NVIDIA DRIVE AGX Thor: Powering the Future of Autonomous Vehicles
As autonomous vehicle (AV) systems evolve, integrating reasoning-capable vision-language-action models, generative AI, and next-gen sensor technologies, developers require tools that are not only high-performance and efficient but also meet rigorous automotive safety standards.
Enter the NVIDIA DRIVE AGX Thor developer kit — now available for preorder with shipments beginning in September. This advanced platform empowers global developers and researchers to accelerate the design, testing, and deployment of AVs and intelligent mobility solutions.
Built for the Next Generation of AV Intelligence
DRIVE AGX Thor is engineered on the latest NVIDIA Blackwell architecture, equipped with Arm Neoverse V3AE CPUs, and runs on the NVIDIA DriveOS 7 software stack. It’s purpose-built for complex multimodal AI models and optimized for automotive development.
Key features include:
- Extensive I/O support for surround cameras, radar, and lidar systems
- Integration with common automotive interfaces like GbE/10GbE and PCIe
- Compliance with critical safety and cybersecurity standards: ISO 26262 (functional safety) and ISO 21434 (cybersecurity)
A Growing Ecosystem Around DRIVE AGX Thor
Leading global automakers have adopted DRIVE AGX Thor, including:
- Passenger Vehicles: BYD, GAC, IM Motors, Li Auto, Volvo Cars, Xiaomi, Zeekr
- Autonomous Trucking: Aurora, Gatik, PlusAI, Waabi
- AV Platforms: DeepRoute.ai, Nuro, WeRide, ZYT
Tier 1 suppliers offering production systems based on DRIVE AGX Thor include Continental Automotive, Desay SV, Lenovo, Magna, and Quanta.
In addition, sensor and embedded tech pioneers such as AdaCore, Lauterbach, OMNIVISION, QNX, and Vector are extending support for the platform.
End-to-End Safety: From Cloud to Car
DRIVE AGX Thor and DriveOS are central to NVIDIA Halos, a comprehensive AV safety framework that unifies hardware, software, and AI research into one safety-driven ecosystem.
Halos provides:
- Technology-level safety: covering platforms, algorithms, and ecosystems
- Development-level guardrails: during design, deployment, and validation
- Computation-level integration: spanning AI training to real-world deployment with:
- NVIDIA DGX for AI training
- NVIDIA Omniverse & Cosmos on OVX for simulation
- NVIDIA DRIVE AGX for vehicle deployment
Get Started Today
- Watch the DRIVE AGX Thor unboxing video
- Join the NVIDIA DRIVE AGX SDK Developer Program
- Explore related platforms: the Jetson AGX Thor developer kit and Jetson T5000 modules — both available now, enabling robotics developers to shape the future of physical AI
