Li Auto's latest financial report shows total revenue for Q3 2025 fell 36.2% year-over-year to 27.4 billion yuan, with a net loss of 624.4 million yuan. This financial performance stands in stark contrast to the company’s decision to increase investment in AI infrastructure, highlighting the inherent tension between technology investment and short-term profitability in the smart vehicle industry. Notably, the company is reshaping its competitive landscape by building a robust AI compute base, a strategic choice that is recalibrating the fundamental logic of the entire intelligent mobility sector.
On the technology front, details on the M100 chip revealed by Li Auto's CTO, Xie Yan, mark a critical breakthrough in the company’s AI chip R&D. The chip’s co-development with a proprietary foundational model compiler could redefine how large language models (LLMs) are deployed in in-vehicle scenarios. According to industry observers, the current AI chip market suffers from a significant performance-to-cost gap. Li Auto's commitment to delivering three times the cost-performance ratio through its self-developed chip could potentially rewrite the rules of computing power competition for smart vehicles.
Significantly, the M100 chip’s system testing phase signals that Li Auto is building a complete AI infrastructure ecosystem. This full-stack approach—encompassing the chip, compiler, and software system—not only enhances the real-time decision-making capabilities of its autonomous driving system but may also reduce the overall cost of deploying AI across the vehicle. Industry analysts point out that this vertical integration model is becoming a key pathway for smart vehicle companies to overcome technological bottlenecks.
From an industry trend perspective, Li Auto’s AI strategy reflects the escalating need for autonomous driving technology to transition from the perception layer to the decision layer. As large models find deeper application in vehicles, the reliance on computing power infrastructure is climbing rapidly. Smart vehicle companies globally are grappling with the challenge of high compute costs. Li Auto’s path of developing proprietary chips may offer the industry a new paradigm for a solution.
Currently, the AI-ification of smart vehicles is shifting from point-specific technology breakthroughs to systemic innovation. Li Auto's practice demonstrates that companies need to build synergistic advantages across chip architecture, large model training, and software ecosystems. This multi-dimensional technological integration is not only crucial for enhancing product performance but will also reshape the business models of intelligent mobility services.