GPU-accelerated AI server

Summary

The tender concerns the supply of one fully assembled, tested and operational GPU-accelerated server or workstation server for large-scale artificial intelligence models, including language models, multimodal models and inference services. The system must support four professional GPUs and meet detailed requirements for the processor, memory, storage, networking, power supply, cooling, Linux and AI software frameworks. Delivery and invoicing must take place by 1 December 2026 at the latest. The bidder must submit a company profile, turnover for the last two financial years, one comparable reference no more than two years old, an exclusion-grounds declaration, and a compliant technical and financial offer. Price carries 70% of the evaluation weight and delivery time 30%.