Any NVIDIA platform
Rack scale or node by node, from one unit up. We buy it against your contract, or you bring it. Either way we place it in Santiago or in the United States and run it for three to five years. Pick the platform, the place and the quantity.
Rack scale
Seventy-two GPUs in one NVLink domain, liquid cooled. The unit is the rack. Positions are reserved at 140 kW and protected to 155 kW.
| Platform | GPUs | GPU memory | NVLink | Power | Cooling | Smallest block | Available from | |
|---|---|---|---|---|---|---|---|---|
| GB300 NVL72Rack scale | 72per rack | 20 TB288 GB per GPU | 130 TB/s1.8 TB/s per GPU | 132 to 140 kW155 kW peak | Liquid | 1 rack, 72 GPUs | Q1 2027, SantiagoOn contract. US on request | Reserve |
| GB200 NVL72Rack scale | 72per rack | 13.4 TB186 GB per GPU | 130 TB/s1.8 TB/s per GPU | 120 to 132 kWper rack | Liquid | 1 rack, 72 GPUs | Q1 2027, SantiagoOn contract. US on request | Reserve |
| Vera RubinRack scale | 72per rack | HBM4not yet published | Not published | Not published | Liquid | 1 rack | On request | Reserve |
HGX and RTX PRO nodes
Eight GPUs in a node, from one node up. Air-cooled nodes at ten kilowatts or less fit an existing hall as it stands, in Santiago or in the United States.
| Platform | GPUs | GPU memory | NVLink | Power | Cooling | Smallest block | Available from | |
|---|---|---|---|---|---|---|---|---|
| HGX B3008-GPU node | 8per node | 2.3 TB288 GB per GPU | 14.4 TB/s1.8 TB/s per GPU | About 14 kWper node | Air or liquid | 1 node, 8 GPUs | December 2026, SantiagoJanuary 2027, US. On contract | Reserve |
| HGX B2008-GPU node | 8per node | 1.4 TB180 GB per GPU | 14.4 TB/s1.8 TB/s per GPU | About 14.3 kWper node | Air or liquid | 1 node, 8 GPUs | December 2026, SantiagoJanuary 2027, US. On contract | Reserve |
| HGX H2008-GPU node | 8per node | 1.1 TB141 GB per GPU | 7.2 TB/s900 GB/s per GPU | About 10.2 kWper node | Air | 1 node, 8 GPUs | Q1 2027, SantiagoOn contract. US on request | Reserve |
| HGX H1008-GPU node | 8per node | 640 GB80 GB per GPU | 7.2 TB/s900 GB/s per GPU | About 10.2 kWper node | Air | 1 node, 8 GPUs | Q1 2027, SantiagoOn contract. US on request | Reserve |
| RTX PRO 6000 Blackwell8-GPU node | 8per node | 768 GB96 GB per GPU | PCIe Gen 5no NVLink | 5.4 to 6.7 kWper node | Air | 1 node, 8 GPUs | Q1 2027, SantiagoOn contract. US on request | Reserve |
NVIDIA's published figures. Tensor Core numbers carry sparsity unless the row says dense; H100, H200 and RTX PRO figures are per GPU because NVIDIA publishes no node total. Power figures are the ones we reserve and protect to, with NVIDIA's own where it publishes one.
The same contract on every platform
What a reservation carries, whichever row you pick.
| Hardware | Ours, bought against your contract, or yours, brought in and operated by us. Say which in the form. |
| Place | Santiago, Chile, or the United States. Say which in the form, or let the workload decide. |
| Term | Three, four or five years. Five is the standard term. |
| Price | Fixed for the term, set in the term sheet. Not published. |
| Commitment | Take-or-pay. The capacity is yours for the term whether you use it or not, and nobody else's. |
| Prepayment | From none to a quarter of the contract at signature. It shapes the price and the term. |
| Start | On contract. HGX B200 and B300 nodes from December 2026 in Santiago and January 2027 in the United States. GB300 NVL72 racks from Q1 2027 in Santiago; in the United States the start is set in the term sheet. Later slots in 2027 and 2028 follow the planned hall and campus. |
| Smallest block | One node of eight GPUs on HGX platforms. One rack of 72 GPUs on NVL72. |
| Included | The hardware, the fabric, the storage you specify, Slurm and managed Kubernetes, commissioning, acceptance, operations and on-call for the term. |
| Yours | The workload, the images, the data, the keys. |
How it works
- Mutual NDASigned within the week
- Technical callWorkload, storage, acceptance
- Term sheetTerm, price, prepayment, start
- ContractTake-or-pay, three to five years
- Cluster liveYours for the term