New Mercedes-Benz Models To Be Software-Defined, Nvidia-Powered From 2024

No automaker has a longer history of technological innovation than Mercedes-Benz. From the first drive that Karl Benz took on his Patent Motorwagen in 1886, the brand has frequently been the first to introduce new technologies to the auto industry including diesel, direct injection, electronic ABS and stability control. But the 21st century has seen new entrants like Tesla TSLA upending the industry with the shift to electrification and continuous updates through software. With the help of Nvidia NVDA , Mercedes-Benz is now aiming to catch up and hopefully surpass its new competitors like Tesla.

Daimler CEO  Ola Källenius and Nvidia CEO Jensen Huang made a joint announcement today from their respective headquarters about the software-defined future of Mercedes-Benz vehicles. Starting in 2024, all new Mercedes-Benz models would incorporate a new electronic architecture powered at least in part by Nvidia’s upcoming Orin system-on-a-chip (SoC). 

Initially announced in December, 2019, Orin is the next-generation SoC designed by Nvidia around the recently revealed Ampere GPU architecture. Nvidia is planning a family of Orin chips ranging from a lower power unit with 10 trillion operations per second (TOPS) performance for driver assist systems to a 200 TOPS unit for highly automated vehicles. 

With this announcement, Mercedes-Benz becomes the first automaker to announce it’s intent to utilize the Orin platform. Mercedes-Benz also plans to utilize this architecture in every redesigned vehicle it builds from 2024. The new electronic architecture will be adopting the trend of consolidating the dozens of smaller electronic control units (ECU) found in most modern vehicles down to a few large computers that can handle virtually all functions. In the case of high-end Mercedes-Benz vehicles, that ECU count can approach 150. 

Current electronic architectures were not designed to support over-the-air (OTA) updates to most systems, but the architecture that will be developed by Daimler and Nvidia will provide this capability. Mercedes-Benz and Nvidia couldn’t provide specific details about the new architecture, except to say that it wouldn’t be a single central compute platform. Most likely, it will be built around three to four main computers. Mercedes-Benz would also only confirm that the computer that runs the automated driving and ADAS features would be powered by Orin. It’s possible that other domain controllers may utilize chipsets from other vendors.   

Many automakers are now moving aggressively direction software-defined vehicle architectures. EV startups like Nio in China have already incorporated this capability but traditional automakers are also launching updateable vehicles this year. Tens of millions will likely be on the road by the time the first example from Stuttgart arrives in 2024. 

GM’s Digital Vehicle Platform (also known as Global B) debuted in late 2019 on the Cadillac CT5 and Chevrolet Corvette C8 and is going into all redesigned GM vehicles going forward. The GM architecture doesn’t utilize centralized compute, but it’s internal networking and security features are designed to support regular updates and the automaker has already committed to updates for its second-generation Super Cruise system. 

Similarly, the Ford Mustang Mach-E that goes on sale later this year has an architecture designed around three domain control computers and the automaker has said that virtually every system in the vehicle can be updated remotely. Just a week ago, Ford announced that its hands-free Active Drive Assist system would be enabled by an OTA update and it is planning a regular cadence of feature updates for Mach-E owners after it launches later this year. Ford is planning to add similar capabilities to other yet to be announced vehicles. 

The partnership between Nvidia and Mercedes-Benz goes well beyond just being a chip supplier. The two companies will collaborate closely on the entire architecture design and Mercedes-Benz will also utilize Nvidia’s full Drive AGX software stack to power its automated driving systems.  This will include partially automated level 2 and 3 systems as well as highly automated level 4 self-parking.

While Mercedes-Benz declined to specify this as the reason for last week’s announcement that it was pausing its automated driving partnership with BMW, it should be noted that BMW has a close relationship with Nvidia’s primary rivals in this segment, Intel INTC and Mobileye. It’s possible that the two German companies could not reach a consensus on the compute platform to use for its planned shared system. 

Utilizing the Drive AGX software and a full suite of sensors that includes cameras, radar and lidar, Mercedes-Benz plans to offer what the companies call “address to address automated driving.” Unlike the separate partnership that Daimler has with supplier Bosch which is targeting the shared services market including robotaxis, this program is focused on retail consumer vehicles. However, even the Daimler-Bosch partnership is using Nvidia compute platforms including the Drive AGX Pegasus.

In Guidehouse Insight’s recent Automated Driving Compute Platform leaderboard report, Nvidia was ranked first overall ahead of Intel/Mobileye and Qualcomm QCOM . Nvidia recently announced that Chinese EV startup Xpeng is using its Xavier SoC for its Xpilot 3.0 driver assist system. Numerous other automakers including Volvo are also using Nvidia chipsets and the company best known for its high-powered computer graphics cards is likely to grab a significant share of the automotive market in the coming years. 

The partnership with Mercedes-Benz is the first that ties Nvidia much more deeply into a production automated driving program that also uses its software capabilities to such a large degree. Besides just automated driving, Mercedes-Benz and Nvidia also highlighted the new revenue opportunities that will be available from a software-defined vehicle platform. 

Just as Tesla has been able to significantly boost its margins over the past five years by selling its “full self-driving” package, other automakers are hoping to leverage vehicles that are connected and updatable to add functionality long after the customer takes delivery. At this stage, it’s too early to tell exactly what added capabilities Mercedes-Benz might offer, but the possibilities are endless. 

Whether customers will actually step up to pay for yet more services in an era where they are already starting to question how many subscription services they are paying for every month is another story. There are certainly some features that would be compelling to have available right in the vehicle such as paying for parking, charging or tolls. Those are all possibilities for automakers to take a slice of the revenue. 

Even though the first software-defined Mercedes-Benz is coming a full dozen years after Tesla launched this effort, the real business opportunities are just starting to get underway now.