Chinese electric-vehicle makers that spent years perfecting car manufacturing are now rushing to rebrand themselves as AI companies, according to a report by Huxiu. The shift reflects a brutal reality: average profit margins in China's car industry have fallen to 1.5%, and domestic sales are shrinking, pushing companies to seek higher valuations in technology and robotics.
Xpeng's quarterly earnings release in August 2026 contained a telling detail in its company description. The firm, which called itself a "smart electric vehicle company" in its 2020 IPO prospectus, changed its label to "global AI car company" in late 2024, then to "embodied intelligence company" in 2025, and finally to "global leading physical AI company" in English this summer. Four changes in four years show a clear attempt to shed the manufacturing label.
Other brands are following similar paths. Geely has built a "full-domain AI system" at CES, iterating from 1.0 to 2.0 in two years. Nio spun off its autonomous-driving chip business, raising 2.257 billion yuan (about $318 million) in its first external round. Li Auto wrote artificial intelligence into its highest corporate vision as early as 2023, though its profits still rest on EREV (range-extender) sales. Chery has replaced human sales guides with humanoid robots on showroom floors, and Seres has launched a new AI brand called AIVA.
From subsidies to survival
The industry's transformation has been painful. After a subsidy scandal in 2016 nearly killed the sector, Nio, Xpeng and Li Auto were founded by internet entrepreneurs who brought software engineering into a heavy-asset industry. The real turning point came in 2019, when Tesla's Shanghai factory opened and the locally built Model 3 forced Chinese rivals to mature their supply chains or die.
By 2026, new-energy vehicles (NEVs) accounted for 62.8% of new car sales in China, but total passenger car sales fell 6.2% in the first half of the year. Domestic sales dropped 20%, offset only by a 60% surge in exports. The price war has been brutal: in early summer 2025, BYD cut prices on 22 models, and within two weeks more than a dozen brands followed. In 2024 alone, 4,419 dealerships closed nationwide.
The financial pressure is intense. In 2025, China's nominal GDP grew 4.0%, mortgage balances fell for eleven consecutive quarters, and over 60% of residents preferred saving over spending. Total passenger car retail value shrank by 170 billion yuan (about $23.9 billion), with the average car price dropping to 170,000 yuan (about $23,900) after six years of increases. Average gross profit per vehicle fell to just 14,000 yuan (about $1,970).
Betting on physical AI
The pivot to AI is partly a response to capital markets that value technology companies more generously than manufacturers. At Nvidia's GTC conference in spring 2025, CEO Jensen Huang outlined four stages of AI evolution, with "physical AI" as the final step—models that understand gravity, friction and stiffness to control machines in the real world. He pegged the addressable market at $50 trillion, an order of magnitude larger than the global auto industry.
Chinese companies are placing different bets. Huawei and Nio back world models, while Xpeng and Li Auto favor VLA (vision-language-action) models. BYD announced in early summer 2026 that it would invest over 100 billion yuan (about $14.1 billion) in AI, targeting zero-accident "super drivers" and "super assistants." Leapmotor unveiled its own humanoid robot at a technology day, with CEO Zhu Jiangming saying, "A robot that can be walked is not impressive; one that can make money is."
Not all bets have paid off. Neta, which won the new-energy startup sales crown in 2022 with 152,100 deliveries, saw its cash run dry three years later, facing 26.58 billion yuan (about $3.7 billion) in claims from 1,631 creditors. WM Motor and HiPhi have also struggled with restructuring and bankruptcy proceedings.
The push into AI extends to self-driving technology. Waymo has imported over 3,000 Zeekr chassis from China since 2024, stripping out autonomous-driving hardware and communication modules for import, then adding them back in Arizona via Magna. Chinese companies have won the hardware orders, but the algorithmic "brains" and commercial operations remain dominated by Western players like Tesla and Waymo, which face their own regulatory hurdles—Tesla's Cybercab robotaxi drew a formal investigation from US safety regulators on its first day of passenger trials.
