Kawasaki Heavy Industries adopts physical AI semiconductor from domestic startup, Nikkei xTECH reports

Nikkei xTECH reported on October 29, 2026, that Kawasaki Heavy Industries adopted a physical AI semiconductor from a domestic startup.

Physical AI News previously reported that Kawasaki Heavy Industries completed a prototype of Home LEO and plans to bring it to market in fiscal 2028. The latest report does not identify which products will use the semiconductor, so it is not possible to determine whether this component selection represents a companywide effort to develop a computing infrastructure.

The report of the semiconductor adoption offers a way to view Kawasaki Heavy Industries' robotics-related activity not only through its machines and applications, but also through the selection of computing components that power them. However, the information provided does not confirm the name or performance of the semiconductor, the products it will be used in, or the timing of its introduction. Therefore, adoption for a specific product cannot be equated with the development of a computing infrastructure common to the company's entire robotics business. The company's product and application developments previously reported by Physical AI News include completion of a prototype of the social robot Home LEO and plans to bring it to market in fiscal 2028. There was also a report that it would adopt Noetra's physical AI for autonomous humanoid robot development. These were separate reports and do not indicate a technical connection to the semiconductor adoption or that the semiconductor will be used in the same model. Going forward, it will be important to determine which products or processes the semiconductor is associated with. In light of a report that the company plans to incorporate physical AI across all its businesses, this development may relate to component selection as a companywide policy is applied to actual machines and control systems. However, the materials provided do not show what organizational structure or common specifications connect the policy to this specific adoption. As the company pursues applications across different products, a key question for understanding its business structure is whether it will extend the component's use across products or limit adoption to specific targets. Points to verify include the adopting company and product name, the models involved, the stage of mass production or implementation, the conditions used to evaluate performance, and how the semiconductor's role compares with other computing platforms. Without this information, the adoption alone does not provide a basis for assessing processing capacity, safety or product competitiveness. See the original article for its technical explanation and details of the adoption terms.