Expandable GPU Server for Local LLM Inference and Agentic AI

Summary

Fraunhofer SCAI is procuring one fully integrated GPU server for local large language model inference and research and development in agentic AI, including confidential industrial and medical data. The server will initially include four NVIDIA H200 NVL GPUs and must be manufacturer-approved and technically prepared for expansion to eight; the later supply of additional GPUs is not included. The procurement also includes delivery and 36 months of hardware service.