Newsletter Signup
Join Our Community!
Stay updated with the latest news, exclusive insights, and more.
Stay updated with the latest news, exclusive insights, and more.
To see how we may use your information, take a look at our privacy policy.
.png)
When people talk about AI today, the conversation is often dominated by names like OpenAI’s ChatGPT, Anthropic’s Claude, Google's Genimi, or Microsoft's Copilot. The spotlight is firmly on the U.S. and Europe, where foundation models, open-source frameworks, and billion-dollar valuations make headlines almost daily.
Japan is also building its own path forward—one rooted in decades of research, a deep cultural relationship with technology, and a unique focus on societal application. While less visible on the global stage, Japan’s AI evolution has been steady, thoughtful, and increasingly relevant in the era of generative AI.
From early machine translation projects in the 1960s to the ambitious Fifth Generation Computer Project in the 1980s, Japan has long invested in artificial intelligence, often with a distinct emphasis on language and logic. Today, that legacy is shaping a new generation of AI startups developing Japanese-specific large language models (LLMs) and practical, industry-focused applications.
In this article, we’ll explore the historical context behind Japan’s AI development—and highlight some of the emerging startups that are quietly redefining the country’s position in the global AI race.
.png)
While the current global AI race often spotlights the U.S. and Europe, Japan’s AI journey also has growth. From early research into machine translation to ambitious national projects and the current wave of large language model (LLM) development, Japan's AI history reveals a consistent focus on language, logic, and societal application.
1950s–1980s: Machine Translation and the Fifth Generation Computer Project
Japan started working on AI in the late 1950s and 1960s. Researchers at the University of Tokyo and Kyoto University were beginning AI research and systems developing. Professor group of Kyoto University played a central role in advancing research on “media information processing” focusing on images, voice, and language. At the University of Tokyo, They were focusing on system for solving puzzles and games (like chess and Othello) with programs and logical reasoning.
In the 1980s, Japan launched a major national project called the Fifth Generation Computer Project (FGCS). It was supported by the government and aimed to build advanced AI systems that could understand logic and make decisions. The project used a programming language called Prolog and focused on new types of computer hardware. Although it didn’t achieve its big goals, the project helped develop many technologies and trained a generation of researchers. It also built valuable R&D infrastructure and the growth of Japan’s computing ecosystem.
1990s–2000s: Quiet Progress of AI Development
After the Fifth Generation Project ended, interest in AI dropped in Japan and other countries. This period is often called the “AI Winter.” Government funding was reduced, and many companies stopped investing in AI. However, AI related research didn’t stop completely. Universities and some major companies like NEC, Fujitsu, and NTT, continued to work on speech recognition, image processing, and robotics.
.png)
2010s: Deep Learning and New Startups
In the 2010s, deep learning became popular around the world. By increased computing power and big data, Japan saw a new wave of AI startups using this technology in industries like manufacturing, healthcare, and retail. Startups like Preferred Networks (PFN) and ABEJA led the way. PFN had a partnership with Toyota and Fanuc, aiming to integrate AI into robotics and smart factories. This period also saw renewed academic activity in AI, with universities expanding research labs focused on deep learning, NLP, and human-machine interaction.
2020s: Generative AI and the Japanese Language Challenge
After the release of ChatGPT and large language models (LLMs), the AI landscape shifted toward generative capabilities. Japan’s response has focused on developing Japan-specific LLMs. Several Japanese startups are building them to being able to handle the nuances of the Japanese language and context.
Some of the leaders in this space are:
Creating AI that works well in Japanese is not easy. The language is complex, and there is less training data than for English. But Japanese companies see this as a chance to lead in a specialized area.
.png)
Promising AI Startups in Japan
Companies and universities have worked on AI for many years. Now, a new generation of startups is bringing fresh ideas and speed. They are redefining how it is built, deployed, and scaled. Here are four startups leading Japan’s new AI movement.
**Sakana AI**Founded in 2023 by ex-Google Brain researchers David Ha and Llion Jones, Sakana AI is building next-generation foundation models in Tokyo. Inspired by natural evolution, the company explores alternative architectures and training methods. Their goal is to build global-level AI from Japan, and they’ve already received strong support from top investors.
**Preferred Networks (PFN)**Preferred Networks is Japan’s most prominent deep learning startup, known for applying AI to manufacturing, robotics, and healthcare. Partnering with Toyota, Fanuc, and others, PFN develops real-world AI solutions with long-term industrial impact. Its technical depth and practical focus make it a pioneer in Japan’s AI commercialization.
ABEJA provides AI-driven business solutions for retail, logistics, and manufacturing. They founded in 2012 and Its platform “ABEJA Platform” helps enterprises use machine learning to optimize operations and decision-making. With years of deployment experience and partnerships with major Japanese corporations, ABEJA was listed on the Tokyo Growth Market in June 2023.
**Jitera**Jitera builds development platform that that utilizes AI to automate and streamline the entire software development process. Their platform uses low-code and no-code technology. Aiming to speed up product launches and cut engineering costs, the startup serves clients in Japan and Southeast Asia.
.png)
Recent Trends and Collaborations
Japan’s AI startups are not working alone. Many of them are actively forming partnerships with universities, large companies, and the government. This focus on collaboration shows highlights Japan’s strength in building long-term, stable innovation ecosystems. At the same time, international investors are paying close attention. Sakana AI’s recent funding round, backed by global VCs, shows that Japan’s AI scene is becoming more visible and more competitive around the world.
In May 2025, Sakana AI signed a multi-year comprehensive partnership with MUFG. The project involves developing AI solutions for the bank sector, starting with the automation of document generation to support MUFG’s transition into an AI-native financial institution.
In May 2025, Preferred Networks strengthened joint development and capital alliances with companies like Sumitomo Life and Shin-Etsu Chemical. They are also expanding joint projects in next-generation materials and healthcare fields using AI.
In May 2025, Jitera formed a strategic partnership with leading consulting firm SIGMAXYZ to support digital transformation and AI development. By combining solutions from both companies, they aim to strengthen support for DX (digital transformation) and system development.
.png)
Japan is building its own path in AI. With strong research, careful innovation, and a growing number of startups, the country is quietly making progress. While many people look at big tech companies in the U.S. or Europe, Japan is showing it has something unique to offer. More partnerships with big companies and support from global investors are helping Japanese AI startups grow faster. From language models to tools for factories and healthcare, Japan’s AI is becoming more active and interesting. For anyone interested in the future of AI, Japan is a country to watch.
.png)
.png)
.png)
.png)