China drives system-on-chip innovation in auto industry

Mainland China is fast becoming a key market for system-on-chip (SoC) technologies, particularly those enabling next-generation digital cockpits and advanced driver assistance systems (ADAS). This momentum is driven by several converging automotive trends in the country: growing EV production, the rapid adoption of autonomous vehicles, and an industry shift toward centralized vehicle architectures.

According to S&P Global Mobility, by 2030, Greater China is forecast to lead global demand for autonomy domain controller (ADC) and front-view camera (FVC) SoCs, accounting for over a third of the total market. Europe is anticipated to follow with about one-fifth and the US with just over a 10% market share. Greater China is also expected to become the biggest market for cockpit domain controller (CDC) SoCs and head-unit SoCs. 

Cockpit and ADAS SoCs are becoming the computational backbone of modern vehicles. In a cockpit, SoCs power a wide array of functions from digital instrument clusters and AI-based voice assistants to large-format infotainment displays and immersive augmented reality (AR) experiences.  

As vehicles evolve into software-defined platforms, cockpit SoCs are critical for delivering seamless user experiences, real-time connectivity and over-the-air (OTA) updates that continuously enhance in-vehicle features.

Similarly, ADAS SoCs enable higher levels of driver assistance and autonomous driving. They are responsible for processing vast streams of sensor data, including cameras, radar, lidar, and ultrasonic sensors in real time to support perception, localization, and decision-making. As the industry moves toward Level 2+ and higher automated and autonomous vehicles, the computational demands of ADAS will continue to soar, requiring more powerful SoCs.  

The demand for high-performance cockpit SoCs in mainland China is intensified by the rapid integration of advanced features such as AI chatbots and large-screen displays, even in vehicles with relatively low price points (less than $20,000). The industry's shift toward larger, higher-resolution displays, including "pillar-to-pillar" panoramic screens and 3D/AR displays, demands significant graphical processing power.

Increasing levels of autonomy — requiring massive compute power to process an exponentially growing volume of sensor data for real-time perception, localization, prediction and planning — is also a driver.

This need is further amplified by the industry's shift toward "end-to-end" (E2E) AI models, where raw sensor data directly feeds into complex neural networks for vehicle control. These models demand incredibly powerful AI accelerators and high memory bandwidth, pushing the limits of current system-on-chip technology. 

As automakers face tighter margins, particularly in the competitive mainland Chinese market, there is an accelerating shift toward integrating consumer-grade SoCs into automotive applications. Consumer-grade refers to system-on-chip technology first developed for consumer electronics use.

For instance, Xiaomi’s recently unveiled YU7 sport utility vehicle features Qualcomm’s Snapdragon 8 Gen3 (4-nanometer [nm]) processor, originally built for Android smartphones. This chip is used in the vehicle cockpit through automotive-grade modifications, including upgraded packaging processes and redundant circuit design. 

The mobile-grade chip costs only half of traditional automotive-grade chips while delivering superior performance. 

Compared to Qualcomm’s Snapdragon 8 Gen2 — also designed for smartphones — the Gen3 offers a 40% performance improvement in central-processing-unit (CPU). Its Adreno 750 graphics processing unit (GPU) supports real-time 3D rendering, enabling the vehicle's infotainment system to achieve smartphone-level smoothness.  

Cost pressure is making automakers consider options such as integrating Qualcomm’s consumer-grade SoCs into automotive-grade system-in-package (SiP) modules or even switching from automotive-grade Snapdragon 8295 platforms to cost-effective alternatives such as MediaTek SoCs.

Mainland China is the only major auto market aggressively pushing toward single-chip solutions. There is a strategic move toward utilizing a single system-on-chip to integrate both ADAS and in-vehicle infotainment (IVI) functionalities. This decision is driven again by heightened cost pressures within the region, which necessitates economical solutions. 

Additionally, this single-chip approach reflects a differing prioritization of safety concerns compared to Western original equipment manufacturers, where safety considerations traditionally receive more emphasis in the technological integration process. In the sub-150,000 yuan segment, OEMs are opting for simplified, single-chip cockpit solutions that reduce bill-of-materials and software integration costs. 

The automotive-grade Snapdragon 8775, Qualcomm’s first system-on-chip made specifically for its Snapdragon Ride Flex platform, is a popular choice here, offering adequate compute for digital cluster, infotainment and some ADAS visualization functions. 

Vehicles priced above 150,000 yuan still use multichip setups, high-performance SoCs separate ADAS and infotainment systems. Configurations such as Qualcomm SA8650 or NVIDIA Orin for ADAS, paired with Qualcomm’s automotive-grade SA8295 for the cockpit, are increasingly common in mid to premium-tier vehicles. 

As vehicle cockpits become more innovative and interactive, the need for AI computing power is skyrocketing. Mainland Chinese OEMs are again pushing the envelope. The Lynk & Co 900, a plug-in hybrid electric vehicle, features two Snapdragon 8295 chips, offering a combined 60 tera operations per second (TOPS) of AI performance. Meanwhile, Xpeng’s G7 SUV pairs an 8295 with a Turing AI accelerator, resulting in an unprecedented 60-plus 735 TOPS, a configuration that supports advanced voice assistants, multimodal interfaces and AI-driven driver monitoring.

Such performance levels, however, are still limited to mid-end vehicles. For mainstream segments, manufacturers are more likely to pursue AI-optimized versions of consumer-grade chips, which deliver “good enough” intelligence at a fraction of the cost. For instance, Xpeng has strategically deployed its self-developed AI chips across as many mainstream models as possible to achieve cost amortization.