As the 2026 Beijing Auto Show draws near, an anonymous executive in charge of automotive intelligence told Huxiu Auto with such emotion. His brand will unveil a high-profile intelligent product equipped with large AI models at this show. Yet the cabin intelligence and autonomous driving teams are still locked in endless disputes, with no clarity on which team shall take charge of rolling out certain functions.

While internal team power struggles persist at this automaker, Li Xiang, CEO of Li Auto, put forward an urgent judgment in an internal speech:

2026 will be the final window for automakers to integrate onboard AI — a consensus shared across the entire industry.

Latest reports indicate that Tesla China’s in-vehicle voice large model service will integrate Doubao and DeepSeek AI models.

The service completed filing approval on April 20.

Tesla Model Y L will be equipped with both Doubao and DeepSeek models, both accessed via Volcano Engine.

In fact, the technical stack for onboard AI large models has long been ready, and the industrial chain fully connected.

Nevertheless, hidden internal struggles among automakers over organizational structure, R&D pace and power distribution have become the biggest variable preventing vehicle AI models from moving from PPT concepts to mass production and delivery.