Fujitsu to Offer Generative AI Platform for Enterprise Data
Fujitsu has announced the launch of the Fujitsu Cloud Service Generative AI Platform. The new service, which combines data confidentiality with the ease of use of the cloud, will be made available in Japan during fiscal year 2025, with a global rollout planned for the future. Fujitsu will begin accepting applications for a trial of this service in Japan starting February.
Generative AI offers significant productivity gains across various business functions from brainstorming and meeting minute generation to automated code creation. However, its adoption in sectors handling sensitive data has been limited due to concerns about data leakage, unintended AI learning, and compliance with internal regulations. Fujitsu’s new platform directly addresses these challenges.
To ensure data security and compliance, the service combines Fsas Technologies Inc.’s Private AI Platform on PRIMERGY, Supermicro’s GPU servers, and Fujitsu’s Takane large language model (LLM) which is the core technology of the Fujitsu Kozuchi AI service and renowned for its world-leading Japanese language capabilities, as well as generative AI security enhancement technologies to offer an on-demand cloud-based generative AI solution through Fujitsu’s data centers. This architecture manages client data in a private area on the cloud in Fujitsu’s data center, allowing for the secure processing of confidential information. In addition to Takane, Fujitsu plans to gradually increase the number of available LLMs from other companies to meet customer needs.
Fujitsu will add the new service to its Fujitsu Computing as a Service (1) portfolio. Fujitsu will provide comprehensive support, from initial implementation to ongoing optimization, enabling businesses to maximize the efficiency and productivity gains of generative AI and drive significant business transformation.
Key features:
1.Sensitive data management for privately
Fine-tuning data, fine-tuned LLMs, and data used for RAG (2) are stored in dedicated private cloud areas for each client. This approach safeguards sensitive information, preventing unintended data usage and ensuring compliance with internal regulations and economic security concerns.
2.Cost-effective shared GPU resources
This service reduces the cost per hour of use by providing a shared GPU server for inference and training while securing data in a private area. This standardized and shared infrastructure significantly reduces initial deployment costs compared to building a dedicated generative AI environment.
For on-premise solutions, Fsas Technologies Inc.’s Private AI Platform on PRIMERGY and Supermicro GPU servers, along with system integration and operational support.
3.Secure on-demand access to generative AI through Fujitsu’ s generative AI security enhancement technology and cloud security measures
The adoption of Fujitsu’s generative AI security enhancement technology allows users to safely and securely use this service. For example, it prevents users from unintentionally accessing unauthorized information via the LLM.
