Nexus GB200 NVL72 Blackwell Rack-Scale Cluster
72 Blackwell GPUs as one giant GPU — exascale-class AI in a single rack.
We help you choose, configure, and deliver the right system — no obligation.
Configuration at a Glance
Tailored per engagement. Full technical overview below.
Configuration Options
Core specifications for this system. Every component is configurable to your workload — request a quote for a tailored build.
72× NVIDIA Blackwell B200 GPUs + 36× Grace CPUs (GB200 NVL72)
NVLink-C2C at 900GB/s per Grace Blackwell superchip
13.4TB unified HBM3e GPU + 17.3TB LPDDR5X CPU per rack
NVIDIA-certified parallel filesystem, GPUDirect Storage
Overview
The GB200 NVL72 connects 36 Grace CPUs and 72 Blackwell GPUs through fifth-generation NVLink into a single 72-GPU domain inside one liquid-cooled rack. Nexus Compute specifies, sources through authorized channels, integrates, and acceptance-tests the rack, XDR-ready fabric, and storage as one warranty-backed system for organizations planning frontier AI build-outs.
Who This Solution Is For
Business Benefits
One rack, one giant GPU
A 72-GPU NVLink domain behaves as a single accelerator, collapsing inter-node bottlenecks for trillion-parameter training and inference.
Real-time at trillion scale
Fifth-generation NVLink delivers dramatically faster real-time inference on the largest models versus prior-generation multi-node designs.
Roadmap-aligned sourcing
We align Blackwell procurement timing, facility liquid-cooling readiness, and staged delivery with your project roadmap and allocation.
Typical Business Use Cases
Trillion-parameter model training and fine-tuning
Real-time inference for very large generative models
Next-generation AI factory build-outs
Frontier research demanding peak per-rack performance
Industry Applications
Technical Overview
The NVIDIA GB200 NVL72 pairs 36 Grace CPUs with 72 Blackwell B200 GPUs via NVLink-C2C, joined by a fifth-generation NVLink switch fabric providing 130TB/s of all-to-all GPU bandwidth in one 72-GPU domain with 13.4TB of unified HBM3e. The liquid-cooled rack scales out over NVIDIA Quantum-X800 XDR InfiniBand and is delivered with a certified parallel storage tier and management stack.
| GPU / Accelerator | 72× NVIDIA Blackwell B200 GPUs + 36× Grace CPUs (GB200 NVL72) |
| GPU Interconnect | 5th-gen NVLink switch, 130TB/s in a 72-GPU NVLink domain |
| Memory | 13.4TB unified HBM3e GPU + 17.3TB LPDDR5X CPU per rack |
| CPU-GPU Link | NVLink-C2C at 900GB/s per Grace Blackwell superchip |
| Networking / Fabric | Scale-out over NVIDIA Quantum-X800 XDR InfiniBand |
| Storage | NVIDIA-certified parallel filesystem, GPUDirect Storage |
| Cooling | Liquid-cooled rack-scale design (~120kW class) |
| Form Factor | Single rack-scale NVL72 unit |
| Warranty | Nexus-backed, NVIDIA AI Enterprise included |
Specifications are indicative and configured to each engagement. Request a quote for a configuration tailored to your requirements.
Warranty, Support & Fulfillment
Every system ships from an authorized channel, configured and tested, with the documentation enterprise buyers need — backed by warranty and a dedicated account team.
Enterprise Warranty
Full manufacturer warranty with optional on-site, next-business-day support and extended coverage.
Authorized Channel
Sourced through Tier-1 distribution and OEM partners — never grey market. Asset & warranty records included.
Lead Time & Deployment
48-hour quotes, then configured, burn-in tested, and delivered on a committed schedule.
Nationwide Fulfillment
Coordinated logistics, rack-and-stack, and delivery wherever your infrastructure lives.
Frequently Asked Questions
Why NVL72 instead of a multi-node H200 pod?
NVL72 turns 72 GPUs into one NVLink domain, so trillion-parameter models that would shard awkwardly across an InfiniBand pod run as if on a single massive GPU — a decisive advantage for the largest training and real-time inference.
Is Blackwell available, and how is delivery handled?
Blackwell supply is allocation-based; we provide a realistic timeline at quote, source through authorized channels, and coordinate staged delivery aligned to your facility readiness.
What facility planning does it require?
NVL72 is a high-power, liquid-cooled rack approaching 120kW, so we engage early on power distribution, coolant loop, and rack planning to ensure the site is ready before delivery.
Hardware Assistance
Configure the Nexus GB200 NVL72 Blackwell Rack-Scale Cluster with Nexus Compute
Tell us your requirements and a hardware specialist will help you specify, configure, and quote the right system — typically within two business days. No obligation.