3 6 Ke·2025-10-16 01:01Core Insights - The recent National Day holiday saw a significant increase in traffic, with an average of 12.5 million new energy vehicles on the road daily, marking a 30% increase year-on-year and a 70% increase compared to regular days [2] - Unlike previous years where companies showcased their intelligent driving (智驾) capabilities during the holiday, this year saw a notable silence from major brands, with only Huawei and Xiaomi releasing relevant reports [3][4] - Major personnel changes occurred in the intelligent driving teams of companies like Xiaopeng and NIO, indicating a shift in strategy from the "Intelligent Driving Year" to a "Universal Intelligent Driving Year" [4][5] Industry Trends - The transition from "Intelligent Driving Year" to "Universal Intelligent Driving Year" suggests a focus on technological advancements rather than mere market penetration [5] - Data from Huawei indicates that during the recent holiday, their intelligent driving models achieved a total driving distance of 294 million kilometers, with 90.8% of users actively utilizing the assisted driving feature [6][8] - The challenge for new energy vehicle companies lies in achieving reliable L3 and L4 level intelligent driving in urban environments, as opposed to highways where conditions are more favorable [8][10] Technological Developments - The limitations of traditional end-to-end models have prompted a demand for innovative approaches in intelligent driving technology [10][11] - Three evolutionary strategies have emerged among leading brands: the "Improvement School" represented by Momenta, focusing on enhancing learning processes; the "Practical School" represented by Li Auto and Xiaopeng, emphasizing optimization of driving details; and the "World Model" (WA) approach, which simulates a virtual world for learning [11][13][17] - The WA model, which aims to provide a deeper understanding of driving logic, is seen as a more advanced but costly alternative to the VLA model, which is already integrated into products like Li Auto's i8 and Xiaopeng's G7 Ultra [21][17] Competitive Landscape - The intelligent driving sector is entering a more competitive phase, likened to a knockout tournament where brands must demonstrate their technological capabilities and ecosystem collaboration [22][24] - Smaller companies face significant challenges due to high costs and the need for integrated capabilities, with many struggling to keep up with the leading players [24][26] - The long-term outlook suggests that while VLA and WA represent different approaches, both are essential for the future of intelligent driving, with companies like Xiaopeng betting on both strategies to attract users and investors [26]