Editor｜Chen Xiaoran

On September 29, Beijing DataCanvas Technology Co., Ltd. (hereinafter referred to as DataCanvas) filed a listing application with the Hong Kong Stock Exchange, with China Galaxy International Securities (Hong Kong) Co., Ltd. acting as the sole sponsor. The company aims to become the “pioneer stock in the new intelligent computing cloud sector.”

Transitioning from AI infrastructure software to intelligent computing cloud services, DataCanvas is bringing a higher-investment, higher-growth business model to the capital market.

Software Company's Evolution

Founded in 2013, DataCanvas initially focused on data science platforms and AI infrastructure software, assisting enterprises in developing, managing, and deploying models.

With the advent of large-scale models, clients' needs have expanded beyond development tools to include training computing power, inference support, and cross-device scheduling. In response, DataCanvas has extended its offerings to include intelligent computing operating systems, intelligent computing clouds, and computing power services.

Presently, DataCanvas offers products such as the DataCanvas Intelligent Computing Cloud (Alaya NeW Cloud), DataCanvas Intelligent Computing Operating System (Alaya NeW OS), and “Computing Power Packages.” The intelligent computing cloud provides model training, inference, and management services for enterprises and developers, while the operating system manages underlying tasks such as computing power scheduling.

This product matrix is already evident in the company's revenue growth. From 2023 to 2025, DataCanvas's operating revenue surged from RMB 106 million to RMB 1.098 billion, achieving a compound annual growth rate of 935.85%. In the first half of 2026, revenue increased by 92.0% year-on-year to RMB 588 million.

However, revenue growth alone does not guarantee business maturity. This listing application cannot be assessed merely through the lens of a software company. The true value of software products lies in clients' willingness to continuously pay for functions and services. Intelligent computing clouds also grapple with challenges such as equipment investment, resource idleness, and electricity and maintenance costs.

While increased client usage can drive revenue growth, if computing power procurement and project expansion outpace demand, financial pressure will mount accordingly.

In 2025, DataCanvas reported an adjusted net loss of RMB 91.27 million, which narrowed to RMB 14.88 million in the first half of 2026. This loss reduction is attributed to revenue growth, yet computing power procurement and project expansion continue to necessitate ongoing investment.

Since its inception, DataCanvas has secured multiple rounds of financing, with the latest round completed in June 2026, raising a total of RMB 500 million.

While capital has fueled technological and business expansion, going public introduces a new challenge: evaluating how much stable revenue has been generated from past investments and determining the capital required for future growth.

Computing Power Transactions

DataCanvas introduces the “Degree of Computing Power” (DCU) as a measurement metric and offers computing power packages, aiming to simplify complex computing power procurement into an easy-to-understand, on-demand service.

This approach holds practical appeal for enterprises new to AI application development. They may require short-term model training but need long-term inference operations. Procuring equipment for peak demand can lead to significant resource idleness during normal periods.

DataCanvas aggregates tasks from various clients on its cloud platform, enhancing equipment utilization efficiency through scheduling and delivering computing power alongside development tools.

Clients prioritize quick project initiation, stable task operations, and reasonable costs. While the DCU metric lowers the purchase threshold, clients' continued payment depends on their actual experience and pricing.

Local intelligent computing centers represent another scenario where DataCanvas demonstrates its delivery capabilities. The Huangshan “Dawei” Intelligent Computing Center, a collaboration between DataCanvas and Huangshan Tourism Group, has an estimated project investment of RMB 260 million and has already launched large model services.

It is crucial to clarify that the RMB 260 million represents the project investment estimate and should not be conflated with DataCanvas's revenue or its own capital expenditures.

Post-project completion, the amount of computing power purchased and cost-sharing among partners determine the operational value for the company.

Regional market data offers additional insights. According to an Analysys report, based on the third-party inclusive intelligent computing cloud market in East China in the second half of 2025, DataCanvas held a 10.2% market share.

This data underscores DataCanvas's presence in a specific niche market but does not allow for extrapolation to estimate its national market share or infer profitability directly.

According to CIC data, based on 2025 revenue, DataCanvas ranks ninth among Chinese intelligent computing service providers and is the sole new intelligent computing cloud provider in the top ten.

Market share provides a snapshot of competition, while client renewal and cash collection represent longer-term challenges.

From products to projects, DataCanvas has established a relatively comprehensive business framework. Moving forward, the capital market will scrutinize this framework: What proportion of revenue comes from cloud services versus projects? Can software-driven value-added enhance overall gross margins? After validating a regional project, how much manpower and capital will be required to replicate it elsewhere? These questions delve deeper into the business's core than merely assessing “how much computing power it possesses.”

Growth Demands Prudent Planning

Opportunities in intelligent computing clouds are expanding from training to inference. While model development necessitates concentrated computing power usage, models in products generate more frequent invocation demands.

DataCanvas, operating at both the underlying scheduling level and as a cloud service provider, has the potential to convert its technical capabilities into cost advantages by enabling resource sharing and reducing idle time across different tasks.

Nevertheless, challenges persist.

Large cloud providers possess significant capital, infrastructure, and an existing client base, while other intelligent computing service providers compete for price-sensitive development teams. Clients can compare platform performance and pricing, while suppliers must ensure equipment updates, stable operations, and ongoing services.

When computing power prices decline, increased usage may not necessarily translate into profit. The continuous emergence of new equipment also alters the economic value of existing equipment.

Moreover, local projects adhere to their own operational rhythms. Signing, construction, delivery, and actual client usage often do not coincide.

The successful launch of a center showcases the company's implementation capabilities. However, the center's ability to secure sufficient paid tasks post-completion determines whether investments can be continuously recovered.

For intelligent computing cloud enterprises, the number of projects, managed scale, and revenue quality must be evaluated separately.

Therefore, the most noteworthy aspect of DataCanvas's listing application is not its ambitious computing power goals but its continuous operating data: which businesses drive revenue growth, how gross margins evolve, whether receivables and operating cash flow keep pace with revenue, whether major clients are concentrated, and whether computing power resource utilization can support the next round of expansion.

Only by collectively examining these figures can we determine whether DataCanvas has transitioned beyond a growth phase solely driven by financing.

DataCanvas's transformation addresses a genuine need in the AI industry: enterprises require more convenient access to and utilization of computing power.

After filing its listing application, investors will focus not only on how much computing power DataCanvas can schedule but also on clients' willingness to continue using its services, the revenue retained per transaction, and the cost of growth.