XPeng conference call: XPeng G6 and G9 orders exceed expectations, trained a cloud-based model with hundreds of billions of parameters

On Tuesday, XPeng released its financial report for the fourth quarter of last year and the full year of 2024. The report shows that XPeng's Q4 revenue increased by 23.4% year-on-year, with a significant improvement in gross margin to 14.4%, and an accelerated layout of its AI strategy.

After the report was released, XPeng pointed out at the press conference that orders for the XPeng G6 and G9 exceeded expectations, and they will make every effort to speed up delivery; XPeng has trained a cloud-based model with hundreds of billions of parameters, unprecedented in China.

Specifically, He Xiaopeng revealed during the Q4 earnings call:

After the launch of the 2025 models G6 and G9, orders have continued to rise and exceeded expectations, while the explosive increase in foot traffic after the new models were launched has also driven demand growth for all XPeng models. XPeng plans to start deliveries of the new G6 on March 21 and is working with the supply chain to fully enhance production capacity.

Regarding AI progress, He Xiaopeng stated during the earnings call:

In the last quarterly report and AI Technology Day, I shared the view that the next decade will be the era of AI, and that the AI large models in 2025 will bring disruptive experiences to autonomous driving. The breakthrough progress in AI technology in China in the first three months of this year has made me even more confident in the correctness of XPeng's strategy and technological route for AI large models in the physical world.

We have trained a cloud-based model with hundreds of billions of parameters, using real driving data of over 100 million kilometers for pre-training, which is unprecedented in terms of model parameter and data volume in China's autonomous driving field. On this basis, we enhance the model's generalization ability and capability to handle long-tail scenarios through reinforcement learning, and deploy it on vehicle models using distillation, pruning, and quantization techniques, achieving more than double the model accuracy.

Continuously updated

Risk warning and disclaimer

The market has risks, and investment should be cautious. This article does not constitute personal investment advice and does not take into account the specific investment goals, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Investment based on this is at one's own risk