Government, DMG MORI and Komatsu to Join Forces on Data Collection for Physical AI, The Nikkei Reports

The Nikkei reported on September 21, 2026, that the government and major machinery makers, including DMG MORI and Komatsu, will collect training data to support the implementation of physical AI that autonomously operates robots and other equipment. The plan is to assign machines in factories common IDs similar to the My Number ID assigned to residents, then gather manufacturing data across company boundaries.

The plan to assign common IDs to machines in factories and collect data across manufacturers appears aimed at addressing the shortage of real-world machine data needed to train physical AI. With the government targeting more than 30% of the global market and a ¥20 trillion scale by 2040, standardization as industrial policy—not merely a research topic—is at stake.

The new element in this report is not a push to improve individual robots or machine tools, but the design of common IDs to collect training data across machines made by different manufacturers. Physical AI requires control models to be trained on field data so that real machines can operate safely. As a result, the competitive conditions will depend less on hardware performance alone and more on which machines provide the data, at what level of detail and how it is shared. This therefore appears to be closer to the creation of industrial infrastructure standards than to product development. In related coverage by this newspaper, Nikkei Business reports on construction physical AI and Nvidia’s growing prominence (2026-09-16) focused on computing infrastructure and the involvement of outside players in construction physical AI. Meanwhile, METI and NEDO’s GENIAC reports selecting 13 companies and groups for robot foundation model development support (2026-09-11) reported that a project plan briefing on September 9 outlined plans to prepare field data and collect work data, including tactile data. This latest report can be read as showing how the government will advance data standardization—the resource most likely to be in short supply—as the effort moves from research and development to implementation. From a competitive standpoint, the participation of industrial machinery makers such as DMG MORI and Komatsu could shift the focus from competition over robot performance alone to competition over ecosystems spanning machine tools, construction equipment, controls and training data. Based on publicly available information, it is unclear which machine categories the common IDs will cover, what data fields will be included, how data will be anonymized or who will manage the standards. These issues are likely to determine whether implementation succeeds. Given Japan’s existing industrial base, including DMG MORI’s machine tools and Komatsu’s construction equipment, a key question is whether these assets can be brought together as sources of training data. Points to watch include: 1) the machine models and participating companies covered by the common IDs; 2) the manufacturing data fields to be collected and the standardization process; and 3) how data-use permissions will be managed and security ensured.

The inclusion of industrial machinery makers with field data, such as DMG MORI in machine tools and Komatsu in construction equipment, appears significant for Japan. If the common ID and data-collection system is implemented, Japan’s existing assets in machine tools, construction equipment, sensors and maintenance could potentially be turned into training data, alongside control software.