StarRing Technology lists on HKEX to fund AI×Data strategy; software licensing revenue falls 45% YoY

For foundational software companies originally engaged in database and big data platform businesses, AI is not only expanding product lines but also transforming software sales and project delivery models.

On September 21, Transwarp Technology (688031.SH, 06727.HK) officially listed on the Hong Kong Stock Exchange, with its H-share IPO price set at HK$49 per share. The company was listed under Chapter 18C of the HKEX Listing Rules as a specialist technology company. On its debut, the stock reached an intraday high of HK$49.5 per share. As of September 23, the closing price stood at HK$42.72 per share.

Transwarp had previously listed on the STAR Market in 2022. In this Hong Kong fundraising round, 30% of the proceeds are planned for enhancing and optimizing R&D and innovation capabilities for its product and solution stack, driving full-stack technology upgrades through "AI × Data." Compared to its earlier listing, the company's technology investment focus has further extended toward AI infrastructure.

From "Selling Software" to "Software + Services"

Transwarp's original business was primarily built on data infrastructure.

According to its prospectus, the company's infrastructure software business currently comprises three main product categories: big data and cloud infrastructure platforms, distributed databases, and data development and AI analysis tools.

Unlike previous disclosures that categorized products by specific types, the Hong Kong prospectus now consolidates core business under "AI and Big Data Infrastructure Software and Related Technology Services," which is further divided into:

• Infrastructure software
• Infrastructure software and related technology services
• Technology services

Corresponding charging models also differ:

• Infrastructure software: Primarily charges a one-time perpetual license fee
• Software and related technology services: Combines software license fees with service fees
• Technology services: Includes recurring fees for maintenance services and project-based fees for other technology services

This shift is already reflected in the revenue structure, with "software product licensing + supporting services" becoming the main driver of revenue growth.

In 2025, Transwarp's main business revenue was approximately RMB 447 million, of which software product and technology service revenue accounted for RMB 362 million, or about 81%. Breaking this down further:

• Standalone software product licensing revenue: RMB 54.009 million, down 7.14% year-on-year
• Software product licensing and supporting services revenue: Increased from approximately RMB 145 million to RMB 212 million, rising from 39.17% to 47.32% of total revenue
• Technology service revenue: RMB 96.4654 million

Delivery methods have also changed. The prospectus notes that standardized software deployment typically takes about one week, while integrated projects combining software and technology services may take weeks to months.

Kuang Yuqing, founder of LensAI, believes that software companies originally focused on data management are logically extending toward AI. As AI changes software development and interaction methods, new software forms such as intelligent agents and AI operations will emerge, driving the transformation of existing software, infrastructure services, and IT services toward AI technology services. However, the barriers to entry for such basic services depend on each company's specific business capabilities.

AI Demand Reflected in Recent Revenue

According to the first-half 2026 financial report, revenue from AI and big data infrastructure software—including basic software products and product-related technology services—increased from RMB 111 million in the same period last year to RMB 139 million. The company attributed this growth to accelerated AI adoption across industries, driving increased procurement demand from government and state-owned enterprise clients. During the same period, overall company revenue rose from RMB 152 million to RMB 174 million, a year-on-year increase of 14.1%.

While demand is rising, how enterprises allocate their AI budgets also affects the market space available to software vendors.

Guo Tao, an angel investor and senior AI expert, noted: "At the demand level, customer procurement priorities still lean heavily toward hardware resources like GPUs. Infrastructure software is often viewed as a supporting component of hardware, with independent willingness to pay not yet established. Software value is difficult to price independently, and customers prefer bundled procurement, which compresses the premium space for software."

As of the end of March 2026, Transwarp had served over 1,800 clients across more than ten industries, including finance, government, energy, healthcare, transportation, and manufacturing. Among these are 105 Fortune China 500 companies. From 2023 to the first quarter of 2026, repeat purchase revenue from existing clients accounted for over 70% of total revenue on average.

The prospectus also shows that approximately 60% of consulting service clients subsequently purchased the company's software products, while about 20% of clients for AI and big data infrastructure software and related technology services were converted from prior consulting engagements.

The Challenge of Revenue and Profitability

With the change in revenue structure, whether the growth of "software + services" can translate into profits remains a key question.

From 2023 to 2025, Transwarp's operating revenue was approximately RMB 491 million, RMB 371 million, and RMB 448 million, respectively. Over the same period, net losses were approximately RMB 289 million, RMB 344 million, and RMB 245 million.

Regarding persistent losses, Transwarp stated: "The company's current operating revenue scale is relatively small and has not yet achieved significant economies of scale, making it insufficient to fully cover various period expenses and cost investments." According to financial reports, the company's R&D investment, overall sales expenses, and substantial investments in back-office construction (including operations management, human resources, financial internal controls, and office space leasing) have contributed to ongoing losses.

In terms of R&D investment, from 2023 to 2025, the company's R&D expenses were RMB 223 million, RMB 227 million, and RMB 204 million, accounting for approximately 45%, 61%, and 46% of revenue, respectively.

In the AI infrastructure software industry, multiple experts interviewed identified "standardization" as a key variable in converting revenue into profit.

An Guangyong, an expert from the Credit Management Committee of the All-China Federation of Industry and Commerce, believes the core constraint on profitability lies in the structural contradiction between "difficulty in product standardization" and "difficulty in independent software pricing."

He further explained:

• Budget concentration on hardware limits software premium: Enterprise IT spending is heavily skewed toward hardware such as GPUs and servers, with software often treated as a bundled accessory lacking independent pricing power and high-margin premiums.
• High heterogeneity from chips to applications: Software vendors are frequently mired in project-based delivery, with high labor and adaptation costs leading to increasing marginal costs as scale grows.

Wu Zewei, a special researcher at Suzhou Bank, added: "First, demand is fragmented. Client needs for AI infrastructure vary greatly—different industry scenarios, models, and deployment methods make it difficult to create standardized products. This results in a high proportion of project-based delivery, making large-scale replication difficult. Revenue growth does not equal profit growth."

According to Gartner data, global AI spending is expected to reach USD 2.67 trillion in 2026, with AI infrastructure accounting for USD 1.48 trillion, but AI software only USD 461.6 billion.

Zhang Li, Honorary President of the Influence Research Institute and Chairman of Guangzhou Haofu Technology Co., Ltd., believes the biggest constraint on scale profitability is the "gap between technical capability and commercial productization capability." Global AI investment is highly concentrated in "buying hardware," leaving a relatively limited revenue pool for software vendors.

This issue is also observable in the gross margins of Transwarp's different business segments.

In 2025:

• Standalone software product licensing gross margin: 88%, down 2.24 percentage points
• Software product licensing and supporting services gross margin: 56.42%, up 2.95 percentage points year-on-year
• Technology services gross margin: 64.64%, up 7.50 percentage points

The faster-growing software licensing and supporting services segment had a significantly lower gross margin than standalone software product licensing.

Meanwhile, the company's operating cash flow remained negative. From 2023 to 2025, net cash flows from operating activities were approximately RMB 365 million, RMB 317 million, and RMB 108 million, respectively. In the first half of 2026, net cash flow from operating activities was RMB 156 million.

Expanding Competitive Landscape

The competitive boundaries of the AI infrastructure software market are also widening. According to Frost & Sullivan data, by 2025 revenue, Transwarp ranked fifth among AI infrastructure software providers in China, with a market share of 2.7%.

As cloud vendors, ICT vendors, large model companies, and traditional foundational software providers all extend into data, models, computing power, and intelligent agent platforms, competition is no longer limited to database or big data products alone.

Zhang Li believes that barriers are shifting from "single-point technology" to "full-stack integration and scenario deployment capabilities." Gartner notes that enterprise clients increasingly prefer to "obtain embedded AI functionality from existing software vendors" rather than purchasing standalone AI platforms. This means the ability to embed into clients' existing workflows matters more than model parameter counts.

"Companies that will capture significant market share in the future need three capabilities: computing resource scheduling, deep industry scenario expertise, and sustainable charging models (shifting from project-based to API calls and subscriptions)," Zhang Li said. "Among ICT vendors and independent software vendors, only those that can complete the transformation of 'AI capability into products, and product revenue into subscriptions' will have a chance to stay in the game."