The trillion-dollar track of intelligent connected vehicles has landed! XUNCE (03317) leverages its ability to tokenize data to seize the opportunity of physical AI, with both business model and profitability capabilities being verified.

On June 8th, XUNCE Technology (03317) announced the signing of a three-party strategic cooperation framework agreement with PATEO and SAIMO. The three parties will jointly develop Token-driven in-vehicle physical AI and world models, and jointly build a "end-side hardware + simulation testing + data infrastructure + Token settlement + commercial operation" integrated closed-loop industrial system. This move marks XUNCE Technology's mature data Tokenization capabilities, from fields such as finance, telecommunications, electricity, energy, and Siasun Robot&Automation training platforms, officially landing in the current largest global physical AI terminal, the intelligent connected vehicles, with a trillion-dollar high-quality market. Full link running for the first time: World models from training to vehicle-side realization In the past, there has been a long-standing invisible barrier in the field of physical AI: those doing algorithms only focus on algorithms, those working on simulations only focus on simulations, and those focusing on commercialization only focus on commercialization, with the three segments operating independently. It is understood that this cooperation, for the first time in the industry, successfully completed the full link loop of world models from training, validation, on-board deployment to realization. The three companies will leverage their respective advantages and deep collaboration to complement each other. PATEO will embed trained models into AI BOX, run them on actual vehicles, SAIMO will provide massive high-value scenario data and a full-stack self-developed simulation testing toolchain. XUNCE Technology will rely on its self-developed full-link AI data infrastructure and TokenOS operating system to provide core services that integrate Token value measurement, dynamic settlement, and standardized billing. Behind this collaboration lies a structural challenge that the physical AI industry has long faced. Unlike large language models, every decision error in physical AI could be irreversible - each judgment of an autonomous vehicle directly affects life safety, with much smaller margin for error compared to traditional AI. At the same time, the gap between the real scenario data needed to train physical AI and the existing data volume is estimated to be of the order of magnitude of tens of millions. It is worth noting that the three parties do not intend to operate this system in a closed manner, but instead plan to jointly expand the "Token World Model Alliance", open to car manufacturers, Tier 1 companies, large model companies, and ecosystem service providers to promote the large-scale deployment of TokenOS operating systems and physical AI world models in various terminal scenarios such as AI vehicles and mobile terminals, exploring a new paradigm of Tokenized industrial collaboration. In addition, all the technology achievements jointly developed by the three parties will be collectively named "BotaiXUNCESaimu TokenOS Enhanced Module", with relevant intellectual property rights and commercial profits shared by the three parties. With more participants, richer scenarios, and more vehicles running, the models will continue to evolve, forming a positive flywheel of data and capabilities. Intelligent connected vehicles with trillion-dollar market, the optimum solution for physical AI to scale landing ahead of time Among the many application scenarios of physical AI, intelligent driving is the most promising direction to achieve large-scale landing ahead of time. Essentially, intelligent driving is a constraint-based embodied intelligence - it only needs to operate in structured road scenes, without the need to adapt to ever-changing open environments like humanoid robots. The technical threshold is relatively convergent. More importantly, the automotive industry already has a complete supply chain and business model, so once the technology matures, it can be quickly and widely promoted. A single intelligent vehicle can generate tens of gigabytes of operational data per day, and the data directly affects driving safety and passenger experience, with strict requirements for real-time and accuracy - this aligns well with XUNCE Technology's core capabilities developed over the past ten years, such as millisecond response time and precise computation throughout the process. Scene data Token is the core cornerstone for the evolution of physical AI and world models. Currently, AI technology is transitioning from the digital virtual space to the real physical world, whereas the positioning of Token has also evolved from traditional semantic measurement units to being the core carrier of value circulation in physical interactive scenarios. Keeping in line with industry trends, the three parties will jointly create specialized models tailored for intelligent cabins and in-vehicle physical AI, focusing on core capabilities such as environment perception, user behavior analysis, travel intention prediction, service recommendations, and multimodal human-machine interaction. This collaboration will accelerate the landing of Token measurement systems in in-vehicle physical interaction scenarios, laying a solid foundation for standardized and traceable Token value measurement in the in-vehicle AI industry. Furthermore, there is a fundamental difference between scene-specific Tokens and general large model Tokens. If general large model Tokens are like "basic utilities" in the AI industry, then scene-specific Tokens are the "core fuel" that drives the operation of physical intelligent devices. The two differ significantly in cost structure, marginal benefits, and pricing logic, which is why scene-specific Tokens can maintain a high premium over the long term. Currently, the pricing range for scene-specific Tokens at XUNCE is between $10 and $100 per million, several times to tens of times higher than general large model Token prices, with market prices still on an upward trend. From the perspective of the industrial chain, simulation platforms and industrial software tools are the core infrastructure for training physical AI and are also value nodes that are often overlooked but crucial. SAIMO's accumulation of resources in the field of autonomous driving simulation testing in China precisely fills this scarce layer, providing a solid guarantee for the data quality and model reliability of the three-party system. Dual validation of business model and profitability, aiming for annual recurring Token revenue in the hundreds of millions For XUNCE Technology, this cooperation is a key breakthrough for its Token payment business model and once again confirms the strong cross-industry adaptability and implementation capabilities of TokenOS operating system. Unlike AI companies that only exist at the conceptual level, XUNCE Technology has both underlying core technology, a mature commercialization path, and stable profitability. Since fully launching the Token payment model in 2026, the related business has entered a fast growth lane, with Token calls corresponding to annual recurring revenue (ARR) increasing by 300% quarter-on-quarter in April, Token payment revenue accounting for over 5% of total revenue, and aiming to increase this ratio to 20%-30% for the full year. As the three-party cooperation project gradually progresses, this cooperation is expected to bring hundreds of millions of annual recurring Token revenue to the company by 2026, showing outstanding growth potential in the medium to long term. On the business implementation level, the three parties are focusing on high-frequency and high user-sticky cabin scenarios such as in-car voice assistants, travel planning, vehicle control services, brand guidance, charging and energy supplementation, parking navigation, and car insurance services, to create tokenized AI Agent applications that can be quickly replicated and billed on demand, completing the full process from technical validation to commercialization. Specific measures include: jointly defining in-car Token billing standards, establishing a compound value measurement system covering Token pay-as-you-go, agent task billing, performance sharing, enterprise proprietary knowledge call billing, training and evaluation billing, data asset Token pricing, etc.; co-building a multi-ecosystem unified Token settlement clearing center, enabling the settlement of different sources of heterogeneous Tokens such as in-car AI models, datasets, and agents in the same system; promoting the realization of sensitive data and professional knowledge assets at the vehicle end, transforming driving behavior, road perception data, and cabin development know-how into standardized and priced data Tokens. This three-party cooperation is also highly compatible with the national industrial policy direction. Recently, the State Administration for Market Regulation and the National Development and Reform Commission jointly issued the "Guidelines for AI Measurement Systems and Capacity Building (2026 Edition)", which clearly requires the establishment of unified measurement standards that are measurable, comparable, and traceable for AI technology performance. The core positioning of XUNCE Technology's TokenOS operating system is to "quantify and price large models", aligning closely with policy guidance and providing solid support for the long-term development of the project. As the penetration rate of intelligent connected vehicles continues to rise, in-car AI applications are entering a phase of comprehensive outbreak. This three-party cooperation not only helps XUNCE Technology successfully enter the trillion-dollar automotive incremental market but also strongly validates the commercial feasibility of Token economy in physical scenarios. Looking ahead, as in-car AI Agents are widely commercialized, Token calls continue to rise, and industrial ecosystem cooperation deepens, the company's leading position in the AI data infrastructure field will be further consolidated.