Fujitsu Research

Fujitsu Research is the artificial intelligence and computing research division of Fujitsu Limited, the Japanese information and communications technology company headquartered in Tokyo. Fujitsu Research operates research labs across Japan (Kawasaki, Tokyo), the United States (Sunnyvale, California; Richardson, Texas), the United Kingdom (London), China (Beijing), and India (Bengaluru), conducting research across artificial intelligence, quantum computing, high-performance computing, materials informatics, and other advanced-computing areas. As of April 2026, Fujitsu Research is one of the principal Japanese industrial AI research divisions, with integration into Fujitsu's broader product line including the Fugaku supercomputer (operated jointly with RIKEN), the Takane Japanese-language large-language-model line, and enterprise AI commercial deployments.

At a glance

• Founded: Roughly 1968 as Fujitsu Laboratories; restructured as Fujitsu Research in 2021.
• Status: Research division of Fujitsu Limited, a publicly traded company (Tokyo Stock Exchange: 6702).
• Funding: Operates within Fujitsu's overall research-and-development budget. Fujitsu reported approximately ¥125 billion (approximately $850 million) in research-and-development spending in fiscal 2024.
• Senior leadership: Hirotaka Hara, Executive Vice President and Head of Fujitsu Research.
• Other notable leadership: Senior research and engineering leadership across the principal lab locations.
• Open weights: Yes, partial. Selected Takane variants and other research code released through Hugging Face and GitHub.
• Flagship outputs: Takane (Japanese-language large-language-model line, with Takane V2 and other variants), AI research output across natural language processing, computer vision, optimization, and other areas. The Fugaku supercomputer (commissioned 2020 jointly with RIKEN; one of the principal Japanese supercomputers).

Origins

Fujitsu Research has its origins in the establishment of Fujitsu Laboratories in 1968 as the corporate research division of Fujitsu Limited. The lab's 1968 to 2020 history included contributions to Japanese semiconductor research, mainframe-and-supercomputer research (including the Fugaku predecessor K computer, commissioned 2011 with RIKEN), high-performance computing applications, and natural-language-processing and computer-vision research output.

The 2021 restructuring of Fujitsu Laboratories as Fujitsu Research consolidated the research division within Fujitsu Limited's broader corporate structure, with emphasis on artificial intelligence, quantum computing, and high-performance computing as principal research priorities.

The 2020 commissioning of Fugaku, the Japanese supercomputer operated jointly with RIKEN, anchored supercomputer-research credibility. Fugaku (using Fujitsu A64FX ARM-based compute) was the world's most powerful supercomputer from June 2020 through November 2021 on the TOP500 leaderboard.

The 2024 to 2026 period has continued AI research output, including the Takane Japanese-language large-language-model line. Takane V2 and other variants have been released open-weights through Hugging Face with Japanese-language capability.

Mission and strategy

Fujitsu Research's stated mission is to advance fundamental research in artificial intelligence and computing for Fujitsu's broader product line. The strategy combines four threads. First, AI research across natural language processing, computer vision, optimization, and other areas. Second, quantum computing research with published research output. Third, high-performance computing research including the Fugaku supercomputer joint operation with RIKEN. Fourth, materials informatics research with industrial applications.

Distribution channels include direct integration into Fujitsu's product line, published research output through major academic venues, the open-weights Takane distribution through Hugging Face, and commercial relationships across Japanese and international enterprise customers.

Models and products

• Takane. Japanese-language large-language-model line. Takane V2 and other variants. Selected variants released open-weights through Hugging Face.
• Fugaku supercomputer. Operated jointly with RIKEN since 2020 commissioning. One of the principal Japanese supercomputers.
• AI research output. Across natural language processing, computer vision, optimization, and other areas.
• Quantum computing research. Continued published research output.
• Materials informatics research. Industrial applications.

Distribution channels include direct integration into Fujitsu's product line, published research output, open-weights Takane distribution, and commercial relationships.

Benchmarks and standing

Fujitsu Research's evaluation framework combines academic-research output (publication count, citation impact) with commercial-product integration metrics. Takane has been characterized in Japanese-language NLP industry coverage as one of the Japanese-language foundation model lines from Japanese industrial-research divisions.

The Fugaku supercomputer has continued supercomputer-performance metrics through 2024 to 2026, although the 2022 to 2026 commissioning of US-based exascale-class supercomputers (Frontier, Aurora, El Capitan) has shifted the supercomputer-leaderboard landscape.

Leadership

As of April 2026, Fujitsu Research's senior leadership includes:

• Hirotaka Hara, Executive Vice President and Head of Fujitsu Research.
• Senior research and engineering leadership across the principal lab locations (Japan, US, UK, China, India).

Funding and backers

Fujitsu Research operates within Fujitsu Limited's overall research-and-development budget. Fujitsu reported approximately ¥125 billion in research-and-development spending in fiscal 2024.

Industry position

Fujitsu Research occupies a distinctive position as one of the principal Japanese industrial AI research divisions, with Takane Japanese-language foundation model line, the Fugaku supercomputer joint operation with RIKEN, and AI research output across the principal lab locations.

Competitive landscape

• NTT, Honda Research Institute, Toyota Research Institute, Sony AI, Preferred Networks, Sakana AI, Matsuo Lab (UTokyo). Japanese AI research peers.
• RIKEN. Japanese research-organization peer with Fugaku joint operation cooperation.
• Samsung Research / SAIT, LG AI Research, Naver, SK Telecom. Korean industrial AI research peers.
• Huawei Noah's Ark Lab. Chinese industrial AI research peer.
• Other Japanese industrial AI research divisions. Japanese corporate AI research peers.

Outlook

• Continued Takane Japanese-language large-language-model iteration.
• Continued AI research output across the principal lab locations.
• Continued Fugaku supercomputer joint operation with RIKEN.
• The continued Japanese government research-funding direction and Japanese AI national strategy.

Sources

• Fujitsu Research official site. Research division reference.
• Fujitsu Limited. Parent company reference.
• Takane on Hugging Face. Open-weights model distribution.
• RIKEN R-CCS / Fugaku. Fugaku supercomputer reference.
• Fujitsu Limited Annual Report. Public-market financial reporting.