At the 2026 Aggregated Intelligence Industry Development Conference, held on September 9th at Wuhan Optics Valley and co-hosted by the Aggregated Intelligence Industry Innovation Center and Hubei Science & Technology Investment, Kaiwang Data CEO Yu Xu shared insights on the industry's evolution. From a long-term perspective, he noted that the industry has progressed from the small-model era of autonomous driving, through the large language model era, and is now advancing toward world models and embodied intelligence. This trajectory closely resembles the mobile internet wave, where virtually every application gets reinvented.

This round of reinvention will not only be driven by large language model-powered intelligent agents but will also incorporate extensive hardware integration, unlocking vast imaginative possibilities. According to Yu Xu, before this major trend fully materializes, the industry will pass through three distinct development phases.

In the first phase, the core objective is to develop a versatile "intelligent brain." This can be achieved by leveraging mobility scenarios such as dedicated rides, express hailing, and carpooling services to rapidly accumulate massive datasets and solidify foundational capabilities. The second phase involves expanding deployment into a broader range of scenarios. Even during the initial rollout of express hailing services, the company had already proactively built up a rich reserve of scenario-specific resources.

During deployment, by incorporating luxury vehicles and various forms of real-machine remote control operations, multi-source data is blended like a cocktail, continuously refined to achieve superior intelligence outcomes. "As capabilities generalize and expand across more scenarios, data will transform into Data Tokens, delivered in this format to empower a wider array of application contexts," she explained.