Last week, NVIDIA and Mercedes-Benz announced their partnership going into the next decade to create fully autonomous vehicles (AVs). With NVIDIA’s global leadership in accelerated computing and Mercedes-Benz’s experience in the manufacturing of premium cars, together, they plan to create a new software-defined vehicle that includes in-vehicle computing systems and AI computing architecture. Mercedes-Benz and NVIDIA plan on rolling out the first fleet of cars with this architecture starting in 2024. On a call with investors, NVIDIA CEO Jensen Huang called this “the single biggest business model transformation” in the history of the company, allowing the two companies to share recurring revenues for future feature purchases and subscriptions.
NVIDIA is primarily known for its gaming GPUs, but it is no stranger to making silicon for autonomous vehicles. At CES 2018, the company revealed its partnership with Volkswagen and Uber to put AI technology and AV capabilities into their cars as well as the release of the Xavier processors as part of its DRIVE AGX Pegasus platform. At CES a year later, NVIDIA unveiled the world’s first commercially available Level 2+ automated driving system.
At GTC this year, NVIDIA announced that it is putting its Ampere technology in the NVIDIA DRIVE platform. The newest SoC, called Orin, will be offered in advanced driver-assistance systems (ADAS), L2+autopilot, and L5 robotaxi systems, as the platform has several scalable solutions. NVIDIA also announced at this event all the partners they’re working with to advance AV, from automotive companies that specialize in trucks to digital mapping companies. Orin is the SoC that Mercedes-Benz and NVIDIA’s software-defined architecture will be based on.
NVIDIA and Mercedes-Benz have an AV leadership mentality
One of the reasons this is such great news is because it puts the infant AV industry into a competitive state. If you have read any of my previous posts on AV, specifically here, you would know that Tesla is largely ahead of the game when it comes to automating its vehicles fully. In the post above, I mention that NVIDIA’s automotive computing solutions (ACS) supply the firepower for other companies like Mercedes-Benz to compete with Tesla. While Tesla is doing the work of NVIDIA and Mercedes Benz in terms of manufacturing its vehicles and ACS, NVIDIA can partner with automotive manufactures to bring the technology to different parts of the industry. With that being said, we have to look at why the partnership between NVIDIA and Mercedes is beneficial.
Mercedes-Benz sold nearly 2.4 million passenger cars last year and more than 438,000 vans worldwide. It is a pioneer in innovation and says that it aspires to be a leader in the fields of connectivity, automated driving, and alternative powertrains. Mercedes-Benz is one of the largest, if not the largest seller of premium vehicles in the world. When we look at new technology, we always have to start somewhere, and that usually means starting from the top. But according to Mercedes, they plan to enable every car in their fleet starting in 2024.
NVIDIA’s founder and CEO Jensen Huang and Chairman of the Board of Management of Daimler AG and Head of Mercedes-Benz AG Ola Källenius, already appear to have a good relationship from previous partnership activities. This deal should build on that rapport.
Variables NVIDIA and Mercedes-Benz have already thought through
There are many problems and speed bumps that NVIDIA and Mercedes-Benz are going to face. Jensen and Ola addressed some of those hurdles in their press release here. Jensen acknowledged that it would take time to make all cars autonomous on the road. It is not an endeavor that will take just a couple of years but could take even a decade. Making all cars autonomous will involve many obstacles that are hard to imagine now.
Ola mentioned that putting autonomous hardware into a vehicle is like making a smartphone on wheels. To that, he is right, and it brings up some interesting points. When we look at the history of smartphone devices, smartphones rarely last longer than two to three years. Vehicles, on the other hand, can last somebody a lifetime. So how can NVIDIA take this differentiation into account to make a car’s autonomous hardware last? That is where the vehicle’s ability to be upgradeable is so essential and what I think is a major reason Mercedes-Benz is intelligently partnering with NVIDIA. NVIDIA has the experience of upgradeability in the gaming industry.
Mercedes-Benz and NVIDIA plan on making their cars perpetually upgradeable, including software that is developed on supercomputers and then uploaded to the vehicle. Ola addressed a question with the upgradeability solution concerning people driving around in 20-year-old designs with the newest hardware. He said it shouldn’t be a problem, and I agree with him on that. Many people upgrade their vehicles for more than just the reason for it not working. Imagine if iPhones were supported for a whole decade rather than only five years. There will be some users who keep their phones, but more often than not, consumers will continuously buy the newest model, as they do now.
Some higher road bumps the AV will have to face
There are industry-changing aspects of this automotive transformation as well. Cars will keep their value as they continuously receive updates rather than losing value as soon as they roll off the lot. It will open up a new market for application and automotive developers. Also, cars will increasingly be on the cutting edge with 5G support. I am interested to hear from Jenson and Ola how 5G will be a factor in all of this. I believe it has the power to implement better safety precautions, more efficient traffic-related routes, updates, etc.
During the press event, Jensen and Ola talked about recurring revenue for both companies from software sales and subscriptions. While they didn’t discuss numbers, this has got to represent a sizable upside in potential revenue. They even intimated that the secondhand owner of one of these supercomputer-enabled cars could continue to purchase updates and add-ons. The question is which updates will be for a one-time payment and which will be subscription-based.
Autonomous vehicles not only need to continue to be more technologically advanced, but they also bring up ethical dilemmas and moral arguments that will be in the heart and the code of the AI. Vehicle safety is going to face heavy criticism until we have complete street autonomy. At that point, will there be laws facing those who desire to drive themselves? Are people infringing on other people’s rights to safety when they desire to drive themselves? I think these problems will be far down the road, but they are challenges that should be thought of and addressed from the beginning.
Going back to ethical dilemmas and moral arguments for safety, how will NVIDIA and Mercedes-Benz program its software when it faces a car crash? Will my car be programmed to always prioritize my life over others in a car crash, all lives even at the possible cost of my life or others’? And, who decides to program our fate and who gets the responsibility of the outcome? These questions are going to have incredible implications in the future of the AV industry. I don’t think these are bumps in the road and problems that NVIDIA and Mercedes-Benz will have to face alone. I do commend them both for being bold enough to be one of the first, however.
During the live-streamed joint press event, Huang said, “This is the biggest partnership of its kind in the transportation industry that’s ever happened.” Indeed, I want to come back to this post four years from now when Mercedes-Benz rolls out its first line of vehicles with NVIDIA’s software-defined architecture. I believe this partnership between NVIDIA and Mercedes-Benz is going to be beneficial for consumers, partners, and even Tesla, as a competitive market always sparks innovation. With NVIDIA’s leadership in accelerated computing and Mercedes-Benz’s global leadership in luxury consumer vehicles, I think autonomous vehicles have a bright future.
Note: Moor Insights & Strategy co-op Jacob Freyman contributed to this article.