As semiconductor giants compete in autonomous driving, how much chance does Intel, which acquired Mobileye, have?

Intel, which stuck to the X86 chip architecture, missed the entire mobile Internet era. Now, it plans to make a comeback in 5G and autonomous driving. However, Intel, which is less than half of its peak market value of 500 billion US dollars, is like the Chinese football team - they don't have much time left.

Intel's comprehensive layout in the automotive field

The automotive industry is undergoing profound changes, and it is certain that cars will become the next smart platform. Compared with smartphones, cars will require more chips, are more difficult to manufacture, have more functions, and involve a more complex business ecosystem. There is no doubt that the automotive industry will be the focus of competition among global semiconductor giants in the future.

Autonomous driving, car networking, and smart cockpits around cars will be development opportunities that Intel and its competitors cannot miss. Intel, which has performed poorly in the era of mobile intelligence, is closely planning to recreate the glory of the PC era.

In order to keep up with the times and seize the initiative in autonomous driving, Intel has been buying and buying non-stop in recent years.

Of course, Intel’s layout in the automotive field is not just acquisitions, it has also made a lot of efforts in independent development.

In October 2016, Intel and Neusoft jointly developed a software-defined cockpit (SDC) platform solution - C-AIfus.

At the end of 2016, Intel established the Autonomous Driving Division. After acquiring Mobileye in 2017, Intel merged the Autonomous Driving Division with it.

In 2017, Intel launched the Intel Go platform, which includes an autonomous driving in-vehicle development platform, an intelligent driving 5G in-vehicle communication platform, and an intelligent driving software development kit.

In terms of 5G Internet of Vehicles, Intel released a 5G modem in early 2017 and launched the 5G baseband chip XMM 8060 at the end of 2017.

In addition, Intel has assembled a test fleet of 100 vehicles.

In July this year, Intel launched the OpenVINO toolkit, which is mainly used to accelerate the development of high-performance computer vision and deep learning vision applications.

In addition, Intel also attaches great importance to China, the world's largest automobile market, and strives to strengthen cooperation with local Chinese companies.

In terms of smart cockpits, Intel has established cooperation with automakers such as FAW, BAIC, and Great Wall. At the CES show earlier this year, Intel announced that it would cooperate with SAIC and NavInfo to develop autonomous driving technology. At this year's Baidu AI Developer Conference, Intel disclosed its in-depth cooperation with Baidu in the fields of autonomous driving and artificial intelligence. In addition, Intel is currently testing 5G-related technologies with Huawei.

Looking back at Intel's actions in recent years, it can be seen that Intel's layout in the automotive field is very comprehensive - from sensors to high-precision maps, from vehicle networks to test fleets, from software development to core chips, smart cockpits, autonomous driving, and vehicle networks, Intel has not missed any of them - except for not participating in car manufacturing itself, it basically involves all the focus of the automotive industry's transformation.

Intel's only trump card is Mobileye

Although it has made extensive layouts in the automotive field, compared with semiconductor giants of similar size such as Nvidia and Qualcomm, Intel's biggest competitive advantage is the large-scale acquisition of Mobileye.

Mobileye is currently the world's largest ADAS technology provider, with more than 27 million cars using Mobileye's products. Currently, most advanced driver assistance systems in the front-end field use Mobileye's solutions.

Intel paid a high price to acquire Mobileye because of its advantages in the ADAS field. After the acquisition, Intel merged its original autonomous driving division with Mobileye. Now, Intel's autonomous driving route is actually the same as Mobileye's previous route.

Intel's approach is to focus on visual technology. Intel's choice of this approach was already evident before the acquisition of Mobileye. Previously, Itseez and Movidius, which were acquired by Intel, were both related to computer vision processing.

Mobileye has risen to prominence through ADAS, and already has a good market and customer base in this field. Most car companies currently do not have the ability to deploy high-level (L4, L5) autonomous driving. Intel's focus on ADAS can both play to its strengths and avoid its weaknesses, and is in line with the strategy of traditional car companies to gradually promote autonomous driving.

Intel's re-vision solution relies on low-cost cameras to perceive the car's driving environment. The advantage of the re-vision solution is not only its low cost, but also its help in promoting the crowdsourcing model of high-precision maps, which is an indispensable condition for realizing autonomous driving.

Intel has reached cooperation with BMW, Volkswagen, Nissan, NavInfo and others. This year, at least 2 million vehicles will be equipped with EyeQ4 chips and officially start collecting road data.

The huge number of users gives Intel an advantage in obtaining map data. Currently, Intel has used its data advantage to form a big data alliance with car companies. On the one hand, this consolidates the cooperative relationship between Intel and car companies and increases the difficulty for latecomers to break through; on the other hand, Intel can use the data shared by car companies to improve its own technology and move towards a higher level of autonomous driving technology.

Waymo, which is at the forefront of autonomous driving, has adopted the LiDAR solution. LiDAR is indeed superior to cameras in performance and can provide more comprehensive information than cameras. This reduces the requirements for artificial intelligence technology and speeds up development. However, the high cost of LiDAR has limited the promotion of this solution.

Although LiDAR is currently expensive, with the advancement of technology and the expansion of application scale, there is still a lot of room for price reduction. It is still unknown which solution will have the last laugh.

Time is running out for Intel

With its acquisition of Mobileye, Intel is currently leading the ADAS market, but in the field of high-level autonomous driving, Nvidia has taken the lead with its GPU.

Intel's CPU is its strength, but its efficiency in deep learning is not as good as Nvidia's GPU, which is its strength. Deep learning is critical for autonomous driving. Through deep learning, the autonomous driving system can acquire image recognition capabilities and can identify and distinguish objects such as paths, pedestrians, traffic lights, obstacles, etc., thereby providing a basis for the car to make judgments and decisions.

NVIDIA was the first to apply GPU clusters to deep learning data processing and is in a leading position in the field of deep learning. NVIDIA's GPU has accelerated the development of AI technology and made NVIDIA's GPU the first choice for AI computing chips. NVIDIA has also won heavyweight customers such as Tesla, Audi, Volvo, Mercedes-Benz, Toyota, Honda, Volkswagen, and Fiat.

Tesla initially used Mobileye's driver assistance chip EyeQ3. However, Tesla was not satisfied with ADAS and wanted to go further and achieve a higher level of autonomous driving. Mobileye was more than capable of handling ADAS, but the computing power of its EyeQ3 was not enough to handle the computing tasks of advanced autonomous driving. Therefore, Tesla abandoned Mobileye and chose Nvidia.

Intel faces great challenges in evolving from ADAS to true autonomous driving.

Although there is only one level increase from L2 to L3, there is actually a qualitative change. L2 is still to assist humans in driving, and the driver always has the dominant power. However, L3 requires the vehicle to take over the entire driving task under certain conditions, which puts very high demands on the vehicle's capabilities, and the requirements for the computing power of the on-board chip are also rising sharply. There are many solutions on the market that support L2 autonomous driving, but there are few solutions for L3. If car companies want to further develop products above L4 level, Nvidia is almost the only choice.

Mobileye is very proud of the ADAS field, but its vision-based solution has no advantages in strategy reasoning and computing power. This limits Intel's development towards higher levels of autonomous driving. To make up for this disadvantage, Intel acquired Altera, which already has FPGA AI chip solutions, in 2015.

FPGA has the advantages of low power consumption, high real-time performance, and flexible programming, making it very suitable for computing tasks related to autonomous driving. Intel's previously launched autonomous driving computing platform Intel Go has already applied FPGA chips.

Intel's layout in the FPGA field has given it the opportunity to compete with Nvidia. However, whether FPGA can eventually compete with GPU or even surpass it in the field of autonomous driving remains unknown.

Semiconductor companies are not the only players in the field of autonomous driving chips. Tesla's autonomous driving part uses Nvidia's chips, but the news of cooperating with AMD to develop its own autonomous driving chips shows that Tesla is not willing to be manipulated by others.

If self-developed chips are successful, Tesla will have a stronger control over core hardware and is expected to form unique advantages in hardware acceleration and other aspects. More importantly, after large-scale production, Tesla can use this to reduce costs. However, not every car company has the courage and strength to do so.

The rapid advancement of 5G technology has provided conditions for the realization of the Internet of Vehicles. Although Intel also has a layout in 5G Internet of Vehicles, Qualcomm obviously has more advantages.

Qualcomm has been very successful in the era of mobile intelligence, but the ARM architecture it uses is based on low power consumption rather than high computing power as its core competitiveness. Although cars still have to consider energy consumption, compared with mobile phones, the energy consumption requirements are not harsh. On the contrary, autonomous driving and car networking require high computing power for cars. Qualcomm's chips are at a disadvantage in terms of computing power.

Intel has gained an advantage in the ADAS field through the acquisition of Mobileye, and the failure of the acquisition of NXP is undoubtedly a major blow to Qualcomm. However, Qualcomm's huge advantage in the communications field is expected to help it make a comeback in the field of connected vehicles.

High-level autonomous driving is actually inseparable from the Internet of Vehicles. The Internet of Vehicles can help vehicles break through sensor blind spots, achieve communication between vehicles and other vehicles, pedestrians, and traffic facilities, and reduce the vehicle's requirements for sensors, artificial intelligence, and chip computing power. Although Qualcomm is at a disadvantage in vehicle terminal computing, the application of the Internet of Vehicles can give Qualcomm room for development in the future.

In fact, there is still a long way to go before true autonomous driving can be realized. In the next few years, ADAS will still be the mainstream autonomous driving in the market. In the next few years, Intel will have an easier time in the field of autonomous driving. However, if Intel cannot catch up in the field of chip computing power and car networking, it is likely to repeat the mistakes of the mobile intelligence era in the more distant future.

As a winner of Toutiao's Qingyun Plan and Baijiahao's Bai+ Plan, the 2019 Baidu Digital Author of the Year, the Baijiahao's Most Popular Author in the Technology Field, the 2019 Sogou Technology and Culture Author, and the 2021 Baijiahao Quarterly Influential Creator, he has won many awards, including the 2013 Sohu Best Industry Media Person, the 2015 China New Media Entrepreneurship Competition Beijing Third Place, the 2015 Guangmang Experience Award, the 2015 China New Media Entrepreneurship Competition Finals Third Place, and the 2018 Baidu Dynamic Annual Powerful Celebrity.