Li Auto's chip bet: margins down, intelligence up
Q1 vehicle margin collapsed to 6.1% while the company put its in-house M100 AI chip and MindVLA model into production — a strategic pivot that trades near-term profits for an embodied-AI moat.
LI · Earnings Call · 2026-05-28
The margin shock that wasn't really a surprise
Li Auto's first-quarter results were ugly on the surface: total revenue fell 11.4% year-over-year, gross margin dropped from 20.5% to 7.9%, and the company posted a net loss of RMB 2.3 billion. The CFO, Tie Li, attributed the decline to "the model refresh cycle" and a "higher mix of i6." Vehicle margin was especially painful — “6.1% versus 19.8% in the same period last year” — Tie Li, Chief Financial Officer · 2026-05-28. Yet the market has heard this story before: management explicitly guided that the launch of the all-new Livis version would absorb most of the hit. The more meaningful narrative is what those dollars are funding.
The real news from the call is that the M100 chip — the company's proprietary, AI-native inference silicon — is no longer a roadmap item. It is in mass production inside the Li L9, paired with the MindVLA model. As CEO Xiang Li put it, “We're the first company in China to deliver full functionalities on a brand-new chip in its first-ever on-vehicle deployment.” — Xiang Li, Founder, Chairman and CEO · 2026-05-28 This is the payoff from years of R&D that was previously discussed only in future tense. Back in August 2025, CTO Yan Xie had promised the chip was “currently ongoing in vehicle testing and everything is in good shape.” — Yan Xie, Management (Chip Design / Technology) · 2025-08-28 Now it is shipping.
Vertical integration as a moat
The economic logic of the M100 is not just performance but cost. On the Q&A, Yan Xie quantified the advantage: “our in-house M100 chip delivers triple the computing power of the previous generation platform at half the cost.” — Yan Xie, Senior Executive in Autonomous Driving or Technology · 2026-05-28 The company removed a separate controller, further cutting per-vehicle expenses. The computing power — the effective multiplier net of cost — is the crux of the strategy. Li's argument is that a deeply integrated chip-model-OS stack creates a structural advantage that competitors cannot copy by simply hiring former employees or using NVIDIA hardware. “It took us 4 years to bring our in-house chips from starting the program to vehicle production,” Li said, implying the time horizon itself is a barrier.
The timing is also deliberate. The new L9 and the upcoming all-new L8 are meant to anchor Li's position in the flagship SUV segment, where the company is targeting >20% share in the RMB 500k+ NEV market. The L Series refresh is the vehicle for monetizing the chip investment. But the market is skeptical because the margin will only recover to ~10% in Q2, and Q1 free cash flow was negative RMB 7.4 billion. Yet the cash pile remains solid at RMB 94.3 billion, and management reiterated a full-year sales growth target of 20%.
Beyond the car: embodied AI and global expansion
Li is positioning itself as more than an automaker. In the Q&A, Li laid out a vision where the next 3-5 years of competition in mid-to-high-end smart vehicles is really a contest of embodied AI. The company is also exploring humanoid robot technology, though he cautioned that full-scale commercialization is still “more than 3 years” away. This echoes prior calls: in March 2026, Li had already framed 2026 as "the most competitive year to date" and promised to “lay a solid foundation over the next 2 years” — a clear reference to the chip and model roadmap.
Overseas is another front. The CEO detailed new distributor agreements in Saudi Arabia and the UAE, entry into Southeast Asia, and plans to bring the BEV i6 to Europe in H2. This is a reiteration of the phased international strategy discussed in earlier calls, but now with concrete signed partners and a dedicated overseas L9 variant. The market will be watching whether the overseas ramp can offset domestic margin pressure.
The takeaway
Li Auto is in a deliberate transition: it is sacrificing near-term profitability to build a vertically integrated AI-vehicle stack. The gross margin collapse is real, but so is the product evidence — 10,000+ orders for the Livis L9 within two weeks, monthly i6 sales stabilizing at ~20,000 units, and a chip that is already generating cost efficiencies. The question is whether the market will wait for the margin inflection in late 2026 or focus on the current cash burn. For now, Li is asking investors to believe that “our capabilities and outputs will no longer be easily replicated by others.” — Xiang Li, Founder, Chairman and CEO · 2026-05-28 The next few quarters will test whether that moat is worth the margin pain.