At the International Consumer Electronics Show 2025 (CES 2025), AI models will be fully applied in intelligent cockpits, vehicle control, and especially in the field of autonomous driving. Intelligent technology has brought disruptive changes to the industry, and the autonomous driving industry is showing a clear trend of recovery, approaching a turning point of technological explosion and commercialization. From the perspective of technology alone, promoting the popularization of autonomous driving requires three major elements: first, leading algorithms; The second is sufficient computing power; The third is rich data. At present, the above conditions are just in place. In terms of computing power, two years ago, the market promotion of autonomous driving chips was mainly based on AI computing power, power consumption, process parameters, etc., but now, the computing power of the main control chip has skyrocketed from single digit TOPS (trillions of floating-point operations per second) to hundreds or thousands of TOPS. Industry predictions suggest that with the acceleration of computing power improvement and the decrease in hardware costs, there may be scenarios in the future where computing power can be enhanced by replacing computing modules. In terms of engineering capabilities, after several years of competition and training, leading autonomous driving companies generally have the ability to transform technology into practical products. Through the accumulation and "feeding" of trillions of kilometers of massive data, comprehensive, low-cost, mass-produced, and vehicle grade autonomous driving solutions continue to emerge. In terms of algorithms, deep learning algorithms are reshaping various industries, especially for the autonomous driving industry. With the emergence of multimodal large models, including new algorithm frameworks such as "end-to-end 2.0" VLA (Visual Language Action Model), there has been a sustained performance improvement on the boundaries of data volume, computational resources, and model complexity. It can greatly reduce duplicate data and computing resources, while reducing model complexity, truly enabling intelligent driving technology to achieve a leap from quantitative change to qualitative change. The author believes that as a complex system development, autonomous driving requires a solid research and engineering team with a persistent attitude, constantly accumulating experience, and being able to absorb the latest technology in a timely manner and understand the boundaries of technology, in order to gradually bring technological innovation into reality. (New Society)