AWS launches open-source Physical AI Toolchain built with NVIDIA's stack

What the toolchain offers

The Physical AI Toolchain on AWS brings together architecture guidance, deployment automation and ready-to-use code. AWS says it is purpose-built for industrial automation, autonomous mobility and humanoid robotics, and inspired by Amazon's own robotics expertise.

The company frames physical AI as machines that perceive, understand and act in the real world. Unlike conventional machines that repeat fixed instructions, such systems sense their environment, reason using cloud-trained AI models and adapt in real time. Operational data flows back to retrain those models.

Uwem Ukpong, vice president of AWS Industries, said customers told AWS that "too much of their engineering effort was going to infrastructure instead of innovation. We want to flip that."

Five pillars and the services behind them

The toolchain spans five core pillars, each feeding the next in a continuous improvement cycle:

• Synthetic Data Generation: AI-generated environments create diverse training scenarios, reducing the need for expensive real-world data collection.
• Model Training: learning from human demonstrations and practice in simulated environments.
• Simulation and Validation: testing machine behavior in virtual environments before deploying to real hardware.
• Edge Deployment: optimized models are pushed to machines in the field, where they decide in real time without constant cloud connectivity.
• Continuous Improvement: operational data from deployed machines generates new training data.

On the AWS side, the stack uses Amazon SageMaker for model training, Amazon EC2 GPU instances for simulation, AWS IoT Greengrass for edge deployment and Amazon Bedrock AgentCore for orchestration. From NVIDIA it draws Isaac Sim for simulation, Isaac Lab for reinforcement learning, Isaac GR00T for humanoid machine training and Cosmos for synthetic world generation.

Customers can adopt the full stack and manage the development cycle from a single control plane, or pick individual components, such as standalone simulation, training or deployment tools, that fit into existing workflows. Fleet management features let them provision, secure and update thousands of machines over the air.

Early users and market context

AWS cites the Global Startup Trends Report on physical AI, which found that one in seven startups globally is now building physical AI, with 72% of builders saying cloud computing is essential to their systems.

Several companies are building on AWS. NEURA Robotics is developing cognitive humanoid robots and aims to bring millions of intelligent robots to market by 2030. RLWRLD is building an 8.1-billion-parameter foundation model for dexterous robotic hands. Config has built a data pipeline capturing more than 200,000 hours of robot action data and uses generative AI to multiply it into varied training scenarios.

NEURA Robotics founder and CEO David Reger said the toolchain helps accelerate the cycle of fine-tuning models, deploying them and scaling from single systems to large fleets. NVIDIA's Amit Goel said the toolchain combines AWS services with NVIDIA's models, tools and libraries into a scalable, end-to-end workflow.

Built on Amazon's robotics experience

Amazon says it has deployed more than 1 million robots across its operations network, handling millions of packages daily alongside hundreds of thousands of employees. The company also notes that Amazon Robotics unveiled its next-generation autonomous Proteus robot earlier this summer.

According to AWS, the toolchain's architecture guidance, reference code and deployment automation are inspired by those learnings and help manufacturers launch physical AI capabilities in weeks rather than the years it would take starting from scratch. Each deployment can be tailored to the customer's hardware, operational environment and use case.

What it means for manufacturers

AWS points to uses such as collaborative robot arms that adapt to new assembly tasks without reprogramming, autonomous robots trained on new parts, and smart factory systems that monitor, predict and optimize production. The company argues that manufacturers can increase throughput, reduce downtime and improve quality, and that data from each machine improves models for the whole fleet.

Customers bring their own domain expertise and hardware, while AWS supplies the foundation. Interested companies can find details on the AWS physical AI page or contact their AWS account manager. The source does not give pricing or a separate availability schedule.