On November 20, at the 2026 China Automotive Software Conference, Shang Jin, Chief Scientist at the Dongfeng Motor Group R&D Academy, stated that vehicle intelligence needs to break through cost barriers. Such features should not be limited to models priced above 200,000 or 300,000 yuan, but should gradually trickle down to vehicles under 100,000 yuan — or even under 50,000 yuan. Image Source: 2026 China Automotive Software Conference To achieve this, Shang believes the industry must rethink how automotive computing resources are allocated. Today's vehicle intelligence relies on sensors, networks, actuators, and computing power. Of these, computing capacity can scale continuously through advancements in chips, models, and communication technologies — eventually pushing beyond the boundaries of individual vehicles. He suggested that extra computing power and models don't all need to be deployed on the vehicle itself. Instead, they can be supported through vehicle-cloud collaboration and, in the future, a broader vehicle-road-cloud-satellite synergy system. This would allow vehicles across different price tiers to share computing resources, driving down the cost of deploying intelligent features. Beyond computing architecture, AI is also reshaping the functional boundaries of automotive software. Shang noted that AI applications in cars shouldn't be confined to autonomous driving and smart cabins. As AI agent technology evolves, automotive software will need capabilities like task decomposition, multi-agent collaboration, model invocation, and execution management — all while maintaining real-time performance and safety. In the safety domain specifically, he argued that AI security should become a core component of the automotive software ecosystem — much like information security is today — and be deeply integrated into features like autonomous driving. Shang also highlighted that AI is expanding from automotive products into areas like material design, engineering R&D, and embodied intelligence. Consequently, the value of automotive software will extend beyond in-cabin feature development, reaching into cross-domain technology applications and the broader industrial ecosystem.