Market Competition and Localization Process of Automotive SoC Chips

In 2024, the global intelligent driving SoC market is expected to exceed $10 billion, reaching an estimated $28.306 billion by 2027, with a compound annual growth rate of 43.11%.

Intelligent Driving SoC Chips

Market Demand

Generally, different vehicle models from manufacturers have different market positioning, which determines the price range of the models. Models at different price points exhibit varying sensitivity to the pricing of functional configurations. Regarding intelligent driving configurations, there are different levels of demand for the main control SoC chips based on the market positioning of the vehicles. Currently, based on the varying AI computing power requirements of different levels of intelligent driving solutions, intelligent driving SoC chips can be roughly divided into three types: low-computing SoC chips (2.5~20 TOPS), mid-computing SoC chips (20~80 TOPS), and high-computing SoC chips (≥100 TOPS).

1) Low-computing SoC Chips

The AI computing power of low-computing SoC chips is typically in the range of 2.5~20 TOPS, supporting product forms mainly as front-view integrated machines or distributed driving or parking controller solutions, characterized by a pursuit of high cost-performance; in terms of functionality, they mainly provide basic L0~L2 level assisted driving functions, with some models possibly offering high-speed NOA functionality, generally priced between 100,000 to 150,000 RMB.

Currently, ADAS functions at L2 and below have entered a rapid growth phase, with front-view integrated machines accounting for about 75% of the ADAS market, remaining the main product form in the current ADAS market. Low-computing SoC chips still possess considerable market space in the future.

Chip Manufacturer	Process Technology	AI Computing Power (TOPS)	Mass Production Status
Mobileye	EyeQ4	28nm	2.5	Mainly applied in front-view integrated machines. In 2018, EyeQ4 was first equipped on NIO ES8, with other models including NIO ES6/EC6, Xiaopeng G3, Li Auto One, SAIC General GL8, Jiangling Ford Lingrui, BMW 5 Series, etc.
TI	TDA4VM	16nm	8	Used in integrated driving and parking domain controller solutions, with equipped models including Chery Exeed LX, Geely Boyue L, Lynk & Co 09 EM-P Pilot Edition, Lantu Zhi Guang, Baojun KiWi EV, and Yue Ye, Nezha S pure electric four-wheel drive version, etc.
Ambarella	CV22AQ	10nm	4eTops	Mainly applied in front-view integrated machines, equipped models include GAC Trumpchi ES9, GAC Haobo GT/HT, GAC Aion S MAX, GAC Trumpchi New Energy E8, Hozon V09, etc.
Horizon	J2	28nm	4	Mainly applied in front-view integrated machines, equipped models include Deep Blue SL03 low configuration version, Changan Qiyuan A05, Changan UNI-V, etc.
J3	16nm	5	Used in front-view integrated machines or integrated driving and parking domain control solutions, equipped models include Deep Blue SL03 high configuration version, Roewe RX5, Deep Blue S7, 2021 Li Auto One, Nezha S pure electric four-wheel drive version, Star Era ES, Qichen VX6, etc.
Black Sesame	A1000L	16nm	16	Applied in integrated driving and parking domain control solutions, expected to be implemented on Hongqi E001 and E202 models in 2024.

Basic information of low-computing intelligent driving SoC chips (Information source: compiled based on public information)

2) Mid-computing SoC Chips

Mid-computing SoC chips typically have AI computing power in the range of 20~80 TOPS, supporting product forms mainly as lightweight integrated driving and parking domain controller solutions; in terms of functionality, they promote features such as “easy to use” high-speed NOA, urban memory NOA, and memory parking, with some models possibly offering city NOA functionality, generally priced between 150,000 to 250,000 RMB.

Overall, the mid-computing SoC chip market is a result of rapid iteration and upgrades of chips. NVIDIA’s Xavier was initially considered a high-computing SoC chip in the intelligent driving market, but with the emergence of NVIDIA’s next-generation chip Orin, and subsequent appearance of Horizon J5, Ambarella CV3-AD, and other chips with hundreds of TOPS, merely pursuing high computing power for some mid-range models does not guarantee their competitiveness in the market, and may instead lead to a loss of cost-performance advantage. Therefore, chips of the same level as Xavier have “downgraded” from the high-computing SoC chip market to the mid-computing SoC chip market.

Chip Manufacturer	Process Technology	AI Computing Power (TOPS)	Mass Production Status
NVIDIA	Xavier	12nm	30	In 2020, the Xavier chip was first equipped on Xiaopeng P7 for mass production delivery.
Orin-N	7nm	84	In September 2023, Orin-N was first equipped on Denza N7 for mass production delivery; additionally, in March 2024, the Xiaomi SU7 Pilot Pro version equipped with Orin-N also started mass production delivery.
TI	TDA4VH	10nm	32	Currently, mass-produced models mainly use DJI’s 7V pure visual intelligent driving solution, including Baojun Cloud Lingxi version, Baojun Yue Ye Plus, and Chery iCAR03, etc.
Mobileye	EyeQ5H	7nm	Mass-produced equipped models include Zeekr 001/009, BMW iX, etc.
Ambarella	CV72AQ	5nm	Supports full-time driving and parking integration function with a single SoC chip, currently in collaboration with customers for development.
Black Sesame	A1000	16nm	58	In November 2023, the A1000 chip was first equipped on Lynk & Co 08 for mass production delivery; other mass-produced models include Hozon V09, Dongfeng eπ007, etc.

Basic information of mid-computing intelligent driving SoC chips (Information source: compiled based on public information)

3) High-computing SoC Chips

High-computing SoC chips typically have AI computing power over 100 TOPS, supporting product forms mainly as high-end integrated driving and parking domain controller solutions, and even cockpit integration solutions; in terms of functionality, they promote features such as “easy to use” city NOA, AVP, and other L2+ level functionalities, with some models considering hardware embedding for L3 and higher-level autonomous driving functionalities, generally priced above 250,000 RMB.

In the development process towards high-end intelligent driving functionalities, new algorithms (Transformer + BEV + OCC) and more advanced vehicle EE architectures (central computing + regional control) are required, all of which necessitate high-computing SoC chips as the “cornerstone” for support.

Chip Manufacturer	Process Technology	AI Computing Power (TOPS)	Mass Production Status
NVIDIA	Orin-X	7nm	254	Equipped models include NIO ET5/ET7, Li Auto L7/L8/L9 Max version, Xiaopeng G6/G9/X9/P7i, Zhi Ji LS7, Xiaomi SU7 Pilot Max version, etc.
Thor	4nm	2000	Mainly focuses on cockpit integration, announced plans for equipped vehicle manufacturers including Zeekr, Xiaopeng, Li Auto, BYD, and GAC Aion, etc.
Qualcomm	SA8650P	5nm	50/100	Second-generation chip of Qualcomm Ride platform, currently, Bosch, Continental, Veoneer, Valeo, Desay SV, and Junlian Zhixing are all designing and developing based on this chip; mass production is expected to be achieved in 2024.
SA8775P	4nm	——	First product of Qualcomm Ride Flex platform, focuses on cockpit integration, expected to achieve mass production delivery by the end of 2024.
Mobileye	EyeQ Ultra	7nm	Expected mass production delivery in 2025.
Ambarella	CV3-685	5nm	750eTOPS	Launched in 2023, mainly targeting L3, L4 level passenger car autonomous driving and L4 level autonomous driving trucks.
CV3-655	5nm	250eTOPS	Launched in January 2024, mainly targeting urban NOA scenarios.
CV3-635	5nm	125eTOPS	Launched in January 2024, mainly targeting high-speed NOA scenarios.
Huawei	Ascend 610	7nm	Huawei’s MDC610 platform based on a single Ascend 610 chip and MDC810 platform based on two Ascend 610 chips, equipped models include Aito M5/M7/M9, Avita 11/12, Nezha S 715 laser radar version, GAC Aion LX Plus, Arcfox Alpha S Hi version, Zhi Jie S7, etc.
Horizon	J5	16nm	Already equipped on Li Auto L9/L8/L7 Air and Pro versions, BYD Han EV Honor version, and has obtained mass production cooperation with dozens of models from nine vehicle manufacturers.
J6P	7nm	560	First batch of mass production models expected to be delivered in the fourth quarter of 2024.
Black Sesame	A1000Pro	16nm	Currently in collaboration with customers for development.

Basic information of high-computing intelligent driving SoC chips (Information source: compiled based on public information)

Market Structure

From the perspective of market size, according to ICV data, the global intelligent driving SoC market size was $3.295 billion in 2022, with the Chinese market size reaching $1.505 billion, accounting for 45.68% of the global total. It is estimated that by 2024, the global intelligent driving SoC market size will exceed $10 billion, reaching an estimated $28.306 billion by 2027, with a compound annual growth rate of 43.11%.

According to statistics from the Gai Shi Automotive Research Institute, in 2023, the number of intelligent driving domain controllers for passenger cars (excluding imports and exports) in the Chinese market reached 1.839 million sets, a year-on-year increase of approximately 70%, with an installation rate of about 8.7%.

Additionally, in 2023, the ranking of intelligent driving domain control chip installation volume in the Chinese market showed that the top-ranked was Tesla’s FSD chip, with a shipment volume of approximately 1.208 million units, accounting for 37%; the second was NVIDIA’s Orin-X chip, with a shipment volume of 1.095 million units, accounting for 33.5%; the third was Horizon’s J5 chip, with a shipment volume of 200,000 units, accounting for 6.1%; the fourth was Mobileye’s EyeQ4H chip, with a shipment volume of about 200,000 units, also accounting for 6.1%; the fifth was Mobileye’s EyeQ5H chip, with a shipment volume of 174,000 units, accounting for 5.4%.

Market Competition and Localization Process of Automotive SoC Chips

2023 China Market Intelligent Driving Domain Control Chip Installation Volume Ranking

From the overall industry structure, currently, the market share of domestic intelligent driving SoC chips is still at a disadvantage. In 2023, foreign chip solutions accounted for a large share, totaling over 80%. Among them, Tesla’s FSD and NVIDIA’s Orin-X accounted for over 70% of the market share. The FSD chip is self-developed and used by Tesla, with two FSD chips standard on each vehicle. NVIDIA’s Orin-X chip is equipped in many models, covering several manufacturers including NIO, Xiaopeng, Li Auto, Zhi Ji, Xiaomi, and dozens of models.

Among domestic intelligent driving SoC chips, the highest shipment volume is the J5 chip, with a shipment volume of 200,000 units in 2023, mainly equipped on Li Auto L7/L8 Air and Pro versions and L9 Pro version; moreover, in February 2024, the J5 chip will be mass-produced on BYD Han EV Honor version.

However, the market structure of intelligent driving SoC is not yet finalized, and domestic chip manufacturers possess their own advantages, still having opportunities to catch up. For example, compared to foreign chip manufacturers, domestic chip manufacturers have stronger local service capabilities and can quickly adapt to changes in local manufacturers’ needs; additionally, the impact of geopolitical factors has also accelerated the pace of domestic chip replacement to a certain extent.

Intelligent Cockpit SoC Chips

Market Demand

Currently, the configuration level of intelligent cockpits has become one of the important reference indicators for consumers when purchasing vehicles. Similarly, intelligent cockpits are also a key area for manufacturers to create differentiation and brand influence. With the increasing number of functions integrated into the cockpit, the hardware resources and computing power requirements will also become higher, high-computing and high-performance SoC chips will become a necessity for intelligent cockpits.

In 2023, the delivery volume of cockpit domain control systems in the Chinese market reached 3.476 million sets, with an installation rate of 16.5%.From the perspective of cockpit main control SoC chip installation volume, foreign chip brands still hold a dominant position, with Qualcomm alone accounting for nearly 60% of the market share.

From the type of chip manufacturers, consumer chip manufacturers currently hold an advantageous position in the intelligent cockpit chip SoC market. Industry experts point out that the reason consumer electronic chip manufacturers can enter the cockpit field is that the technical barriers for upgrading consumer electronic SoC chips to cockpit SoC chips are not high — the technical requirements of the two are highly similar, and the special requirements for automotive-grade mainly reflect in aspects like lifespan and adaptation to automotive environments; however, the difficulty for consumer electronic chip manufacturers to pass these automotive-grade certifications is not particularly high. At the same time, consumer electronic chip manufacturers already possess sufficient design capabilities on the consumer side, enabling them to design automotive-grade cockpit chips in the automotive field.

Moreover, compared to traditional automotive chip manufacturers, leading consumer electronic chip manufacturers like Qualcomm, AMD, and Samsung have cost advantages and iteration speed advantages in the intelligent cockpit SoC chip field that traditional chip manufacturers cannot match.

1) Cost Advantage

Leading consumer chip manufacturers can maximize their production and sales capabilities in the consumer sector to amortize the overall design costs of the chip. Therefore, when they transfer consumer chips to the cockpit field, it becomes a dimensional reduction attack on traditional chip manufacturers in terms of cost.

Cockpit SoC chips generally include CPU, GPU, NPU, DSP, etc., with these IP designs and licenses generally coming from third-party companies such as ARM, Imagination, etc. For traditional automotive chip manufacturers, the licensing fees for these IPs are very high. However, for leading consumer chip manufacturers like Qualcomm and AMD, the scale effect established in the consumer sector helps them obtain more favorable IP architecture licensing costs.

2) Iteration Speed Advantage

The cockpit SoC chips of consumer-level chip manufacturers not only have significant advantages in process advancement and computing power but also have a faster iteration speed. Their chip iterations can be built on the foundation of consumer chip iterations, thus their cockpit SoC chip iteration speed is naturally much faster than that of traditional automotive chip manufacturers. Currently, Qualcomm has released four generations of cockpit chips, all following the underlying logic of “consumer-level chips first, intelligent cockpit chips later.”

Basic information of Qualcomm’s four generations of cockpit platforms (Information source: compiled based on public information)

4.2.2 Competitive Landscape

According to relevant statistics, in 2022, the global intelligent cockpit SoC chip market size was $3.092 billion. Among them, the number of intelligent cockpit SoC chips installed in new vehicles in China reached 7.505 million units, with a market size of $1.486 billion, accounting for approximately 48% of the global total market share. It is expected that by 2025, the penetration rates of intelligent cockpit configurations in China and globally will reach 78% and 59%, respectively, while the global intelligent cockpit SoC chip market size will exceed $5 billion.

Currently, the market structure of main control SoC chips for cockpit domain controllers has gradually become clear: in the mid-low-end market — traditional automotive chip manufacturers are the main force, such as Renesas, TI, and NXP; in the high-end market — consumer electronic chip manufacturers are the main force, such as Qualcomm, Samsung, Intel, and AMD.

Currently, the market share of intelligent cockpit SoC chips is mainly concentrated in several overseas chip companies, including Qualcomm, AMD, Renesas, Intel, Samsung, etc. From the perspective of the Chinese market, according to Gai Shi Automotive Research Institute data, in 2023, Qualcomm’s cockpit SoC chip market share was the highest, with a shipment volume of 2.26 million units, accounting for 59.2%; AMD ranked second with a market share of 15.1%; Renesas ranked third with a market share of 8.6%; Intel ranked fourth with a market share of 4.6%; and Samsung ranked fifth with a market share of 4.0%; the top five accounted for over 90% of the market share, indicating a highly concentrated market for intelligent cockpit SoC chips.

2023 China Market Share of Cockpit SoC Chips (Excluding Imports and Options)

At this stage, although the penetration rate and market size of intelligent cockpit SoC chips in China’s passenger vehicles are relatively high, the market share of domestic cockpit SoC chips is not high, below 10%. The main reason is that domestic cockpit SoC chip manufacturers started relatively late. However, they currently occupy advantageous positions in terms of timing (rapid development of the domestic new energy vehicle industry and trend of going overseas) and location (domestic chip replacement), indicating that domestic cockpit SoC chips have significant growth potential in the future.

Related posts

Leave a Comment Cancel reply