When will AMD's Kria AI robotics platform be available?

The Ryzen AI Embedded X100 processors began sampling with early-access customers in June 2026. AMD expects the Kria AI SOM and developer platform to reach broader availability through device-maker partners in the fourth quarter of 2026; pricing has not been disclosed.

How is Kria different from AMD's Instinct AI chips?

Instinct is AMD’s data-center accelerator line for training and running large AI models in the cloud. Kria is aimed at embedded, on-robot compute — real-time sensing and control inside a physical machine rather than a data center.

Does this actually beat Nvidia's robotics chips?

AMD claims 3.4x better real-time control reliability and 2.3x more concurrent AI-agent capacity than Nvidia’s Jetson T5000, but these are AMD’s own benchmarks; independent verification is still pending.

AMD Unveils Kria AI Platform to Challenge Nvidia in Robotics

AMD introduced a new hardware line built for robots on July 23, 2026, unveiling the Kria AI system-on-module (SOM) and an accompanying Kria AI Robotics Developer Platform at its Advancing AI 2026 event, according to the company’s official announcement. The launch is AMD’s most direct push yet into physical AI — AI systems that sense and act in the physical world — a category Nvidia has dominated with its Jetson line of embedded computers.

What AMD announced

The Kria AI SOM pairs up to 16 Zen 5 CPU cores with an RDNA 3.5 integrated GPU and a neural processing unit, built on AMD’s newly launched Ryzen AI Embedded X100 Series processors. AMD says the module can sustain more than 8,000 control decisions per second and complete vision-language-action reasoning — the loop that lets a robot see, interpret an instruction and act — in under 100 milliseconds.

The developer platform bundles the module with a robotics carrier card and evaluation kit, giving robot builders a single, pre-integrated stack instead of assembling compute, sensing and real-time control separately. AMD is calling it the first open, turnkey platform of its kind for robotics hardware. “For decades, AMD adaptive computing technologies have powered the sensing, safety and real-time control systems at the heart of industrial and surgical robots,” said Salil Raje, senior vice president and general manager of AMD Embedded. “Today, we’re extending that leadership from the robot body to the robot brain.”

Aiming at Nvidia’s turf

AMD pitched the new hardware directly against Nvidia’s Jetson T5000 modules, claiming 3.4 times better real-time control reliability, 1.6 times more available CPU headroom and support for 2.3 times more concurrent AI agents. Nvidia’s chips remain the default choice for most robotics developers — the company recently rallied Japanese manufacturers around its own physical AI push — and AMD’s performance figures have not been independently verified.

The launch extends AMD’s broader campaign to challenge Nvidia across the AI stack; its Instinct accelerators already target the same rival in data-center AI training and inference.

AMD said the Ryzen AI Embedded X100 processors began sampling with early-access customers in June 2026, with the Kria AI SOM and developer platform reaching broader availability through original device manufacturer partners in the fourth quarter of 2026. The company did not disclose pricing.

What AMD announced

Aiming at Nvidia’s turf

Frequently asked questions