迅策 (03317.HK) 2026智通财经夏季路演大会
文章语言：
简
繁
EN
Share
Minutes
原文
会议摘要
With the ability of tokenized data, enterprises can achieve efficient, accurate and measurable model calls and build competitive advantages. In the face of many domestic and foreign suppliers, only a few can provide full-link services, overseas Partier for its main competitors. The transition to the by token charging model is highly recognized by customers, especially in terms of improving ROI or risk control. The company is open to investment and mergers and acquisitions, aiming to expand new industries and enhance the industrial chain. As a AI real-time data infrastructure and analysis service provider, Xunce Technology started from the financial industry since its establishment in 2016 and now covers telecommunications, electricity, high-end manufacturing and other fields. Its revenue will reach 1.28 billion in 2025 and is expected to reach 20 to 2.5 billion in 2026. Diversified industry revenue accounts for more than 80%. The company adopts the transaction project system, subscription system and by token charging model, is committed to promoting the application of AI technology in enterprise scenarios, and will continue to expand new industries and global layout in the future.
会议速览
Xunce Technology: Diversified Industry Expansion and Value AI Real-Time Data ServicesThis paper introduces the development process of Xunce Technology as a AI real-time data infrastructure service provider. It started from the financial industry and gradually expanded to diversified industries such as telecommunications, electric power and high-end manufacturing. By providing millisecond-level data processing and analysis services, it builds an enterprise data base, outputs SaaS scenarios and AI digital employees, thus realizing the solution of cross-industry complex scenario problems, emphasizing the growth trend of high customer retention and cooperation amount.
Comprehensive Analysis of Integration of Enterprise Data Governance and AI Industry ChainThe ability of enterprises to call private domain data through natural language, combined with local rules and know how, to achieve decision-making data presentation is discussed. It emphasizes the reusability of cross-industry data governance and its position as the core link of data processing and analysis in the AI industry chain, and demonstrates the full link service from data acquisition to model optimization, especially highlighting the importance of one-stop, safe and accurate data processing for major customers.
Smart Energy and Vertical Token: A New Model for Reshaping Enterprise AI ApplicationsBy showing the case of smart energy management and virtual power plant, the core value of scene vertical Token in enterprise AI applications is expounded, that is, encapsulating private data and local rules, solving enterprise-level data processing problems, optimizing energy allocation, predicting power price trends, and achieving more efficient and personalized service delivery.
Token OS: Promoting Enterprise Data Standardization and Industrial AI LandingThis paper introduces the Token OS system, which aims to realize the standardization and tokenization of enterprise data, covers nine major links of data processing, calls vertical token through intelligent body, connects multiple computing power and large model, realizes the quantifiable, pricing and compliance review of data, and helps the large-scale application and value mining of AI in enterprise scenarios.
Corporate Financial Growth and New Industry Expansion StrategiesThe company's revenue continues to grow and is expected to reach 20-2.5 billion in 26 years. The revenue structure is diversified, and the proportion of subscription and token fees is increasing. High gross margin, new industry expansion brings high returns. Operating costs in research and development-oriented, high profit potential, RP model innovation.
Analysis of Corporate Efficiency Improvement and Financial HealthThrough automation and AI technology to improve efficiency, personnel downsizing but high output, low debt, sufficient cash flow, accounts receivable management optimization, is expected to achieve self-blood in the future, showing sound financial and market expansion capabilities.
Data Quality and AI Model: New Dimension of Enterprise Competition and Global LayoutDiscusses the central role of data quality, governance, and tokenization in enterprise competition, as well as the application of AI models to solve specific scenario problems. Mention the demand for high barriers in the financial industry, the demand for high-quality data in the AI era, and the increase in token calls driven by open cloud. Looking ahead to the future development of Xunze, including industry expansion, fee-based upgrades, product technology capabilities, strategic cooperation with head manufacturers, and a global layout under the premise of compliance. Emphasize the tokenization of data assets to improve the freedom and value of data use.
General and vertical Token price trend analysis: enterprise-level scenario demand drives prices up.This paper discusses the different properties and price trends of general Token and vertical Token, and points out that the price of general Token decreases, while the price of vertical Token increases due to the increase in demand for enterprise-level complex scenes, emphasizing the importance of high-quality data processing and tokenization.
Competitive landscape and industry expansion strategy discussion: domestic and overseas competition analysis and investment mergers and acquisitions considerations.The competition pattern in the field of data infrastructure at home and abroad is discussed, and it is pointed out that some overseas suppliers can provide full-link services, but there is no direct competitor in China. Mention the company's unique advantages in implementing model call metering and payment through the by token charging model. At the same time, it expressed an open but cautious attitude towards investment mergers and acquisitions, especially in the expansion of new industries to seek opportunities for industrial chain extension.
The company's transition to the Token charging model: double verification of customer acceptance and business growth.Discusses the company's transition from a subscription or project-based to a Token fee model, emphasizing customer acceptance of the pay-per-use model and how this model validates the value of the product and market leadership, while mentioning revenue growth and increased customer trust.
要点回答
Q：What does Xunce Technology do and what is its core business?
A：Xunze Technology is a leading service provider of AI real-time data infrastructure and analysis in China, with AI data agent as the core, focusing on one-stop products and services of real-time processing and analysis at the millisecond level. Founded in 2016, the company initially started in the financial management industry, mainly serving customers such as quantitative hedging, achieving millisecond-level processing in speed and 100% accuracy in accuracy, providing decision support for customers.
Q：What are the changes in business areas and products and services after ten years of development?
A：In the past ten years, Xunze Technology has expanded from a single service to the financial industry to a diversified industry and cross-industry development. By the end of 2025, the share of financial revenue in total revenue will fall to 20 per cent, with the remaining 80 per cent coming from telecommunications, electricity, high-end manufacturing, urban operations and emerging areas such as aggregate intelligence, low-altitude economy, medicine and health care. At the same time, the company has accumulated more than 300 modules to solve complex scenario problems in various industry scenarios, which can bring higher reusability, and has established long-term cooperative relationships with more than 230 enterprises, with high customer retention. At the product level, Xunce Technology provides a data butler service that is deeply deployed in the customer's local cloud or public cloud. Through standardization and labeling, it processes data at different levels, builds an enterprise's own data base, and outputs SaaS solutions or AI digital employees according to the needs of business scenarios.
Q：What is the positioning of Xunce Technology in the industrial chain?
A：In the AI industry chain, Xunze Technology does not compete with GPU manufacturers, model manufacturers and cloud manufacturers, but acts as a collaborative partner to efficiently play the core role of the data link. Specifically, Xunce, as the downstream of GPU computing power, is responsible for data processing and analysis. It has a downstream relationship with model manufacturers, and customers can freely choose to use data or vertical token that has been cleaned, standardized and injected with energy consumption. Maintain cooperative relations with cloud vendors to build a data center or data-driven decision-making platform for customers in the local cloud native environment.
Q：How is the strategy positioned in the implementation, infrastructure and analysis of the data domain? What is the strategic positioning of the strategy in the AI ecology?
A：Training in the data area has covered the entire link from data acquisition, cleaning and standardization to tuning and connecting large models. We have the advantage of being fast enough, accurate enough and deep into the industry at the tool layer, and it takes us one to three years to dig deep and co-create with our customers every time we enter a new industry. In addition, we provide a one-stop service in data governance capabilities, especially focusing on the needs of super large customers for absolute security, accuracy and one-stop solutions. Training Policy's strategic positioning is in the AI ecology, for data analysis, we are committed to providing large customers with one-stop, efficient and secure data processing solutions, and the ability to connect large models.
Q：How is Training's performance in delivering products in the Jusheng Intelligence industry?
A：In an industry case of Jusheng Intelligence, we provide a robot control platform for real-time monitoring and intelligent decision-making, which can manage the operation, early warning, inspection tasks and process design of all equipment in the park, and carry out centralized management and intelligent decision-making according to the needs of enterprises.
Q：What are the innovative solutions in the field of smart energy?
A：In the field of smart energy, we have realized the efficient allocation of electric energy in the region and predicted the trend of electricity prices through intelligent virtual power plants, and solved the difficulties of regional peak and low peak pain points of urban energy storage facilities.
Q：What changes have taken place in the pricing and sales model of Training Policy?
A：Before 26 years, it was mainly based on transaction and subscription-based projects. Starting in 25 years, the subscription fee model accounts for 10% and 90% comes from the transaction project system. In 26 years, we introduced the by token trading model, breaking the single revenue ceiling, and adjusting the module price and token call times according to the needs of industry vertical categories and specific scenarios, thus realizing finer price control.
Q：How does training solve the problem between enterprise private domain data and large models?
A：As a connector, the training policy encapsulates the private domain data, local rules and know how of the enterprise into a scene vertical token, and then inputs it into the large model, which not only protects the privacy and compliance of the enterprise, but also enables the large model to provide solutions for specific scenarios.
Q：What is the role of the token OS system proposed by the training policy?
A：Token OS is a AI native enterprise-level token operating system, which covers the whole process of data from nine links to five application layers. First, data standardization is transformed into data token, standardizing the data token capability of different enterprises, and injecting industry know how into the middle layer for data modeling and real-time stream computing.
Q：What are the characteristics of your token OS system?
A：The token OS system can flexibly link and adjust various computing resources, including GPU, CPU, etc., and supports access to different types of high-quality large models. It provides the interface of the whole system, monitors in real time how many tokens are generated in which link of the business, and ensures that the tokens are fully measurable, traceable and compliant, forming a set of core capabilities of the data token base.
Q：What is your core goal in terms of enterprise data tokenization?
A：Our core goal is to help all enterprises realize data tokenization and solve the problem that the value of enterprise data is difficult to price and confirm rights. By tokenizing and standardizing the private domain data of enterprises, it promotes broader data windowing capabilities, so as to realize the measurable and priceable data, and provides a key way for the industrialization of AI in specific enterprise scenarios.
Q：What is the company's financial position and revenue structure?
A：The company's finances continue to grow rapidly, with revenue reaching 1.28 billion in 2025 and expected revenue of 20 to 2.5 billion in 2026. In terms of income structure, the proportion of diversified industries has reached 80% and is still improving. In terms of revenue share, subscription revenue remained at 10%, transaction project revenue fell from 90% last year to 60% to 70% this year, while by token fee revenue grew from zero to 0 in 25 years to 20% to 30% this year. In addition, the company is actively expanding into new industries, and although gross margins are high (40-50%) in the early stages of entering new industries, they will gradually converge to a steady state (around 70-80%) as the deployment period progresses. At present, the overall gross profit margin is stable at about 60%, and the adjusted net profit has achieved break-even, and it is expected to turn losses into profits for the whole year this year.
Q：What is the company's R & D-driven characteristics and capital expenditure?
A：As an R & D-driven company, the company's R & D investment continues to grow, and the annual capital expenditure due to entering new industries is about 1 to 0.15 billion. Since the company mainly relies on high-quality data or vertical tokens to create value for customers, the cost of sales is extremely low and there is little need for large-scale marketing. The company's RP has gone beyond the simple enterprise software system pricing model, and the amount of cooperation with customers continues to grow with the value-added of high-quality data responsible for business decisions, which provides the company with a huge growth space and a flexible ceiling.
Q：How does the company perform in terms of staffing and human effectiveness?
A：The company has doubled its revenue while reducing the number of personnel. Internal through the extensive use of robots and AI for automated processing, effectively improve the efficiency and per capita output.
Q：How is the company currently expanding and staffing in new industries?
A：With the continuous accumulation of product modules, the horizontal expansion of new industries and new customers is accelerating, and the staffing tends to be streamlined. Although we are expanding into new industries this year and the number of people has increased, single-person output is high, whether compared to traditional SaaS companies or model companies.
Q：What is the company's balance sheet?
A：The company as a whole has no gaming debt and remains low. In terms of economic data, cash flow has narrowed significantly since 2025 and is expected to enter a self-made ecology in 2026 or 2027. At the same time, the company is currently selling sufficient cash, close to 1.1 billion yuan.
Q：How is the company's accounts receivable management?
A：Between 2022 and 2024, accounts receivable increased due to the impact of the outbreak and business operations. However, starting in 2024, the Company has strengthened the management of quality customers and established a stronger account warning and tracking mechanism to continuously improve its account management capabilities.
Q：What is the value of data in the future AI era?
A：In the AI era, the value of data will grow indefinitely, because the product of model performance and data quality will become the key. In the model performance, the high-quality scarcity of data, the quality of data governance, and whether it can be tokenized to assist business reasoning and analysis are the core.
Q：What is the impact of changes in market demand on the development of Xunze?
A：The change in market demand has driven the development of Xunce, from the initial focus on the financial industry to the increased demand for high-quality, vertical and local rules in the AI era, which has led to a significant increase in token calls and provided a new value enhancement path for enterprises.
Q：What are the strategies for future development plans?
A：In the future, Xunce will continue to increase the number of customers and deepen the degree of cooperation by expanding the space for industry expansion, deepening cooperation with enterprises and strengthening product technology upgrading. At the same time, it actively participates in the global layout, paying particular attention to the needs of Chinese head customers to go overseas, and is committed to promoting the off-the-mouth and wider use of data assets.
Q：What is the impact of this round of DPC token price reduction on us?
A：In fact, I have always mentioned that our token and token are completely two concepts, even completely two price trends. As you can see, the future of universal token must be a public utility. In fact, more people can lower the threshold and more enterprises can use large models with lower threshold.
Q：In the application of large models, why do the price trends of general tokens and vertical scene tokens reverse?
A：This is because when a generic token is used on a large scale, its price may go down all the way due to the needs of data governance and pre-processing, while the vertical scenario token can meet the needs of enterprises to use large models in specific scenarios, and its price will continue to rise as more complex scenarios are applied. High-quality data is transformed into a tokenized form after governance and preprocessing, becoming the smallest unit that large models can directly understand and invoke, so companies with the ability to process and provide this data will be welcomed by the market and willing to pay for it.
Q：What are the main competitors at home and abroad?
A：In the field of data infrastructure and governance analysis, there are many domestic and foreign suppliers, but most of them focus on the product provision of a certain link or several links. There are more domestic enterprises in the application layer, and there are separate suppliers in each link of the data governance standardization process. There are some overseas companies that can provide full-link services, but compared with us, we have unique advantages in the by token charging mode, data token capability, and fast, accurate, and direct model calls.
Q：Is the company considering industry replication and expansion through investment mergers and acquisitions?
A：The company is open to investment and mergers and acquisitions, and actively seek opportunities for industrial chain extension and new industry expansion. For entering the new industry, if there is a good team, the company will be willing to contact. However, in the process of investment and M & A, the company will maintain a cautious attitude to ensure that all operations can bring good industrial chain integration or new industry expansion effect.
Q：When the company transformed into token billing mode, did it encounter customer resistance and how to help the original old customers switch to token payment mode?
A：The company's transition to the by token invocation model did not encounter significant resistance, but some customers took the initiative to adopt this model. Because customers have become accustomed to using the model directly and understand the concept of token charging. The advantage of the by token call mode is that it can reduce the initial investment when customers are uncertain about the effect, and the company has confidence in its own products and can help customers improve ROI and optimize risk control by providing high-quality data that accurately supports business decisions in real time. Judging from the data growth in April and the expected doubling of by token revenue this year, customers have a high acceptance of this transformation.
