Dedicated NVIDIA Blackwell GPU clusters for training and inference.
Both are built to order on power Helios already holds. Reserve a cluster, or bring your own hardware.
GPU Cloud
Dedicated NVIDIA Blackwell, reserved for one to five years. Your GPU, node count and orchestration.
USD, 3-year reservation. Clusters are built for you, not drawn from stock.
Colocation
Prefabricated Helios modules with power, liquid cooling and connectivity built in.
Rates depend on cooling and density. Estimate yours in the colocation calculator.
How it works
Energy is the longest lead time in any data center. Helios reserves it ahead of demand, then builds a GPU cluster or a colocation hall on that power to your specification.
Washington, Wasatch Front & Vernal
946MW reserved
across 7 sites
Long-term supply agreements across 6 states, reserved before a customer signs.
A custom cluster or hall on prefabricated 2.5 MW modules, sized to the workload.
From contract to compute, subject to configuration and site readiness.
A target of about three months.
Timing depends on configuration and site readiness. Confirm the schedule in your agreement.
Confirm your capacity, hardware and site requirements. Set the delivery schedule in your agreement.
Plan power, cooling and connectivity around the hardware and its operating requirements.
Prepare the facility and bring the electrical, cooling and network systems together.
Commission the infrastructure, complete acceptance checks and hand over the deployment.
Energy supply and cooling.
Planned alongside your compute.