Ahead of CES 2026 and AMD CEO Lisa Su’s keynote address, AMD announced the expansion of its embedded processor lineup with the new Ryzen AI Embedded P100 Series, followed by the X100 series with a higher core count.
This new product brings the latest Zen 5 architecture and XDNA AI engine to the field of edge computing, emphasizing that it provides up to 50 TOPS of NPU computing power in a low power consumption range of 15W to 50W, targeting application markets such as industrial HMI (human-machine interface), automotive digital cockpit, medical equipment, and AI robots.
Among them, the Ryzen AI Embedded P100 series offers versions with up to 6 computing cores to provide an immersive experience, which are expected to be available to the public starting today. Another version with up to 12 computing cores will be available in the first half of 2026, mainly designed for industrial automation applications.
The X100 series targets the application needs of physical AI (such as robots and self-driving cars) equipped with autonomous driving systems, offering a design with up to 16 computing cores, and is expected to be launched in the second half of 2026.
Zen 5 architecture moves to the edge and adopts BGA packaging.
Unlike socket-based designs, the Ryzen AI Embedded P100 series uses BGA (Ball Grid Array) packaging, resulting in a smaller size and suitability for harsh vibration environments. In terms of hardware specifications, the P100 series utilizes TSMC's advanced 4nm process technology and features 4 to 6 Zen 5 architecture cores.
AMD emphasizes that this processor uses a "heterogeneous computing" architecture, which integrates the CPU, GPU, and NPU into a single chip (APU):
• CPU (Zen 5):It is responsible for control logic and complex calculations, bringing a significant improvement in IPC compared to the previous generation architecture.
• GPU:It offers immersive graphics processing, with official data showing a 35% increase in rendering speed and support for 4K to 8K resolution displays, making it suitable for high-end medical monitors or in-vehicle entertainment systems.
• NPU:Designed specifically for AI inference, it provides 30 to 50 TOPS of computing power and can handle tasks such as speech recognition, gesture control, or defect detection without having to send data back to the cloud.
Tailor-made for "digital cockpit" and "industrial HMI"
In terms of application scenarios, AMD specifically highlighted two major areas: automotive electronics and industrial automation.
For the automotive market, the P100 series (such as the P132a model) will support automotive-grade certification (AEC-Q100), enabling it to simultaneously drive the driver's dashboard and passenger entertainment screen, and to realize in-vehicle voice assistant and driving monitoring system through NPU.
In the industrial sector, the P100 series supports 10GbE w/TSN (Time-sensitive Networking), ensuring that commands from robotic arms or automated production lines can be transmitted instantly and accurately. Furthermore, addressing the "long-term" usage requirements most important to embedded products, AMD promises up to 10 years of product lifecycle support and 24/7 reliability.
Mass production in Q2 2026, with the entire software ecosystem open source.
On the software side, in order to lower the development threshold, AMD stated that the P100 series will support multiple operating systems such as Yocto, Ubuntu, Android Automotive, and RTOS, and will adopt an open source strategy, allowing developers to use familiar TensorFlow or PyTorch frameworks to deploy AI models.
According to AMD's plan, samples of the Ryzen AI Embedded P100 series chips have already been provided to some customers for testing, and mass production is expected to begin in the second quarter of 2026. The related development reference board will be available in the second half of the year.
Edge AI enters the "truly practical" stage
In the past, embedded systems mostly aimed for "good enough," but with Industry 4.0 requiring real-time defect detection and automotive systems needing to run large-scale voice assistants, traditional MCUs or older architecture CPUs are becoming inadequate. AMD's P100 series has a clear strategy: solve all problems with a single chip.
The 50 TOPS NPU computing power is a sweet spot, sufficient to run Llama or speech models with moderate parameters on the ground without requiring a power-hungry discrete graphics card. For industrial computers or robots with limited space and demanding cooling conditions (such as fanless designs), this high-performance APU will be a strong competitor to traditional x86 platforms or Arm solutions.
