The autonomous driving sector is moving from early technical validation toward commercial rollout at scale. For CaoCao Mobility, that shift is taking shape through RoboX, a new strategy built around driverless vehicles, artificial intelligence-based operations, and a broader range of mobility services.
Unlike upgrades that focus mainly on driving assistance technology or new vehicle models, RoboX is designed as a systemwide upgrade across products, technology, and operations, according to CaoCao. The company plans to expand from robotaxis into robovans, robobuses, and robotrucks. Its autonomous driving technology will continue to advance toward Level 4 capabilities, meaning vehicles that can operate without human intervention under defined conditions. Its operating system will also introduce an AI-driven “brain” intended to connect demand forecasting, supply matching, and service fulfillment.
The significance extends beyond CaoCao’s own business. If the company can execute, RoboX could become a test case for the commercial use of AI in the physical world, where revenue is generated through transportation and logistics services rather than digital tools alone.
Three years after the global wave of large language models began, the industry is looking for more examples of real-world implementation. 36Kr spoke with CaoCao CEO Gong Xin about how the company sees the next phase of physical AI.
Over the past three years, advances in AI have accelerated the development of intelligent driving. In China’s new energy vehicle industry, a category that includes battery-electric, plug-in hybrid, and fuel cell vehicles, adoption of Level 2 and above assisted driving has risen quickly. These systems can control functions such as steering and acceleration under human supervision, while capabilities have moved toward broader use across more road scenarios.
Level 4 autonomous driving is also moving closer to commercial deployment in 2026. According to 36Kr, policy signals in China have strengthened. The Ministry of Industry and Information Technology has issued a draft mandatory national standard for Level 4 autonomous driving for public comment, while China’s Ministry of Public Security has introduced rules related to autonomous vehicles’ use of public roads. 36Kr also reported that 26 Chinese cities have opened qualification access for paid commercial robotaxi operations without in-vehicle safety operators. These developments make autonomous driving one of the more tangible areas for AI deployment.
“The point at which the industry crosses from human-driven to driverless is whether it can solve corner cases,” Gong said. He noted that driverless mobility vehicles are shared spaces. “We have remote safety operators. If an unsolved corner case occurs, a remote safety operator can intervene to ensure safety. But with private cars, it would be impossible to allow someone in the cloud to watch your car every day.”
For CaoCao, driverless mobility is not just a strategic narrative. It is also part of the company’s business logic, and one reason it believes it can secure a meaningful position in the commercialization of autonomous driving.
But deploying driverless vehicles in mobility services will not be easy.
In traditional ride-hailing, competition centers on vehicle supply, pricing, and traffic scale. In the driverless era, vehicles may move beyond their role as transportation tools and become mobile living spaces, urban data nodes, and intelligent mobility terminals. Competition will no longer be decided by a single dimension. It will depend on full-system, closed-loop capabilities.
“Smart driving technology determines whether the model is feasible. Deeply customized vehicles determine the basic capabilities. Platform operating efficiency determines commercial success,” Gong told 36Kr. “The three complement each other. Only together can they push RoboX into reality.”
As a key commercialization vehicle for Geely’s RoboX strategy, CaoCao has spent years building capabilities in mobility operations. With strengths in technology, products, and operations, the company is positioning itself for the new competitive rules of driverless mobility.
Gong also believes that, in the AI era, a company’s competitiveness will not depend on owning an app-like entry point. It will depend on how effectively it connects AI with the physical world.
“What CaoCao Mobility wants to do is not fight to become the entry point,” Gong said. “No matter which entry point users come through, they can use RoboX to enjoy smart ride-hailing, delivery, and even low-altitude mobility services. At the same time, we also have an in-cabin agent that can meet more of users’ needs after they get out of the vehicle, such as choosing a restaurant and ordering food.”
CaoCao’s progress did not happen overnight. Its work in driverless mobility began earlier than many outside the company may have realized. While many players were still testing single robotaxi use cases and focusing on technology trials, CaoCao had already begun accumulating technology, refining products, and iterating its operations.
The RoboX strategy rests on three pieces: driving capabilities, customized vehicles, and intelligent operations. Other companies may be able to pursue any one of these areas. CaoCao’s bet is that integrating all three into one operating network will create a more durable advantage.
First, driverless mobility depends on advanced autonomous driving capabilities. The industry broadly understands that Level 4 autonomous driving will not improve through larger algorithmic models alone. It also requires continuous iteration using real-world scenario data.
CaoCao has nationwide ride-hailing operational data at scale. According to company data cited by 36Kr, its average monthly active users reached 41.3 million in 2025, while it completed more than 1.9 billion mobility services that year. Its road condition coverage spans a broad range of real-world mobility scenarios, including complex urban roads, extreme weather, morning and evening rush hours, and unusual road conditions.
“Our logic is very clear,” Gong told 36Kr. “We use real operational data to support algorithm iteration, then feed it back into real-vehicle road testing. Once this flywheel starts spinning, operations make the system safer, smarter, and lower-cost.”
CaoCao also has customized vehicle capabilities, an area where it sees an advantage over pure technology companies focused on driving assistance.
The company previously launched a purpose-built robotaxi model. According to CaoCao, the vehicle was designed from the start for autonomous driving and shared mobility, with a dedicated spatial layout, safety configurations, and intelligent interaction features. It supports battery swapping and has a longer theoretical service life than ordinary passenger cars, the company said.
Gong told 36Kr that as early as 2021, when autonomous driving was starting to gain momentum, CaoCao had already defined customized vehicles as a strategic direction.
“What exactly is the difference between an operating vehicle and a private car? These are not questions that can be resolved in a day or two. We spent five to six years, made many mistakes, and deployed so many vehicles before we figured it out.”
Autonomous driving technology may eventually converge. If that happens, the endgame in driverless mobility will still come down to the mobility experience. The foundation of that experience is the vehicle, making CaoCao’s investment in customized vehicles central to its long-term strategy.
Under RoboX, CaoCao’s vehicle customization capabilities will move beyond a single robotaxi scenario and toward a broader portfolio of intelligent mobility vehicles.
CaoCao has also entered a strategic partnership with Farizon Auto to advance robovan applications, including models such as the Shentong T6. The company plans to gradually roll out businesses such as robovans for urban on-demand delivery, long-haul logistics, and smart public mobility. Its goal is to build a full-scenario driverless mobility capacity system spanning passenger and cargo transport, commuting and logistics, and private and public use.
That would move the industry beyond several current constraints, including single-track technology development, limited use cases, and fragmented products. RoboX is not intended to be only a driverless ride-hailing service. CaoCao wants it to become an intelligent mobility solution that covers a wider range of urban transportation demand.
If customized vehicles and autonomous driving are the hardware, intelligent operations are the system that makes that hardware run.
“In driverless mobility, asset operation capabilities determine the gap between companies’ business performance,” Gong told 36Kr. In the future, he said, driverless vehicles will become corporate assets. “How do you make these assets safer, deliver a better experience, and achieve the highest operating efficiency and lowest cost? Behind this are not only intelligent driving and vehicle manufacturing, but also maintenance, cleaning, preparation, quality inspection, and other processes in operations.”
CaoCao already has 11 years of shared mobility operating experience, dispatching capabilities for million-scale orders, and a mobility network covering 195 cities, according to 36Kr. Under RoboX, the company plans to rebuild these capabilities around AI.
The new AI-integrated operating system is designed to coordinate and dispatch different categories of driverless mobility capacity. It will dynamically adjust capacity distribution based on urban passenger flows, road conditions, and order demand, addressing problems such as idle driverless vehicles, mismatched capacity, and delayed responses. This operating layer could become one of the main factors determining whether driverless mobility services can scale profitably.
In the first half of 2026, CaoCao continued to optimize its business structure and reduce noncore businesses, according to 36Kr. The company reorganized around AI, increased investment in autonomous driving, AI operating systems, and intelligent mobility capacity networks, and promoted the integration of AI capabilities into product R&D, operations management, and business systems. The goal is to accelerate its transformation into an AI-native company.
After releasing RoboX, CaoCao said it aims to deploy 100,000 robotaxis and 100,000 robovans by 2030, while building an open robovan ecosystem.
CaoCao’s upgraded RoboX strategy points to a broader shift in autonomous driving. Companies are moving beyond the exploratory stage of validating robotaxi technology and into the commercial deployment stage. A new cycle brings new rules. The companies best positioned for this phase are likely to be those that combine autonomous driving capabilities, full-vehicle capabilities, and operating capabilities.
Across each stage of the mobility sector, from traditional taxis to internet ride-hailing, from human-driven ride-hailing to robotaxi-based ride-hailing, and now to RoboX, CaoCao has tried to position itself around moments of industrial change. That positioning is less a matter of timing than the result of a long-term industrial strategy.
As AI moves from the digital world into the physical world, mobility and freight may become important infrastructure in the AI era. CaoCao’s target of 100,000 robotaxis and 100,000 robovans by 2030 is a statement of intent. AI should not remain in demos. It needs to be tested against real mobility gaps, real operating costs, and real demand.
KrASIA features translated and adapted content that was originally published by 36Kr. This article was written by Xiao Xi for 36Kr.