Chinese car-standard chip series (9): Domestic smart cockpit chips emerge

With the development boom of smart electric vehicles, computing power is replacing horsepower and becoming the new standard for measuring the power of automotive products. As the main body carrying computing power, the importance of chips has been elevated to unprecedented heights.

Hello everyone, this is Gasgoo.

In the previous videos, we have introduced in detail the overall situation and competitive landscape of the global and Chinese chip industry chain development. This video will focus on smart cockpit chips and talk about the industrialization process and localization development in this field. Condition.

In the field of smart cockpits, chips are the core driving force for the upgrade of cockpit intelligence . The performance of the chips directly determines the level of cockpit intelligence. But like many other smart car core technologies, the early smart cockpit chip market was mainly dominated by foreign giants. In recent years, with the rapid development of China's new energy intelligent connected vehicles and the complex and ever-changing market environment, it has provided an unprecedented window period for local chip companies to independently break through. Domestic chip manufacturers have gradually begun to emerge.

High-performance chips drive the development of smart cockpits

According to statistics from the Gasgoo Automotive Research Institute, the overall penetration rate of smart cockpits in the domestic market has exceeded 58% in the first half of this year. The definition of "smart cockpit" here is a central control screen of more than 8 inches, voice interaction, Internet of Vehicles and OTA. Cockpit with great core functions.

Image source: Gasgoo

From the perspective of different price ranges, the adoption rate of models priced between RMB 100,000 and RMB 300,000 has increased most significantly.

If we look at the overall configuration rate, the market penetration rate for models above 300,000 yuan is relatively high, generally above 70%. Among them, for models above 600,000 yuan, the smart cockpit penetration rate even exceeds 90%.

Analyzing the reasons behind this, in the past few years, end-market consumption has continued to upgrade, and the car-buying group has become younger. At the same time, out of the need for differentiated competition, car companies are also rushing to use smart cockpits as the core selling point to enhance product strength, and are sparing no effort. Market education has made users' demand and willingness to pay for smart cockpit-related functions continue to increase. After all, when more and more automobile channels enter the supermarket model, the smart cockpit is not only the appearance, but also the most important and core area that is easiest for consumers to experience and perceive.

Looking at the evolution of smart cockpits in the past few years, from large and multi-screen hardware to multi-modal human-computer interaction, the complexity of in-cabin functional applications cannot be separated from the support of chips, especially It is a high-performance cockpit chip.

The data carrying capacity, processing speed and image rendering capabilities of the smart cockpit chip not only directly determine the number, quality and application richness of screen displays in the cockpit, but also affect the smooth operation of the entire cockpit system. It can be said that the performance of the cockpit chip directly determines the upper limit of smart cockpit functions and user experience.

Image source: Gasgoo

In recent years, the significant improvement in the performance of smart cockpit chips is the most direct proof. Currently, the smart cockpit chips that have been mass-produced on the market have high CPU and GPU computing power, exceeding 200 KDMIPS and 3TFLOPS respectively, and the AI ​​computing power has increased to as much as 30 TOPS, such as Qualcomm ’s fourth-generation Snapdragon automotive digital Cockpit platform 8295. For early cockpit chips, the CPU computing power was mostly between 20 and 50KDMIPS, the GPU computing power was generally less than 500GFLOPS, and they had almost no AI computing power.

It is not difficult to predict that as smart cockpit functions are further enriched and application scenarios become more diversified, the performance requirements for chips are expected to further increase.

The cockpit chip landscape has changed, and it’s time for domestic products to break through

The early smart cockpit chip market was mainly dominated by foreign giants. Traditional car-grade chip manufacturers such as Renesas Electronics, NXP Semiconductors ( NXP ), and Texas Instruments (TI) are all important players in this field.

For example, Renesas Electronics launched the R-Car H3/M3 chip for cockpit applications as early as 2015, and has successively achieved large-scale batch shipments on multiple models of Toyota, Honda, Great Wall, Geely and other brands.

NXP’s cockpit products are mainly i.MX6 and i.MX8 series, and its core customers include Ford, Changan, SAIC, etc. In addition, TI's Jacinto6 also occupies a part of the smart cockpit domain control market.

In recent years, the rapid development of smart electric vehicles has made it possible to integrate automobiles with other consumer electronics across terminals . Following this trend, consumer electronics chip manufacturers represented by Samsung, Qualcomm, Intel, AMD, Nvidia, etc. have also begun to Cross-border layout of cockpit chips.

It is particularly worth mentioning that Qualcomm has launched four generations of smart cockpit chips so far, namely the Snapdragon 602A with the first-generation platform 28nm process, the Snapdragon 820A with the second-generation platform 14nm process , and the third-generation platform 7nm process. The Snapdragon SA8155P and the fourth-generation platform 5nm process Snapdragon SA8295P. Among them, the third-generation 8155 chip has almost become the standard in the domestic mid-to-high-end smart electric vehicle market in the past two years. The 8295 is currently focusing on promoting mass production. It has previously taken the lead in the Jiyue 01 and Jikrypton 001 FR models. carry.

According to relevant analysis data, in 2022, NXP, Renesas, Qualcomm, TI, and Intel have a market share of 24.3%, 21.6%, 18.8%, 14.5%, and 9.3% respectively in the domestic passenger car smart cockpit chip market. Among them, Qualcomm’s Growth has been particularly rapid.

While foreign giants are vigorously deploying cockpit chips, a number of local manufacturers have taken advantage of the outbreak of domestic smart electric vehicles and have begun to rush into the market. The main players include Huawei , Xinqing Technology, Xinchi Technology, Jiefa Technology, Quanzhi Technology, Rockchip, Unisoc , etc.

Among them, Huawei's smart cockpit chips are mainly Kirin series, including 710A, 980A, and 990A. The Kirin 990 A, which is currently widely used , has been installed in many models such as the Wenjie M5, Avita 11, BAIC Cube, and BAIC Polar Fox Alpha S. In addition, this chip is also used in some BYD models.

Founded in 2018, Xinqing Technology, as a representative of new local start-ups, successfully mass-produced the first smart cockpit chip "Dragon Eagle One" at the end of 2022. This chip uses an advanced 7nm process and has a CPU computing power of up to 100K DMIPS, a GPU computing power of 900GFLOPS, and an NPU computing power of 8 TOPS. It is worth mentioning that "Dragon Eagle One" is also China's first 7nm car-grade smart cockpit chip. It has been delivered on a large scale on the Lynk & Co 08, and more mass-produced models will be launched in the future.

Xinchi Technology was also established in 2018. In 2020, it officially launched the "Coin of the Cabin" smart cockpit chip X9, which uses a 16nm process and can support "one core and ten screens", covering many mainstream cockpit applications. In April this year, Xinchi Technology released an upgraded version of the smart cockpit chip X9SP. Compared with the previous generation product, the CPU performance was improved by 2 times and the GPU performance was improved by 1.6 times. It also integrated a new NPU with an AI processing power of 8TOPS. Currently, models equipped with X9 series chips from SAIC, Chery, Changan, GAC, BAIC, Dongfeng Nissan and other car companies have been put into mass production.

According to official reports, Jiefa Technology’s first-generation entry-level smart cockpit AC8015 has shipped more than one million units, covering dozens of mass-produced models. On this basis, Jiefa Technology is promoting mass production of the next-generation high-performance smart cockpit domain control chip AC8025. This chip was successfully lit in June this year and was previously planned to be officially mass-produced by the end of the year.

It can be found that although in the smart cockpit chip market, foreign giants still occupy the majority of the share. However, local manufacturers have begun to occupy a place in the smart cockpit chip market by virtue of their continuous new breakthroughs in core technologies, more open cooperation models, and the advantages of providing nearby services and responding to customer needs anytime and anywhere.

The next competition point, cross-domain integration

If smart cockpit chips compete for intra-domain function integration in the first half, then cross-domain integration will be an unavoidable topic in the second half.

Currently, as vehicle functions become more and more complex, the limits of traditional electronic and electrical architecture are constantly being exceeded. On the one hand, the function expansion based on the traditional distributed architecture will bring huge challenges to both cost control and design and assembly due to the continuous increase in the number of ECUs . On the other hand, the scalability of the distributed architecture is ultimately limited. The functional complexity of the entire vehicle has increased significantly, which will inevitably exceed the endurance limit. What's more, the distributed architecture also naturally has disadvantages such as information islands in the car, waste of computing power, and deep coupling of software and hardware. Changes are imperative.