Wanli has taken a small step, what is the domestic large computing chip?
Author:Drive China Time:2022.07.01
In addition to autonomous driving the focus of competition between giants, automobile chips, especially autonomous driving chips, the next battlefield for giants.
In order to accelerate the development of autonomous driving chips, the ideal has changed from Mobileye to cooperation with Nvidia, and Weilai also reached a cooperation with Nvidia.
With the role of autonomous driving chips in the industrial chain, chip manufacturers have become capital darlings. However, front -loading is the only criterion for testing the commercialization capabilities of these "black technology" enterprises. Although domestic autonomous driving chips still rely on foreign manufacturers, fortunately, a number of domestic companies are already paying more attention.
Recently, Black Sesame Smart and Lantu Automobile and Industry partners held a autonomous driving chip exchange meeting. At the end of May, Black Sesame Smart reached a platform -level strategic cooperation with Jiangqi Group.
In addition, Horizon and SAIC's first models have been mass -produced after the cooperation with SAIC; head enterprises represented by Huawei, Xinchi Technology, and Zero Running Automobile have launched independently developed "cores".
This may be just a small step for the Long March, but it is undeniable that local companies are breaking the competition barriers of car -level chips and are expected to usher in a period of high -speed development.
Large computing chips ushered in the "Golden Age"
With the conversion of traditional cars to autonomous vehicles, the proportion of electronics and software in the car also increased, and cars are more like an electronic product. Looking back at the development of the electronics industry, whether it is PC or smartphones, the performance of hardware is a necessary way to go. The rapid iteration of chip performance has also become a support basis for software and applications.
The automobile industry has entered the era of computing power competition. Recently, among the newly released models, including Weilai, Xiaopeng, and ideal, the auxiliary driving has become standard and can achieve higher -level autonomous driving. In the selling point, car companies are constantly emphasizing how high the computing power of new cars is in the release of new cars.
Why do smart cars need large computing chips? This may be the consideration of the car company for the new business model: After the car company performs hardware in the car, it will make money through software upgrades. How much composition does it take for autonomous driving in the future? Although there are no standard answers, car companies can continue to carry out software upgrades through hardware embedding, providing more possibilities for continuous OTA upgrades.
For example, Tesla's FSD (completely autonomous driving optional package), according to Anxin Securities data, Tesla's cumulative cash revenue of Tesla's FSD is expected to reach US $ 1.26 billion, and it is expected to exceed $ 16 billion per year by 2030. This is the huge potential brought by software definition cars. At the same time, this is inseparable from the support of large computing chips.
The epidemic has enlarged the "core" problem of the global automobile industry, and has also brought huge challenges to the smart car supply chain. Facing the supply chain problem that cannot be resolved in the short term, the more difficult computing chips are looking for opportunities in risks. The safety and stability of the car company seeking the security and stability of the supply chain has become a top priority.
Vehicle regulatory chips require "high reliability, high safety, and high stability", and it is necessary to enter the car supply chain after two or three years of strict certification. In strict technical standards and the 5-10-year supply cycle, chip companies and vehicle companies can form a strong binding supply chain relationship.
This is why, although China has achieved good results in the first half of the self -driving track, but the car chip as its core device has fallen into a time. According to statistics, in 2019, the global automotive chip market size was about 35 billion U.S. dollars, while China's independent automotive chip industry was less than 15 billion yuan, accounting for about 4.5%of the world's.
Chip manufacturers such as Nvidia, Qualcomm, MOBILEYE, etc. have held key technologies. Domestic car companies face the problem of being "stuck in the neck" all year round. In addition to BYD, Chinese manufacturers' market share is almost 0.
Although the products provided by foreign chip manufacturers have high computing power and more complete ecology, if they want to maintain such advantages for a long time, the chip will be used in design, not entirely for autonomous driving development. The "lack of cores" accidentally gave domestic chip manufacturers more opportunities to make them appear in a more focused way.
"Ambitionist" on the chip track
As the core of the next smart car electronic architecture, the chip has seized the development of the core chip and seized the core of the smart car industry chain.
In response to such a key market gap, traditional car chip giants, car companies, domestic chip manufacturers, Internet giants, etc. have aimed at the autonomous driving chip market.
However, the product threshold of the car computing chip is extremely high. In the definition and production of the product, it is necessary to strictly connect with the standards of the automotive industry. The computing power of the chip must be turned over, which has led to the lack of mature products in the domestic "independent chip".
With the improvement of the technical level, domestic car AI chip manufacturers have achieved achievements in breaking through large computing bottlenecks. For example, manufacturers such as horizontal lines, black sesame intelligence have entered the harvest period. From the latest autonomous driving chips, it can be found that computing power and algorithms are still the most important upgrade direction.
Take the horizon as an example. Recently, 5 chips are equipped with a journey on the car. Its single chip AI computing power is as high as 128TOPS. It supports 16 -way camera perception calculation, millisecond -level collaboration, supports multi -sensor fusion, prediction and planning control required for autonomous driving, etc. need.
The domain controller platform created based on the journey 5 supports up to 8 journey 5 chips combinations to form a self -driving domain controller, and the maximum computing power can reach 1024TOPS. At present, the shipping volume of the horizon journey chip has exceeded 1 million pieces, and has obtained more than 40 models of more than 40 models. Ecological partners have exceeded 100. As mentioned earlier, the Black Sesame Huashan No. 2 A1000 series chip, which is mentioned in the new model of Jiangqi Group, is currently the most computing power of the domestic computing power and the strongest performance. The required car certification and software supporting facilities.
In addition, Black Sesame Intelligence is developing the A2000 series chips, and it is expected that single -chip computing power exceeds 300TOPS. According to media reports, this large computing chip will be released this year.
It is still not enough to have high computing power. To gain the favor of more customers, the development ecosystem created around the chip must also be improved.
In this regard, Horizon also launched a self -developed intellectual driving solution, including three solutions including Horizon Matrix Mono forward -auxiliary driving, Horizon Matrix Pilot leading driving, and Horizon Halo's intelligent interaction.
Through the optimization of software and hard work, the horizon will continue to expand the leading advantage of algorithms, and continuously accelerate the iterative efficiency of product to achieve the entire scene capabilities of smart cars.
The black sesame intelligence positioned in Tier2 is different from the introduction and absorbing foreign advanced technology in the past. Black Sesame Intelligence also hopes to develop truly pioneering technologies, including hardware structures, hardware equipment, deep learning and algorithm and other soft hardware solution.
For car road collaboration scenarios, black sesame intelligent releases a new generation of vehicle road collaborative road side computing platform FAD EDGE, sink the cloud calculation to the edge layer, and complete most of the calculations on the edge computing node to meet the ultra -low extension of the road collaboration. Demand.
It can be seen that domestic autonomous driving chip manufacturers are rising in different postures. They have achieved breakthroughs from 0 to 1 in the chip and extension fields, but now that domestic replacement may be too early.
Domestic chip manufacturers' break chips
As the chip is becoming more and more important for smart cars, car companies have also begun to pay attention to this core technology, and want to further grasp the core competitiveness and dominance of their own software through self -developed chips.
For car companies, they want chips that need to maintain adaptability, compatibility and openness. When the players on the autonomous driving track swarm in, differentiation has become the key to car companies breakout from it. However, most chip manufacturers on the market currently have great limitations in order to meet the chip's generality, which is not conducive to increasing functions in the future. Once you want to expand the function, you need to take a longer development cycle to redesign.
This also makes car companies more inclined to choose chips related to self -developed autonomous driving functions. However, the chip industry is not suitable for the entry of car companies. Compared with the relatively low profit car brands, the risk tolerance is relatively weak. For chips, high risk is its most obvious industrial characteristics.
Therefore, car companies cannot complete the design of autonomous driving chips with their own strength, which also puts forward higher needs for chip manufacturers.
For chip manufacturers, they want to change from the identity of cooperation to the supplier. In this regard, innovative chip manufacturers such as hierarchical lines and black sesame intelligence adhere to the role of their Tier 2, provide car companies with high -reliable, high -performance autonomous driving chips, and provide corresponding solutions for car company needs on this basis. Essence
This may be a way to maximize commercialization value in the current car supply chain. In the process of cooperation, car companies still have the dominance of smart car design, and also provide the development of domestic innovative chip manufacturers. window.
At present, many car companies have chosen to innovate chip manufacturers, and rely on their own accumulated data and algorithms to make smart cars a faster landing. For example, the new Roewe RX5 is equipped with three horizontal lines 3 chips to form its own AI algorithm platform, becoming the first fuel model with auxiliary driving technology in the industry; Jiangqi Group is based on the Black Sesame Smart Huashan A1000 series chips to build a row. Polo -in -one intelligent driving platform.
From the perspective of the security of the supply chain, car companies embracing domestic chips are also a good choice. With the continuous expansion of demand, especially in the context of "core panic", choosing domestic chips can more effectively prevent the risk of disconnecting supply.
Strengthening technological innovation in auto companies is the only way to promote the production and landing of smart cars, and "chips" are one of them. At present, as high -level autonomous driving is entering the mass production period, it can be seen from the new smart model recently announced that open cooperation and value co -creation have become the first options for smart cars.
Chip shortages are difficult to solve in the short term. It is foreseeable that in the future intelligent connected car competition, the localization of chips may become a general trend. Under this wave of intelligent automobiles, the automotive industry has shifted from speed competition to computing power competition. Whoever can stand out in the competition of this "eight immortals", it remains the test of time.
- END -
Chongqing Public Security Promotion Promoting 10 Measures and Interpretation of Logistics Transportation and Transportation of Automobile Consumption
In order to conscientiously implement a policies and measures of the State Council and the municipal government on the solid and stable economy, the Chongqing Public Security Bureau has launched a ser
If the international oil price exceeds $ 130 per barrel, the price of domestic refined oil products will not be raised for the time being
National Development and Reform Commission: International oil prices over 130 US dollars per barrel of domestic refined oil prices will not be raised for the time beingThe National Development and Ref