HPE has booked its first order for AMD's Helios rack-scale AI system: a $1.2 billion deal with cloud provider Vultr, announced September 30 in Sunnyvale. Vultr will deploy AMD Helios AI Rack by HPE systems, equipped with purpose-built HPE Networking scale-up switching, at its U.S. data center locations to serve model training and inference workloads for service provider and enterprise customers.

The order is a concrete proof point for HPE's integrated strategy following its acquisition of Juniper Networks. The Vultr relationship itself builds on nearly three years of collaboration with Juniper that, according to HPE, continues to expand now that Juniper sits inside the company. The deal pairs HPE compute and rack-scale engineering with Juniper-derived switching, exascale liquid cooling, and HPE's global services organization — a stack HPE pitches as an open, Ethernet-based alternative in the scale-up AI market.

"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."

The Helios rack is a dense, standardized unit aimed at trillion-parameter training and high-volume inference. Each rack integrates 72 AMD Instinct MI455X GPUs paired with AMD EPYC "Venice" CPUs, AMD Pensando Vulcano AI NICs for networking, and the AMD ROCm software stack. AMD says the unified design targets power, cooling, and serviceability while supporting open, interoperable rack-scale fabrics such as UALink over Ethernet (UALoE).

HPE's differentiation shows up in the fabric. Each Helios rack ships with six HPE Juniper Networking QFX5252 scale-up Ethernet switch trays, connecting all 72 GPUs over high-bandwidth, low-latency, standards-based Ethernet. That matters commercially: Helios by HPE gives AMD a complete rack-scale reference design with an established channel, while giving HPE a vehicle to sell Juniper silicon into AI clusters rather than only into campus and service provider networks. HPE Services adds end-to-end deployment and liquid-cooling expertise for large AI clusters, which the company positions as a way to reduce project and operational risk.

Vultr's motivation is capacity, plain and simple. "Demand for high-performance AI infrastructure continues to outpace available capacity, and our work with AMD and HPE enables us to bring Helios rack-scale compute and scale-up networking online for customers faster," said J.J. Kardwell, CEO of Vultr. "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 AMD Helios AI Rack by HPE sits within the HPE AI Data Center Solutions portfolio — a production-ready infrastructure stack co-engineered with AMD on open standards and tuned, in HPE's words, for the demanding economics and scale of the AI lifecycle. Vultr's deployment counts as the first customer win for that stack.

For AMD, the deal extends its challenge to Nvidia's dominance in rack-scale AI systems with a design that leans on open interconnect standards rather than proprietary fabrics. For HPE, the $1.2 billion order validates the Juniper acquisition thesis in the highest-growth segment of the infrastructure market. HPE did not disclose deployment timelines or rack counts, so actual shipment schedules remain to be seen — but with Vultr citing demand that outruns supply, competition for confirmed AI rack orders is likely to intensify as AMD's Helios-based systems ramp against established alternatives.