Skip to main content

Nvidia and Mercedes Team Up to Make Next-Gen Vehicles

Mercedes plans to use Nvidia's Orin system-on-chip within its next-gen vehicle fleet.
  • Author:
  • Publish date:

Nvidia  (NVDA)  has landed Daimler's  (DMLRY)  Mercedes-Benz as a client for its newest processor for autonomous and semi-autonomous vehicles.

On Tuesday, Nvidia and Mercedes announced that Nvidia’s Drive hardware/software platform will underpin a “new software-defined architecture” that will be standard for Mercedes’ next-gen vehicle fleet. The cars will begin rolling out in 2024 and are promised to support “state-of-the-art automated driving functionalities," including the ability to “automate [the] driving of regular routes from address to address.”

Mercedes also plans to use Nvidia’s data center offerings to support its Drive-related R&D efforts. And -- borrowing a page from Tesla  (TSLA)  -- the German luxury automaker says that customers buying cars supporting its new software architecture will be able to “add capabilities, software applications and subscription services through over-the-air software updates during the life of the car.”

The deal comes 17 months after Nvidia and Mercedes announced a partnership to develop autonomous driving and “intelligent cockpit” systems for future Mercedes cars. A year before that, the companies announced that Nvidia’s silicon is powering Mercedes’ MBUX infotainment system.

Notably, Mercedes’ software architecture will be powered by Nvidia’s Orin system-on-chip (SoC), which was unveiled in December and further detailed in May. Orin packs 12 ARM CPU cores and a GPU based on Nvidia’s new Ampere architecture and is declared to be able to process 200 trillion deep learning operations per second (TOPS) while consuming just 45 watts of power. For comparison, Orin’s predecessor, known as Xavier, delivered only 30 TOPS while consuming 30 watts.

Nvidia's Orin system-on-chip. Source: Nvidia.

Nvidia's Orin system-on-chip. Source: Nvidia.

Nvidia is also offering a less powerful Orin SoC for advanced driver-assistance (ADAS) applications -- it only delivers 10 TOPS, but consumes just 5 watts of power and can fit behind car windshields. And for self-driving robotaxis, Nvidia is offering an 800-watt, 2000-TOPS system that contains two Orin SoCs and two discrete Ampere GPUs.

Along with Mercedes, Nvidia has Drive-related engagements with Audi, Volvo and Toyota, as well as with Baidu’s  (BIDU)  Apollo self-driving unit and with auto parts Bosch and ZF. The company has also indicated that some of the automakers that aren’t using its in-vehicle offerings are still major buyers of its data center solutions.

The Mercedes deal, which involves developing cars that support “automated driving” capabilities but don’t fully eliminate the need for a human driver, arrives at a time when near-term expectations for autonomous driving have been dialed back by both automakers and various third parties.

In April, Ford  (F)  said that it’s postponing its plans to launch an autonomous ride-sharing service until 2022. And last year, General Motors’  (GM)  Cruise self-driving unit called off plans to start a ride-sharing service by the end of 2019.