On June 8th, XUNCE (03317) announced the formal signing of a tripartite strategic cooperation framework agreement with PATEO and Simulate Tech. The three parties will jointly research and develop token-driven in-vehicle physical AI and world models, collaboratively building an integrated, closed-loop industrial system encompassing "edge hardware, simulation testing, data infrastructure, token settlement, and commercial operations." This move signifies that XUNCE is officially deploying its mature data tokenization capability from sectors like finance, telecommunications, power, energy, and robot training platforms into the intelligent connected vehicle (ICV) sector—currently the world's largest physical AI terminal market and a multi-trillion-dollar premium growth avenue.
A Pioneering End-to-End Process
Historically, a significant invisible barrier has existed in the physical AI field: algorithm developers, simulation specialists, and commercialization teams often operated in silos. This collaboration marks the industry's first complete, closed-loop process for a world model, covering training, validation, vehicle deployment, and monetization. The three companies will leverage their respective strengths in deep, complementary synergy. PATEO will integrate trained models into AI BOX units for real-vehicle operation, while Simulate Tech will provide massive volumes of high-value scenario data and a fully self-developed simulation testing toolchain. XUNCE will utilize its self-developed, full-stack AI data infrastructure and TokenOS operating system to deliver core services integrating token value measurement, dynamic settlement, and standardized billing. This collaboration addresses a persistent structural challenge in the physical AI industry.
Unlike large language models, every decision error in physical AI can be irreversible—each judgment by an autonomous vehicle directly impacts safety, leaving far less room for error than traditional AI. Concurrently, the gap between the real-world scenario data required to train physical AI and the existing data stockpile is estimated within the industry to be as high as millions of times. Notably, the trio does not plan to operate this system in a closed manner. Instead, they will jointly expand a "Token World Model Alliance," continuously opening it to automakers, Tier 1 suppliers, large model companies, and ecosystem service providers. This aims to promote the large-scale deployment of the TokenOS operating system and physical AI world models across scenarios like AI-powered vehicles and mobile terminals, exploring a new paradigm for tokenized industrial collaboration.
Furthermore, all jointly developed technological achievements will be collectively named the "PATEO·XUNCE·Simulate TokenOS Enhancement Module," with related intellectual property and commercial benefits shared among the three parties. In the future, as more participants join, scenarios diversify, and more vehicles operate, the models will continuously evolve, creating a positive feedback loop for data and capabilities.
The Intelligent Vehicle Arena as a Prime Launchpad
Among the numerous application scenarios for physical AI, intelligent driving is the most promising direction for achieving large-scale implementation first. In essence, intelligent driving is a constrained form of embodied intelligence—it operates within structured road environments and does not need to handle the ever-changing open world like humanoid robots, making its technical challenges relatively more manageable. More importantly, the automotive industry already possesses a complete supply chain and business model, allowing for rapid, large-scale promotion once the technology matures. A single intelligent vehicle can generate tens of gigabytes of operational data daily, with this data directly linked to driving safety and user experience, demanding stringent requirements for real-time performance and accuracy. This aligns perfectly with the two core capabilities XUNCE has honed over a decade: millisecond-level response times and precise full-process computation.
Scenario data tokens are the foundational cornerstone for the iterative evolution of physical AI and world models. As AI technology transitions comprehensively from digital virtual spaces to the real physical world, the role of tokens is also evolving—from a traditional semantic measurement unit to a core carrier for value transfer in physical interaction scenarios. Aligning with industry trends, the three parties will jointly develop vertical, specialized models tailored for intelligent cockpits and in-vehicle physical AI, focusing on core capabilities such as environmental perception, user behavior analysis, travel intent prediction, service recommendation, and multimodal human-machine interaction. This cooperation will accelerate the deployment of the token measurement system in in-vehicle physical interaction scenarios, establishing a standardized, traceable token value measurement foundation for the in-vehicle AI industry.
Additionally, there is a fundamental difference between vertical scenario tokens and general-purpose large model tokens. If general-purpose model tokens are the "basic water supply" for the AI industry, then tokens for vertical scenarios are the "core fuel" powering physical intelligent devices. They differ entirely in cost structure, marginal returns, and pricing logic, which is the core reason vertical scenario tokens can maintain a high premium long-term. Currently, XUNCE's vertical scenario tokens are priced between $10 to $100 per million, several times to over ten times the price of general-purpose model tokens, with market quotes still trending upward.
From an industry chain perspective, tool layers like simulation platforms and industrial software are core infrastructure for physical AI training and a crucial, yet often overlooked, value node. Simulate Tech's accumulated resources in China's autonomous driving simulation testing field precisely fill this scarcity, providing solid assurance for the data quality and model reliability of the tripartite system.
Validating Both Business Model and Profitability
For XUNCE, this cooperation represents a critical breakthrough in the real-world application of its token-paid business model, further validating the TokenOS operating system's powerful cross-industry adaptability and deployment capability. Unlike AI companies remaining at the conceptual stage, XUNCE possesses underlying core technology, a mature commercialization path, and stable profitability. Since fully launching its token-paid model, the company's related business has entered a fast growth track. In April, the Annual Recurring Revenue (ARR) corresponding to token calls surged 300% quarter-over-quarter. Token-paid revenue now exceeds 5% of total revenue, with a full-year target to increase this proportion to 20%-30%. As the tripartite project progresses, the cooperation is expected to bring hundreds of millions in token-based ARR to the company, highlighting significant medium- to long-term growth potential.
On the commercial deployment front, the three parties will focus on high-frequency, essential, and high-user-retention cockpit scenarios—such as in-vehicle voice assistants, travel planning, vehicle control services, brand-guided shopping, charging, parking navigation, and auto insurance services. They aim to create rapidly replicable, usage-based tokenized AI Agent applications, completing the full cycle from technical validation to commercial closure. Specific initiatives include: jointly defining in-vehicle token billing standards, establishing a composite value measurement system covering pay-as-you-go tokens, Agent task billing, performance sharing, billing for enterprise private knowledge calls, training and evaluation fees, and data asset token pricing; co-building a multi-ecosystem unified token settlement and clearing center to enable the settlement of heterogeneous tokens from different sources (like in-vehicle AI models, datasets, Agents) within the same system; and promoting the assetization and monetization of anonymized vehicle data and professional knowledge, transforming driving behavior, road perception data, and cockpit development know-how into standardized, priceable data tokens.
This tripartite cooperation also deeply aligns with national industrial policy direction. Recently, the State Administration for Market Regulation and the National Development and Reform Commission jointly issued the "Guidelines for the Construction of Artificial Intelligence Metrology Systems and Capabilities (2026 Edition)," explicitly requiring the establishment of unified metrology standards for AI technical performance that are measurable, comparable, and traceable. The core positioning of XUNCE's TokenOS operating system is precisely to "make large model capabilities quantifiable and billable," highly synchronized with policy guidance and providing solid support for the project's long-term development.
As the penetration rate of intelligent connected vehicles continues to climb, in-vehicle AI applications are entering a comprehensive breakout window. This tripartite cooperation not only successfully positions XUNCE within the trillion-dollar incremental automotive market but also strongly validates the commercial feasibility of the token economy in real-world physical scenarios. Looking ahead, with the large-scale commercial use of in-vehicle AI Agents, the continuous rise in token call volumes, and the deepening of industrial ecosystem cooperation, the company's leading position in the AI data infrastructure sector is set to be further solidified.
Comments