peopleimages.com - stock.adobe.c
Fujitsu targets private AI with Japanese LLM on Nutanix
Fujitsu’s Takane large language model, optimised for Japanese language and business use, is aimed at supporting private AI deployments
Fujitsu has teamed up with Nutanix to make its Japanese language-optimised large language model (LLM), Takane, available on the Nutanix Enterprise AI (NAI) platform.
The validation means Takane is now a certified LLM for deployment via NAI, which supports generative AI (GenAI) models both on-premise and in public clouds, as well as on the Nutanix Cloud Platform (NCP) hybrid multicloud platform.
Fujitsu claimed that Takane offers superior performance for Japanese language tasks compared with general-purpose LLMs, addressing specific linguistic challenges such as mixed character sets (hiragana, katakana and kanji), omitted subjects, and nuanced honorific expressions common in Japanese business communication.
The move targets enterprises, particularly in Japan, that are looking to leverage GenAI but are facing challenges with the accuracy of global models or needing to deploy AI workloads in private environments due to data sensitivity, regulatory compliance or latency requirements.
This marks the first time a Japanese-language-enhanced LLM has been validated for Nutanix’s AI platform.
The Takane model running on NAI will initially be offered through Fujitsu’s PrimeFlex integrated virtualisation platform for Nutanix. Fujitsu said this will support the creation of reliable AI infrastructure for organisations where public cloud use is restricted.
Addressing the need for hybrid deployments, where development might occur in the public cloud but production runs on-premise or at the edge, Fujitsu noted that compatibility with NAI and NCP simplifies the migration and integrated management of AI applications and data across these different environments.
Read more about AI in APAC
- Singapore researchers are using Google DeepMind’s AlphaFold to understand the intricate protein interactions at play in Parkinson’s disease, paving the way for potential diagnostics and treatments.
- Southeast Asian super app Grab names AWS as its preferred cloud provider to optimise costs, enhance infrastructure and support its growing AI workloads.
- India has earmarked significant investments in AI, global capability centres and digital public infrastructure while boosting training and education in its latest budget.
- Australia’s utilities sector is exploring and implementing AI to enhance grid stability, manage rooftop solar and prepare for the influx of electric vehicles.
Fujitsu plans to offer Takane on NAI as a managed service starting in July 2025. This will be complemented by its existing Fujitsu Cloud Managed Service offering, designed to optimise and manage hybrid cloud operations, including environments running Takane on NCP.
The company positioned the offering as part of its Fujitsu Uvance business initiative, which focuses on solving societal and business challenges, particularly in high-security sectors such as finance, government, and research and development, where Takane’s capabilities are expected to be most beneficial in private or hybrid cloud settings.
Japan’s sovereign AI push
The Fujitsu-Nutanix collaboration comes on the heels of a significant push by Japan to develop domestic LLM capabilities. Driven by linguistic needs, data sovereignty concerns and national economic strategy, Japanese tech giants and research institutions have been investing in foundation models trained predominantly on Japanese data.
Notable efforts include NTT’s “tsuzumi” LLM and SoftBank’s investments in GenAI development, including a joint venture with OpenAI to develop AI agents tailored to the unique needs of Japanese enterprises.
Last December, NEC launched its enhanced “cotomi” model, which has been touted for its accuracy and high performance in Japanese language benchmarks. In addition, the company claimed to have developed technologies that can double the computational efficiency of graphics processing units while maintaining the performance of GenAI.
The Japanese government has also signalled strong support for building a domestic AI ecosystem, viewing it as crucial for the country’s competitiveness and ensuring AI models align with Japanese cultural and business contexts.
Fujitsu’s move to make Takane deployable on an enterprise hybrid cloud platform like Nutanix can be seen as a key step in operationalising these domestic AI ambitions, providing Japanese enterprises with options for leveraging GenAI securely in their own infrastructure.
