As long as the global technology giants related to "chips" have almost invested a lot of money and energy in the field of automatic driving, GPU giant NVIDIA is naturally the first to bear the brunt.
In fact, NVIDIA's depth in the field of autonomous driving may be much higher than many people think, autonomous driving platforms, development software and hardware, simulation engines, simulation software, AI chips, AI servers and so on. It can even be said that as long as the car companies are willing, they can only provide a shell, and Nvidia is responsible for the development of the entire autonomous driving system.
Since getting involved in autonomous driving, the list of car companies cooperating with NVIDIA has been continuously expanded: Mercedes-Benz, Volvo, GM, Weilai, Xiaopeng, Ideal, Weima, Zhiji, SAIC... Now, there is a car company that Chinese familiar with this list - BYD.
Recently, the chip giant announced that BYD will adopt its drive Hyperion 8 platform and drive Orin chip.
The Hyperion platform is an autonomous driving hardware architecture designed by NVIDIA, including 12 cameras, 9 millimeter wave radar, 1 lidar, and 3 in-car cameras, which can give cars autonomous driving, automatic parking and other capabilities. The DRIVE Orin chip is one of the most powerful vehicle-grade chips at present, with a computing power of 254 TOPS, so it can process data collected by multiple sensors at the same time.
ASD, as the other party of cooperation, has grown into a force that cannot be ignored in the global electric vehicle market. According to the statistics of The CleanTechnica website, in the 2021 global new energy vehicle sales list, BYD ranks second, second only to Tesla. However, although BYD has a reputation in the fields of car manufacturing and batteries, its automatic driving ability has always been a major shortcoming. Now, after achieving a relatively stable market position, BYD has finally begun to sprint towards automatic driving.
In the past two years, BYD has also made frequent moves in China. In February last year, BYD invested in the domestic self-driving chip startup Horizon, and the two sides will also cooperate in the research and development of chips, algorithms and tool chains; in December last year, BYD and the self-driving startup Momenta established a joint venture to jointly develop autonomous driving technology; in February this year, BYD made a strategic investment in radar manufacturer Sagitar Juchuang.
A more heavyweight partner is Baidu. At the end of February, it was reported that BYD and Baidu had reached a cooperation in the field of autonomous driving, and would use the an integrated ANP intelligent driving product and human-machine co-driving map provided by the latter. In fact, as early as 2015, before Baidu officially established the autonomous driving division, BYD had reached a cooperation agreement with them, so today's cooperation is expected for both parties with good relations and able to take what they need.
However, although the relationship is good, the cooperation effect and prospects of BYD and Baidu have yet to be verified. In 2018, Baidu announced that it will provide BYD with a complete solution for L3-level intelligent driving, and three years have passed, Baidu's L4-level automatic driving mileage test has exceeded 16 million kilometers, but BYD's automatic driving ability has not seen much improvement, and its 3 millimeter wave radar + monocular camera hardware configuration is also somewhat backward compared to the new car-making forces.
Therefore, relying only on Baidu and several domestic potential stocks, I am afraid it is difficult to meet the needs of BYD to quickly make up for shortcomings, so they also need partners with more mature products and stronger technical capabilities, such as NVIDIA.
NVIDIA's attraction to car companies, in addition to its strong technical strength, also comes from the openness and upgradability of its platform. At present, although NVIDIA has the ability to provide full-stack automatic driving systems, it does not force "bundle sales", and car companies can either purchase part of the hardware or directly let NVIDIA design a complete set of solutions for it. Today's large car companies are particularly sensitive to data ownership, so this more flexible and open approach to collaboration is also acceptable to them.
At the same time, the DRIVE Hyperion platform can also achieve cross-generational compatibility of chips. Previously, Nvidia released a more powerful DRIVE Atlan chip in addition to the DRIVE Orin chip chip, and is expected to be mass-produced in 2025. For automakers, if the heavily purchased platform and chip fall into a comprehensive backwardness after 2-3 years, this is undoubtedly unacceptable, and this feature of the DRIVE Hyperion platform can lengthen the life cycle and reduce the cost of replacement, which will undoubtedly make the car company eat a reassuring pill.
But even so, the sensitive question cannot be avoided: Isn't BYD worried about giving its "soul" to someone else? BYD will certainly not ignore such obvious problems, but perhaps in their plans, it is the most important thing to have a certain degree of autonomous driving capabilities.
If you want to summarize the development characteristics of the new energy automobile industry in one word, it is "fast".
Tesla is a prime example. As an important promoter of the trend of new energy vehicles, slow work has never been Tesla's style: the Model 3 has been in constant condition for 5 years, and the reliability of FSD has been repeatedly questioned, and it has also been investigated by the US government. But at the same time, it is this method of efficiency first, even at the expense of a certain quality, that makes it a leader in the new energy vehicle and autonomous driving industry.
This style of behavior may bring Tesla a lot of moral and legal pressure, but if they really polish each product and then go on the market, who can guarantee that they can have the status they have today? If you are more pessimistic, if Tesla really acts like a conscientious enterprise, perhaps when people hear the word "Tesla" today, they still think of the genius scientist who designed the alternating current system.
Nowadays, the penetration rate of new energy vehicles has exceeded 20%, and with the proposal of carbon neutrality goals in various countries, the penetration speed will only become faster and faster, and the inflection point of the automatic driving outbreak is about to come. Therefore, it is more important to "inject the soul" into your car first than to focus on the obsession with the control of the "soul".
Moreover, judging from BYD's practice of wooing partners in the past two years, they are not unaware of the risk of handing over their "souls" to others, otherwise they can find the strongest manufacturers on the market and directly purchase the full-stack solutions provided by the other party. All car companies want to develop their own full stacks, but in the face of no one's time, this way of putting eggs in multiple baskets may already be the best choice that BYD can make in urgent need of autonomous driving capabilities.