Intel Series 3 Processor Secures Major Robotics Design Wins to Enable Integrated Physical AI and OpenVINO Software Framework for Industry Adoption
Intel is making great strides in the areas of robotics and edge computing, already locking down over 130 design wins with the Intel Series 3 processor. These deals are meant to take experimental robot designs out of labs and into service for businesses. Integrated physical AI will play a significant role, enabling robots to see, reason, and act in the real world with high precision timing that is measured in milliseconds.
To prove the utility of these processors in real world situations, Sensory AI decided to base Ella, which is publically running in the 1st multi agent physical AI store, on the Intel Core Ultra Series 3 platform. This transition will effectively replace a hardware configuration with a discrete CPU and discrete accelerators with a single Intel SoC, enabling this system to handle both real time mechanics and demanding AI operations with no need for additional processors. The integrated design can support 3 separate software agents, known as Avatar, Guardian, and Ella Agent, which control customer interactions, physical store functions, and business intelligence, respectively. A deterministic orchestrator is responsible for controlling the actual hardware of the robot, allowing a single system to eliminate the need for a full category of secondary processors, simplify overall software architecture, and offer greater return on investment. This technology is currently on display near the main entrance at Computex.
Intel realized that the fragmentation of robot software systems will greatly hinder adoption. To solve this problem, the company has released Physical AI Studio and OpenVINO Physical AI. Dan Rodriguez, Corporate VP for the Edge Computing Group at Intel, explained their rationale behind the integrated software stack:
Physical AI models are making great strides in the world of robotics, but their adoption is being slowed by complex and inconsistent software stacks that requires custom integration for every robot. Physical AI Studio coupled with Intel Core Ultra Series 3 and OpenVINO Physical AI offers an integrated, open, and extensible pathway from robot experimentation to production systems delivering high performance inference backed by hardware.
Developers can use Physical AI Studio for data collection, fine tuning, and quantization of models and export these verified vision language action models. OpenVINO Physical AI acts as an open source robotics software library featuring an inference engine that is optimized for Intel processors. OpenVINO provides a standardized framework that translate robot policies and multi modal models into actual robot movements while supporting standard open source frameworks such as LeRobot.
Companies in manufacturing and retail are utilizing the automation that these technologies enable to overcome issues such as labor shortages and the increasing cost of operations. These systems, however, previously required unique software for each specific robot, integrating everything from sensor inputs to codec processing to mechanical activation. This forced each operation to purchase a specialized, overbuilt dual compute system that was difficult to maintain. However, using the combination of Intel Core Ultra Series 3 and OpenVINO software creates a standard stack which is much cheaper for companies to operate and makes it possible to reuse robot software across all factories, warehouses, and retail centers.
These new software resources are a part of the larger Intel Robotics AI Suite and are available for use with Intel partner development kits. Physical AI Studio is available to developers starting immediately, however, the preview of OpenVINO Physical AI can be found on GitHub, and full commercial release will follow in the 2H 2026.
