The car chip war begins with the "Intel-branded" BMW

Nokia's history tells us that every change of a generation of mobile communication technology is a bloody business war, with former giants "failed" and newcomers becoming leaders. As the king of the PC era, Intel has provided 8086 chips for IBM's first computer in 1981, and has swept the personal PC and server markets with chips in the next two or three decades. However, missing out on mobile Internet began to shake Intel's "status in the arena." Seeing that another "Nokia-style tragedy" may be staged, 5G and autonomous driving are emerging at this time, and Intel has embarked on a new journey of "continuing greatness." According to the original plan, BMW self-driving cars equipped with Intel chips may be launched in 2021.

30 seconds quick reading:

1. Intel has transformed from a PC-centric business to a data-centric business. Cars are one of the largest data sources in the future.
　　

2. Autonomous driving platform and smart cockpit business are the two major focuses of Intel's automotive layout.
　　

3. Qualcomm, Nvidia, and Intel have all launched autonomous driving computing platforms, and the competition in the automotive field is fierce.

On April 9, at the 2020 Intel China Annual Strategic Online Media Exchange Conference, Intel said that its subsidiary Mobileye's EyeQ chips have sold 54 million units, and high-precision maps cover more than 300 million kilometers of roads around the world. However, in this new journey, former rivals Qualcomm and Nvidia are still with Intel. Intel-BMW-Mobileye formed an alliance, Nvidia-Volkswagen- Bosch joined forces, and Qualcomm launched the Snapdragon Ride autonomous driving platform. Decades of chip industry grievances will continue in the automotive field...

● Transformation to be “data-centric”, cars are the source of data

"Intel had always wanted to pursue the mobile phone business, but actually took a detour." Yang Xu, Intel's global vice president and president of China, once openly explained their failures in the mobile era, and finally found that their core skills were "processing data", "storing data" and "transmitting data". In the future, various smart devices will generate geometrically growing amounts of data, and various connections, storage and calculations will collide in them. Whether these data operations occur on the terminal or in the cloud, Intel can "get a piece of the pie".

Intel is there wherever there is a need for data processing. By 2018, Intel's transformation had paid off, with the proportion of Intel's "data-centric" business equaling that of its "PC-centric" business, each accounting for 50%.

Automobiles happen to be the largest application scenario of "data-centric", and autonomous driving and intelligent networking are the source of automotive big data. Previously, Intel made an optimistic forecast that in 2020, an autonomous vehicle will generate 4TB of data every 8 hours of operation, and the scale of the global vehicle system, data and service market will reach US$70 billion (about RMB 490 billion) in 2030.

At Intel's online media exchange meeting, Yang Xu shared another set of data: from 2025 to 2035, the compound annual growth rate of autonomous vehicle shipments will reach 47.5%; by 2050, the passenger economy spawned by autonomous driving will reach 7 trillion US dollars. This shows that automobiles are a business that Intel has to compete for.

Is Intel still a chip company? Yang Xu said Intel is no longer a simple chip company. Its product lines cover cloud, network, and terminal, and promote future innovations such as quantum computing and neuromorphic computing. All of these business tentacles will have more in-depth interactions with cars.

● Using chips to drive two “computing platforms”

How does Intel outline its idea for cars? Simply put, it is "chip + two computing platforms", one is the smart cockpit platform, and the other is the autonomous driving platform.

A notable trend in the smart cockpit of cars is "one chip, multiple screens interacting". More and more screens are being "stacked" in cars, and various interconnected functions are complementing each other. The functional chips MCU that were originally scattered everywhere are gradually replaced by a more powerful main control chip. The new Mercedes-Benz A-Class released in 2018 integrates independent infotainment systems, HUD and other display screens on a single-chip multi-domain controller.

Mercedes-Benz A-Class cockpit domain controller

The hardware foundation of Intel's smart cockpit business is the specially developed Atom Apollo Lake in-vehicle processor, which is conducive to the chip driving the entire in-vehicle infotainment system. The smart cockpit system jointly developed by Intel, Neusoft and Hongqi has been mass-produced on Hongqi HS7. The automobile products of WEY brand under Great Wall are also equipped with Apollo Lake. The in-vehicle entertainment system launched by Volvo's high-end sub-brand Ploestar also uses Intel's solution, with better application performance and smooth interactive experience in its class.

After 2015, one of the biggest changes in the field of autonomous driving is alliances. Intel-BMW-Mobileye is one level, Nvidia-Volkswagen- Bosch is another level, and General Motors (Cruise)-Honda-Softbank is another. The Intel-BMW-Mobileye alliance established in 2016 has a clear division of labor: Intel is responsible for processing data, Mobileye contributes its proprietary EyeQ5 vision processor, and BMW provides the entire vehicle platform.

According to the initial idea of ​​the three parties, all BMW models in 2018 and beyond will be equipped with Mobileye solutions. The three parties jointly developed an autonomous driving concept car based on BMW I Vision Future Interaction and plan to launch a mass-produced model in 2021.

BMW I Vision Future Interaction

Intel faces an awkward situation if it rushes into autonomous driving. First, it does not have deep enough technical experience and business foundation in the automotive field. Second, it is difficult to change the competitive landscape on its own. Acquiring Mobileye is a very good choice. The latter occupies 70% of the global ADAS (Advanced Driver Assistance) market share and has served car companies such as BMW and Tesla. Its industry status and technical strength are impeccable.

In 2017, Intel spent $15.3 billion (about 107.7 billion yuan at the current exchange rate) to acquire Mobileye. This acquisition, like the $32.2 billion acquisition of ARM by Masayoshi Son the previous year, shook the entire chip industry. You should know that Mobileye's total revenue for the whole year of 2016 was only $358 million.

Is the premium acquisition equivalent to 40 times the revenue too high? Even though Intel is a "chip giant", the $15.3 billion has a significant impact on its finances. Facts have proved that high premium acquisitions are effective. Intel has stepped into the leading camp of autonomous driving in one step, and also received more than 20 vehicle partners of Mobileye around the world. By 2019, Mobileye's annual sales were nearly $1 billion, and the cumulative shipments of EyeQ series chips reached 54 million, covering more than 50 million vehicles worldwide. This acquisition is very cost-effective no matter how you look at it.

● New game of the old three giants

Intel, Qualcomm, and Nvidia, the three old rivals in the chip industry, are still confronting each other in the automotive field, and the three are each other's biggest competitors.

Many people think that Mobileye is just a company that makes ADAS, but in fact it also has accumulated autonomous driving technology. However, Intel has chips, and Mobileye also has chips. How to coordinate the relationship between these "two chips"?

In the CP of Intel and Mobileye, Mobileye is responsible for processing the visual data of the self-driving car, while Intel chips serve as the "brain", responsible for processing the data processed by the former and deciding the next action of the car. Intel + Mobileye actually builds a computing platform of CPU + FPGA + EyeQ + AI, and combines it with communications to form a complete solution from car to cloud.

Qualcomm's approach to entering the automotive field is similar to Intel's, hoping to make up for its shortcomings through mergers and acquisitions. When Qualcomm's plan to acquire NXP for $44 billion failed, it also meant that its pace of autonomous driving was slightly behind its two "rivals." As a last resort, Qualcomm had to do it itself and launched the Snapdragon Ride autonomous driving platform in January this year, which has scalability to meet the needs of L1-L5 autonomous driving. Qualcomm's differentiated competitiveness lies in its successful advantages in 5G and V2X, forming systematic solutions such as data processing, V2X, and 5G to realize its ambition to penetrate the autonomous driving market.

Nvidia's approach is different from that of Qualcomm and Intel. It mainly provides "plug-and-play" solutions for car companies. After all, Nvidia is a "super-powerful" presence in the field of general-purpose chip GPUs, maintaining a market share of more than 70% for many years. With its outstanding graphics processing capabilities, Nvidia has successively launched autonomous driving computing platforms such as Drive PX and Drive Xavier. The latest generation of computing platform Drive AGX Orin has a computing power of 200TOPS (trillions of times per second), which is 7 times that of Xavier.

The three companies have firmly "occupied" the leading position in the autonomous driving industry with their senior industry influence, strong financial resources, and super resource integration capabilities. They are secretly "competing" with each other and trying to "win over" the industry's top auto partners. However, the market situation they all face is that the development of autonomous driving is not as rapid as expected, and some companies have postponed the commercialization time of L3 autonomous driving. In addition, although AI chips are developing rapidly, they are still in their infancy. There is no clear distinction between the advantages and disadvantages of the "rivals". The key to whether Intel can take the lead lies in how many automakers and Tier 1 manufacturers it can get to stand in its camp.