The era of physical artificial intelligence — robots that perceive their surroundings, handle objects and make decisions on their own — is arriving faster than expected. Nvidia declared a "ChatGPT moment for physical AI" at CES earlier this year, and CEO Jensen Huang framed humanoids and labor automation as a $40 trillion market. Nvidia looks positioned to become the core beneficiary of that wave.
Amazon sharply expanded its partnership with Nvidia in August. Amazon Web Services agreed to deploy an additional 2 million graphics processing units from the Blackwell Ultra, Rubin and Rubin Ultra generations over 2027 and 2028. Amazon Robotics also adopted Nvidia's full-stack physical AI platform — spanning Jetson, Omniverse, Isaac and Cosmos — to develop and train robots. Amazon, which has been building its own chips to reduce reliance on Nvidia, chose Nvidia for the foundational layer of its robotics work.
Nvidia's push into physical AI rests on the full-stack strategy it proved in the data center. From AI factories that train models, to Omniverse and Cosmos for simulating the physical world, to GR00T, a general-purpose foundation model for robots, to Jetson chips that put intelligence inside machines, the company has bundled the entire path from training to deployment into a single ecosystem. That is why CUDA's grip on the data center looks increasingly likely to be replicated in robotics.
Moving away from Nvidia also remains difficult. Even as Amazon develops its own Trainium chips, it decided to use Nvidia's NVLink for the scale-up network at the heart of rack-scale connectivity. Through NVLink Fusion, Nvidia is pulling even custom silicon into its own ecosystem. That means wider adoption of custom ASICs may not translate directly into lost revenue for Nvidia.
For investors, the question is who can monetize the growth of physical AI most broadly. Nvidia is expanding beyond selling finished GPUs into a full-stack physical AI company covering intellectual property, components and software for robots.
Valuation remains attractive. Nvidia trades at 18.7 times forward 12-month earnings, below the S&P 500's 19.9 times, while its projected annual earnings-per-share growth of 52.7% for 2026 through 2028 far outpaces the index's 20.1%. As physical AI extends beyond the data center into real physical space, Nvidia's position is unlikely to prove a passing benefit. That is why an overweight allocation remains warranted.
