NVIDIA GB200 NVL72
The first rack-scale Blackwell system. Same 72-GPU NVLink domain, lower power.
- 72 GPUs per rack
- 13.4 TB GPU memory
- Liquid cooled rack
- Available Q1 2027
Configurations
Pick the footprint that fits. Every one is sold on the same contract.
- 72 GPUsOne rack
- One NVLink domain. The smallest unit.
- 144 to 792 GPUsTwo to eleven racks
- Joined by an InfiniBand or Spectrum-X fabric.
- 864 GPUsA full hall
- Twelve racks in one existing Santiago hall.
Specifications
- GPUs
- 72 Blackwell per rack
- GPU memory
- 13.4 TB, 186 GB HBM3e per GPU
- NVLink
- 1.8 TB/s per GPU, 130 TB/s domain
- CPUs
- 36 Grace, 2,592 Arm cores
- System memory
- 17 TB LPDDR5X
- Networking
- ConnectX-7, 400 Gb/s per GPU; BlueField-3
- Power
- 120 to 132 kW
- Cooling
- Liquid cooled rack
- Fabric at Fulmis
- InfiniBand or Spectrum-X Ethernet, sized to the workload
- Software
- Slurm and managed Kubernetes
- Location
- A data centre in the Santiago Metropolitan Region
- Available
- Q1 2027
Performance
Per rack, as NVIDIA publishes it.
- FP4 Tensor, dense
- 720 PFLOPS
- FP8 / FP6 Tensor, dense
- 360 PFLOPS
- FP16 / BF16 Tensor
- 360 PFLOPS
- TF32 Tensor
- 180 PFLOPS
- GPU memory
- 13.4 TB HBM3e, 576 TB/s
- Fast memory, GPU and CPU
- 30 TB
- NVLink bandwidth
- 130 TB/s
- Scale-out network
- ConnectX-7, 400 Gb/s per GPU
- Rack power
- 120 to 132 kW; NVIDIA's rack guide says about 120 kW
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.
How we deploy it
GB200 NVL72 is the first-generation rack-scale Blackwell system, with the same 72-GPU NVLink domain as GB300 and a lower power draw. It is the right platform where a workload is already tuned to Blackwell and the Ultra memory is not needed.
A rack is the smallest unit. We quote it in racks, with the fabric, storage and software sized to the cluster you describe.
Fulmis places the purchase order, or takes delivery of the fleet you already own. Either way we bring it into Chile, commission it against the vendor baseline and operate it for the term. You reserve the capacity, or your own hardware; the workload and the data stay yours.
Terms
- Term
- Three to five years
- Price
- Fixed for the term, in the term sheet
- Commitment
- Take-or-pay
- Prepayment
- None to 25% at signature
- Smallest block
- One rack, 72 GPUs
- Start
- Q1 2027, first slot
Other platforms
| Platform | GPUs | GPU memory | NVLink | Power | Cooling | Smallest block | Available |
|---|---|---|---|---|---|---|---|
| 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 |
| 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 |
| 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 |
| 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 |
| 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 |
| 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 |
| Vera RubinRack scale | 72per rack | HBM4not yet published | Not published | Not published | Liquid | 1 rack | On request |