GPU Hosts

Accelerator-dense servers for machine learning, MLOps, and other workloads that require dedicated GPU capacity.

• GPU · 2× 2U

  Two accelerators in a wide chassis. Your first taste of the MLOps customer market.

• GPU · 4× 4U

  Four accelerators per box. Where serious ML training workloads start.

• GPU · 8× 7U

  Eight accelerators in a taller, hotter chassis than the 6U — more cores, more memory, fatter uplinks. The top of the GPU ladder.