NVIDIA HGX B200

An HGX baseboard with eight Blackwell GPUs that system builders design into x86 servers. NVIDIA's own configuration is the DGX B200.

Specifications at a glance

As published by NVIDIA, checked 24 September 2026. Configurations vary by system builder.

System: 8 Blackwell SXM GPUs on one HGX baseboard, in x86 servers from system builders
GPU memory: 180 GB HBM3E per GPU; 1.4 TB per 8-GPU system
Memory bandwidth: 7.7 TB/s per GPU in NVIDIA's Blackwell datasheet; NVIDIA's DGX B200 materials state 8 TB/s
FP4 compute: 144 PFLOPS with sparsity, 72 PFLOPS dense, per 8-GPU system
Interconnect: Fifth-generation NVLink: 1.8 TB/s per GPU; ConnectX-7 at 400 Gb/s
Power: Configurable up to 1,000 W per GPU; about 14.3 kW maximum for a DGX B200 system
Cooling: DGX B200 is air-cooled; other HGX B200 servers follow their builder's design
Status: Shipping

HGX B200 servers vary by builder, so the server model defines the power, cooling and rack plan.

Each server's power and cooling follow the system builder's design.
NVIDIA's DGX B200 draws up to about 14.3 kW per system and is air-cooled.
Document the server model, the quantity and the supplier, or fleet records for deployed servers.

Document the configuration, the supplier and the deployment so the request can be assessed against the actual equipment.**

System configuration and quantity
Supplier and purchase order
Or fleet details and ownership information for deployed servers.
Deployment site and colocation agreement
Power and cooling readiness
Customer contract or documented on-demand revenue