Japan’s manufacturers see factory floors as AI’s next frontier

Japan is searching for its place in the global AI race. While American and Chinese companies dominate AI models and computing infrastructure, Japanese companies believe their expertise in robotics could help pioneer AI in real world tasks.

On May 13 Japanese industrial equipment maker, Fanuc, announced a partnership with Google that aims to create factory robots that can understand spoken and handwritten instructions and carry out factory tasks autonomously.

Fanuc, founded in Japan in 1956, is one of the world’s largest industrial machinery manufacturers. It’s developed an AI system with the help of Google Gemini that can be operated without programming skills. It plans to make all its robots compatible with Google software.

In December 2025, Fanuc also announced a collaboration with NVIDIA which will see it open its previously closed robot software systems. At a press conference on May 13, Senior Managing Officer Kenishiro Abe said the partnership stems from the limitations of developing an entire AI ecosystem in-house. It plans to incorporate AI systems from a host of different companies.

What is physical AI?

Factories are set to benefit the most from physical AI. While robots are already used extensively, they remain limited to repetitive tasks.

Physical AI is the practical application of AI. These AI systems are trained to perceive the real world, reason with it, act autonomously in real time as well as learn and collaborate from humans. They excel at handling complex and unpredictable tasks.

What is the winning formula?

For decades, Japanese factories have been shaped by knowledge that was never written down. Now, Japanese companies are trying to teach that knowledge to machines.

According to a Nomura Securities report, Japan’s decades-long manufacturing expertise and factory-floor data could power industrial humanoid robots.

In the 1990s Japanese manufacturers made up 80 percent of the global industrial robot market, according to the International Federation of Robotics. The figure has since fallen to roughly 40 percent.

As of 2024, Chinese companies such as Estun Automation and Inovance Technology are gaining ground and account for 40 percent of the global humanoid robot market.

But many Chinese companies still rely on Japanese machinery components. Nomura Securities predicts that Japan’s expertise in motion control technologies, industrial datasets, precision manipulators (i.e. robot hands) and semiconductor equipment could drive growth in a post-2030 economy.

Japan’s gaping ‘digital deficit’

Fanuc’s decision to open its source robotics software is a significant pivot from the Japanese manufacturing sector’s emphasis on hardware.

The country trails behind the U.S. and China in AI digital transformation (DX). Japanese companies rely heavily on software from U.S. tech giants resulting in a massive ‘digital deficit’ in which payments for digital services flow overseas.

The Ministry of Economy, Trade and Industry (METI) recorded a $4.9 billion digital services deficit in 2023. The U.S. on the other hand, posted a $173.7 billion surplus while China logged a $40.4 billion digital surplus.

As companies integrate AI into manufacturing, the Japanese government anticipates that rising demand for industrial robots will support the growth of Japanese industrial machinery companies.

Japanese technology company ARUM Inc has developed a fully automated, AI-enabled production line for metal part manufacturers. Its TTMC system costs approx. $2.3 million each. At Tokyo Sushi Tech Expo 2026, the company said it will install 100 units across Japan and has received enquiries from South Korea and the United States.

“We are not simply selling machines. We are connecting them through the cloud and building infrastructure,” said Takayuki Hirayama, CEO of ARUM Inc.

ARUM Inc believes that AI-driven manufacturing automation can solve global labor shortages and changing career preferences.

“Even in younger countries like India and Southeast Asia, skilled manufacturing workers are disappearing because IT and tourism are seen as more lucrative.”

The Japanese government wants to lead the robotics AI race

At a New Years press conference, Japanese Prime Minister Sanae Takaichi announced plans to accelerate physical AI innovation and expand the technology globally. She stated that AI-powered robots will learn from high-quality domestic data, in particular, Japan’s long established factory know-how.

The initiative builds on remarks made in December 2025 when Takaichi directed the government to support domestically produced general purpose AI models which are an essential component of physical AI. METI is set to launch a one trillion yen funding package (approx. $6.45 billion) over five years to help develop Japanese physical AI.

CEO Masato Fujino of Japanese industrial devices company, Fairy Devices Inc, believes that the challenge is no longer using AI within computers but bringing AI into the real world.

The company has produced wearable AI devices that prevent technicians from missing important checks. They are built with cameras, microphones, sensors and communications capabilities. The devices have accumulated large volumes of data and have trained the company’s vision language model which aims to replace experts such as air conditioner repair technicians.

At Tokyo Sushi Tech Expo 2026, Fujino said specialized data directly from skilled workers is indispensable for industrial AI systems.

“Google Gemini is powerful because Google owns Youtube. But when it comes to highly specialized industrial tasks, such as repairing industrial equipment, that data does not exist on Youtube.”

What role can Japan play in the physical AI sector?

Japan’s answer to AI is not frontier models but industrial data. Despite fierce competition for low-cost, high-quality physical AI, Japanese industry leaders are optimistic about Japan’s trajectory.

In their eyes Japan’s reputation for manufacturing excellence and proven track record in factory automation is difficult to replicate anywhere else.

If you're reading this, you’re already ahead. Stay there with our newsletter.