The “Autonomous Driving SoC Research Report, 2025” has recently been added to ResearchAndMarkets.com, providing a comprehensive analysis of the evolving autonomous driving semiconductor (SoC) landscape in China and globally. The report delves into market trends, OEM strategies, chip vendor offerings, and projections through 2030, highlighting the accelerating adoption of high-level intelligent driving systems in passenger vehicles.
Rising Penetration of High-Level Intelligent Driving
In 2024, China’s domestic passenger car market achieved total sales of approximately 22.84 million units, marking an 8.6% year-on-year increase. This growth exceeded the 5.8% increase observed in 2023 by nearly three percentage points. A notable feature of this market expansion is the rapid proliferation of L2 and above intelligent driving SoCs, with more than 14.56 million passenger cars equipped with these chips, corresponding to a 63.8% installation rate—an increase of roughly six percentage points over 2023. While overall car sales growth has begun to moderate, the deployment of intelligent driving SoCs continues to accelerate, reflecting growing consumer demand for advanced driver-assistance features.
Intelligent Driving Levels and Market Segmentation
L2.9 SoCs
Most models priced between 250,000 and 300,000 yuan now come standard with L2.9 SoCs, with penetration exceeding 50% during 2024 and Q1 2025. For high-end vehicles above 500,000 yuan, L2.9 SoC adoption surpassed 30% during the same period. Key market leaders include NVIDIA Orin-X, Huawei Ascend 610, and Tesla FSD, while new entrants expected to reach mass production in 2025 include NVIDIA Thor-X, Thor-U, Orin-Y, Qualcomm 8797/8397, Horizon J6P/J6M, Black Sesame Technologies A2000/C1236, RHINO Guangzhi R1, XPeng ‘Turing’, and NIO Shenji NX9031.
L2.5 SoCs
The mid-tier L2.5 market—covering models priced above 200,000 yuan—shows an upward adoption trend. Mainstream SoCs include Horizon J5/J6E, NVIDIA Orin-X/Orin-N, Black Sesame Technologies A1000, TI TDA4VH/VM, and Mobileye EyeQ5H. These chips support enhanced automated features, bridging the gap between basic driver assistance and higher-level autonomous capabilities.
L2 SoCs
For entry-level vehicles under 100,000 yuan, L2 SoCs dominate, typically supplied by Horizon, Axera, and Mobileye. While these chips offer lower computational performance, they provide essential features such as adaptive cruise control, lane-keeping assistance, and basic driver monitoring.
2025: Replacement Cycle and OEM Self-Development
The year 2025 marks a significant replacement cycle for intelligent driving SoCs. Leading OEMs are increasingly developing self-owned SoCs, including NIO and XPeng, while others like Li Auto continue incremental upgrades to existing platforms. This shift represents a broader industry trend: OEMs seeking greater control over performance, security, and differentiation in autonomous driving capabilities.
Case Study: Li Auto ‘Turing’ AI Chip
Li Auto’s Max and Ultra versions of its L-series vehicles are set to adopt the NVIDIA Thor-U chip in 2025. This chip offers a single-chip computing power of 730 TOPS, enabling support for large VLA models. Key features of VLA include:
- Language Interactive Driving: Voice commands perform complex operations such as navigating congestion or locating parking, enabling anthropomorphic decision-making.
- Scene Generalization: Diffusion models optimize driving trajectories for rare or challenging scenarios, including unmarked roads and complex intersections.
Additionally, the Li Auto L6 Pro will integrate a lidar system, with its SoC upgraded from Horizon J5 (128 TOPS) to J6M (256 TOPS), supporting up to 12 cameras, six radars, and lidar units. Following these hardware upgrades, the AD Pro system is expected to match AD Max in active safety capabilities.
Chip Vendor Strategies: High Computing Power and Full-Stack Deployment
Vendors continue to advance high-performance chips and promote full-stack algorithm IP deployment, reducing technical barriers for OEMs. Notable trends include:
- NVIDIA: Thor series chips offer computing powers up to 2,000 TOPS, with mass production starting in 2025. Models like Li Auto MEGA, Lynk & Co 900, Zeekr 9X, and Xiaomi YU7 have confirmed installations, with potential partnerships spanning BYD, GAC, Mercedes-Benz, Hyundai, and Jaguar Land Rover. Thor variants include:
- Thor-U: 730 TOPS, widely adopted in passenger vehicles.
- Thor-X: 1,000 TOPS, intended for Robotaxi and L3 autonomous driving.
- Thor-X-Super: Dual Thor-X chips, using MCM/Chiplet technology, priced $1,000–$1,300.
- Thor-Jetson: Designed for industrial robotics, 1,000 TOPS, 100GB Ethernet bandwidth, priced $400–$500.
- NDAS (NVIDIA DRIVE AV Solution): NVIDIA’s end-to-end L3 autonomous driving system launched in April 2025, covering sensor simulation, traffic modeling, synthetic data generation, and model deployment. The dual Thor highway version L3 is expected in early 2027, with the urban version arriving by late 2027.
- Qualcomm: Promotes Snapdragon Ride and Snapdragon Ride Flex platforms, aiming for large-scale mass production of over 20 ADAS/AD projects globally between 2025–2026. Collaborating with OEMs such as Leapmotor, FAW, Geely, Volkswagen, GM, Honda, Mercedes-Benz, and BMW, Qualcomm achieves smooth operation of large AI models (up to 14B parameters) locally, integrated by Tier-1 suppliers like Bosch, Autolink, and Visteon.
- Black Sesame Technologies: Introduced the Huashan A2000 family, targeting next-generation AI models and high computing power for autonomous vehicles.
- Horizon: Released Journey 6P (J6P), a major upgrade from Journey 5, increasing transistor count by 85% and integrating five computing modules—CPU, GPU, NPU, MCU, VPU—through its BPU ‘Nash’ architecture, significantly boosting performance for complex driving algorithms.
Key Report Coverage
- China Passenger Car Intelligent Driving System and SoC Strategy
- Definitions and performance metrics of intelligent driving SoCs
- Vehicle automation taxonomy and installation rates (L1-L4) 2022–2030E
- Market share analysis for L2 and L2.5 SoCs across different price segments
- Passenger car sales by price range, 2023–2024
- Intelligent Driving SoC Market Research
- Classification by automation level
- Comparative analysis of ultra-high computing SoCs (>500 TOPS)
- Installation volumes and market shares 2022–2024
- Cost breakdowns and shipments by nanometer process
- Market size and revenue for domestic and global vendors
- OEM Deployment and Self-Research Strategies
- Detailed strategies of OEMs including NIO, XPeng, Li Auto, Leapmotor, FAW Hongqi, GAC, SAIC, BAIC, Changan, Great Wall Motor, Dongfeng, Geely, Chery, BYD, Xiaomi, Tesla, BMW, Volkswagen, Audi, Daimler Benz, Toyota
- Foreign Intelligent Driving SoC Vendors
- NVIDIA, Qualcomm, Mobileye, TI, Renesas, Ambarella, Intel
- Chinese Intelligent Driving SoC Vendors
- Horizon, Black Sesame Technologies, and other emerging player
