Li Auto Forms Chip Unit, Eyes External Supply of In-House AI Chips

China’s corporate registry platform Qichacha shows that Li Auto has recently established Xinchuang Zhixin (Shanghai) Technology Co., Ltd., with a registered capital of RMB 100,000 ($14,760).

The new company is legally represented by Li Auto Co-Company Secretary Wang Yang. It is wholly owned by Li Auto and ultimately controlled by the company’s CEO Li Xiang.

Its business scope covers integrated circuit (IC) design, semiconductor R&D and product sales.

Among the listed businesses, “sales of integrated circuit chips and related products” has attracted particular market attention, as it is widely viewed as signaling the possibility of future external chip supply.

Although Li Auto has not disclosed the company’s specific role, its name and registered business scope suggest the new entity will serve as the primary operating platform for Li Auto’s in-house semiconductor business.

According to industry reports, Li Auto’s automotive AI chip program will eventually be consolidated under the new company, covering the entire development process for the Mach M100 series, including architecture design, compiler development, IP design, tape-out and automotive-grade validation.

Li Auto’s self-developed Mach M100 AI chip has already entered mass production in the latest Li L9, L8 and L6 models.

Manufactured using TSMC’s N5A automotive-grade process, a single Mach M100 chip delivers 1,280 TOPS of computing power, while the dual-chip configuration provides a combined 2,560 TOPS.

Li Auto describes the Mach M100 as the world’s first dynamic dataflow AI chip, claiming a computing efficiency of 82%.

The processor is designed primarily for high-performance applications such as intelligent driving and large AI foundation models.

Beyond the chip itself, the new company’s business scope may be the more significant development.

In an interview in June, Li Auto CTO Xie Yan said the company does not rule out supplying its self-developed chips to external customers.

According to Xie, the key to commercializing automotive AI chips lies in establishing a mature software toolchain.

Once Li Auto’s internal development platform and supporting tools become fully mature, opening the technology to external customers would become a natural next step.

He added that several robotics companies have already approached Li Auto about adopting its in-house AI processors.

China’s leading new energy vehicle manufacturers have increasingly begun operating semiconductor businesses independently.

NIO previously established Anhui Shenji Technology Co., Ltd. to oversee chip development and commercialization.

Its Shenji NX9031 processor has already entered external technology licensing, while the company has also formed a joint venture with Axera Technology.

XPeng’s self-developed Turing AI chip has secured a production nomination from Volkswagen, marking its entry into external customer programs.

Li Xiang has also responded to skeptism that developing proprietary chips is simply an expensive industry trend.

He argued that the strategy is not intended to showcase technological capability, but rather to solve AI computing bottlenecks that existing suppliers cannot adequately address through tighter hardware-software integration.

In his view, future AI competition will not be determined by individual chips or standalone technologies, but by the ability to optimize chips, operating systems, software and vehicle platforms as an integrated system.

As intelligent driving, foundation AI models and robotics continue to advance, demand for automotive-grade AI processors is expanding rapidly.

For Li Auto, establishing an independent semiconductor company suggests its chip business is evolving beyond supporting internal vehicle programs toward broader industrialization and commercial deployment.