AMD Helios secures $1.2B deployment with Vultr through HPE

AMD will power a $1.2 billion order placed by cloud infrastructure firm Vultr for AMD Helios AI Rack systems built by HPE. The deployment, announced on September 30, 2026, marks the first order for the AMD Helios platform and will roll out across Vultr data centers in the United States.

AMD designed the Helios system as a high-density architecture tailored for trillion-parameter model training and high-volume AI inference workloads. Each rack integrates 72 AMD Instinct MI455X GPUs, AMD EPYC Venice CPUs, AMD Pensando Vulcano AI NICs, and AMD ROCm software.

The setup uses open rack-scale fabrics, including UALink over Ethernet (UALoE). HPE completes each rack with six HPE Juniper Networking QFX5252 scale-up Ethernet switch trays, connecting the 72 AMD GPUs over standards-based Ethernet alongside direct liquid cooling.

The agreement expands on nearly three years of collaboration between Vultr and Juniper Networks, which is now part of HPE. Vultr plans to use the infrastructure to serve growing enterprise demand for AI training and inference capacity.