Blackwell / AI Factory

NVIDIA B200 / GB200 for AI factories

B200/GB200 is usually sourced as a system configuration: server, NVLink domain, network, rack, power and cooling are verified before pricing.

Request priceView specification

Specs

Procurement specifications

The table helps qualify the request quickly. For a commercial offer we verify datasheet, OEM SKU, availability and warranty region.

ArchitectureBlackwell
Memory per GPU180 GB HBM3e
Memory bandwidthup to 8 TB/s per GPU class
System formatGB200 / HGX / NVL by platform
Coolingliquid-cooled rack designs are common
WorkloadsAI factory / large-scale training / inference
NVIDIA GB200 NVL72

Procurement profile

Where this model makes sense

Project-based supply for AI factories, hyperscale inference and large training clusters.

  • large LLM and inference factories;
  • GB200, HGX and NVL platforms;
  • clusters with InfiniBand/Ethernet fabric and NVLink.

RFQ

What to verify before quoting

  • platform: HGX, GB200, NVL72 or OEM system;
  • liquid cooling or reinforced thermal design;
  • power density, PDU, rack and data-center readiness;
  • networking, storage, management and support contract.

Critical limitation

NVIDIA B200 / GB200

B200 is not a simple PCIe card for a standard server. Engineering infrastructure, liquid or reinforced cooling and industrial power must be checked.

Specification sources

The table helps qualify the request quickly. For a commercial offer we verify datasheet, OEM SKU, availability and warranty region.

NVIDIA GB200 NVL72

Internal links

Related NVIDIA pages

These links keep model-level pages inside the current catalog architecture without overloading the global menu.

Data Center GPUs

NVIDIA B300 / GB300 Blackwell Ultra

 Data Center GPUs

NVIDIA DGX B200

 Data Center GPUs

NVIDIA H200 for LLM and AI infrastructure

Back to section

RFQ

Request price and availability

Attach SKU, BOM or model list. For server systems, include site, power, cooling, network and compatible server details.