Nvidia Drive
System on a chip family for autonomous vehicles
Nvidia Drive is a series of systems on a chip designed for autonomous driving and driver assistance. Introduced in 2015, it uses deep learning to process sensor data in cars from manufacturers like Tesla and BYD.
- Introduced at CES in Las Vegas in January 2015
- Tesla used Drive PX 2 for Autopilot from 2016
- Thor SoC delivers 1000 Sparse INT8 TOPS
- Uses ARM CPU cores and Nvidia GPU microarchitectures
- Lynk & Co 900 is the first car with Thor
What it is
Nvidia Drive is a computer platform built around systems on a chip. It provides the processing power needed for autonomous cars and advanced driver assistance systems. The hardware relies on deep learning algorithms to interpret data from cameras and sensors. Nvidia introduced the first version at the Consumer Electronics Show in Las Vegas in January 2015. The platform has evolved through several generations, each named after Nvidia GPU microarchitectures such as Maxwell, Pascal, and Blackwell. It serves as the central brain for vehicle navigation, safety features, and digital cockpits.
How it came about
The project began with the launch of Drive CX and Drive PX in 2015. These early models used the Maxwell architecture. A year later, Nvidia released Drive PX 2 based on Pascal. This version gained prominence when Tesla Motors included it in vehicles manufactured from late 2016. The platform was rebranded as DRIVE AGX in 2017 with the introduction of Xavier. Nvidia announced the cancellation of the Atlan project in 2022 and replaced it with Thor. Each iteration aimed to increase computational power while managing energy consumption for automotive use.
Inside the chip
The chips integrate ARM CPU cores with GPU units for parallel processing. Early models like Drive PX 2 combined Tegra X2 processors with Pascal GPUs. Later designs, such as Xavier, added a Deep Learning Accelerator to boost performance per watt. The Thor SoC uses an Arm Neoverse V3AE CPU and a Blackwell GPU. It supports 8-bit floating point operations for efficient AI inference. Multiple chips can connect via NVLink-C2C to scale performance. The system handles inputs from lidar, cameras, and radar through specialized drivers.
How it was made
The Drive PX 2 AutoCruise board used a modified single-chip configuration with an added MXM Module GPU. The Pegasus system combined two Xavier devices with two Turing discrete GPUs. Thor represents the latest generation, featuring high-density integration for sparse tensor operations. The hardware is packaged into reference boards for developers to test and adapt for specific vehicle architectures.
Where it went
Tesla used the Drive PX 2 for its Enhanced Autopilot and full self-driving features. The Roborace autonomous racing series also utilized this hardware. Modern electric vehicle manufacturers including BYD, XPENG, Li Auto, and ZEEKR have adopted the DRIVE AGX Thor. The Lynk & Co 900 is the first production car to feature the Thor SoC. The platform supports Level 5 autonomous driving capabilities in high-performance configurations like Pegasus.
Versions and family
The lineup includes Drive CX for digital cockpits and Drive PX for autonomy. Drive PX 2 had AutoCruise and AutoChauffeur variants. DRIVE AGX Xavier offered a balance of power and efficiency. DRIVE AGX Pegasus targeted full autonomy with higher power draw. The cancelled Atlan was set to use Ada Lovelace graphics. DRIVE AGX Thor is the current flagship, offering significant gains in teraflops. Software bundles like Hyperion match hardware generations with operating systems.
Why it matters
The platform helped bring deep learning to mass-market automobiles. By providing a standardized hardware and software stack, Nvidia enabled faster development of self-driving features. Third-party companies like DeepRoute.ai use the Drive SDK as a foundation. The shift from general-purpose computing to specialized AI accelerators in cars reflects broader industry trends. The inclusion in Tesla vehicles marked a major milestone for automotive AI adoption.
Questions people ask about Nvidia Drive
What is Nvidia Drive used for?
It is a platform for autonomous driving and driver assistance. It processes sensor data using deep learning to help cars navigate safely without human input.
Which cars use Nvidia Drive chips?
Tesla vehicles from late 2016 used Drive PX 2. Newer models from BYD, XPENG, Li Auto, ZEEKR, and the Lynk & Co 900 use the DRIVE AGX Thor system.
How powerful is the latest Drive chip?
The DRIVE AGX Thor delivers 1000 Sparse INT8 TOPS and 1000 Sparse FP8 TFLOPS. It uses an Arm Neoverse CPU and a Blackwell GPU for high performance.
What software runs on Nvidia Drive?
The system uses DRIVE OS, which supports Linux, Android, and QNX. Developers use the Drive SDK, which includes DriveWorks, DRIVE AV, and DRIVE IX components.
Good reading and sources
- Checking your browserAnandTech, Article
- NVIDIA: End-to-End Platform for Robotaxis & Autonomous VehiclesNvidia, Article
The facts on this page come from the Wikipedia article "Nvidia Drive", written by its contributors and used under the CC BY-SA 4.0 licence, and from its Wikidata record (CC0). EveryChip rewrote the text in its own words, with AI assistance, and compares every page with the article to make sure no passage is copied. It may be reused under the same CC BY-SA 4.0 licence with a link back.
It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.