Source: Red Interstellar Technology, Intelligent Vehicle Technology, Publicly Available Online Information

Multiple media outlets have reported that Jin Yihua, the head of software R&D in Li Auto's chip division, has recently resigned. This marks the second high-profile technical executive departure from Li Auto's chip team in just a few months, following the resignation of SoC R&D lead Qin Dong in March of this year.

Public records indicate that Jin Yihua is a graduate of Fudan University and previously worked at Alibaba before joining Li Auto's chip division as Senior Director of R&D and Head of the Computing Unit Software Department. His primary responsibilities included overseeing the development of compilers, AI toolchains, and other foundational software for in-house chips. He was also deeply involved in the software development for Li Auto's first in-house smart driving chip, the Mach M100.

Departures of Key Hardware and Software Leaders

Jin Yihua and Qin Dong, who resigned in March, had distinct roles within the chip division. Qin Dong oversaw the hardware development of the chips, including leading the hardware development of the Mach M100. Jin Yihua, on the other hand, focused on foundational software, leading the R&D of compilers and AI toolchains. Their successive departures leave Li Auto's in-house chip development without core leaders in both hardware and software.

The Mach M100 is Li Auto's first in-house 5nm automotive-grade dataflow AI chip, boasting a single-chip computing power of 1280TOPS. The project was launched in November 2022, successfully taped out in 2024, and officially entered mass production in 2026, with its first application in the new-generation Li L9. Currently, Li Auto models equipped with Mach M100 chips include the L9, L8, and L6 series, all of which have achieved batch deliveries.

The timing of this personnel change is particularly noteworthy. Just one month before Jin Yihua's resignation—on July 13, 2026—Li Auto fully registered and established Chip Innovation & Intelligence Core (Shanghai) Technology Co., Ltd. The new company's business scope explicitly includes the "sales of integrated circuit chips and products," which is widely seen as a clear signal that Li Auto is moving towards independent operation and market-oriented output of its in-house chip business.

Li Auto's founder, Li Xiang, and CTO, Xie Yan, have both publicly stated this year that they do not rule out the possibility of supplying their in-house chips externally. In a media interview in mid-June, Xie Yan explicitly said, "The external use of (chip) toolchains is very important. Once we have perfected the toolchains internally, supplying them externally will be a natural progression. (Currently) many robotics companies are interested in our chips."

And compilers and toolchains are precisely the key to external chip supply.

Compilers play a fundamental role in chips by translating high-level programming languages used by engineers into binary machine code that chips can recognize and execute. This bridges the gap between algorithm models and chip hardware instructions. More critically, compilers are essential for the "last-mile" efficiency of deploying algorithm models onto chips. They need to encapsulate the instruction sets, memory models, and other details of different chips through a unified adaptation layer, enabling existing algorithm models to migrate to new chips with minimal modifications. The compiler for Li Auto's Mach M100 chip was developed under Jin Yihua's leadership.

Challenges for Automakers, Opportunities in Robotics

When new energy vehicle makers consider supplying their in-house chips externally, they have two primary paths: selling to other automakers or to robotics companies.

Selling to other automakers is the first path. However, in the current highly competitive automotive industry, this route is extremely challenging. Other automakers are naturally inclined to avoid relying on direct competitors for core computing components in their procurement decisions.

The second path is selling to embodied robotics companies. Currently, the embodied robotics sector is in its nascent stages regarding chips and domain controllers. While many embodied robotics companies currently use NVIDIA chips, NVIDIA has not yet demonstrated the same level of dominance in this field as it has in autonomous driving. For Li Auto, this represents a promising market opportunity.

Notably, a number of former Li Auto executives have ventured into the embodied robotics sector to start their own businesses. These include companies with Li Auto backgrounds such as Zhijian, Kunlun Xing, Wujie Power, Xieyue Intelligence, and Vital Power. Xia Zhongpu, the former head of Li Auto's end-to-end intelligent driving, has joined Wujie Power. Chen Wei, the former Chief AI Scientist at Li Auto, and Zhang Xiao, the former head of the second product line, co-founded Xieyue Intelligence. Lang Xianpeng, the former head of Li Auto's intelligent driving, has also ventured into the embodied intelligence sector to start his own business.

The exploration of these former Li Auto executives in the embodied intelligence sector objectively expands the potential application scenarios for Li Auto's chip and AI capabilities. These companies are seen as potential seed customers for Li Auto.

However, the current shipment volume in the embodied robotics market is still limited, necessitating long-term strategic planning.

Li Auto is not alone in advancing towards external supply of its in-house chips. NIO initiated chip R&D as early as 2021, and its in-house chip, Thor NX9031, has already begun external technology licensing and established a joint venture with Axera. NIO also released the "Rui Dong" development platform for embodied intelligence during the World Artificial Intelligence Conference in July 2026. XPENG Motors' Turing AI chip has secured a mass production nomination from Volkswagen.

Against the backdrop of slowing sales growth and sustained price competition in the automotive industry, automakers are facing varying degrees of financial pressure. Chip R&D requires massive investment, with the development cycle for an automotive-grade chip from R&D to mass production typically spanning several years. External supply to amortize high R&D costs has become a common choice among new energy vehicle makers.

At this critical juncture where Li Auto's chip business is transitioning from internal use to market-oriented output, the company's top priority is to promptly fill team vacancies and consolidate organizational capabilities. This will help transform its first-mover advantage into sustained market competitiveness. This, perhaps, holds more long-term value than dwelling on the departure of any single individual.

- End -

Disclaimer:

Any works marked "Source: XXX (non-Intelligent Vehicle Technology)" in this official account are reprinted from other media for the purpose of transmitting and sharing more information. This does not imply that this platform endorses their views or assumes responsibility for their authenticity. All copyrights belong to the original authors. If there is any infringement, please contact us for deletion.