He Xiaopeng: Methodologies, Key Decisions & Mental Models

He Xiaopeng

Redefining the car with software thinking — the tech-first founder behind Xpeng's intelligent EV roadmap

Born in 1977 in Guangdong, He Xiaopeng founded UC Browser (UCWeb) after graduating from Guangzhou University's computer science program, selling UC to Alibaba in 2014 for approximately $250 million. In 2017, he formally took the helm at Xpeng Motors as CEO, using his engineer's systems thinking to bet fully on in-house intelligent driving technology (XNGP), making Xpeng the most technologically aggressive new energy vehicle startup in China. After a sales crisis in 2023, Xpeng underwent large-scale organizational restructuring, struck a strategic partnership with Volkswagen, and advanced its long-term AI-defined vehicle strategy.

Methodologies

Software-Defined Hardware Architecture Method - At the outset of hardware product design, treat software upgradeability as a constraint condition; reserve sufficient computing power and interfaces to ensure the product can continuously evolve through OTA throughout its lifecycle rather than becoming static.
Full-Stack In-House Technology Moat Construction Method - In core technology domains, insist on full-stack in-house development rather than procurement; build technology barriers difficult for competitors to rapidly replicate by controlling key technology nodes (computing power, algorithms, data).

Key decisions and timeline

Co-Founded UCWeb, Entering Mobile Internet Entrepreneurship - Building technical barriers early in a nascent track and waiting for market scale explosion is the classic path for internet entrepreneurship
Sold UCWeb to Alibaba for Approximately $2.5 Billion - Sometimes choosing to integrate into a larger ecosystem creates more value than competing independently — the key is judging where the boundaries lie
Became Xpeng Full-Time CEO, Formally Pivoting to Electric Vehicles - Switching from a familiar track to an unfamiliar one requires extreme self-confidence and clear awareness of how core capabilities can transfer

Beliefs and mental models

Software Will Define the Core Value of Next-Generation Vehicles - He Xiaopeng believes cars are undergoing a fundamental transformation from mechanical products to intelligent terminals. The biggest differentiation in future vehicles will come from software capability rather than hardware — this is the core logic behind his commitment to massive investment in proprietary AI and intelligent driving.
Intelligent Driving Must Be Developed In-House; Supplier Dependency Is Fatal - Unlike some automakers that purchase third-party intelligent driving solutions, He Xiaopeng insists on full-stack in-house development — from chips and algorithms to data loops — believing that outsourcing core technology is equivalent to surrendering future competitiveness. Better to be slower short-term than weak long-term.
The Data Flywheel Is the Deepest Moat in Intelligent Driving - He Xiaopeng believes the essence of intelligent driving competition is data competition — more vehicles on the road collecting more scenario data, driving rapid model iteration, which feeds back into product experience, creating a scale barrier. Xpeng invests over 5 billion RMB annually in this flywheel.
Software-First Architecture Model - Build software upgradeability and data collection capability into hardware architecture from the beginning of product design, rather than as an afterthought, ensuring the vehicle continues to evolve throughout its lifecycle.
Full-Stack AI In-House Moat Model - Develop the complete technology chain in-house, from basic data labeling to training frameworks, end-to-end models, and chip adaptation — each layer forms a barrier, preventing any single point from being replaced by competitors or suppliers.
Internet Product Methodology Applied to Automotive - Transplant internet's rapid iteration, user feedback-driven, and data-driven product development methodology into automotive development cycles, dramatically shortening software iteration cycles while respecting the baseline constraints of automotive safety and quality.