GPU server for AI inference and training

Summary

The University of Osnabrück is procuring one fully assembled and tested GPU server for AI inference and training, equipped with four GPU accelerators, each with at least 140 GB of HBM memory, for installation in an existing 19-inch server rack. The server must be new and delivered to the university’s data centre by 10 December 2026. Separately priced options include on-site warranty service, an additional GPU where supported, and a dual-port 100 GbE network adapter; the total budget is capped at EUR 121,800 excluding VAT, including delivery and other ancillary costs.