Cluster · NVIDIA · rack
NVIDIA GB200 NVL72
72 Blackwell GPUs + 36 Grace CPUs in one liquid-cooled rack, sharing a single NVLink 5 domain. The current flagship training rack.
Key facts
- Chips per instance
- 36
- Power (typical)
- 132 kW
- Interconnect
- NVLink 5 (1.8 TB/s per GPU, 130 TB/s aggregate)
- Cooling
- water
- Announced
- 2024-03-18
- GA / release
- 2024-12-01
Chips inside
Where you'll find it
Data centers publicly running the chips inside this design. Per-rack counts are rarely disclosed, so these are presence rows: this campus runs these chips.
- Meta Prometheus Data Center (New Albany) · NVIDIA Blackwell training capacity at the 1 GW Prometheus site
- Cologix Johnstown Campus · NVIDIA Blackwell capacity for AI-training tenants at the Johnstown campus
- Fermi America HyperGrid (Project Matador) · Planned Blackwell capacity for the Fermi HyperGrid campus (2 GW target, four planned SMRs)
- HUMAIN (Riyadh & Dammam flagship data centers) · Up to 500,000 Blackwell GPUs committed across the HUMAIN Riyadh/Dammam campuses (multi-year)
- Homer City Energy Campus · Planned 4.5 GW gas-fed AI campus targeting NVIDIA Blackwell tenants
- Stargate Abilene · GB200 NVL72 racks part of the Blackwell buildout
- Microsoft Cheyenne Data Center Campus · Blackwell-generation training capacity at Cheyenne
- Microsoft Fairwater Atlanta / QTS Fayetteville Campus · Second Fairwater campus targeting Blackwell-generation compute
- Microsoft Fairwater Data Center (Mount Pleasant) · Blackwell-generation AI datacenter (Fairwater flagship); Microsoft's next-generation closed-loop cooled AI DC design
- Project Kilby (Chevron / Microsoft) · Microsoft-anchored AI capacity at the Chevron gas-turbine-fed campus
- DataVolt NEOM Oxagon Net-Zero AI Factory · 1.5 GW zero-carbon AI factory targeting NVIDIA Blackwell tenants
Vendor page: https://www.nvidia.com/en-us/data-center/gb200-nvl72/
See also: all rack designs · chip catalog · NVIDIA.