news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

Article | Semiconductor Industry Overview

On July 27, at the 2024 NIO Innovation Technology Day, Li Bin, Chairman of NIO, announced that the world's first automotive-grade 5nm smart driving chip Shenji NX9031 has been successfully taped out, and both the chip and the underlying software are independently designed.

According to reports, this chip uses a 32-core CPU architecture, built-in LPDDR5x, 8533Mbps RAM, has 6.5GPixel/s pixel processing capability, and a processing delay of less than 5ms.

Li Bin said that the Shenji NX9031 has more than 50 billion transistors. Whether in terms of comprehensive capabilities or execution efficiency, a self-developed chip can achieve the performance of four industry flagship chips.

This time, Li Bin's speech caused some controversy in the semiconductor industry because he claimed that the Shenji NX9031 is the world's first automotive-grade 5nm smart driving chip. However, before that, 5nm chips that can be used for smart driving systems have been launched, typically NXP's S32N55 processor and Ambarella's CV3 series domain controller.

01Where is the controversial point?

Li Bin said that the Shenji NX9031 is the world's first 5nm smart driving chip, which is worth discussing.

First, let’s take a look at the 5nm automotive chips launched by NXP and Ambarella.

It can be said that NXP is the first company in the industry to announce the use of 5nm process technology to produce automotive chips. As early as June 2020, the company announced this news, and its wafer foundry partner is TSMC.

The S32N55 processor, which uses a 5nm process, integrates 16 Arm Cortex-R52 real-time processor cores and runs at a frequency of 1.2 GHz, which can meet the high computing power requirements of software-defined vehicles. The S32N55's Cortex-R52 core can operate in split or lock-step mode and can support ASIL ISO 26262 functional safety levels. Two pairs of auxiliary lock-step Cortex-M7 cores support system and communication management.

As the central vehicle controller solution of the S32 CoreRide platform, the S32N55 processor integrates advanced network technology, with a time-sensitive network (TSN) 2.5 Gbit/s Ethernet switch interface, a CAN hub for efficient internal routing of 24 CAN FD buses, 4 CAN XL interfaces and a PCI Express Gen 4 interface, enabling efficient communication and collaboration between various systems in the vehicle. In addition, the "core to pin" hardware isolation and virtualization technology of the S32N55 enables its resources to be dynamically partitioned to adapt to the ever-changing vehicle functional requirements.

At the beginning of 2022, Ambarella released the CV3 series chips with a 5nm process, which can support the development of ADAS and L2+ ~ L4 systems. This series of chips is based on the scalable and energy-efficient CVflow architecture and can achieve 500 eTOPS computing power, which is 42 times higher than Ambarella's previous generation automotive-grade CV2 series.

The "e" in eTOPS stands for equivalent. Because CVflow is not equivalent to any GPU, the counting unit of the CV3 chip's AI computing power is different from the TOPS of the commonly used GPU. The "e" is added here to indicate that it can achieve equivalent performance compared to the general chip architecture. Nvidia's Orin chip has a computing power of 254TOPS, and NIO ET7 has achieved 1016 TOPS computing power through cascading 4 Orin chips. If 4 CV3 chips are cascaded, 2000 eTOPS computing power can be achieved.

In February 2023, Ambarella announced that it would use Samsung's 5nm process technology to produce CV3-AD685.

Following NXP and Ambarella, Qualcomm's automotive chips also began to adopt 5nm process. At this time, we have to mention Nvidia and Intel's Mobileye. The smart driving chips of these two companies mostly use 7nm process, while Tesla's HardWare 3 chip uses Samsung's 14nm process. Not long ago, the supply chain reported that Tesla's new HW4.0 chip will switch to TSMC's 4nm/5nm process.

It can be seen that before NIO, NXP, Ambarella, and Qualcomm all taped out 5nm process automotive chips. However, there are some differences in the chip types and applications of these companies. From the above introduction, it can be seen that NXP's S32N55 belongs to the control chip, while Ambarella, Nvidia, Tesla and NIO are computing chips, and Qualcomm's is a smart cockpit chip, which is more control-oriented.

Here we need to briefly introduce the types of automotive chips, which can be divided into computing, control, analog, power, communication, sensor, power, and storage. Among them, computing and control belong to digital chips, which have the highest requirements for process. With the rise of intelligent driving, the requirements for chip computing power are increasing day by day. At this time, the computing power of computing chips has become a very critical indicator, followed by control chips.

In summary, the first companies in the industry to use 5nm process technology to manufacture smart driving chips should be NXP or Ambarella. NIO is the first company in China to use 5nm process technology to manufacture smart driving chips.

So why did NIO develop such a high-end chip by itself? It all started with Nvidia.

At present, the industry's most widely used flagship smart driving chip is Nvidia's Orin-x, which has a single-chip computing power of 508TOPS. In addition, Nvidia has also released a DRIVE Thor chip with a single-chip computing power of 2000TOPS, which will not be mass-produced until 2025.

In 2023, NIO purchased a lot of NVIDIA smart driving chips, accounting for 46% of NVIDIA's shipments, with a total amount of US$300 million. This is a huge expense. For NIO, which has been losing money on R&D, reducing costs and increasing efficiency is a must in the increasingly involuted Chinese auto market. Based on this, it is natural to develop its own smart driving chips. One Shenji NX9031 is equivalent to four NVIDIA Orin Xs, which can save a lot of chip expenses.

02 The value of using 5nm process to manufacture smart driving chips

Traditionally, automotive chips have low requirements for process technology (mostly 20nm or above), but high requirements for chip stability and reliability. In other words, cars must use automotive grade chips.

Automotive-grade chips are those designed and manufactured specifically for automotive applications and meet stringent automotive industry standards. Such chips need to maintain stable and reliable performance in harsh environments such as extreme temperature ranges, high vibration, high voltage, high humidity, EMI, etc., and usually have to pass automotive industry quality standards such as AEC-Q series certification.

Due to the extremely high application requirements for automotive safety and reliability, any chip failure may lead to serious safety accidents. For this reason, compared with consumer-grade or industrial-grade chips, automotive-grade chips have higher quality requirements. Such chips are widely used in engine control, braking systems, safety systems, in-vehicle entertainment information systems, ADAS and other vehicle subsystems.

Although advanced processes (16nm and below) can improve chip performance and reduce power consumption, they also bring some challenges. For example, the smaller the process node, the higher the chip production cost. In addition, chips with small feature sizes require more sophisticated production equipment and technology, which also increases costs.

Therefore, automotive chip manufacturers and automakers need to find a balance between chip performance, cost and reliability. They need to choose the appropriate process technology according to the purpose, performance requirements and cost budget of the vehicle. For some high-end models, manufacturers may use more advanced processes to improve vehicle performance. For some economical models, manufacturers will choose more economical processes to reduce production costs.

In general, the mainstream process of automotive chips is between 40nm and 16nm.

However, with the popularization of intelligent driving, the manufacturing framework of traditional automotive chips has been broken as computing power begins to dominate automotive applications.

In fact, practitioners know that piling up computing power will inevitably lead to waste. However, compared with invisible software algorithms, the actual computing power indicators can be easily judged. Users' pursuit of hardware capabilities is fully reflected in mobile electronic products, and now the same situation continues to smart cars.

Based on this, various marketing rhetoric has emerged in the market. For example, some media have likened the computing power level of chips to "house utilization rate", using the difference between dense computing power and sparse computing power to calculate completely different computing power conclusions. Nowadays, volume computing power has become a hurdle that car manufacturers and related chip companies cannot overcome. More and more newly launched smart driving chips have proved that increasing computing power is the most effective way to improve market evaluation.

At present, the computing power of many new SUVs priced above 300,000 has exceeded 100 TOPS, and some brands of cars have even exceeded 1,000 TOPS. Even if there is a lot of redundancy, it seems that no one will refuse higher computing power.

As chip computing power easily exceeds 500 TOPS or even 1000 TOPS, other indicators of the chip are bound to attract public attention, such as process technology. Although there is no ultimate pursuit of process technology for automotive-grade chips, in the field of smart driving and smart cockpits, chip processes have obviously begun to advance towards 5nm or even smaller process nodes. Compared with 7nm, TSMC's 5nm process has a 20% increase in processing speed and a 40% decrease in power consumption. Migrating to 5nm will help automakers differentiate their cars through enhanced functionality, simplify the increasingly complex architectural challenges of cars, and easily deploy powerful computing systems.

Therefore, the value of 5nm process for automotive chips is highlighted.

Under such market demand, TSMC has also started to play "hunger marketing". In July 2023, Paul de Bot, general manager of TSMC Europe, said at the "27th Automotive Electronics Conference" held in Germany that for a long time, the automotive industry has been regarded as a technological laggard and only focused on mature processes, but in fact, some automotive chip suppliers have started using 5nm process technology since 2022, which is only two years away from the official mass production of 5nm. Due to Samsung's 5nm process yield, TSMC has become almost the only wafer foundry that can mass-produce 5nm process chips. As a result, the company's production capacity is in short supply. TSMC said that it is impossible to reserve idle production capacity for the automotive industry, and automotive chips need to accelerate the transition to advanced processes. Paul de Bot believes that it is absolutely necessary for automakers to plan and control the number of orders in a forward-looking manner, and the transition of some automotive chips from the original mature process nodes to advanced processes is also an important means to ensure supply.

Compared with consumer electronics and server applications, the automotive chip foundry market accounts for a smaller proportion (less than 10%), but the unit price is higher, which is an extremely profitable market. From TSMC's perspective, during the COVID-19 pandemic, TSMC's annual automotive chip business has grown by about 40%, and the leading foundry hopes to retain and expand this customer base in the future, especially for advanced processes.

03 Smart cockpit chips also require 5nm

The above-mentioned ones are all smart driving chips, which are more computing-oriented. Another major category is smart cockpit chips, which are more control-oriented and have an increasingly urgent demand for advanced processes.

The smart cockpit has multiple functional blocks, including high-definition display, instrumentation, active safety alarm, real-time navigation, online infotainment, emergency rescue, Internet of Vehicles, and human-computer interaction system (speech recognition, gesture recognition), etc. Its main function is to improve the driver and passenger experience by changing the human-computer interaction mode. At this time, the importance of artificial intelligence (AI) technology is highlighted, and the performance requirements for related chips are also improved.

The typical representative of smart cockpit chips is the Qualcomm Snapdragon 8155 that we often hear about, and the updated 8295 chip.

At the end of 2021, Qualcomm released the Snapdragon 8295 using a 5nm process. Compared with the 8TOPS computing power of the previous generation 8155 (7nm process), the computing power of the 8255 reaches 30TOPS, and the 3D rendering capability has increased by 3 times. It has added integrated electronic rearview mirrors, machine learning, passenger monitoring and information security functions. One chip can drive 11 screens.

In addition to the cockpit chip, the core SoC of Qualcomm's Snapdragon Ride smart driving platform is also built on a 5nm process and integrates core components such as high-performance CPU, GPU and AI engine, with a maximum computing power of 700TOPS. However, compared with several other smart driving chip manufacturers (Nvidia, Intel's Mobileye, Tesla), Qualcomm's smart driving chip presence is relatively weak.

In addition to Qualcomm and other industry giants, Chinese local SoC companies are also moving towards advanced process smart cockpit chips, and have now progressed to 7nm. If there were not so many international trade restrictions, they would definitely adopt the 5nm process. At present, Horizon Robotics, Heima Intelligent, CoreChi Technology, and CoreJing Technology have all released related products. Among them, CoreQing Technology's self-developed "Dragon Eagle No. 1" is the first domestic automotive-grade 7nm chip and has been installed on the car. Heima Intelligent launched the first self-developed 7nm chip Wudang C1200. Horizon Robotics' Journey 6 series chips also use 7nm process technology. The flagship chip has a single computing power of 560TOPS and will be mass-produced in 2024. When Journey 7 or Journey 8 is mass-produced, it is expected to further improve the process technology.

04 Automotive chips are moving towards 3nm process

As the intelligence level of automobiles increases, related chips are also moving towards more advanced process technology.

NVIDIA's latest smart driving chip DRIVE Thor has a computing power of 2000 TOPS and uses the Hopper architecture. It will adopt a 4nm process and go into production in 2025. NVIDIA said that Chinese auto brands such as BYD, Aion, Xpeng, Ideal, and Zeekr will use DRIVE Thor.

Tesla is more radical and is ready to launch a 3nm process chip foundry plan. It will continue to enhance the speed and power consumption performance based on TSMC's N3E and plans to start production in 2024, but whether it can obtain production capacity remains to be seen.

After seeing Qualcomm's success in the field of smart cockpit applications, MediaTek could no longer sit still and began to enter the automotive chip market, especially smart cockpits. It plans to launch the "Dimensity Car Platform", which will be built using a 3nm process.