On September 6, Baidu officially launched three clouds of group cloud, network connection cloud and supply chain collaboration cloud for the automotive industry. Not only Baidu, but in June last year, ByteDance was exposed to form an auto cloud team; in November last year, Tencent Cloud officially announced the launch of the auto cloud, and in June this year, it announced a strategic upgrade and proposed "auto cloud integration"; in May this year, HUAWEI CLOUD announced that it won the first place in China's automotive cloud market; at the end of August, Alibaba Cloud and Xpeng Motors jointly completed the "swinging" driving intelligent computing center...

So far, leading companies including Alibaba, Tencent, Huawei, and ByteDance have all entered the "automobile cloud" battlefield. It is worth pondering why the car cloud has become a battleground for electric vehicles? What role will the car cloud play in the second half of electric vehicles?

"Crazy" car cloud, "Crazy" tech giant

With the development of electric vehicles, the demand for cloud services is increasing exponentially, and accessing cloud services has become an inevitable choice for new electric vehicles.

According to the "2021 China Automotive Cloud Market Tracking Report", the overall market size of the automotive cloud industry in 2021 will reach 33.52 billion yuan, and in 2026, the size of China's automotive cloud market may exceed 80 billion yuan. Cars are transforming from mechanical tools to intelligent mobile devices, and the core competitiveness of the automobile industry is also changing, from mechanical mobility capabilities to software service capabilities.

From "machine-defined car" to "software-defined car", Zhang Jian, secretary general of the China Software Industry Association's Intelligent and Connected Vehicle Industry Branch, gave figures that can be quantified and compared. He said that the amount of code for a smart car produced in 2025 is expected to be It will reach 700 million rows, a 2.3-fold increase compared to 2022.

What drives this line of code to run is computing power, and computing power is naturally inseparable from the help of cloud computing. With the acceleration of automobile intelligence, the automobile cloud will play the role of the digital base and infrastructure of the automobile industry, improve the intelligent and ecological connection capabilities of automobiles, and further improve the user experience of intelligent automobiles.

Specifically, car intelligence, one is the wide application of intelligent cockpit, and the other is the rapid development of automatic driving The cloud computing needs of smart cockpits are relatively easier to meet, and the key lies in autonomous driving.

The development process of autonomous driving is also a process of data explosion in the automotive industry.

To achieve autonomous driving, it is inseparable from the acquisition of road condition information, which relies on high-precision sensors such as cameras, millimeter-wave radars, and lidars. OEMs often adopt multi-sensor fusion solutions.

As the technological strength of bright automotive products, sensors have become an important dimension of competition among car companies. At present, the number of sensors in mainstream models has exceeded 30. Even sensors have formed a chain of contempt among car companies. What? If you carry two lidars, then I will install three, and then a car brand speaks: If you have less than four, please don’t speak.

Smart cars may be equipped with more and more sensors in the future. There are reasons behind the development of autonomous driving technology, as well as product marketing considerations. But no matter what the purpose is, the final result must be that the data generated by the car is like an explosion, which will gradually expand and eventually exceed the ability and control scope of the car company.

For car companies, a lot of data is a good thing, but it’s also a headache, especially for traditional car companies, which naturally brings the demand for car cloud.

So, why did the car cloud choose such a time node to start erupting today? The reason may be that self-driving technology has reached a critical period from assisted driving to completely unmanned driving.

Note that it is possible not to transition, but to actually cross. At the 2022 World Artificial Intelligence Conference (WAIC) on September 1, Li Yanhong judged that the first commercial use after L2 is likely to be L4, not L3. Indeed, including traditional OEMs and new car-making forces, many car companies have announced to abandon L3 and directly leap to L4.

The leap in autonomous driving technology will also bring about a leap in computing power.

Lu Wenliang, a distinguished researcher in the automotive industry at the Industrial Science and Technology Innovation Center of the Chinese Academy of Sciences, once said, "At present, 80% of the technical problems of autonomous driving have been solved, and the remaining 20% of the long tail problems are often referred to as corner cases.

The key difference between high-level autonomous driving technology and low-level autonomous driving technology is that it can solve a large number of special and complex long-tail scenarios, which is inseparable from the help of the car cloud.

For example, the simulation test of autonomous driving scenarios, that is, to abstract the actual traffic system at a certain level with the help of computer virtual technology.

The industry's general view is that an autonomous driving system needs at least 10 billion miles (about 16.1 billion kilometers) of test drive data to ensure the safety of vehicles on the road. It is difficult to do this with test vehicles alone, and simulation testing becomes automatic An important part of driving research and development. According to the Tianyancha professional version APP, there are as many as 18,721 pieces of information about "automobile simulation" related companies.

Currently, almost every autonomous driving company or company involved in the corresponding business is doing simulation-related work. The most famous is Tesla simulation, and its shadow mode and fleet learning have always been talked about. In the future, the mainstream solution in the industry will be to use cloud resources for parallel computing and large-scale cloud simulation to improve test efficiency.

In short, if the computing power is insufficient, it may cause the development of electric vehicles to stagnate in the fully driverless stage and enter an infinite loop.

The "Triangle Model" of the Automotive Cloud Industry

For tech giants such as Alibaba, Baidu, and Huawei, developing an automotive cloud is no longer an optional question, but a must-answer question.

First of all, judging from the financial report data of these big manufacturers, due to the continuous strengthening of external supervision and the peak of Internet traffic dividends, the main business is generally faced with the dilemma of slowing growth, and the cloud computing business has become the driving force for the growth of corporate performance.

According to Ali's Q1 financial report for fiscal year 2023, Alibaba Cloud has become Ali's second largest source of revenue, and its market share cannot be ignored in China's cloud computing market; Baidu's 2022 Q1 financial report shows that the growth rate of intelligent cloud business is 45%, a high Compared with the industry average; Tencent's 2022 Q2 financial report shows that the To B business is the only business segment that is growing, and has become Tencent's second largest revenue pillar after value-added services.

The development paths of cloud computing and autonomous driving technology are very similar. The common points of both parties are: first, it burns money, and second, it is difficult to make profits in the early stage.

Compared with building a car, the level of money-burning of cloud computing is also not much less. Moreover, if cloud computing wants to maintain its leading edge, it not only requires high-intensity investment, but also needs to maintain it for a long time.

For example, in 2020, Alibaba Cloud proposed to invest 200 billion yuan in the next three years. Amazon said at this year’s Q1 earnings conference call that it expects infrastructure to account for half of its capital investment in 2022. As of the 12 months of Q1, Amazon’s capital investment totaled $61 billion, of which about 40% was invested in the foundation of AWS. facility area.

In the early stage of the development of the domestic cloud computing industry, it also experienced significant price reductions and "one-dollar bids". For example, Alibaba Cloud, the largest in China, had a dozen rounds of price cuts in 2016. In 2016 and 2017, there were even events such as China Mobile winning the bid for the Wenzhou government affairs cloud platform project with one yuan, and Tencent Cloud winning the Xiamen government affairs cloud project with one cent.

The high cost of investment coupled with the fierce competition in the early stage has made cloud computing manufacturers eager to find profit points to recover their investment. Fortunately, compared with car building, domestic cloud computing manufacturers have passed the stage of industry pioneering and have reached the threshold of profitability. Alibaba Cloud, for example, achieved annual profitability for the first time in fiscal 2022.

Cloud computing has become a key development goal of technology companies, and there are obvious cloud computing needs and computing power gaps in the development of automotive intelligence.

According to the development status of the automobile cloud, Tan Qing said that AI has built an industry triangle model for technology companies to enter the automobile cloud, and can judge the advantages and disadvantages of technology companies in the automobile cloud industry from these three dimensions.

The first and foremost is the dimension of security and compliance.

A large number of road tests of autonomous driving are related to domestic urban street data. After the mature implementation in the future, it will also bring a large amount of consumer travel data. The importance of data security cannot be underestimated. Cloud computing, as the underlying infrastructure, faces higher compliance concerns.

Therefore, when car companies choose the car cloud, they must consider in advance how to ensure the safe and compliant use of data and how not to cross the red line. In this regard, Amazon Cloud and Microsoft Cloud are obviously unable to compete with domestic cloud computing companies, while e-Surfing Cloud, Mobile Cloud, China Unicom Cloud and other operator clouds with state-owned assets are obviously more advantageous.

At this stage, under the background of the saturation of the traditional business market, the traditional businesses (mobile and broadband business) of the three major operators are also facing growth problems, and they are also actively developing cloud computing. Although its cloud computing business is still small, it is growing rapidly.

Secondly, it is the understanding of cloud computing manufacturers for the automotive industry. This includes not only the understanding of the digitalization of the traditional automobile manufacturing industry, but also the layout of cloud services for the intelligent development of automobiles. From this perspective, the dominant companies are Ali and Baidu.

On the R&D and manufacturing side, Alibaba Cloud provides CAE simulation and industrial big data solutions, and uses hybrid cloud to uniformly dispatch computing resources to enable efficient R&D and production of vehicle models. According to Alibaba's official disclosure, Alibaba Cloud has served several industrial segments such as photovoltaics, rubber, new energy, and steel. Therefore, it also has enough experience to help the digital transformation of the automotive industry to reduce costs and increase efficiency.

Baidu Cloud OS had previously suspended operations, and then Baidu released the "Apollo" platform based on Baidu Smart Cloud, which began to elevate cloud computing to a strategic level. It can be said that Baidu's cloud computing business is more like a resurrection to build smart cars.

The last evaluation dimension is whether the auto cloud enterprise is involved in car manufacturing. The "car-making" here refers to all other business related to car-making except for the car cloud.

This is still about data security at the core of the enterprise, but mainly in the field of business competition. If an auto cloud company stops building cars at the same time, it is equivalent to being both a referee and an athlete. The auto company may worry about the loss of its own technology, so it is difficult to trust this auto cloud company.

In this regard, it is clear that Tencent and Ali have more advantages. Ma Huateng once said, "Tencent wants to play the role of a small assistant in 2B. In the automotive industry, Tencent also uses this role to cut in."

Based on Tencent's understanding of the needs of car manufacturers, Tencent Cloud has launched the TFS (Tencent Service Framework), a micro-service framework for car companies, which combines the characteristics and models of car factories' related businesses. Cycle management and monitoring.

Finding the "cornerstone" of the second half of electric vehicles

According to Tan Qingshuo AI, the transformation of the automobile cloud to the automobile industry is essentially divided into two parts: the implementation of digital industrialization and the digital transformation of the industry.

Take the three clouds released by Baidu, namely the Group Cloud, the Networking Cloud and the Supply Chain Collaboration Cloud, for example.

According to reports, the Group Cloud is oriented to the construction of the digital base of the car company group itself, covering the whole process of vehicle research and development, production, delivery, marketing, etc.; the supply chain collaboration cloud is to open up the car companies and upstream and downstream supporting enterprises to ensure the integrity of the industry chain. Safe and stable; Connected Cloud provides car companies with intelligent cloud solutions at the level of autonomous driving and smart cockpits, helping car companies perform data collection, vehicle status monitoring and remote upgrades, and improve the car’s intelligent interaction and cloud collaboration capabilities.

It can be seen that both the group cloud and the supply chain collaboration cloud are the digital transformation of the automobile industry. By applying digital technology and data resources, the automobile industry can increase output and improve efficiency, thereby integrating digital technology with the real economy; The cloud fully serves the intelligent implementation of automobiles, which is equivalent to the implementation of digital industrialization in the automotive field, providing cloud solutions that rely entirely on digital technology and data elements.

According to the current trend, the implementation of digital industrialization to drive the digital transformation of the industry may become the most effective path for the digital transformation of the automotive industry.

So, how important will the role of auto cloud manufacturers in the development of the auto industry be? Tan Qing said that AI believes that it is comparable to the position of lithium ore manufacturers in the electric vehicle industry.

In the first half of the development of electric vehicles, lithium batteries have become an important cornerstone of vehicle electrification.

In recent years, the development of electric vehicles has led to a surge in the demand for lithium batteries. The result of the shortage of supply is that the market sentiment continues to heat up, resulting in the rising price of upstream lithium resources. Even the lithium battery company CATL can only pass the pressure on to downstream car companies.

So much so that some car companies have shouted loudly, angering that the battery price is too high, saying that they are working for a lithium battery manufacturer. It can be seen that in the electric vehicle industry chain, most of the profits flow to the upstream lithium ore manufacturers and corresponding processors.

Next, we can think about a question, who will be the cornerstone that supports the intelligentization of electric vehicles in the second half? Which part of the automotive industry chain will it appear in? The answer may be automotive cloud manufacturers.

Vehicle intelligence depends on three elements: data, algorithms, and computing power.

As mentioned above, the development of the auto industry in the future will be a process of data explosion. Therefore, there will be no shortage of data in the future of automobile intelligence. The key lies in whether car companies can effectively use data; as for algorithms, from the current domestic automobile industry From the perspective of the current situation of the development of enterprise intelligence, at least in the autopilot technology, the main event of automobile intelligence, the algorithms are actually similar, and the gap is small.

Therefore, computing power may be the key to the competition in the second half of electric vehicles, and computing power mainly comes from the cloud. According to Tan Qingshuo AI, the development of electric vehicles is actually two resources, one is lithium and the other is computing power.

The essence of the car cloud is equivalent to a block of computing power mines, but compared with lithium mines, cloud resources can be artificially generated. The importance of the cloud is self-evident, and it is no surprise that the automotive cloud is at war.

In the second half of electric vehicles, the competition logic has undergone dramatic changes, from the competition in the traditional automobile era around power, control, and space to the competition in the first half of electric vehicles around batteries and lithium resources. , intelligent chips, automotive data and other intelligent elements put forward higher requirements. Next, around computing power, algorithms, and data, a new industrial chain ecology of electric vehicles will be formed.