HPE가 Vultr와의 $1.2 billion 규모 거래에서 첫 AMD Helios 주문을 확보했으며, Vultr의 미국 데이터센터에 HPE 네트워킹이 탑재된 차세대 AMD Helios AI Racks를 배포한다.

Vultr has placed a $1.2 billion order with HPE to deploy AMD Helios AI Rack systems across its U.S. data centers, marking HPE's first customer order for the new platform. The systems will support Vultr's service provider and enterprise customers running model training and inference workloads.

Each rack integrates 72 AMD Instinct MI455X GPUs paired with AMD EPYC "Venice" CPUs, AMD Pensando Vulcano AI NICs, and AMD ROCm software. HPE provides purpose-built networking through six Juniper QFX5252 scale-up Ethernet switch trays per rack, delivering high-bandwidth, low-latency connectivity to all 72 GPUs using standards-based Ethernet. The design supports open, interoperable rack-scale fabrics including UALink over Ethernet (UALoE). HPE's exascale liquid cooling and end-to-end deployment services manage the operational complexity of large AI clusters.

"AI is driving the largest infrastructure buildout in history and together with AMD and Vultr, we are building an open foundation to accelerate the next phase of AI," said Antonio Neri, president and CEO of HPE. "AMD Helios by HPE demonstrates HPE's unique ability to engineer the complete AI architecture with rack-scale compute, scale-up networking, and direct liquid cooling, all delivered by our global services."

Vultr's CEO J.J. Kardwell noted that demand for high-performance AI infrastructure continues to outpace available capacity. "Our work with AMD and HPE enables us to bring Helios rack-scale compute and scale-up networking online for customers faster. Together, we're delivering the memory capacity, interconnect performance, and energy efficiency customers need to run the largest and most demanding AI training and inference workloads."

The order builds on nearly three years of collaboration between Vultr and Juniper Networks, which became part of HPE in 2024. The AMD Helios AI Rack is engineered as a high-density system for trillion-parameter model training and high-volume inference, designed using open standards and optimized for the economics and scale of production AI workloads.