Fujitsu and NVIDIA partner to build full stack AI infrastructure for industry

Fujitsu and NVIDIA expanded their strategic collaboration to develop a "full-stack" artificial intelligence infrastructure, aiming to accelerate the adoption of AI agents across key industries like manufacturing, healthcare, and robotics. The partnership focuses on delivering a computing platform that overcomes the challenges of deploying sophisticated AI, such as high costs and technical hurdles, which have limited AI's benefits mostly to large enterprises.

The collaboration centres on two main areas: a co-developed AI agent platform and next-generation computing infrastructure.

A Platform for Self-Evolving AI Agents

The companies plan to build an AI agent platform designed for industry-specific applications. This platform will use Fujitsu's Kozuchi AI technology, including its workload orchestrator, with the NVIDIA Dynamo platform.

This structure intends to allow AI agents and models to continuously learn, evolve, and be customised for customer needs in specific sectors. For example, in manufacturing, the goal is to leverage 'physical AI,' including robotics and digital twins, to automate operations and address labour shortages. These developed AI agents will become available as NVIDIA NIM microservices to simplify and speed up AI adoption for customers. Fujitsu also plans to enhance its multi-AI agent technologies using NVIDIA NeMo, including optimising its Takane AI model.

Next-Generation Computing Infrastructure

A key part of the expanded collaboration is co-developing high-speed computing infrastructure. This system will directly integrate Fujitsu’s upcoming FUJITSU-MONAKA CPU series with NVIDIA’s high-performance GPUs. The connection will use NVIDIA NVLink-Fusion, a technology designed to tightly link third-party CPUs and NVIDIA GPUs.

The companies are working to create a silicon-level optimised AI computing platform intended for zetascale performance, making high-end AI processing more accessible to industry. Fujitsu will provide a full high-performance computing (HPC) and AI ecosystem by combining its high-speed software technology for Arm with NVIDIA CUDA, offering consolidated support for AI transformation.

Fujitsu CEO Takahito Tokita stated the collaboration will "accelerate AI-driven business transformation in enterprise and government sectors," initially targeting manufacturing, where Japan is a world leader. NVIDIA CEO Jensen Huang emphasised the significance of the partnership for the region, noting, "The AI industrial revolution has begun, and we must build the infrastructure to power it, in Japan and across the globe."

By realising a "human-AI co-creation cycle" and continuous system evolution, the companies aim to achieve transformative automation and intelligence across all sectors. This full-stack approach is projected to become an essential foundation for Japan's digital society by 2030, driving enterprise AI market growth and creating substantial social value.