During the 2025 World Robot Conference in August, Wang Xingxing, CEO of Unitree Technology, stirred up considerable debate by labeling the widely embraced VLA (Vision-Language-Action) model as a "relatively simplistic architecture." As a fervent advocate of the VLA approach, Ideal Auto made history in September by integrating the world's first VLA driver large model into its vehicles. Lang Xianpeng, the Head of Autonomous Driving at Ideal Auto, penned a detailed response to Wang Xingxing's skepticism. He contended that the model's true value lies in its seamless integration with embodied intelligent systems, with data playing a pivotal role. Drawing from his practical experiences, Lang highlighted that VLA represents the optimal model solution for autonomous driving, and that the competition in embodied intelligence hinges on comprehensive system capabilities. Ideal Auto is well-positioned to gather data, leveraging both its existing data repositories and the daily interactions of its vehicle owners. He further revealed that Ideal Auto's VLA is not confined to its current automotive offerings but is also geared toward future automotive-like embodied robots. The company's pioneering implementation of VLA can be attributed to four core advantages. Moreover, Lang believes that in the VLA era, inferential computing power will take center stage, and if it can achieve 1000MPI (Million Instructions Per Second) by next year, VLA will experience a breakthrough moment reminiscent of ChatGPT's rise.