Autonomous Vehicle Model Development # Build Smarter, Safer Driving Models [Get Started](https://developer.nvidia.com/drive/infrastructure) Overview ## From Fleet Data to Reasoning-Based Driving Models A production-ready driving model is built on large-scale data and the reasoning capability to handle complex, real-world driving. NVIDIA brings together data curation and synthetic data generation to build training data in volume and coverage, including rare, long-tail events. NVIDIA’s open reasoning VLA models learn from that data and refine through reinforcement learning to make interpretable driving decisions that advance safer autonomy. ### Alpamayo 2 Super, The Frontier Open Model for Robotaxis and Autonomous Vehicles, Now Available The frontier open reasoning model for autonomous vehicle and robotaxi development is here — commercially licensed and ready to build on. [Read the Blog](https://blogs.nvidia.com/blog/alpamayo-2-super-open-model-now-available) [Watch the Video](https://www.youtube.com/watch?v=VeYZ-MrJv3A) ### The World Is Building Robotaxis on NVIDIA DRIVE Hyperion Global automakers, software partners, and mobility leaders are bringing level 4-ready fleets to market on DRIVE Hyperion—NVIDIA's robotaxi-ready platform. [Read the Press Release](https://nvidianews.nvidia.com/news/nvidia-drive-hyperion-becomes-the-global-platform-for-a-robotaxi-ready-world) Benefits ## Why Model Development Matters for Autonomous Vehicles Model development bridges the gap between a capable prototype and a production-ready AV through models that reason, data that covers the long tail, and continuous iteration on edge cases. ### High-Throughput Data Processing AVs generate terabytes of multimodal data from cameras, lidar, radar, and sensors. This data has to be ingested, reconstructed, curated, and labeled at scale before it can be used to train AI models. ### Continuous Improvement AV systems need to improve continuously, learning from new data, rare events, and edge cases to refine perception, prediction, and planning. ### Synthetic Data Generation at Scale Optimize for high‑throughput synthetic data generation of real‑world drives and scalable scene reconstruction. This enables efficient validation of changes and broad scenario coverage from fleet data. ### Safety-Grade Data Curation Ensure the right data, not just more data, is used to train and validate safety-critical systems. Technology ## Dataset Preparation and Model Training ### NVIDIA DGX™ * Unified AI training platform combining software, infrastructure, and expertise for enterprise-scale model development * High-performance model training and fine-tuning at data center scale [Learn More About DGX](https://www.nvidia.com/en-eu/data-center/dgx-platform.md) ### NVIDIA Alpamayo * Open family of VLA models, simulation frameworks, and datasets for reasoning-based AVs * Human-like reasoning to interpret complex driving scenes and explain decisions * Causal reasoning label generation for driving clips at scale * Available in 10B and 34B parameters [Learn More About Alpamayo](https://www.nvidia.com/en-eu/solutions/autonomous-vehicles/alpamayo.md) ### NVIDIA Cosmos™ for Data Factory * Open platform for physical AI with WFMs, video data processing libraries, video evaluation, and post-training frameworks * Large-scale dataset processing and metadata generation * Petabyte-scale data search and curation * Synthetic data generation, video quality scoring and evaluation at scale [Learn More About Cosmos](https://www.nvidia.com/en-eu/ai/cosmos.md) ### NVIDIA AI Enterprise * Essential tools for streamlining the development and deployment of AV software * Includes everything from data preparation and training to optimizing for inference and deploying at scale * Direct access to NVIDIA AV experts for NVAIE subscribers—the deepest level of technical guidance available to optimize your NVIDIA software deployments [Learn More NVIDIA AI Enterprise Software Suite](https://www.nvidia.com/en-eu/data-center/products/ai-enterprise.md) ## Customer Stories ### AI Factory Across the Automotive Industry ### Hyundai Hyundai Motor Group will tap into NVIDIA’s data-center-level computing and infrastructure to efficiently manage the massive data volumes essential for training its advanced AI models and building a robust AV software stack. [Read the Blog](https://blogs.nvidia.com/blog/hyundai-motor-group-ces/) ### Wayve Wayve’s partnership with NVIDIA enables seamless deployment of autonomous driving, AI training, and fleet learning. This collaboration accelerates scalable, high-performance adoption of AI-driven automotive systems. [Read the Blog](https://blogs.nvidia.com/blog/wayve-generative-ai/) ### Volvo Cars and Zenseact Volvo Cars and its software subsidiary, Zenseact, are investing in NVIDIA DGX systems for model training in the cloud. This will help ensure that future fleets are equipped with the most advanced and well-tested AI-powered safety features. [Read the Blog](https://blogs.nvidia.com/blog/volvo-cars-accelerated-computing-ai/.) ### Waabi Waabi trusts NVIDIA hardware to run complex simulations and trains its AI models. They'll also use NVIDIA Cosmos for data curation for software development and simulation. [Read the Blog](https://blogs.nvidia.com/blog/waabi-autonomous-trucking/) ### NIO NIO, a smart electric vehicle manufacturer, is using the NVIDIA DGX AI platform to improve training efficiency and GPU utilization of perception models for autonomous vehicles. [Read the Blog](https://developer.nvidia.com/blog/perception-model-training-for-autonomous-vehicles-with-tensor-parallelism/) Previous Next 1. First teaser 2. Hyundai 3. Wayve 4. Volvo Cars and Zenseact 5. Waabi 6. NIO ## Accelerate Your Development Unblock data bottlenecks with the NVIDIA Physical AI Dataset, an open-source dataset for autonomous vehicle, robot, and smart space development. The unified collection is composed of validated data used to build NVIDIA physical AI—now available to developers on Hugging Face. [Start Building](https://huggingface.co/collections/nvidia/physicalai-67c643edbb024053dcbcd6d8) #### Resources ## Breakthroughs in AI, Accelerated Computing, and Simulation 1. News 2. Sessions Load More [View More Sessions](https://www.nvidia.com/en-us/on-demand/playlist/playList-822dc11f-f1e2-4a75-ac11-035c09fec6bf/) ## Next Steps ### Ready to Get Started? Use NVIDIA open datasets, models, and frameworks to curate driving data, generate training sets, and train autonomous driving models. [Start Developing](https://developer.nvidia.com/drive/infrastructure) ### Request a Consultation Talk to an NVIDIA automotive specialist about building reasoning-based models with high-quality training data. [Let’s Talk](https://www.nvidia.com/en-eu/self-driving-cars/self-driving-cars-contact-us.md) ### Stay Informed Get the latest news on NVIDIA automotive technologies, breakthroughs, and more. [Subscribe](#stay-informed)