LIVETracking 971 AI models from 268 providers.
Beta
NVIDIA logoNVIDIA

NVIDIA DGX GB200 NVL72

Full rackShippingBlackwell2025

NVIDIA's flagship liquid-cooled rack. 72 Blackwell GPUs + 36 Grace CPUs connected via NVLink 5.0 in a single 72-GPU domain. Designed for trillion-parameter model training. The most powerful commercially available AI system, requiring liquid cooling infrastructure and 120+ kW per rack.

72
GPUs per system
720 FP8 PFLOPS
All systemsCompute hub
GPU model
NB
NVIDIA B200
GPU count
72x
CPU model
NVIDIA Grace
36x
Memory type
H
HBM3e
Total HBM
13.8 TB
Host memory
17.3 TB
Interconnect
NVLink 5.0
130 TB/s
Networking
3200 Gbps
Storage
120 TB NVMe SSD
Form factor
Full rack (42U)
Weight
1,360 kg
Rack units
42U

Manufacturer datasheet values · aggregate system compute

FP4 PFLOPS1,440
FP8 PFLOPS720
FP16 PFLOPS360
BF16 PFLOPS360
Training effective PFLOPS540

Thermal envelope · cooling requirements · efficiency

Rack power
120 kW
Per GPU
1000 W
Cooling
liquid
PUE estimate
1.1
Power draw relative to tracked systems120 kW / 2500 kW max
6 FP8 PFLOPS per kW · average across all systems is 4.81

Hardware amortized over 3 years · power at $0.05/kWh

List price
$3,000,000
Per GPU effective
$41,667
Cost per GPU per month
$1,157
TCO per PFLOPS per year
$1,469
PFLOPS per kW
6
84% below the average TCO of $9,046/PFLOPS/year across all tracked systems
Available from
NVIDIA logoNVIDIA
D
Dell
H
HPE
S
Supermicro
L
Lenovo

Disclosed in press releases, SEC filings, and conference talks

Microsoft Azure logoMicrosoft Azure
Quantity
tens of thousands of GPUs
Source
Microsoft Build 2025
OC
Oracle Cloud
Quantity
65,000 B200 GPUs
Source
Oracle Q4 2025 earnings
CoreWeave logoCoreWeave
Quantity
large-scale deployment
Source
CoreWeave IPO S-1
xAI logoxAI
Quantity
Colossus cluster expansion
Source
xAI Memphis datacenter

Every data point on this page is reproducible

Compare across the system landscape

Google logoTPU v5p Pod
8960x Google TPU v5p · 8100 PFLOPS
Pod / clusterShipping
NVIDIA logoDGX GB300 NVL72
72x NVIDIA B300 · 1440 PFLOPS
Full rackAnnounced
Google logoTPU v6e Pod
256x Google TPU v6e · 230 PFLOPS
Pod / clusterShipping
Microsoft logoMaia 100 Rack
32x Microsoft Maia 100 · 96 PFLOPS
Full rackRamping
NVIDIA logoDGX B200
8x NVIDIA B200 · 80 PFLOPS
Server nodeShipping
AMD logoMI325X Platform
8x AMD MI325X · 48 PFLOPS
Server nodeShipping
Back to all systemsCompute hub