NVIDIA GB200 NVL72 Rack System
Brand: NVIDIA | Category: Servers
SKU: NVIDIA-GB200-NVL72 | Part #: GB200-NVL72 | MPN: GB200-NVL72
Contact for Pricing — Request a Quote
About the NVIDIA GB200 NVL72 Rack System
The NVIDIA GB200-NVL72 is a 2U rack-mounted system engineered for data centre deployments requiring massive GPU acceleration in a compact footprint. This configuration pairs 72 × Blackwell B200 GPUs with your choice of Intel or AMD processors, delivering 13.8 TB of HBM3e memory and 1374.7 TBps aggregate memory bandwidth. Built to the EIA-310 19-inch rack standard with redundant hot-swap power supplies and liquid cooling with passive backplates, this system is purpose-built for AI infrastructure teams and network engineers preparing next-generation compute clusters.
The GB200-NVL72 (part number GB200-NVL72) addresses the most demanding workloads in machine learning, scientific simulation, and large-scale data processing. All 72 GPUs connect via NVLink 4.0 for all-to-all 100 Gbps interconnectivity, ensuring minimal latency between accelerators. Configurable hot-swap NVMe, SAS, and SATA storage options, paired with industry-standard management interfaces (iDRAC, iLO, XCC, or IPMI 2.0), enable seamless integration into existing enterprise environments. The system operates within 10–45 °C and supports configurable warranty options including 3-year ProSupport, Foundation Care, and Premier plans. Whether you are evaluating single-unit performance or planning multi-rack deployments, NVIDIA delivers the architecture and flexibility your infrastructure requires.
Key Specifications
- Manufacturer: NVIDIA
- Part Number: GB200-NVL72
- Form Factor: 2U Rack (EIA-310 19-inch)
- GPU Count: 72 × Blackwell B200
- Total GPU Memory: 13.8 TB HBM3e
- Aggregate Memory Bandwidth: 1374.7 TBps
- Network Interconnect: NVLink 4.0 (all-to-all 100 Gbps)
- Power Redundancy: Redundant hot-swap PSU
- Cooling: Liquid-cooled with passive backplates
- Operating Temperature Range: 10–45 °C
Contact Omnixon Global to request a quotation and configure the NVIDIA GB200-NVL72 for your enterprise infrastructure.
Technical Specifications
| Brand | NVIDIA |
| Category | Servers |
| SKU | NVIDIA-GB200-NVL72 |
| Part Number | GB200-NVL72 |
| Condition | New |
| Form Factor | Rack-Mount |
| Interface | NVLink |
| Chassis Standard | EIA-310 19-inch rack |
| PSU Redundancy | Redundant hot-swap supported |
| Drives | Hot-swap NVMe / SAS / SATA (configurable) |
| Management | iDRAC / iLO / XCC / IPMI 2.0 (vendor-specific) |
| Warranty | 3-year ProSupport / Foundation Care / Premier (configurable) |
| Manufacturer Part Number | GB200-NVL72 |
| Form Factor | 2U Rack |
| Rack Standard | EIA-310 19-inch |
| Rack Units | 2U |
| GPU Count | 72 x Blackwell B200 |
| Total GPU Memory | 13.8 TB HBM3e |
| Aggregate Memory Bandwidth | 1374.7 TBps |
| Processor | Intel or AMD (configurable) |
| Network Interconnect | NVLink 4.0 (all-to-all 100 Gbps) |
| Power Redundancy | Redundant hot-swap PSU |
| Cooling | Liquid-cooled with passive backplates |
| Operating Temperature Range | 10–45 °C |
Frequently Asked Questions about NVIDIA GB200 NVL72 Rack System
Can you ship the NVIDIA GB200 NVL72 Rack System configured to my BOM?
Yes. Submit your CTO BOM (CPU SKU, memory layout, drives, NIC, GPU, OS preload) in the RFQ and we return a configured-to-order quote with realistic lead time. Standard CTO turn is 4-6 weeks vendor-side.
Do you offer Dell ProSupport / HPE Foundation Care alongside the hardware?
Yes — we attach the matching support contract during quoting. Available tiers: ProSupport / ProSupport Plus / Mission Critical (Dell), Foundation Care / Proactive Care 24x7 (HPE), Premier (Lenovo). Recommend 3y minimum for production.
Is rack installation available in Dubai / GCC?
Yes for Dubai (same-week scheduling) and on quotation for the broader GCC via certified partners. Includes rack-and-stack, cable management, OS install, and post-install verification.
How does pricing compare to going direct to the OEM?
For SMB and mid-market quantities, channel pricing is typically equal to or better than direct OEM, with faster quote turn (24h vs 2-3 weeks). For multi-million-dollar projects, we partner with the OEM directly and pass through a transparent margin.