· Research & technology· 日本語で読む
Intel and SAP Reportedly Conduct Warehouse Proof of Concept for Physical AI and Autonomous Robots
Intel and SAP reportedly conducted a proof of concept using physical AI in warehouse operations. The setup connected SAP EWM requests to work performed by autonomous robots and sent progress and completion status back to the system. The reported account does not establish the scale of the trial or its quantitative effects.
What happened
automatedwarehouseonline.com reported on October 27, 2026, that Intel and SAP AG conducted a proof of concept (PoC) using physical AI in a warehouse. According to the report, the trial tested a system that linked requests from SAP Extended Warehouse Management (SAP EWM) to work performed by autonomous robots, then sent the status and completion of transport tasks back to SAP EWM.
Why it matters
The reported PoC showed a setup that linked requests from a warehouse management system to robot transport tasks and returned completion information to the same business system. It is an example of how, when considering automated transport in warehouse operations, integration of instructions and status with SAP EWM is a condition for deployment alongside the robots’ capabilities.
Analysis
The focus of the trial was less on the autonomy of the robots themselves than on connecting warehouse management system requests to work on the floor and returning the results to the business system. The reported sequence was “SAP EWM request → instructions to robots using Intel edge AI devices → material transport → status and completion updates sent back.” This suggests that when incorporating physical AI into operations, the design connecting existing business information with the execution status on the floor is also a subject for evaluation, in addition to the robots’ movements. However, the reported benefits were described qualitatively as reductions in transport distance and repetitive work. Without details on scale or comparison conditions, the magnitude of the benefits and the scope of applicability cannot be assessed.
In our previous coverage of Intel, we reported on Intel develops “EMIB-T,” an advanced packaging technology for AI semiconductors (August 19, 2026). That article covered semiconductor packaging technology, while this report concerns operations on the floor, including edge AI devices, and their connection to business systems. The two cannot be treated as adoption examples of the same technology, but manufacturing technology for computing infrastructure and operational design for using computing resources on the floor are issues at different layers.
Separately, Tesla reportedly outlines a “robot school” concept for Optimus (October 22, 2026) covered views on securing training data for robots. The work status and completion information returned in the warehouse PoC were described as operational records; the report did not say they would be used to train robots. Operational progress tracking should not be conflated with the collection of training data, and how the two might be connected requires separate confirmation. Going forward, the scale of the trial environment, the conditions used to measure reductions, responses to unexpected events and the scope of status information returned will be key issues in assessing a transition to live operations.
Sources
Translated from the Japanese original: /news/intel-to-sap-souko-de-fijikaru-ai-to-jiritsu-robotto-no-jiss-32ae98. The Japanese version is authoritative if they differ.