On the afternoon of May 13, the stock price of XUNCE (03317.HK), known as the "first Token stock," surged over 26% during the session, reaching a high of HK$289.
In terms of news, a Morgan Stanley report clarified that the focus of China's AI development has shifted from technology training to scenario inference, and from conceptual exploration to profit realization.
The AI industry is moving beyond the first half of technology pursuit and entering a critical phase of value monetization. Amid this industry wave, XUNCE, with its Token payment model as the core engine, has charted a new curve for business growth, also providing a replicable new sample for AI commercialization.
**AI Shifts from Training to Inference, XUNCE Takes the Lead as the "Data Hub"**
Currently, the development logic of China's AI industry is undergoing a fundamental transformation. Previously, the industry focused on technical dimensions such as model parameters and computing power competition. Now, the core objective has shifted to scenario implementation and commercial profitability.
This trend has been confirmed by multiple sources. Morgan Stanley pointed out that the core of the new phase in China's AI is capturing practical value, with inference demand replacing training demand as the main growth driver. The People's Bank of China also emphasized in its monetary policy report that AI will inject new momentum into high-quality economic development through technological iteration, industrial integration, and market expansion.
Within the AI triangle composed of computing power, algorithms, and data, computing power and algorithms are becoming homogenized. How to obtain high-quality data has become the "secret" to enhancing AI model performance. Simply put, the AI model that enables enterprises to "use effectively, affordably, and profitably" is the future core competitiveness of the industry, and XUNCE has already taken the lead in this track.
XUNCE's management stated that from the perspective of the behavioral logic of the industrial ecosystem, the company connects computing power to algorithms, becoming the "data hub," and has a naturally symbiotic relationship with the entire industry chain. For example, the company has a natural upstream-downstream cooperative relationship with large model manufacturers, not a competitive one. The company's role is to sit on the client's side. If a client chooses Model A, the company optimizes according to Model A's data requirements; if the client chooses another model, the company optimizes for that other model. Each model differs in input/output formats, debugging methods, and inherent attributes. XUNCE's value lies in helping clients achieve optimal results with their specified model. The more models a client uses, the more service opportunities for XUNCE, the more value created for the client, and the higher the resulting revenue.
**Token Payment Becomes New Growth Engine, Significant Future Growth Potential**
The foresight of the business model is directly reflected in XUNCE's impressive financial data. In 2025, XUNCE Technology delivered high-growth results: annual revenue reached 1.285 billion yuan, a year-on-year increase of 103.28%, achieving a doubling of revenue.
The growth exhibited a clear "stronger second half" characteristic: revenue in the first half was only 198 million yuan, soaring to 1.087 billion yuan in the second half, a quarter-on-quarter surge of 449%, sending a clear signal of concentrated release of market demand. Profit quality was equally outstanding, with the annual gross margin stable at a high of 61.66%, far exceeding the average level of computing power and model manufacturers during the same period, demonstrating significant profitability advantages.
Customer value also saw substantial simultaneous improvement: in 2025, customer ARPU increased from 2.72 million yuan to 5.59 million yuan, a year-on-year growth of 105.4%; revenue per employee reached 2.87 million yuan, a year-on-year increase of 135.25%, achieving dual breakthroughs in single-customer value and operational efficiency.
Entering 2026, XUNCE underwent a key transition in its business model, with the Token payment model becoming the core variable driving growth. Data shows that in April 2026, XUNCE's Token call ARR surged 300% quarter-on-quarter, and the proportion of Token payment revenue has exceeded 5%, with the annual target aimed at 20%-30%.
This marks XUNCE's shift from traditional subscription and project-based models to a new "Token payment + value sharing" model. Different from simple "charging per word," XUNCE's Token payment pricing integrates three dimensions: single-call price, actual call frequency, and module embedding depth. It is essentially "value binding," highly aligned with the trend of AI penetrating enterprises' core businesses. Management stated that within the next two to three years, the Token payment model will gradually become the core component of the company's revenue.
**Significant Premium for Vertical Tokens, High-Value Sectors Continue to Expand**
Traditional data product development involves high costs and long cycles, severely limiting enterprises' willingness to transform raw data into tradable data products. In the AI inference stage, the core of industry competition is high-quality real-time data, not massive, coarse-grained historical data. This is precisely the core advantage XUNCE has cultivated over 10 years.
Taking the biopharmaceutical and medical device industry as an example, current pain points in the gene therapy field include data fragmentation, high R&D costs, and weak personal data control. As a low-cost tool, AI supports gene therapy Token factories in processing multi-source data, paired with personal health and longevity archives for autonomous data control, and integrated with specialized small models. This not only reduces enterprise development costs and ensures data compliance but also provides individuals with precise, customized health management and longevity therapy services, offering a practical path for empowering data elements in the biopharmaceutical and medical device field.
XUNCE has deeply cultivated 10 high-growth sectors including finance, robotics, and commercial aerospace. Centered on AI Data Agent, it has built a full-chain technical system covering data acquisition, cleaning, real-time computing, and model optimization. Leveraging high-barrier, industry-specific private data accumulated over a decade, it has become the "high-quality data fuel" for the AI application and model layers, forming a difficult-to-replicate competitive barrier.
The explosion in industry demand has further amplified the value of XUNCE's barriers. Currently, China's daily Token call volume exceeds 140 trillion, growing over a thousandfold in two years; institutions predict that AI inference Token consumption will surge 370 times by 2030, and the AI Agent market will grow 139% in five years. The proliferation of intelligent agents is driving sustained soaring demand for vertical Tokens.
Relying on its data barriers, XUNCE achieves a high premium for vertical Tokens: pricing reaches $10-100 per million Tokens, more than ten times that of similar overseas products. The company's management has publicly stated, "In the future, the company will also expand into new billion-level tracks such as robotics training data, commercial satellite computing, and aviation industry computing, continuously opening up growth space."
As the penetration rate of the Token payment model increases and high-value vertical sectors continue to materialize, XUNCE is expected to achieve a dual uplift in performance and valuation, growing into a benchmark enterprise in the era of AI commercialization, with its sights set on a market capitalization of 200 billion!
Comments