🐯虎嗅•Stalecollected in 20m
Li Auto Bets on Robots Amid Auto Slump
💡Li Auto's 2560TOPS chip + robot pivot amid EV woes—key for auto AI hardware watch
⚡ 30-Second TL;DR
What Changed
Q4 vehicle ASP falls 2.7k RMB to 25k; i6 low-price mix drags margins to 16.8%
Why It Matters
Shifts focus from struggling EVs to AI chips/robots for growth amid price wars. High cash reserves (917B net) support R&D pivot to embodied AI.
What To Do Next
Benchmark Li Auto's 2560 TOPS M100 against Nvidia for edge AI inference costs.
Who should care:Founders & Product Leaders
🧠 Deep Insight
Web-grounded analysis with 8 cited sources.
🔑 Enhanced Key Takeaways
- •Li Auto's M100 chip prototype production was completed in Q1 2025, followed by functional, performance, and stress tests within two weeks, with small-batch road testing on prototype vehicles starting shortly after[1][2][3][4][5].
- •The project is led by CTO Xie Yan, formerly of Intel, Alibaba, and Huawei, who is driving an integrated hardware-software strategy with investments in NPUs, SoCs, and software optimization[5].
- •Until M100 mass production, Li Auto continues using chips from Nvidia and Horizon Robotics in current models[1][2][4].
📊 Competitor Analysis▸ Show
| Company | Chip | Computing Power Benchmarks |
|---|---|---|
| Li Auto | M100 | 2560 TOPS; 1 chip = 2x Nvidia Thor-U (LLM tasks), 3x Thor-U (vision/CNN tasks)[1][2][3][4] |
| Nio | Shenji NX9031 | 1 chip = 4x Nvidia Orin X; ET9 uses 2 chips[1][2] |
| Xpeng | Turing | 1 chip = 3x mainstream chips; G7/P7 use 3 chips for >2200-2250 TOPS[1][2][3][5] |
🛠️ Technical Deep Dive
- •M100 designed in tandem with Li Auto's base model, compiler, and VLA software system for next-gen autonomous driving[2].
- •Demonstrates superior efficiency via co-optimization of hardware (NPUs, SoCs) and software scheduling[2][5].
- •Performance: Effective computing power equivalent to 2 Nvidia Thor-U chips for LLM tasks; 3 Thor-U chips for CNN-based vision tasks like image recognition[1][2][3][4].
🔮 Future ImplicationsAI analysis grounded in cited sources
Li Auto's M100 will deliver 3x performance of current high-end chips at comparable prices
Company management stated the chip, integrated with VLA model system, achieves this via efficient AI inference and execution in next-gen platform[2].
⏳ Timeline
2025-01
M100 chip completes prototype production
2025-08
M100 passes functional, performance, and stress tests; begins small-batch road testing on prototypes
2025-11
Li Auto confirms M100 commercialization starting 2026
📎 Sources (8)
Factual claims are grounded in the sources below. Forward-looking analysis is AI-generated interpretation.
- youtube.com — Watch
- cnevpost.com — Li Auto M100 Smart Driving Chip Into Cars 2026
- chinaevhome.com — Li Auto Tests Own Autonomous Driving Chip Topping Nvidia Thor U in Some Calculations
- cnevpost.com — Li Auto Key Advance in House Chip
- stocktwits.com — Chsc6y5rdlx
- autotechinsight.spglobal.com — Li Auto Advances M100 Chip to 2026 Mass Production
- kr-asia.com — 156956
- automotiveworld.com — Li Auto Progresses M100 Chip to 2026 Mass Production
📰
Weekly AI Recap
Read this week's curated digest of top AI events →
👉Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: 虎嗅 ↗