AMS-XGS

One AI fabric. Multiple campuses.
AMS-XGS connects distributed AI data center campuses across the Amsterdam region through high-bandwidth, low-latency infrastructure — enabling GPU clusters in different locations to operate as one coordinated compute environment.

Built around the Spectrum-XGS scale-across concept, the architecture extends AI networking beyond a single data center. It combines high-performance Ethernet, optical interconnects, congestion control and telemetry to improve GPU-to-GPU communication across distributed sites, while keeping latency and network contention predictable for large-scale AI workloads.