He Xiaopeng Shocked After Testing Tesla FSD 14.2: The Gap is Like Old and New Eras
He Xiaopeng, chairman of XPeng Motors, expressed strong shock after personally test-driving Tesla's FSD 14.2 in Silicon Valley, stating that one cannot truly understand the extent of its improvements without experiencing it firsthand.
He pointed out that this is not an ordinary upgrade, but the "beginning of a new era." Tesla's FSD is iterating at speeds 5 times, 10 times, or even dozens of times faster, representing a fundamental difference between the previous generation of cars and the next.
Source: Public Information
ABAB AI Insight
He Xiaopeng has previously publicly benchmarked against Tesla, and this personal test drive of FSD 14.2 continues the strategic pressure on XPeng to transition from "multi-sensor fusion" to "pure vision end-to-end." In earlier years, XPeng's XPilot was ahead domestically, but after 2025, Tesla's rapid iteration has made domestic car manufacturers feel the generational gap.
On the capital front, XPeng is accelerating investment in end-to-end large models and massive video data training, while publicly signaling urgency to internal and capital markets through executive statements. Resources are shifting from hardware redundancy to data loops and algorithm efficiency, motivated by the need to seize the Robotaxi and smart driving commercialization window, avoiding further widening of the gap with Tesla.
Similar to how executives like Li Xiang from Li Auto and Lei Jun from Xiaomi adjusted their strategies after test-driving FSD in 2024, and the industry shock caused by Tesla's FSD v12 end-to-end transformation in 2023, the current Chinese smart driving industry is in a fierce mid-stage transition from "function stacking" to "general AI driving."
Essentially, this is a technological replacement: traditional smart driving relies on rules + multi-sensors, while FSD achieves end-to-end neural network decision-making through pure vision + large-scale real scene training, with an exponential iteration speed. Mechanically, this makes data scale and training efficiency decisive barriers, shifting pricing power from the traditional automotive supply chain to AI platforms that possess massive driving data, accelerating the global automotive industry’s transition from "hardware-defined" to "software + AI-defined."
ABAB News · Law of Cognition
If you haven't experienced it, you'll always underestimate the gap of the next generation technology.
When iterating 10 times, catching up is no longer an upgrade, but a rebirth.
Once the data flywheel starts, traditional car manufacturers see not competitors, but an era.