AMD Unveils Kria Suite of Solutions and New Partner Network to Advance Physical AI

AMD, in addition to providing the infrastructure that allows the development and execution of artificial intelligence, also promotes physical AI and robotics, hand in hand with the set of Kria solutions and a network of partners that seeks to break the dependence on proprietary and closed environments.

To boost the body-to-brain capabilities of robots, the semiconductor firm has unveiled at the ‘Advancing AI’ event the Kria suite of solutions, designed to rapidly prototype, build and deploy next-generation autonomous robotics and physical AI systems.

It includes the Kria AI SOM module, which powers the robot’s brain by enabling capabilities such as perception, reasoning, autonomous decision-making and action through artificial intelligence. To do this, it uses Ryzen AI Embedded X100 Series processors, which integrate CPU, GPU and NPU.

It also offers real-time deterministic control with more than 8,000 control decisions per second, while providing vision-language-action (VLA) AI reasoning in less than 100 milliseconds.

As highlighted by AMD, performance for robotics workloads includes up to 3.4 times better real-time reliability for real-time control, up to 1.6 times more free CPU cores for spare computing capacity, and support for up to 2.3 times more concurrent agents for agentic AI compared to the Nvidia Jetson T5000.

For physical AI development, AMD Robotics Partner Network has announced, a new network of validated hardware and software partners that helps customers streamline development, validate systems before deployment, reduce integration complexity, and accelerate the process from prototype to production without relying on proprietary runtime environments or closed development environments.