Tokyo, February 14, 2024 – Fujitsu today announced details of a new company-wide AI strategy that focuses on deepening human-AI collaboration, presenting a vision of AI as a trusted assistant that supports and augments productivity and creativity.

As the first major step of this new strategy, Fujitsu plans to offer Fujitsu Data Intelligence PaaS as part of its Fujitsu Uvance portfolio and launch consulting services for AI and related technologies starting March 2024 for users in Japan and from April 2024 to the global market.

Fujitsu will further commercialize Fujitsu Kozuchi, Fujitsu’s dedicated AI platform that allows users to accelerate the testing and deployment of advanced AI technologies, as one of its key AI commercial offerings and will focus on the integration of AI capabilities into solutions for the four cross-industry vertical areas of Fujitsu Uvance.

In addition to Fujitsu Kozuchi, Fujitsu Data Intelligence PaaS will also incorporate Fujitsu Track and Trust, Fujitsu’s blockchain offering that enables data linkage and traceability across companies and industries, and support for data platforms including Palantir and Microsoft Azure.

As a global leader in AI, Fujitsu has been promoting the research and development of innovative AI and machine learning technologies for over 30 years, and continues to embrace recent breakthroughs in areas like generative AI. To date, Fujitsu boasts a track record of more than 7,000 AI customer use cases in fields including manufacturing, retail, healthcare, public safety and more.

Fujitsu’s AI strategy: past milestones and the road ahead

In addition to R&D on a distributed training method for LLM using supercomputer Fugaku (1), Fujitsu in December 2023 further developed a special LLM model based on Open LLM (2) to improve Japanese language performance, achieving top-class outcomes against key performance benchmarks (3).

Fujitsu is also currently developing its own generative AI technology specializing in image and code generation, as well as generative AI mixture technology that integrates Fujitsu’s generative AI models with generative AI models from partner companies without the need to retrain models. Fujitsu confirmed that the accuracy of these generative AI models is equal to or higher than that of state-of-the-art generative AI models including GPT-4V and HuggingGPT, and further confirmed improved cost effectiveness.

Fujitsu is further linking its own knowledge graph technology that enables the verification of relationships across more than 1 billion nodes and generative AI with trustable internet technology, which enables users to determine the reliability of data. By combining these technologies, Fujitsu will work on the development of a generative AI trust technology that eliminates output instability, an ongoing task in generative AI, and ensures accurate output in compliance with large-scale and complex legal and internal regulations.

Starting April 2024, Fujitsu will offer these generative AI mixture technologies, the generative AI trust technology, and LLMs that can be used securely in local environments to users via the Fujitsu Research Portal, an online resource that that provides registered users free-of-charge access to trial versions of Fujitsu’s advanced technologies.

To encourage creative experimentation and further development of AI technologies, Fujitsu has also created an environment in which all of its 124,000 global employees can securely use generative AI. Since May 2023, Fujitsu has been actively promoting internal use in areas including programming code generation and internal report generation. These in-house use cases will allow developers to explore the many potential uses of generative AI and ultimately lead the way to the commercialization of new solutions for its customers.