The automotive industry is witnessing a significant shift towards AI-oriented cockpit SoCs, with the localization rate of intelligent cockpit SoCs exceeding 10% in the Chinese market. While established vendors like Qualcomm, Renesas, and AMD still dominate the market, domestic players such as SemiDrive, Huawei HiSilicon, and SiEngine are rapidly gaining traction.
According to ResearchInChina, the localization rate of intelligent cockpit SoCs is expected to continue rising, with a forecast that AI-oriented cockpit SoCs will become mainstream within the next 2-3 years. The evolution of automotive intelligent cockpit SoC chips is driven by advancements in technology, with key developments including the transition from 7nm to 4nm and below processes.
In 2024, chips with 7nm and below processes accounted for 36%, a figure expected to exceed 65% by 2030. The next generation of chips will shift towards 4nm and 3nm processes, offering improvements in transistor density, performance, and power efficiency. This will better support high-throughput AI computing tasks for AI cockpits across various applications.
Integrated cockpit-driving SoCs that support AI cockpits and high-level autonomous driving are also emerging, such as the NVIDIA DRIVE Thor, Black Sesame’s “Wudang” C1200 series, Qualcomm’s SA8795P/SA8775P series, and MediaTek’s CT-X1 (MT8678). These advancements signal a shift towards AI-oriented cockpit SoCs becoming mainstream in the near future.
One of the key players in this space is SemiDrive, which recently unveiled its next-generation AI cockpit chip X10 at the 2025 Shanghai Auto Show. The X10 series chips, scheduled for mass production in 2026, feature an advanced 4nm process, supporting the on-device deployment of a 7B-parameter multimodal large model.
The X10 series products boast impressive specifications, including an ARMv9.2 CPU architecture optimized for AI computing, a 1800 GFLOPS GPU, a 40 TOPS NPU, and a 128-bit LPDDR5X memory interface with a bandwidth of 154 GB/s. These features address the performance requirements for deploying 7B multimodal models on-device, a significant challenge for AI cockpits.
In addition to hardware advancements, the development toolchain for the X10 includes functions such as compilation, quantization, simulation, and performance analysis. This toolchain aims to reduce the cycle for model deployment and performance optimization, providing a standardized model invocation interface for simplified development and migration of AI applications.
Furthermore, the integration level of automotive intelligent cockpit SoC chips continues to increase, with vendors like Qualcomm and MediaTek integrating 5G modems, WiFi 7, Bluetooth, and V2X modules into smart cockpit SoCs. This convergence of high-speed connectivity and intelligent computing capabilities on a single chip enhances real-time performance, multitasking capabilities, and user experience in vehicles.
SIP packaging modules for cockpit SoCs are also gaining traction, addressing challenges in PCB reliability, thickness control, and warpage management. Companies like Qualcomm offer SIP modules directly, while module manufacturers like Quectel provide solutions based on advanced SiP technology, reducing hardware design complexity.
Overall, the automotive industry is at the forefront of innovation with the development of AI-oriented cockpit SoCs, signaling a shift towards intelligent, connected vehicles in the near future.
