-quarter reports, the divergent landscape of this track has been fully presented: leading companies saw their profits multiply several times, while small and medium-sized "players" are struggling with "revenue growth without profit growth."

Performance exploded in the first three quarters

Leading computing power companies saw profits multiply several times

The computing power industry chain includes core segments such as servers, switches, optical modules, computing power chips, and computing power leasing. Among them, computing power leasing directly faces the application market and is the link in the industry chain that is "closest to money."

Since the beginning of this year, with the rapid growth of the domestic computing power market, the computing power leasing industry has ushered in explosive development.

Wind data shows that among listed companies in the computing power leasing sector, more than 30 saw year-on-year growth in net profit attributable to parent company shareholders, while more than 20 saw year-on-year declines.

The largest revenue growth was at Xingyun Technology, a computing power enterprise focused on global cross-border digital trade, Token factories, liquid cooling technology, and other businesses. In the first three quarters, it achieved revenue of 1.175 billion yuan, a year-on-year increase of 1896.66%, an increase of nearly 18 times, making it the company with the highest revenue growth in the computing power leasing industry. At the same time, Xingyun Technology achieved net profit attributable to parent company shareholders of 269 million yuan in the same period, compared with a loss of 72.32 million yuan in the same period last year. The company has announced a cumulative total of 18.316 billion yuan in computing power and storage orders this year.

Hongjing Technology, with the second-highest revenue growth rate, achieved revenue of 2.872 billion yuan in the first three quarters, a year-on-year increase of 181.17%; it achieved net profit attributable to parent company shareholders of 312 million yuan, a year-on-year increase of 189.93%; among this, single-quarter revenue in the third quarter increased by 239.92% year on year. In 2025, the company's computing power service revenue was 904 million yuan, already accounting for 76.53% of operating revenue.

Although the computing power leasing business has exploded, not all companies can achieve profitability.

Dawei Technology, located in Guangdong, mainly provides customers with services such as colocation, cloud business, and computing power servers. In the first three quarters, the company achieved total operating revenue of 402 million yuan, up 28.51% year on year, and achieved net profit attributable to parent company shareholders of -102 million yuan, down 367.84% year on year. The company said this was mainly due to phased losses in the initial production period of newly built data center projects. In addition, many small and medium-sized enterprises such as Sugon Dataenergy, Xinju Network, and Orient Guoxin all experienced revenue growth but net profit losses.

Why is there a situation where leading enterprises are profitable while small and medium-sized computing power leasing enterprises are losing money?

In this regard, Zhu Keli, founding dean of the Guoyan New Economy Research Institute and chief expert on the intelligent economy, said in an interview with a reporter from Dahe Finance Cube that the current computing power leasing industry presents a typical structural differentiation pattern, and the core barrier has shifted from "whether there is computing power" to "whether there is high-quality computing power and whether there are stable long-term orders."

"Leading enterprises rely on reserves of high-end computing power supply, large-scale cluster deployment, and long-term price-locked orders to effectively amortize fixed costs such as depreciation, electricity, and operations and maintenance, supporting high profitability with high utilization rates," Zhu Keli said. By contrast, small and medium-sized manufacturers generally have problems such as lower-tier supply sources, insufficient asset utilization, and a lack of long-term orders as a backstop. Against the backdrop of rapid hardware iteration and high energy consumption costs, the cash flow benefits brought by revenue growth cannot cover rigid depreciation and financial costs, ultimately resulting in revenue growth without profit growth, or even sustained losses.

Contract liabilities rose by up to more than 41 times

Advance receipts have become the "decisive factor" in computing power leasing

Computing power leasing is a typical asset-heavy industry. Equipment procurement and computer room construction all require upfront capital investment, and hardware depreciation, electricity bills, and loan interest are rigid expenditures. Therefore, contract liabilities (customer advance payments) directly determine a company's funding pressure and risk resistance.

A reporter from Dahe Finance Cube noted that in the first three quarters, the contract liabilities of computing power leasing operating companies ranged from 13.76 million yuan to 5.779 billion yuan, a difference of about 420 times. Behind this are two different business models: one in which customers pay first and enterprises then invest in building assets; the other in which enterprises first advance huge sums to build plants and wait for payment collection.

Dongyangguang is a typical example of "customers pay first." As of the end of the third quarter, the company's contract liabilities were 5.779 billion yuan, an increase of 5.683 billion yuan from the beginning of the year. The company disclosed that "contract liabilities increased to 5.8 billion yuan, an increase of 5.7 billion yuan from the beginning of the year," with advance receipts mainly from long-term computing power order customers. Financial report data shows that under Dongyangguang's computing power service orders, the scale of high-performance server assets was 19.653 billion yuan, and customer advance payments of 11.762 billion yuan had been received, with advance receipts covering about 60% of hardware investment. Abundant advance receipts directly drove a sharp strengthening in cash flow. In the first three quarters, Dongyangguang's net cash flow from operating activities was 12.365 billion yuan, of which the single quarter in the third quarter was 9.853 billion yuan, compared with only 220 million yuan in the same quarter last year.

In addition, there are a number of growth-oriented targets in the sector whose contract liabilities are climbing rapidly. Litong Electronics' contract liabilities surged from 26.64 million yuan in the same period last year to 1.130 billion yuan, a year-on-year increase of 4139.94%, making it the company with the highest growth rate of contract liabilities in the sector; Xingyun Technology's contract liabilities reached 242 million yuan, a sharp increase of 959.09% from 22.91 million yuan in the same period last year. As large five-year orders are gradually implemented, customer advance deposits continue to thicken, and the cash flow safety cushion rapidly strengthens. The common point of the two companies is that orders have just entered the delivery period, and advance receipts are accumulating from zero.

As one of the representative companies pursuing heavy-asset, self-funded expansion, Runze Technology reported revenue of 5.666 billion yuan in the first three quarters, up 42.45% year on year, with revenue expanding steadily, but net profit attributable to the parent company was only 1.803 billion yuan, plunging 61.68% year on year; net profit in the third quarter alone was 599 million yuan, down 84.32% year on year, making the pattern of revenue growth without profit growth particularly prominent.

As of the end of the third quarter, Runze Technology's balance of construction in progress was 15.172 billion yuan, ranking first in the sector; however, contract liabilities were only 13.76 million yuan. The company relied on its own funds and bank borrowings to invest in infrastructure, with an asset-liability ratio of 71.07%, and net cash flow from operating activities of 4.508 billion yuan in the first three quarters.

In this regard, Zhu Keli said that in the computing power leasing track, the higher the contract liabilities and the more ample the advance receipts, the better a company can withstand a fixed-asset depreciation cycle of as long as three to five years; for manufacturers without the support of advance receipts who expand production purely through loans, even if short-term revenue rises sharply, profits are highly likely to be hollowed out by depreciation and interest.

Industry competition intensifies

The era of "earning easy money by renting out cards" draws to a close

Behind the industry boom, risks should not be overlooked. The asset depreciation pressure brought by rapid hardware iteration is a sword hanging over the head of every computing power leasing company.

Industry insiders said that Nvidia's hardware iteration rhythm is continuously accelerating, from H100 to B200, and the next-generation Rubin architecture is already being planned and advanced, with a new generation of computing power chips launched almost every year. After a new generation of hardware comes to market, rental prices for the previous generation of GPUs will decline significantly. For small and medium-sized leasing companies, when servers are procured through three-year debt financing, it is very likely that before the hardware loan is repaid, rental income from the old cards will no longer be enough to cover even electricity and interest.

In addition, valuation correction signals have already appeared in overseas markets, with the U.S. stock computing power leasing sector seeing a valuation selloff, and share prices of star companies such as CoreWeave falling sharply.

Zhu Keli judged that differentiation in the A-share computing power leasing industry will further intensify: leading companies with long-term orders in hand and stable supply chains will continue to release profits; for small and medium-sized manufacturers lacking supply guarantees and with high asset vacancy rates, the revenue growth without profit growth shown in the third-quarter reports is only the beginning. The entire industry is bidding farewell to extensive scale expansion, formally ending the easy-money era of "stockpiling cards and subletting them, simply earning the price difference," and entering a new stage of quality improvement, reshuffling and value competition.

The reporter noted that leading companies have already begun upgrading their business models: Saiyi Information plans to carry out no more than 20 billion yuan in computing power server financial leasing business, creating integrated delivery of "high-performance GPU computing power leasing + AI scenario-based services"; Dongyangguang proposes upgrading from simply supplying hardware resources to exporting high-value comprehensive computing power capabilities; and Paratera adopts an asset-light scheduling route, connecting 62 computing power centers through "computing network + computing sea."

"Companies that can continue to occupy the high ground of profit in the future must possess four major elements: the ability to lock in long-term orders, reserves of high-end computing power, liquid cooling energy-saving and cost-reduction technology, and MaaS scenario-based services; small and medium-sized manufacturers of low-end computing power will continue to come under pressure, and market concentration will rise rapidly," Zhu Keli said.

(Source: Dahe Caifang) Middle of the main text: embedded advertisement