On June 20th, Huatai Securities published its inaugural coverage report on XUNCE (03317), a leading domestic real-time data company often compared to Palantir, assigning a "Buy" rating and a target price of HK$240.08. The firm expresses confidence in the company's ability to sustain high earnings growth and elevate its long-term growth trajectory and valuation ceiling, leveraging its robust real-time data technology, cross-industry expansion opportunities, and a token-based billing model. The report explicitly states that as China's leading scenario token provider akin to Palantir, XUNCE is capitalizing on the high technical and client barriers established in the asset management sector to replicate its solutions across various industries. Concurrently, its TokenOS platform facilitates a business model upgrade to token-based billing, driving revenue growth.

Clear Industry Tailwinds: Low Real-Time Data Penetration and Palantir's Validated Growth Model

China's real-time data infrastructure market remains in its early stages, offering substantial room for growth. According to Frost & Sullivan data, the overall penetration rate of real-time data infrastructure in China was only 3.6% in 2024, with the asset management sub-sector at just 2.8%. Huatai Securities notes the industry is still nascent with significant incremental space. Policy support and the acceleration of AI Agent industrialization are creating synergistic momentum. In June 2026, the National Data Bureau issued an implementation plan to promote the construction of high-quality industry datasets, explicitly advocating for dataset commercialization and assetization, and exploring a token-based value system. Frost & Sullivan forecasts the Chinese real-time data infrastructure and analytics market will grow from RMB 2.24 billion in 2024 to RMB 5.05 billion in 2029, a CAGR of 22.0%, with the asset management sub-market expanding from RMB 2.1 billion to RMB 11.6 billion, a CAGR of 40.7%.

Huatai Securities posits that the ongoing penetration of AI Agents will transform larger volumes of data from offline assets into real-time production materials, potentially shifting real-time data infrastructure from a big data construction cycle to a new AI-driven cycle. The development path of the overseas benchmark, Palantir, provides a crucial reference for AI software companies built on data foundations. The report highlights that Palantir, leveraging its Ontology data architecture and AIP platform, has continuously enhanced client stickiness and value per client, leading to a significant acceleration in revenue growth. In 2025, Palantir's revenue grew 56.6% year-over-year, with net profit surging 251%. Huatai believes Palantir's trajectory offers an important roadmap for data-foundation AI software firms and further validates the long-term value of the scenario token and real-time data infrastructure sector. The report emphasizes that combining enterprise data foundations with large model Agents can continuously expand application scenarios within existing client workflows, increasing value per client and platform monetization capabilities.

Core Transformation: TokenOS Reimagines Commercialization with Significant Scenario Token Premiums

XUNCE's revenue was previously based primarily on subscription and transaction models. The report notes that the subscription model has a clear revenue ceiling, limited by fixed ARPU and unable to fully capture incremental contributions from high-value scenarios. In May 2026, the company officially launched TokenONE, the world's first TokenOS operating system, building capabilities for tokenizing enterprise private data, generating scenario tokens, metering usage, and tracking business value. Huatai Securities views the company's transition from project-based and subscription models to a token-based transaction model as aligning with the characteristics of AI data services, offering stronger potential for scalable growth.

The TokenOS establishes a vertical data token production system, packaging non-standard private data into scenario tokens embedded with industry know-how, ready for large model invocation. Unlike general large model tokens, which primarily correspond to base model inference calls, the company's scenario tokens emphasize real-time data, industry knowledge, and business scenario adaptation, forming a basis for differentiated pricing. In April 2026, XUNCE's Annual Recurring Revenue (ARR) from token calls grew 300% quarter-over-quarter, with token-based billing revenue exceeding 5% of total revenue, targeting 20% to 30% for the full year. Huatai expects that, referencing Palantir's platform monetization path, the token billing model will help standardize, quantify, and monetize data value. As the proportion of token billing revenue increased in Q1 2026, the company's long-term growth potential and valuation flexibility are expected to expand further.

Notably, while market concerns exist that open-source large models could depress general token prices and pressure the company's business, Huatai's report offers a differentiated view. It clearly distinguishes the different pricing logics between general tokens and scenario tokens. General tokens mainly correspond to base model inference capability, with prices more affected by model openness, inference cost optimization, and vendor competition. XUNCE's scenario tokens are built on enterprise private data, undergoing processes like data acquisition, cleansing, standardization, real-time computation, industry rule encapsulation, and model adaptation to serve specific business scenarios such as investment decision-making, risk control, energy dispatch, and healthcare. Their value derives more from data quality, real-time requirements, scenario complexity, and business outcomes, rather than solely from underlying model inference costs.

XUNCE's scenario token pricing logic primarily considers three dimensions: the scarcity and acquisition difficulty of industry data; client requirements for data processing timeliness; and industry and business scenario complexity—higher difficulty and business value-add command higher pricing. The company's scenario token call prices range from $10 to $100 per million tokens, over ten times the price of general large model tokens. The report further suggests that price reductions for open-source models could actually improve the environment for XUNCE's token billing business. Lower underlying model inference costs would reduce the overall investment for clients piloting and scaling AI applications, improving the business return per token used, making clients more willing to adopt usage-based or outcome-based billing models across more business processes. Huatai therefore believes that falling general token prices, primarily affecting the base model inference layer, could enhance the return on investment for client AI applications, driving more business scenario adoption and growth in token usage.

Dual Growth Engines: Solid Asset Management Foundation and Multi-Industry Expansion

In 2024, XUNCE held an 11.6% revenue share, ranking first in China's asset management real-time data solutions market. The company already covers China's top ten asset management institutions, with its leading client base providing crucial support for business expansion. The report believes the asset management sector has endowed the company with difficult-to-replicate technical barriers. Its long-term focus on high-compliance, high-timeliness scenarios in asset management has honed its capability for millisecond-level heterogeneous data processing, cleansing, and analysis, building an end-to-end service system covering data collection, governance, real-time computation, and AI adaptation. By the end of 2025, the company had developed over 300 standardized data modules, employing a flexible, building-block delivery model that allows for rapid assembly and fine-tuning based on different industry attributes and client needs.

Huatai judges that the asset management industry has the highest demands for data accuracy and timeliness, where any data delay, inconsistency, or human error can be quickly amplified into a risk event. The barriers formed by this financial-grade real-time technology and industry know-how are difficult for cloud providers to replicate with standardized products. Having started in the high-barrier asset management sector, the company is now replicating its mature products across diverse industries. The report explicitly states that diversified industries have now become the company's largest revenue source and core growth engine. The revenue contribution from diversified industries surged from 25.6% in 2022 to 79.6% in 2025, with growth rates exceeding 100% from 2023 to 2025. Its products have been deployed across ten high-value sectors, including telecommunications, urban governance, industrial manufacturing, power, healthcare, commercial aerospace, and robot training.

In terms of delivery efficiency, leveraging its 300+ standardized modules and self-developed TokenOS, the company is continuously advancing cross-industry deployment in telecommunications, advanced manufacturing, commercial aerospace, and more. For new industries, only the initial industry modeling and know-how accumulation require a 1-3 year adaptation period. Once a benchmark scenario is proven and industry modules are solidified, replication for other clients within the same industry can be scaled rapidly.

Regarding client value, the company's average revenue per user continues to rise. From 2022 to 2025, its total paying clients increased from 182 to 230, while ARPU grew from RMB 1.58 million to RMB 5.59 million, a 105.5% year-over-year increase in 2025. The report suggests that as clients demand more comprehensive services spanning data access, governance, analysis, and AI application adaptation, the depth of platform usage per client is likely to increase, strengthening renewal and expansion momentum. In terms of profitability, cross-industry gross margins are continuously improving. For instance, the urban management sector's gross margin rose from 37.0% in 2022 to 72.1% in 2024, demonstrating the replicability of the model. Huatai believes that, referencing Palantir's path from high-barrier government and defense sectors to multi-industry expansion, as the company's industry modules continue to accumulate and reuse rates increase, it is well-positioned to penetrate various high-value sectors, building a long-term second growth curve.

Financial Outlook: Expecting Strong 2026 Revenue Growth and 2027 Profit Inflection

Supported by the triple drivers of continued expansion in diversified industries, steady growth in transaction-based fees, and the rapid rise of token billing, the company's revenue scale is expected to keep expanding. Huatai Securities forecasts XUNCE's revenue for 2026/2027/2028 to be RMB 2.357 billion / RMB 3.915 billion / RMB 5.940 billion, representing year-over-year growth of 83.4% / 66.1% / 51.7%. Adjusted net profit is projected at RMB 2 million / RMB 177 million / RMB 553 million for those years, corresponding to adjusted net margins of 0.1% / 4.5% / 9.3%. The total target market capitalization is RMB 66.787 billion, equivalent to HK$77.473 billion, translating to a target price of HK$240.08 per share, representing a 2027 P/E of 17.06x. Coverage is initiated with a "Buy" rating.

In conclusion, the convergence of AI Agent industrialization and data element policies is transforming real-time vertical data from a traditional auxiliary tool into a core production material for AI decision-making. XUNCE, leveraging its millisecond-level real-time data processing capabilities honed in asset management and its self-developed TokenOS, has pioneered a business model transition "from traditional software delivery to extending into usage-based and business value-based billing." Its diversified industry expansion and token billing model together form a powerful dual-engine for growth.