Intel Corp. said today it has locked in more than 130 design engagements for its Series 3 processor family, which targets edge artificial intelligence and edge computing workloads. It also introduced an open-source framework called OpenVINO Physical AI, aimed at solving what it describes as a deployment gap between robotics models built in the lab and fleets operating on factory floors. The announcements, made ahead of the Computex trade show in Taipei, position Intel’s edge silicon and robotics software stack as a single alternative to the fragmented mix of central processing units and discrete accelerators that have dominated robot design. The Series 3 family — which includes Intel Core Ultra Series 3 and Intel Core Series 3 processors — launched at CES in January as the first product built on Intel’s 18A manufacturing process.
Single-chip retail robot replaces dual-compute setup
One of the anchor design wins is SensoryAI Inc., which has moved its multi-agent retail robot Ella to Intel architecture. Ella is described by the company as the first multi-agent physical AI store running in public commercial service.
It previously used a separate CPU paired with a discrete accelerator.
SensoryAI swapped that arrangement for a single Intel Core Ultra Series 3 platform that handles both real-time control and AI inference.
Three specialized AI agents now run concurrently on the same system-on-chip: an Avatar agent handling customer conversation, a Guardian agent overseeing system operations, and an Ella agent providing store-level business intelligence.
Related: Blue Origin’s New Glenn rocket explodes during test
A deterministic orchestrator issues commands to the robot itself.
The chipmaker said the consolidation removes a class of components, reduces software complexity and offers a cleaner path to scaling future robot designs.
Other design wins span industrial generative AI, AI vision defect detection, rugged onboard computers, general-purpose humanoids, conversational AI for quick-service restaurants, agentic AI for infrastructure security, AI-enabled self-checkout, digital avatars and multimodal AI for medical imaging.
The list covers a wide range of edge applications.
It is not just robotics.
Related: The Evolving Software Landscape: What to Expect by 2026
OpenVINO Physical AI targets fragmented robot software stacks
The second part of the announcement is OpenVINO Physical AI, an extension of Intel’s OpenVINO toolkit, which the company launched in 2018 for computer vision at the edge. It is calling the new framework the first open-source robotics library with a silicon-optimized inference runtime.
The framework is designed to give developers a consistent manner to take robot policies and multimodal models from experimentation into working robot systems while maximizing inference performance. It integrates with open-source robotics model development environments, including Intel’s own Physical AI Studio and the open-source LeRobot project.
The company argues the framework tackles a specific industry problem.
Deploying physical AI models at scale has required customized pipelines for each robot to handle sensors, codecs, inferencing loops and actuation, often locking customers into dual-compute solutions that are expensive to deploy and maintain. Pairing Intel Core Ultra Series 3 with OpenVINO Physical AI, the company said, allows customers to lower total cost of ownership, reuse more code across robot types and scale fleets across factories, warehouses and retail environments.
“Physical AI models are transforming robotics, but deployment has been slowed by fragmented software stacks and one-off integrations for every robot,” said Dan Rodriguez, corporate vice president of the Edge Computing Group at Intel. “With Intel Core Ultra Series 3 and OpenVINO Physical AI, we provide a unified, open and scalable path from AI experimentation to production-grade robots delivering hardware-accelerated, high-performance inference.”
Related: 7 Surprising Facts About Software You Probably Didn’t Know
Intel pitches cost advantage over Nvidia’s Jetson modules
The chipmaker is also pushing the Series 3 family on price.
In benchmark figures shared with media, it put its Core Ultra X7 358H against Nvidia Corp.’s Jetson AGX Orin and Jetson Thor T5000 modules. It claimed competitive performance with Thor on medium-sized vision-language-action models at roughly half the system cost and lower latency than AGX Orin on a three-camera Pi0.5 Droid workload. Intel pegs Thor’s relative system cost at twice that of its own platform.
Those figures are Intel’s own benchmarks, and independent verification is not yet available. Nvidia’s Jetson line remains a strong competitor in the edge AI market, and design wins are not the same as volume shipments. Still, the consolidated approach gives Intel a narrative that may appeal to manufacturers looking to simplify robot hardware.
Physical AI Studio is available now.
OpenVINO Physical AI is available in preview on GitHub, with general availability targeted for the second half of 2026.
