Chip companies are strengthening their capabilities in software and looking to move further from Tier2's role. For auto OEMs, they also hope to control software capabilities in their own hands.
However, for some traditional car companies that do not have strong software capabilities but want to transform, the software capabilities provided by chip companies are extremely important.
Changes in the model of cooperation between suppliers and car companies are confirming this. Recently, NVIDIA and Jaguar Land Rover announced that they will fully cooperate on autonomous driving systems.
Since Nvidia released a new generation of autonomous driving chip products Orin in 2019, many of the world's leading automakers and domestic new forces have appeared on NVIDIA's customer list. But this time is different, Nvidia and Jaguar Land Rover, together with software experts from both companies, will develop AI autonomous driving and connected services for future cars built on NVIDIA DRIVE.
This collaboration is an attempt at a new model.
Not long ago, car companies also expressed their position to take the software into their own hands, just like GM and Qualcomm, Ford and Mobileye, and other car companies that cooperate with Nvidia. Therefore, this cooperation with Jaguar Land Rover to open a full model and full ecology is of great significance to NVIDIA.
Touching the brain of car companies Nvidia goes deep into the hinterland
Some time ago, the twists and turns of NVIDIA's acquisition of ARM failed, and its core problem involved the competition of intelligent ecology. From the news in August last year to the announcement of abandonment in January this year, Nvidia has been too much hindered.
One of the obstacles is Tesla, Tesla's self-developed self-driving software and hardware, in front of many intelligent driving chip manufacturers, Tesla's most obvious opponent is NVIDIA.
Mobileye, which initially cooperated with Tesla, was the first to start researching autopilot chips, and also collaborated with many new energy vehicle companies such as Weilai, but was eventually abandoned in the car factory because of the software packaged black box solution. On the contrary, NVIDIA's software infrastructure and complete tool chain give customers greater initiative, which opens up the situation of intelligent electric vehicles and wins the fixed points of many car companies in one fell swoop.
In 2021, Mercedes-Benz and NVIDIA will cooperate in the development of high-end chips to perform complex and advanced functions such as autonomous driving assistance and intelligent car-machine systems; provide support for Volvo to build a new generation of core computing systems through the NVIDIA DRIVE platform architecture; and cooperate with new domestic car companies to support chip hardware systems.
And holding hands with Jaguar Land Rover brings a different model.
All new models from 2025 onwards for Jaguar and Land Rover will be built on the NVIDIA DRIVE software-defined platform, which will provide them with a variety of active safety, autonomous driving and parking systems, as well as driver assistance systems. The system will also support a variety of AI functions in the car, such as driver and occupant monitoring systems and advanced visualization of the vehicle's surroundings.
At this year's CES show, NVIDIA showcased the NVIDIA DRIVE Hyperion platform, a platform for mass production of autonomous vehicles. With a complete software stack for autonomous driving (DRIVE AV) and driver monitoring and visualization (DRIVE IX), drive Hyperion is capable of over-the-air updates, adding new features and functionality throughout the vehicle's lifecycle. The next-generation platform, Hyperion 8, is now open for sale and is available for the 2024 model.
The full-stack solution developed by Nvidia in collaboration with Jaguar Land Rover is based on NVIDIA DRIVE Hyperion and incorporates the DRIVE Orin Centralized Autonomous Vehicle Computing System, DRIVE AV and DRIVE IX software, a variety of occupant safety, network security and network connectivity systems, and sensors. DRIVE Orin is the CAR's AI "brain" that runs Jaguar Land Rover's operating system, while DRIVE Hyperion is the car's "central nervous system."
Jaguar Land Rover will also train AI models using a self-developed data center solution as well as NVIDIA DGX, and real-time physics-accurate simulations using DRIVE Sim software built on NVIDIA Omniverse. With Jaguar Land Rover's software-defined capabilities and its end-to-end testing and validation architecture, vehicles will receive novel assistive and autonomous driving services through OTA software updates throughout their lifecycle.
The product matrix formed by hardware + software and infrastructure makes NVIDIA have a developed autonomous driving product and solution system. But the "brain" and "nerve" that really touch the car companies are the most exciting for chip suppliers.
Is software payment the next wealth password for the car?
At the same time as announcing the opening of a new partnership with Jaguar Land Rover, NVIDIA released its fourth quarter and full year financial reports. The latest financial performance is eye-catching: revenue for the full fiscal year 2022 was $26.914 billion, an increase of more than $10 billion from the previous fiscal year, an increase of 61% year-on-year, and profit increased by 106% year-on-year.
In the fourth quarter of fiscal 2022, NVIDIA's automotive business achieved revenue of $125 million, down 14% year-over-year. This is related to the lack of cores in the entire automotive industry, affecting the overall industry shipments. But increased demand for autonomous driving and AI cockpit solutions led to a 6% year-over-year increase in revenue.
The automotive business is not a large share of NVIDIA's revenue, less than 2%, but machine learning-based artificial intelligence and autonomous driving will be Nvidia's future decisive market winning weapons. Driven by the intelligence of automobiles, autonomous driving is the future of NVIDIA's computing power and represents the next market for chips. Nvidia hopes to generate at least $8 billion in revenue from the automotive industry over the next 6 years.
Huang Jenxun said, "We have always been a software-driven company, but for the first time, we have a software package that can be used for licensing. Our AI-related software for enterprises is based on the number of servers, there are currently about 25 million servers in the world (except for the cloud), we firmly believe that in the future all of these servers will be equipped with artificial intelligence software, we want to provide the most reliable and advanced artificial intelligence tools. ”
"Regarding NVIDIA Drive (self-driving car hardware), we have a similar business model with the software business, whether it is parking software or in-car artificial intelligence software, whether it is prepaid or in the form of monthly membership fees, it is similar. Since we can benefit from the entire product cycle of the car, assuming that there are 10 million vehicles with an average lifespan of 10 to 20 years, the potential is huge."
Cars make money by "hardware embedding + software upgrades", and car companies are becoming more and more aware that hardware suppliers also want to eat this wave of dividends.
Starting Feb. 17, Tesla's FSD price for customers in the U.S. was raised above $12,000. When FSD was first introduced in 2015, a set cost just $2,500. Currently, Tesla owners can subscribe to the FSD feature on a monthly basis for $199/month. Musk said the monthly subscription price will also rise when the FSD is widely available.
Western Securities expects the overall Tesla FSD selection rate to increase to 50% in 2025, accounting for about 14% of Tesla's revenue.
As early as xiaopeng disclosed the revenue of the automotive software business for the first time, He Xiaopeng said, "The realization of XPILOT software will become a continuous source of revenue for us in addition to the sales of vehicle hardware." ”
Chip suppliers such as NVIDIA are developing sophisticated toolchains to provide a platform for these car companies that have the ability to develop their own software. In addition, they still have a lot of opportunities to serve non-self-developed software car companies.
A Qualcomm spokesperson once said: "For some car manufacturers, if the software development capabilities are weak, and the development costs of third-party traditional Tier1 suppliers are too high, we can directly provide a complete ADAS turnkey solution." ”
The collaboration between NVIDIA and Jaguar Land Rover, then, opens up a third approach, not just providing a hardware platform or submitting a software solution directly, but on the NVIDIA DRIVE software-defined platform, the two parties jointly build and develop a series of functions.
Che Yun summary
Reshaping the structure of the industrial chain has not yet come out of a clear route, but car companies have an increasingly high demand for software, and the ability to develop software and realize payment is mainly based on the first echelon of new cars such as Tesla. Secondly, traditional car companies rely on the platform of hardware suppliers and jointly develop software functions in the development of intelligence, which is a more efficient way.
The absolute chain relationship between suppliers and OEMs is transitioning to a mesh relationship, but in general, car companies will still stand in the leading position for a long time, and how suppliers participate in this value chain is also a great test.