As AI becomes a dominant theme, more companies are discussing large models, intelligent agents, and real-world applications. However, for capital markets, what truly matters is not just what is said, but what is actually achieved. On March 23, Qingsong Health Group (02661) released its full-year 2025 results, providing a concrete case study for the industry. In 2025, the company reported revenue of RMB 1.256 billion, a year-on-year increase of 32.9%. Revenue from health-related services reached RMB 926 million, surging 50.2% and accounting for 73.7% of total revenue. Adjusted net profit was RMB 92 million, up 9.0% year-on-year.
Viewed purely from a numerical perspective, this is simply a report showing steady growth and optimized structure. But a deeper look reveals the report's true significance: AI has moved beyond technical narrative and is now actively embedded in content production, user operations, health services, and industrial collaboration, directly contributing to revenue composition.
When AI is integrated into the revenue structure, five core modules build a solid moat. The most obvious signal comes from changes in the customer base. In 2025, pharmaceutical clients contributed 67.9% of Qingsong Health's revenue, up from 55.3% the previous year. Revenue from this segment grew from RMB 523 million to RMB 853 million, an increase of 63.2%, almost single-handedly driving the group's overall growth. Meanwhile, revenue from insurance clients remained largely flat in absolute terms, with its contribution falling from 35.0% to 26.6%.
What drives this change? The answer points directly to AI capabilities. Pharmaceutical companies are willing to pay for Qingsong Health's services partly due to its user base of nearly 168.6 million registered users and approximately 63.8 million followers on WeChat. More importantly, the company translates its AI capabilities into services that pharmaceutical firms actually need: digital marketing for precise patient outreach, scalable health education content production, and digital tools covering the entire clinical research lifecycle.
Full-year digital marketing revenue reached RMB 713 million, up 52.1%, accounting for 56.7% of the group's total revenue. This growth is not merely due to an increase in client numbers but is directly linked to the scaling of AI-driven content production. In 2025, the platform produced 163,331 digital marketing videos, a 55.2% increase year-on-year, while related content views surged 238.1%. The gross margin also improved from 10.3% to 13.6%, indicating enhanced scale effects and bargaining power.
From a technical perspective, Qingsong Health's self-developed AIcare technology platform consists of five core modules. While appearing as a platform system, these modules correspond to specific business functions that directly impact revenue. The i-Odin content creation engine has cumulatively produced over 1.2 million health education pieces, including more than 970,000 articles and 260,000 videos. The 238.1% growth in content views is supported by i-Odin's generative AI capacity—a rate unachievable by human editors. This directly supports the RMB 713 million in digital marketing revenue.
The i-Galaxy AI marketing platform has created over 910 feature tags and registered 23 invention patents and 2 software copyrights. It addresses a core challenge in digital health: delivering content, products, and services precisely to the right users in scenarios involving large user bases, long service chains, and diverse needs. The 63.2% growth in pharmaceutical client revenue is underpinned by i-Galaxy's data engine.
The i-Phoenix enterprise service platform provides insurance partners with smart operations, risk control, and monitoring services. It directly supports the RMB 221 million in insurance technology services revenue, which boasts a 97.0% gross margin—a near-zero marginal cost model typical of AI-driven service profits, serving as a stable profit anchor for the group.
The i-Centaurus user platform supports AI health management services within comprehensive health service packages. In 2025, it reached 44,147 users, with 211,508 interactions, representing year-on-year increases of 26.3% and 52.8%, respectively. Corresponding revenue from comprehensive health service packages was RMB 79.4 million, with gross margin rising from 27.3% to 36.7%.
The i-Magellanic medical digitalization platform offers pharmaceutical clients full-lifecycle services from research design and physician education to clinical and post-market studies. Related digital medical research support services generated revenue of RMB 41.7 million, up 29.5%, with gross margin jumping from 8.1% to 20.5%. A total of 46 projects were initiated.
These five modules are not laboratory demonstrations but operate in real revenue-impacting areas: content, outreach, management, research, and operations. When their effects are traceable in financial reports, AI is no longer a futuristic concept but an integral part of current business operations.
Unlike many companies still at the "model demonstration" or "product trial" stage, Qingsong Health follows a clearer AI development path: instead of first building a standalone AI product and then seeking application scenarios, it embeds AI into existing operations, uses business growth to validate AI effectiveness, and then promotes capability productization and platformization.
This also explains why the launch of QSevidence and the MedClaw collaborative framework naturally aligns with this financial report. In March 2026, Qingsong Health introduced QSevidence, a medical AI agent based on evidence-based methodology, designed for real clinical settings to provide doctors with explainable, verifiable, and traceable medical decision support. The simultaneous launch of the "QSevidence·MedClaw Collaborative" integrates QSevidence with the OpenClaw multi-agent framework, covering scenarios such as evidence-based clinical decision support, multi-guideline consistency checks, case discussions, and research assistance, enabling a full closed-loop from task decomposition and evidence retrieval to guideline comparison, conclusion generation, and process archiving.
From a product timeline perspective, this is a natural extension of prior AI implementation. The i-Odin, i-Galaxy, i-Centaurus, and i-Magellanic modules in the financial report provide evidence of validated underlying capabilities, while QSevidence and the MedClaw collaborative aim to apply these capabilities to higher-threshold, higher-value medical decision-making scenarios.
For capital markets, this progressive approach is more meaningful than launching an isolated AI product. It demonstrates that Qingsong Health's AI evolution is continuous—from content and marketing to user management, research support, and now medical AI agents. The earlier phase proved AI can enhance efficiency and drive revenue; the next phase explores whether AI can become a standalone high-value product or service.
In the current market environment, telling an AI "story" is easy; turning AI into business, revenue, and products is hard. The significance of Qingsong Health's 2025 report lies in providing a real-world example: while many companies are still talking about what AI might bring, Qingsong Health has already incorporated AI into its financial statements, revenue structure, and product roadmap.
Notably, the company's financial stability supports this strategy. As of the end of 2025, cash and cash equivalents increased to RMB 862 million, with no bank borrowings throughout the year. Net cash flow from operating activities remained at RMB 82.089 million. Approximately HK$502.3 million in net proceeds from the global offering remains unused and will be allocated according to the prospectus for technology enhancement, medical research, and brand building.
This means Qingsong Health has ample "ammunition" for further AI productization, from QSevidence to broader explorations. If 2025 was the year Qingsong Health wrote AI into its business, 2026 may see it embedding AI into more specialized healthcare scenarios with the launch of QSevidence and the MedClaw collaborative. The market's real concern may not be who first mentioned AI, but who first turned it into a verifiable reality.
Comments