Physical AI News (Japan)

Physical AI, robotics and compute — reported from Japan.

日本語
· Research & technology· 日本語で読む

Panasonic Holdings Announces Faster R&D in “Physical AI Originating on the Factory Floor”

Panasonic Holdings will accelerate R&D that combines AI with technologies and expertise developed on manufacturing floors. It will use AI to improve materials and manufacturing processes for glass-type perovskite solar cells and equipment for advanced semiconductor packaging. The company will also advance its “Manufacturing Commons” initiative to share expertise across industry, government and academia.

Key facts

  • Panasonic Holdings said it would promote “Physical AI originating on the factory floor” as an important direction for its technology strategy and accelerate R&D.
  • For glass-type perovskite solar cells, business activities through sales of products from initial production runs began in September 2026, and the company aims to begin full-scale mass production in April 2029 at its mass-production technology development site in Nagahama, Shiga Prefecture.
  • For equipment used in advanced semiconductor packaging, the company launched the “AI Semiconductor Packaging Project,” which uses AI analysis of process data and inspection images for quality prediction and condition searches.
  • The project name for the initiative to share technologies and expertise across industry, government and academia is “KMC.”

What happened

In an announcement published on November 10, 2026 (dated November 11, 2026 in the document), Panasonic Holdings said it would accelerate R&D, identifying “Physical AI originating on the factory floor” as an important direction for its technology strategy. The company aims to advance research, design and manufacturing processes by linking AI with technologies and expertise accumulated on manufacturing floors. Specific initiatives include using AI in materials and manufacturing processes for glass-type perovskite solar cells and developing equipment for advanced semiconductor packaging.

Why it matters

Panasonic Holdings said it plans to use AI analysis of process data and inspection images for quality prediction and condition searches, and to reflect the results in equipment specifications. The effort connects the use of AI on manufacturing floors not only to analysis but also to equipment and process design, directly addressing challenges in developing equipment for semiconductor packaging.

Analysis

The announcement’s focus is not just the slogan of introducing AI to manufacturing floors, but the framework it sets out for linking manufacturing technologies, process data, AI analysis, and improvements to equipment and materials. The published examples cover materials development and process-condition optimization for solar cells, as well as quality prediction and condition searches for semiconductor packaging. In other words, the announcement presents a cycle in which AI outputs are applied to design and manufacturing, and the results are fed back to the factory floor, as a common framework for R&D. It is important to distinguish what is new from what builds on existing efforts. The company describes materials, devices, sensing, control and manufacturing processes as technologies it has developed over time, so linking them with AI extends its existing assets. Meanwhile, plans to share expertise across organizations and companies through KMC and create AI and robotics reference models originating in Japan, along with the “AI Semiconductor Packaging Project,” which integrates R&D and business development, were presented as frameworks for putting this strategy into practice. Articles published later by this newspaper—Nikkei reports Panasonic Holdings will open a physical AI collaboration hub in Tokyo, its second after Osaka, Nikkei Xtech reports development of a model to predict screen changes at Panasonic Holdings, and Panasonic Holdings and others jointly develop T-Rex, a two-handed task AI that responds instantly to touch—covered a hub, screen-change prediction and two-handed tasks involving touch, respectively. These articles were published after this announcement and cannot be treated as results available at the time. They do, however, provide context for distinguishing the broad policy of AI originating on the factory floor from individual research and collaboration efforts. The implication for industry structure is that companies with manufacturing expertise are seeking to take on the role of translating AI analysis into processes and equipment specifications on the factory floor. In semiconductors, the focus is on quality and conditions derived from inspection images and process data; in solar cells, it is on materials and manufacturing conditions. The types of data and processes involved differ. A key question will be what can be standardized and what requires custom design for each product and process. The cross-organizational sharing proposed under KMC will also depend on how much expertise can be standardized and how each company handles information specific to its processes. The announcement did not specify the quantitative effects of AI adoption, the scope of the processes covered, which organizations will participate in KMC or the operating terms for data sharing. The next points to watch are how AI analysis is reflected in actual equipment specifications and manufacturing conditions, the reproducibility of processes as solar cells move toward full-scale mass production, and the scope of application for common models. Progress on the initiative should be assessed based on whether these implementation conditions and the procedures for returning expertise to the factory floor are made public.

Details

For glass-type perovskite solar cells, the company plans to use informatics and AI in materials development and manufacturing processes to improve materials development and optimize process conditions. The cells use the company’s proprietary materials technology as well as inkjet coating, laser processing and sealing technologies developed in display manufacturing. Business activities through sales of products from initial production runs began in September 2026, and the company aims to begin full-scale mass production in April 2029 at its mass-production technology development site in Nagahama, Shiga Prefecture. For equipment used in advanced semiconductor packaging, the company will draw on technologies for plasma surface modification, high-precision bonding, printing, inspection and transport to develop equipment solutions supporting technologies such as 3D stacking of semiconductor chips. The “AI Semiconductor Packaging Project,” which uses AI analysis of process data and inspection images for quality prediction and condition searches, has also been launched. The company is advancing its “Manufacturing Commons” initiative (project name: KMC) to share technologies and expertise across industry, government, academia and companies, and to create AI and robotics reference models.

Implications for Japan

The KMC initiative aims to share technologies and expertise from Japan’s manufacturing industry across organizational and corporate boundaries, and to create AI and robotics reference models originating in Japan. If realized, a concrete issue for the cross-industry use of domestic manufacturing technologies will be how far the process expertise of individual companies can be made shareable.

Sources

Translated from the Japanese original: /news/panasonikku-hd-fijikaru-ai-no-kenkyuukaihatsu-soshiki-wo-tac-f7cdf1. The Japanese version is authoritative if they differ.