Amazon is placing a serious bet on the idea that AI’s next frontier isn’t generating text or images. It’s understanding the physical world around us.

The company has been quietly assembling a portfolio of investments, partnerships, and internal initiatives aimed at what the industry calls “Physical AI,” the kind of artificial intelligence that can perceive real-world environments and make decisions about them in real time. The centerpiece of that effort is a startup called Archetype AI, which raised a $35M Series A round in November 2025 with participation from Amazon’s Industrial Innovation Fund.

What Archetype AI actually does

Archetype AI built a model called Newton. Think of it as a foundation model, but instead of being trained on internet text, it’s trained on the messy, noisy data that comes from the physical world. Newton processes multimodal sensor data, including text, audio, and video, to interpret and interact with physical environments in real time.

The model is already integrated with AWS services and listed on the AWS Marketplace, which means any company running infrastructure on Amazon’s cloud can plug it in.

The applications here stretch across manufacturing, logistics, energy, and any industry where physical processes need monitoring and optimization.

Amazon’s broader Physical AI strategy

Archetype AI isn’t Amazon’s only move in this space. In September 2025, Amazon partnered with NVIDIA and MassRobotics to launch the Physical AI Fellowship, a program designed to nurture startups building simulation-based robotics and automation solutions. The fellowship provides resources, mentorship, and access to compute infrastructure. Applications are already open for the 2026 cohort.

NVIDIA’s Isaac Sim platform, which runs on AWS, lets companies build and train robots in simulated environments before deploying them in the real world. Physical AI initiatives using NVIDIA Isaac Sim on AWS have reportedly delivered a 25% efficiency gain in Amazon’s own supply chain operations.

The Amazon Industrial Innovation Fund, which led the investment in Archetype AI, specifically targets startups working on technologies that improve supply chain operations, logistics, and industrial processes.

Why simulation matters more than you think

Simulation technology lets companies train their models in realistic virtual environments where mistakes are free and data is abundant. Once the model performs well enough in simulation, it graduates to the real world.

It’s worth noting that not every physical AI project orbits Amazon directly. Project Prometheus, for instance, focuses on engineering AI for physical-world applications but operates independently from Amazon’s corporate backing.

What this means for investors

By investing in startups through its Industrial Innovation Fund, hosting their products on AWS Marketplace, and co-sponsoring fellowships that create pipeline for future acquisitions and partnerships, Amazon is building a flywheel of physical AI development on its infrastructure.

Watch for how quickly Archetype AI’s Newton model gains adoption on the AWS Marketplace, and whether the Physical AI Fellowship’s 2026 cohort produces startups that attract meaningful follow-on funding.