Where are automotive chips used in the car?

First, due to the increasing demand for autonomous driving and new energy vehicles, the market size of automotive storage chips will continue to expand, with DRAM and NAND being the focus of demand. It is estimated that the number of DRAM and NAND per vehicle will increase fivefold/tenfold from 2021 to 2025, and the value of each vehicle will increase by more than four times, which will drive the overall automotive storage market size to US$8.8 billion in 2025.

Storage chips can be divided into flash memory and memory. Flash memory includes NAND FLASH and NOR FLASH, and memory is mainly DRAM. From the application form, the specific products of NAND Flash include USB (U disk), flash memory card, SSD (solid state drive), and embedded storage such as (eMMC, eMCP, UFS). Storage chips are widely used in smart cars and electric vehicles. The realization of system functions such as smart cockpits, Internet of Vehicles, autonomous driving, infotainment, dashboards, black boxes, and power transmission requires storage technology to provide parameters. Autonomous driving places higher requirements on the computing power and storage capacity of automotive storage chips; new energy vehicles need to increase charging speed and ensure endurance, which also requires continuous upgrading of storage chip technology.

As the level of autonomous driving increases, the demand for automotive chip computing power and storage capacity continues to rise. According to the data from the Forward Industry Research Institute, the code of the autonomous driving system of a high-end car currently exceeds 100 million lines, and the computing capacity of the autonomous driving software reaches the level of 10 trillion operations per second, far exceeding that of airplanes, mobile phones, and Internet software. With the improvement of the penetration rate and level of autonomous driving in the future, the number of lines of automotive system code will show exponential growth, requiring high-bandwidth DRAM for reading and writing, and DRAM will move from DDR32/4GB to LPDDR48GB. In the process of automotive automation, the number of various sensors used to collect vehicle operation and surrounding data will gradually increase, and the external environment data of the vehicle needs to be collected at all times, which is bound to generate a large amount of data processing and storage requirements. In addition to the data generated by the car itself, real-time information sharing, interactive communication, and data exchange between cars will also generate huge storage requirements.

Smart cars' pull on memory chips mainly comes from four major areas: in-vehicle infotainment systems (IVI), advanced driver assistance systems (ADAS), in-vehicle information systems, and digital instrument panels. Among them, IVI accounts for about 80% of the total storage product usage, and ADAS accounts for about 10%. Since autonomous driving is generally based on Level 1/2/2+, the demand for storage capacity is still very limited. Some high-end models are equipped with at most 12GB DRAM and 256GB UFS, which is comparable to current flagship smartphones; in mid-range models, 2/4GB DRAM and 32/64GB eMMC are common configurations; in low-end models, DRAM and eMMC capacity requirements are relatively low, only 1/2GB and 8/32GB are required. According to Canalys data, a total of 11.2 million L2 autonomous driving vehicles were sold worldwide in 2020. Calculated based on an average capacity of 8GB, L2 autonomous driving vehicles consume about 90PB NAND Flash.

Secondly, the improvement of the level of autonomous driving will also promote the continuous increase in the number and pixels of automotive cameras, and the corresponding automotive CIS will continue to increase in volume. It is expected that the number of cameras per vehicle will increase from 2 in 21 years to 6 in 25 years, and the value of each vehicle will increase from US$18.8 to US$57.8. It is estimated that by 25 years, the global automotive CIS market size will reach US$5.75 billion. The intelligence of automobiles has led to the multi-faceted application scenarios of automotive CIS, and the automotive CIS market has broad space. The application scenarios of automotive CIS in smart cars are very wide, mainly divided into three categories: vision, in-cabin and forward processing of advanced driver assistance systems (ADAS). Vision includes reversing images, front view, rear view, bird's-eye view, panoramic parking images, and mirror replacement. For in-cabin applications, there are passenger monitoring, fatigue driving monitoring, dashboard control, driving recorder (DVR), and airbags. Forward processing of advanced driver assistance systems (ADAS) includes forward collision warning, lane departure warning, automatic high beam control, traffic sign recognition, pedestrian detection, adaptive cruise control, night vision, etc. The expansion of automotive CIS application areas has created a broad market prospect.

The improvement of the level of autonomous driving drives the increase in both quantity and price of automotive CIS. In terms of quantity, autonomous driving can be divided into L0-L5 levels. As the level increases, the number of cameras equipped in the vehicle ADAS system increases. The L1 level has 1 camera, the L2 level has 5-8 cameras, the L3 level has more than 8 cameras, and the L4/L5 stage is expected to require 10-15 cameras. In terms of price, from L0-L5, the requirements for the pixels and anti-interference of the vehicle camera are constantly increasing. It is necessary to capture clear images of objects during high-speed movement of the vehicle. The demand for continuous iteration of technology has pushed up the value of automotive CIS. Generally, an L1-level autonomous driving car equipped with 1 rear-view camera usually requires a CIS pixel of less than 2M, 3~8 US dollars per camera, L2/L3 is 2~3M, L4/L5 is 3~8M, and the unit price is 10~20 US dollars.

In addition, with the intelligentization of automotive electronics, the demand for automotive MCUs has doubled (about 70 MCUs for traditional cars, 100-200 for new energy vehicles, and more than 300 MCUs for L2 and above vehicles). The global automotive MCU market size was US$6.6 billion in 2020 and is expected to reach US$8.8 billion by 2023. The intelligentization of automotive electronics + the Internet of Everything + "chip shortage" make automotive MCU the most promising chip market segment. Traditional cars use dozens of MCUs. With the development trend of automotive automation and electrification, the electronic and electrical architecture has been reconstructed. MCUs are widely used in dashboards, climate control, entertainment information, body electronics and chassis, and ADAS systems. The demand for MCUs has doubled.

Automotive MCU is the world's largest application field for MCU products, and MCU products with different bit numbers are suitable for different application scenarios. In 2020, automotive electronics, industrial control, consumer electronics, and medical health accounted for 35%, 24%, 18%, and 14% of the global MCU market, respectively. Among them, MCU is most widely used in the field of automotive electronics. The market demand for automotive MCUs is mainly concentrated in 8-, 16-, and 32-bit microcontrollers, and the application scenarios are various subsystems of the vehicle body, powertrain and body control, dashboard control, entertainment information systems, ADAS, and safety systems. The complexity of the application scenarios increases with the increase in the number of bits of MCU products.

The trend of automobile electrification and intelligence has led to an exponential increase in the demand for automotive MCUs. MCUs are mainly used in the core aspects of safety and driving, control of autonomous driving (assistance) systems, display and computing of central control systems, engine, chassis and body control, etc., and have a wide range of applications. A traditional fuel vehicle requires about 70 MCU chips, while a new energy vehicle requires 100-200 MCU chips. Intelligence will also drive the growth of the automotive MCU market. L2 and below ADAS hardware solutions still use traditional distributed architectures, and sensors and MCU processors coexist in edge hardware modules. The increase in the penetration rate of autonomous driving vehicles below L2 will inevitably bring about the demand for automotive MCUs. At the same time, the entertainment information system, network system and autonomous driving system in smart cars also have a large demand for MCU chips, and each smart car is expected to use more than 300 MCUs overall.