エフサスとPFN、国産LLM対応のオンプレミスAI基盤を検証完了

On July 14, Fujitsu Technologies and Preferred Networks announced that they have completed operational verification of PFN's domestically developed language models on Fujitsu's Private AI Platform on PRIMERGY, an on-premises generative AI infrastructure platform. The two models verified are PLaMo 2.2 Prime, a general-purpose large language model, and LLM PLaMo for Japanese translation, which specializes in Japanese-language translation tasks. With this verification complete, customers can now deploy these Japanese-language AI models within their own on-premises environments.

Fujitsu has been providing Private AI Platform on PRIMERGY to government agencies, research institutions, financial institutions, and manufacturing companies—sectors where high security requirements and the need for stable operations independent of overseas conditions are critical. Under the partnership structure, Fujitsu supplies the hardware and provides nationwide maintenance support, while PFN supplies the language models and their usage licenses. This arrangement allows organizations to begin using secure, Japanese-developed language models in their on-premises environments immediately upon hardware delivery, without relying on external cloud services.

Both verified models are designed to safely process highly sensitive information—administrative documents, research data, customer information, and production data—without transmitting it outside the organization. PFN is applying its knowledge of generative AI and AI model deployment to ensure high-speed inference and stable operation of the language models within the execution environment. Looking ahead, Fujitsu and PFN plan to expand this offering beyond government, research, financial, and manufacturing sectors to include social infrastructure, energy, and healthcare—fields that demand both sophisticated AI capabilities and rigorous information governance.

Fujitsu and Preferred Networks have aligned their strengths to address a critical need in Japan's highly regulated sectors. Fujitsu brings its established relationships with government, financial, and manufacturing customers alongside its PRIMERGY hardware platform and nationwide support infrastructure. PFN contributes expertise in generative AI and language models, having developed the PLaMo series specifically for domestic use. The verification completion signals that the integration is production-ready, enabling organizations to deploy Japanese-language AI capabilities entirely on their own infrastructure.

The timing reflects growing awareness among Japan's institutional buyers that data sovereignty and independence from overseas AI platforms matter for competitive and national-security reasons. By pairing Fujitsu's trusted hardware delivery with PFN's Japanese LLM and licensing model, both companies position themselves as enabling secure, domestically controlled AI for sectors where confidentiality is mandatory—administration, finance, manufacturing, and soon infrastructure and healthcare. The division of labor (Fujitsu hardware + support; PFN model + licensing) also clarifies commercial responsibility and simplifies deployment for customers requiring certified security.

No comments yet. Be the first to share your thoughts!