On January 26 Li Xiang, founder and chairman of Li Auto, spent nearly two hours addressing the company in an online all‑hands that departed from cars and landed squarely on artificial intelligence. The meeting outlined a shift in corporate ambition: Li wants the company to evolve into an "embodied intelligence" enterprise, building not only in‑vehicle AI but full software and hardware stacks that extend to humanoid robots.
Li set out a multi‑layered technical agenda that spans embodied perception, base models, inference chips and an operating system. He framed those pieces as mutually reinforcing components of a larger strategy to produce end devices — from smart glasses to robots — powered by a shared base model and on‑device reasoning. Internally, hiring moves already signal the pivot: urgent postings for embedded software, mechanical design and modules related to dexterous hands and bipedal locomotion have appeared on Li Auto’s careers site.
The announcement is the latest example of a broader trend: China’s consumer hardware companies and automakers are repositioning around AI as a route to higher margins and new product categories. Investors in Beijing increasingly argue that automobile manufacturing is a tough, capital‑intensive business with narrow returns, while AI and software platforms can scale more profitably. Li’s public framing — that a large model and capable agents could become "production tools" deployed like human labor — captures that calculus.
Technically, Li Auto is targeting what researchers and start‑ups call "agents": systems that perceive, plan and act autonomously rather than simply responding to user prompts. Li described an Agent as "a digital hand" capable of proactive operation, and said the company will keep training system‑level Agents as part of its software ecosystem. That ambition aligns with activity across the industry, where firms race to commercialize multimodal models that can control hardware and assist users in everyday tasks.
The company’s pivot is not entirely surprising. Li has long spoken about AI, intelligent driving and robotics as strategic priorities, and leading figures in China’s chip and AI ecosystem voiced support for his thinking. Still, turning an automaker into an AI and robotics firm presents material challenges: recruiting and integrating roboticists, designing custom inference chips, and developing robust base models and operating systems are time‑consuming and expensive undertakings.
There are also hard engineering obstacles at the frontier of autonomy. Experts caution that building a truly capable general Agent requires overcoming technical complexity in decision‑making, filling gaps in multimodal understanding, and ensuring reliable interfaces with hardware and operating systems. Even when systems appear to work in prototype form, APIs, permissions and system integration issues can limit real‑world utility, as early deployments of assistants have shown.
For Li Auto the strategic stakes are high. Success would open new revenue streams beyond vehicle sales and create a defensible software‑hardware stack that competitors without similar investments would find costly to replicate. Failure or long delays, however, would entail heavy R&D expenditures and risk distracting from the core business at a time when competition in China’s EV market remains intense.
The move also reshuffles talent markets. Li said the company will recruit top robotics talent and bring back engineers who left to join start‑ups; job listings marked "urgent" suggest an immediate scramble to build teams. That dynamic will intensify competition for people who can bridge machine learning, embedded systems and mechanical design — skills that remain scarce and expensive.
Regulatory and societal questions follow. Humanoid robots raise new safety, privacy and employment questions that regulators in China and abroad are beginning to explore. How Li Auto designs safeguards, data governance and human oversight into its agents will be as politically important as the underlying technical breakthroughs.
In the short term the announcement hardens a familiar narrative: Chinese technology firms are betting that AI, not hardware alone, will determine the next cycle of consumer and industrial innovation. Whether Li Auto can translate its automotive engineering know‑how into a credible robotics and AI stack will test both the company’s management and the broader ecosystem’s ability to deliver integrated, production‑grade agents and embodied systems.