MCU and SoC are mainly responsible for computing and processing in automobile chips, among which MCU is widely used, and most of all commercial, industrial and consumer electronic devices will use MCU. In all application fields, automobiles have the highest requirements for the performance, reliability and safety of MCU.
In 2021, in the MCU market, the proportion of vehicles reached 38%-40%. At present, the market size of automobile MCU is about 8 billion US dollars, and CAGR is 11% from 2022 to 2025, which is higher than the average level of MCU industry. Among all precision MCU, 32 bits are the mainstream, accounting for nearly 77%, 16 bits about 18%, and 8 bits about 5%. 32-bit MCU accounts for 77% of revenue and about 40% of shipments, so 32-bit MCU accounts for the largest proportion in the automobile market, with a market size of about 5.8 billion US dollars.
In terms of price, ASP of automobile MCU is significantly higher than other applications, reaching $3.1 in 2021. Since 2020, due to the shortage of supply, the price of automobile MCU has increased by 16%, and in 2021, it has increased by 22%, with the largest price increase among all application types of MCU. Prediction from Yole, the price of automobile MCU will remain at a high level in the future.
At present, with the rise of new energy vehicles and the improvement of intelligent level, the consumption of SoC has increased greatly, even exceeding MCU, so how should automotive MCU develop in the future?
Present and Future of Automobile MCU
Before the popularization of automobile electrification and intelligence, each functional block of automobile was controlled by ECU, and MCU was the core of ECU. Besides MCU, ECU also integrated memory (ROM, RAM), input/output interface (I/O), analog-to-digital converter (ADC), which monitored various automobile running data (braking, shifting, speed, heading angle, etc.) and various automobile running states (acceleration, skid, fuel consumption, distance from front car, etc.) at any time, and calculated the information sent by various sensors according to pre-designed program logic. After processing, each parameter was sent to relevant execution modules to execute various predetermined control functions. This architecture is commonly referred to as distributed.
With the deepening and popularization of intelligent, networked and electrified applications in automobiles, the electronic and electrical (E/E) architecture of automobiles gradually moves from distribution to centralization, so as to reduce vehicle wiring harnesses and improve the efficiency of internal information flow. Under the distributed architecture, each functional module of the automobile is independent of each other, and only MCU is needed to meet the required computing power. When the electronic and electrical architecture evolves to centralized, the computing power tends to be centralized, and it is difficult to meet the computing requirements only by traditional MCU, which accelerates the development of automotive SoC chips.
At present, SoC chips composed of AI accelerators such as CPU, GPU and NPU are the mainstream chips for automatic driving and intelligent cockpit, which are integrated in domain controllers. Domain is to further concentrate the traditional ECU control and form several functional blocks, which can be summarized as vehicle control domain (VDC), intelligent driving domain (ADC) and intelligent cockpit domain (CDC). In the future, on the basis of the centralized domain-based architecture, it will also develop towards the ribbon architecture of domain convergence (central integration), which further simplifies the architecture and makes the functions more centralized.
Under the domain control architecture, the control chip will develop towards MCU + SoC, and SoC chip cannot replace all MCU. On the one hand, it is not necessary for all systems to access SoC, such as the control mode of making turn signals shine. If MCU scheme is not used, all access to SoC will form a star network, which will not only increase the number of wires, but also greatly increase the difficulty of management; On the other hand, some MCUs are needed as the security redundancy of SoC chips.
At present, the automobile market is still dominated by fuel vehicles, and the development momentum of pure electric vehicles is very strong, but it will take time for the market share to exceed that of fuel vehicles. In this case, the amount of MCU is still considerable, especially in intelligent cockpit, high-precision map, body electronics and other applications, the demand for MCU has greatly increased, and the number and unit price of MCU required have increased. MCU is needed to control everything from power system, body control, motor drive control system, instrument panel, car audio-visual entertainment system, advanced safety system and ADAS to windows, wipers, electric seats, reversing radar and car keys. At present, the amount of MCU in an electric vehicle can reach dozens, especially the switch control system, automatic parking, advanced cruise control, anti-collision system, etc., and the demand for 32-bit MCU will increase significantly.
In the past few years, the development of assisted driving is in the ascendant, with MOBILEYE, NVIDIA and traditional MCU manufacturers as the main manufacturers. People's ideal automatic driving function is still in its early stage (assisted driving), that is, L2 level, and only when it reaches L4 level can it be truly called automatic driving. At this stage, the electronic and electrical architecture of automobiles is still mainly distributed, and the intelligent driving demand can be realized by using intelligent forward-looking integrated machines, and the demand for chip computing power is not high. MOBILEYE occupies most of the L1-L2 visual ADAS chip market at this stage, while the chips of traditional MCU manufacturers, such as RENESAS and TI, are installed in BOSCH system solutions, and also occupy a large market share. Based on the general GPU architecture, NVIDIA launched Tegra Parker SoC in 2016 for TESLA HW 2.0 platform, bringing GPU-based self-driving SoC to the market. However, the iteration speed of SoC technology at this stage is still slow, and MCU still has room for development.
In the short and medium term, L1/L2 assisted driving smart cars will still account for a large proportion of the market. Due to the lack of path planning function and the limited number of sensors, the integration and decision-making tasks can be completed only by MCU on the sensor side, and the distributed architecture is still the mainstream. Therefore, the accelerated popularization of low-end ADAS will drive the increase of MCU usage. In L2 + and higher-level smart cars, SoC chips will gradually replace MCU, but some chassis interactive high-real-time tasks still need MCU to complete, and ADAS domain controllers will still be equipped with at least one MCU to ensure system functional safety.
In addition to the auxiliary driving system, other functional domains, such as cockpit, instrument, power, body control, etc., have different requirements for MCU usage. As the cockpit becomes more and more intelligent, the amount of MCU will decrease. Intelligent cockpit has many functions, including infotainment, human-computer interaction, etc. In order to realize these advanced functions, higher performance chips are needed, which makes MCU's status decline. Taking dashboard and head-up display (HUD) as examples, the performance improvement of dashboard makes MCU's master control position replaced by high computing power processor. HUD function, especially AR-HUD, needs to process a large amount of information, and the processor needs system on a chip SoC.
In the power domain, the power system of traditional fuel vehicles mainly includes engine and gearbox, which each has an MCU, an engine master MCU and a transmission master MCU. The power system of pure electric vehicle includes three parts: vehicle control module, motor controller module and battery management module. The power domain controller centrally controls the above three parts. This system needs more MCUs, and it is estimated that each vehicle will use at least five more than traditional fuel vehicles. However, the number of MCUs used in the car body control system is relatively stable and has little change. The reason is that the car body domain technology is mature and the service life cycle is long. To realize these functions, the requirements for chip computing power are low, and the price of MCUs used is also low.
In a word, MCU still has a room in the control application of traditional functions, while the consumption in cockpit and automatic driving will be significantly reduced. With the further development of automobile electronic and electrical architecture, cockpit domain and automatic driving domain also have a trend of integration, until the whole vehicle central computer control architecture is realized, and there may be cloud computers to control automobiles in the future.
Forecast of Automotive Semiconductor Market Segment Scale (US $100 million, sorted according to the scale in 2021)
It can be seen from the table that in the foreseeable future, although the growth rate of various automotive chips, especially SoC, is very high, MCU is still the largest in the overall market. After that, as the proportion of fuel vehicles decreases, the proportion of pure electric and intelligent vehicles increases greatly, and the overall consumption of MCU will decline.
At present, Tesla Electric Vehicle is a typical representative of centralized architecture. Under the domain architecture, TESLA integrates many small ECU functions into the regional controller. Therefore, the number of ECU is less than that of ID.4/Mach E, and the ECU consumption of Model Y, ID.4 and Mach E is 26, 52 and 51 respectively. This is much less than the ECU of traditional fuel vehicles, and the decrease of ECU number leads to the decrease of MCU consumption.
Generally speaking, the application of MCU changes with the development of automotive electronic and electrical architecture, and the consumption will go through a process from less to more, and then from more to less. SoC chips will integrate some low-end MCU functions, and with the improvement of automotive SoC computing power and increasingly powerful functions, more and more MCU functions will be integrated into SoC. With the development of automotive electronic and electrical distributed architecture to domain control, the amount of MCU per vehicle will gradually increase from an average of 30-40 to 70-80. In the future, with the popularization of centralized architecture and the concentration of computing power to the whole vehicle computing platform, the amount of MCU per vehicle will gradually decrease to about 50-60.
Market structure of automobile MCU
For auto parts and vehicle manufacturers, the update iteration speed of MCU is slow and the service life is long, so they tend to be suppliers who can provide stable solutions. To a great extent, this determines the long-term stable pattern of the global automotive MCU market, with RENESAS, NXP and INFINEON, followed by STMICROELECTRONICS, TEXAS Instrument, ONSEMI and MICROCHIP.
Different manufacturers have different emphases. For example, INFINEON is good at chassis and power domain control; NXP builds an open MCU platform based on Arm architecture, which is suitable for small and medium-sized customers. In addition to ADAS camera control, NXP has significant advantages in connection, network and sensor control, and its application coverage is wide; RENESAS relies on Japanese automobile manufacturers, with products covering high, middle and low end, and applications covering body, chassis, power, intelligent cockpit, etc.; In ADAS, RENESAS is better at camera control, INFINEON is good at central security MCU, and NXP is good at radar control (including millimeter wave radar and ultrasonic radar).
All of the above are IDM manufacturers, and automobile MCU foundry also occupies a large market share. TSMC accounts for about 70% of the global shipments of automobile-level MCU foundry, and the global head MCU manufacturers are highly dependent on TSMC. In the past two years, affected by the epidemic situation, TSMC has reduced the production capacity of vehicle-level MCU. With the rapid increase of market demand, the related MCU is in short supply.
China's market power has doubled
At present, China's automobile chips are still highly dependent on foreign enterprises, with the import rate of automobile chips as high as 95% and the self-sufficiency rate less than 10%. The key chips used for power system, chassis control and ADAS are monopolized by foreign giants. On the demand side, China's automobile market accounts for about 30% of the global share, which is the market with the largest demand for automobile-level chips.
Due to the high certification threshold, long certification period and high requirements for reliability, safety, consistency and life, China's automobile MCU chip market has been occupied by foreign manufacturers for a long time. In recent years, many domestic manufacturers have been laying out automotive grade MCU, such as CHIPON Microelectronics, AUTOCHIPS, SEMIDRIVE Technology, BYD Semiconductor, GIGA DEVICE Semiconductors and so on.
In recent years, Chinese mainland manufacturers have started to develop high-end MCUs needed for automobile intelligence, such as intelligent cockpit and ADAS, from the low-end MCUs with low correlation with safety performance, such as wiper, window, remote controller, ambient light control and dynamic running water lamp. At present, GIGA DEVICE Semiconductors, CHIPSEA Technologies, CCORE Technology, BYD Semiconductor and other manufacturers all have MCU products that have passed the vehicle regulation verification, and SINO WEALTH Electronic automotive grade MCU has also been released.
Forecast of Automotive MCU Market Scale in China and the World
It is conservatively estimated that the market size of automotive MCU in China from 2022 to 2025 is US $32.92, US $36.02, US $3.930 and US $4.274 billion respectively, and the global market size of automotive MCU from 2022 to 2025 is US $85.59, US $93.66, US $10.219 and US $11.112 billion respectively. Under the expectation of penetration rate of 25% and 30%, the CAGR of China's automotive MCU market in 2021-2025 is 9.24% and 11.22%, respectively, and the market scale of vehicle-level MCU has great room for growth.
At present, China's car-level MCU industry is in the introduction period. On the basis of relatively mature consumer-level and industrial-level MCU technologies, the large-scale demand of car-level MCU market will stimulate domestic manufacturers to increase investment in research and development. Manufacturers who can give priority to meeting technical requirements and obtaining customer orders will quickly occupy the market, and domestic substitution potential is huge.
The popularity of the automotive MCU market continues unabated, and the domestic substitution potential is huge-China exportsemi.com
