As smart cars enter the critical cycle of "cost reduction and efficiency improvement", the impact on the upstream industry chain, especially chips, is also continuing to ferment.
This week, Nvidia released its third quarter financial report data as of October 29, 2023. Overall business revenue was US$18.12 billion, an increase of 206% over the same period last year and an increase of 34% over the previous quarter.
Among them, the data center business's revenue in the third quarter reached a record $14.51 billion, an increase of 41% from the previous quarter and an increase of 279% from the same period last year. The two business segments of games and professional design visualization solutions also achieved a year-on-year growth rate of more than 80%.
NVIDIA third quarter financial report
However, As a business segment with huge future potential, Nvidia’s automotive business revenue in this quarter was US$261 million, an increase of only 3% from the previous quarter, and a year-on-year increase of only 3%. Growth was only 4%.
Prior to this, Nvidia CEO Huang Jenxun’s future revenue expectations for the automotive business were to reach the tens of billions of dollars. Obviously, changes in the automotive industry are constantly revising the company's expectations.
"The decline in automotive business revenue growth may be due to the excess inventory of Chinese automakers (in the field of smart driving, Chinese automakers are NVIDIA's largest customer group) and the downward revision of sales expectations for mid- to high-end models. "In the view of the Gaogong Intelligent Automobile Research Institute, this means that market demand is being reconstructed.
In fact, clues can also be seen from the Guangzhou Auto Show that opened last week. Most car companies are beginning to focus on the smart electric market segment with prices below 300,000 yuan. For them, this volume market, which accounts for nearly 80% of new car sales throughout the year, is the ticket to stay in the battlefield.
Up to now, NVIDIA has launched a number of high-computing chip products in the automotive industry, such as the Tegra series, Parker, Xavier, and Orin, and has also won large-scale mass production projects from many leading car companies.
According toHigh-tech Intelligent Automotive Research Institute monitoring data shows that in the Chinese market only, from January to September this year, NVIDIA Xavier/Orin’s The number of vehicles delivered reached 300,000 vehicles (corresponding to 743,200 chips), a significant increase of 269.16% year-on-year.
From the perspective of performance comparison, on the one hand, Nvidia's traditional cockpit product projects have gradually entered the model upgrade cycle, and Qualcomm platforms have become mainstream; on the other hand, it also reflects the uncertainty of subsequent order delivery for smart driving projects.
In the Chinese market, since this year, local manufacturers including Horizon, Black Sesame Intelligence, and Aixin Yuanzhi have been continuously grabbing customers from car companies. At the same time, in the global market, Mobileye and Qualcomm continue to impact the high-end smart driving market where NVIDIA is located.
For example, in April this year, Black Sesame Intelligence officially launched the Wudang series C1200 smart car cross-domain computing chip. As an "All in one" chip, C1200 focuses on the integration of comprehensive computing types, oriented to multi-domain fusion cross-domain computing, and a single chip covering high-end intelligent driving scenarios.
A more prominent advantage is that this chip also provides the industry's highest MCU computing power of 32K DMIPS; this means that true single-chip NOA can be achieved. Compared with the existing inefficient domestic solutions that use at least three chips, the board level is the simplest.
The latest data released by the Gaogong Intelligent Automotive Research Institute shows that from January to September 2023, 377,300 new cars were delivered in the Chinese market (excluding import and export) with standard passenger car front-end equipment (software and hardware) NOA, an increase of 151.20% year-on-year. %.
However, in terms of front-end installation rate, it is only less than 3% (2.55%). At the same time, from the perspective of specific brand models, except for Li Auto (NOA's installation volume accounts for about 60% of the overall market), the installation volume of other brands is still subject to the influence of sales volume and configuration rate.
At the same time, the average delivery price of NOA’s standard new cars is still as high as 360,900 yuan; this has led many car companies to start looking for alternatives to NVIDIA. This is one of the reasons why the latter started to promote the low-end version of Orin last year.
Just in September this year, Denza officially announced that on the basis of the previously provided Orin-X version, a new Orin-N configuration (with a computing power of 84TOPS) was added; and the corresponding high-speed smart driving package for new car owners is priced at 12,000 yuan, which is 11,000 yuan cheaper than the previous X version.
Three years ago, Nvidia even planned to launch the low-end Orin intelligent front-view all-in-one solution, with a computing power of only 10TOPS, to compete with Mobileye and other competitors in the entry-level ADAS market.
After all, Nvidia had to make concessions in the face of the cruel reality of the market.
In addition, in the cockpit part, Nvidia chose to cooperate with MTK this year to integrate GPU chips into MTK’s next-generation smart cabin SoC. It hopes to use the latter’s customer resources and mass production experience to continue to maintain or even seek new business breakthroughs.
At the same time, Nvidia has focused its core resources on intelligent driving. This year, it recruited Wu Xinzhou, the former vice president of autonomous driving of Xiaopeng Motors, to lead the research, development and mass production of full-stack autonomous driving software and hardware systems.
For Nvidia, the automotive chip (hardware) business has gained momentum in the past few years, but its contribution to the company's overall performance has not yet reached a critical point.
Data shows that in 2021, Nvidia's automotive business revenue will be US$566 million, accounting for 2% of the company's total revenue, and in 2022, revenue will be US$903 million, accounting for approximately 3%. Although NVIDIA has more than 10 billion US dollars in orders on hand, it is still unknown whether it can be fulfilled.
Previously, Huang Renxun also proposed the concept of "Software 3.0" (relying on data, algorithms and computing engines). Combined with the company's strategic transformation, he believed that "Nvidia hopes to sell more software that only runs on its GPUs."
However, the house leaked and it rained all night.
Just this week, overseas media revealed that Valeo has sued Nvidia for stealing trade secrets. The reason is that a former Valeo employee recruited by Nvidia in 2021 is suspected of stealing a large number of confidential documents from the company (Valeo), including the source code of the APA automatic parking system.
Previously, after joining NVIDIA, the employee was found to have a window containing Valeo software source code and "ValeoDocs" displayed on his computer desktop during a conference call with Valeo.
Currently, this employee has been criminally convicted by the court in September this year. Previously, Valeo and Nvidia signed a contract with an automobile manufacturer to develop an APA system.
Valeo claims in the lawsuit that the stolen information was shared with other Nvidia software engineers who had access to and were using Valeo's trade secrets. "This means that Nvidia can save millions or even hundreds of millions of dollars in development costs."
Nvidia has not yet commented on the lawsuit; however, Valeo is seeking an injunction from the court to prohibit Nvidia from continuing to use the leaked software code, plus an unspecified amount of damages.
As we all know, currently, the number of suppliers that can provide mass-produced automatic parking solutions globally is far less than that of driving ADAS. Valeo, Bosch, and TTE are the major share holders in the global APA market; in China, there are only a few manufacturers such as Desay SV, Zhihua Technology, and Moshi Intelligence.
For high-end intelligent driving, integrated driving and parking is a major trend. For NVIDIA, which is currently stuck in software system development and has been unable to mass-produce and deliver its products, what it lacks is engineering and mass-production delivery experience.
The latest monitoring data from Gaogong Intelligent Vehicles shows that Nvidia is currently recruiting a large number of autonomous driving solution software engineers in the Chinese market. The offices are located in Beijing and Shanghai. The goal is to be deeply involved in verifying, adapting and improving the performance of the software stack in China.
Recruitment directions include high-precision maps and positioning, behavioral decision-making, planning and control, system solutions and automatic parking system solutions, and a "full employee shareholding" plan has also been launched.
NVIDIA has previously launched the NVIDIA DRIVE™ open source software stack, which can help developers efficiently build and deploy a variety of applications, including perception, positioning and mapping, planning and control, driver monitoring and natural language processing.
However, this is still a long way from mass production delivery.
Three years ago, Nvidia and Mercedes-Benz announced that they would work together to develop fully autonomous vehicles over the next decade. The two parties will use their respective experience in high-performance computing and high-end automobile manufacturing to jointly create a new software-defined car.
This cooperation is regarded by Nvidia as "the largest single business model transformation" in the company's development history. Both parties will be able to share revenue from future user purchases of features and subscription services. Behind the scenes, NVIDIA provides full-stack software and hardware solutions and is jointly developed with Mercedes-Benz.
To this end, Huang Renxun also specifically emphasized that "in the next few years, new cars sold at cost prices will no longer be a fantasy, because profits will mainly come from software." Public data shows that Nvidia sells $1 of hardware, software, etc. Related system providers will receive $8 in revenue.
However, the strategic cooperation project between Nvidia and Mercedes-Benz has been delayed due to the former's lack of full-stack software capabilities, coupled with unexpected factors such as the epidemic in recent years. In addition, Mercedes-Benz is even seeking help from other suppliers for its projects in China.
Now that Valeo is taking it to court, Nvidia's full-stack automotive business layout may become even more difficult. Previously, Valeo won orders for domain controllers and parking systems from the BMW (Qualcomm platform) project, and Qualcomm is Nvidia's direct rival.
Previously, the BMW Group and Valeo announced a new strategic cooperation, which is a further deepening of the existing long-term cooperative relationship between the two parties. This further highlights the competitiveness of automotive Tier 1 suppliers, especially in an industry context where the new car research and development cycle has been significantly shortened.
Following the announcement of cooperation in the development of ADAS domain controllers, sensors and parking and control software for BMW's upcoming new generation platform "Neue Klasse", the BMW Group and Valeo announced their cooperation in the development of fully automated parking technology up to L4 level .
In addition, as Nvidia's main customers in the Chinese market, new forces such as Lideal and Xpeng are also preparing for self-research of chips. At the same time, Huawei has introduced its own chips on a large scale relying on its smart selection model.
In particular, it is a foregone conclusion that Tesla FSD is about to enter the Chinese market, and it will accelerate the follow-up of China's leading smart electric car companies; in the view of these companies, software and hardware collaboration (to achieve independent controllability of chips) is the next cycle improvement Intelligent experience and the key to building differentiation.