LEVIATHAN SYSTEMS
← Back to Glossary

What Is NVL72?_

NVL72 refers to NVIDIA's rack-scale GPU system containing 72 GPUs and 36 Grace CPUs connected by NVLink in a single liquid-cooled rack. The GB200 NVL72 and GB300 NVL72 are the two current NVL72 platforms, drawing approximately 120 kW each and requiring 100% direct liquid cooling.

Technical Details

The NVL72 architecture represents a fundamental shift from traditional server-based GPU deployment. Instead of discrete servers in a rack, the NVL72 integrates 72 GPUs and 36 Grace CPUs into a unified rack-scale system connected by NVLink 5.0 with 130 TB/s total interconnect bandwidth. NVSwitch chips inside the rack provide all-to-all GPU connectivity without external NVLink cables. The external network connections (400GbE or InfiniBand to leaf/spine switches) are still dense, with 200+ fiber connections per rack. The rack weighs approximately 1,360 kg, requires verified floor loading capacity, and uses mandatory direct liquid cooling with CDUs, manifolds, and quick-disconnect fittings.

How Leviathan Systems Works with NVL72

Leviathan Systems has direct deployment experience with both GB200 NVL72 and GB300 NVL72 systems, including site preparation, rack positioning, liquid cooling integration, network cabling, and commissioning.

Related service

GPU Rack Assembly & Rack-and-Stack →

Related Terms

Direct Liquid Cooling (DLC)NVLinkGrace CPUCoolant Distribution Unit (CDU)

Appears In

Transceiver Breakout & Splitter Cabling for GPU FabricsNVIDIA Spectrum-X vs Quantum InfiniBand: The Cabling and Optics ViewRear-Door Heat Exchangers vs Direct Liquid Cooling: Choosing Your Path to High DensityRaised Floor vs Slab for High-Density GPU HallsScaling from Pilot to Production GPU Cluster: What Breaks and What to PlanMEP Coordination for AI Data Halls: Where Mechanical, Electrical, and IT CollideHow to Choose a GPU Deployment Partner: A Buyer's ChecklistH100 / HGX H100 Deployment Checklist: Power, Cooling, and Cabling DemandsGPU Data Center Deployment in Ohio: Staffing the Physical BuildGPU Data Center Deployment in Northern Virginia: Who Does It and How to HireGPU Data Center Deployment in Georgia: Who Builds It and How to HireGB200 NVL72 Deployment Deep Dive: Liquid Loop, Busbar, and SpineFloor Loading for NVL72 Racks: Will Your Slab Hold ~1.4 Tonnes?Fiber Certification Workflow: From Splice to Signed Acceptance ReportFacility Water Loop Design for AI Halls: TCS to FWS Done RightDe-Racking & Decommissioning a GPU Cluster for MigrationDAC vs AOC vs AEC: Choosing GPU Rack Interconnect CablesContainment & Pathway Build-Out for a GPU RowArmored vs Standard Fiber in the GPU Data Hall: When Ruggedization PaysSingle-Mode vs Multimode for AI Fabric: The Cable-Plant DecisionPatch-Panel & Cassette Design for GPU Halls: Breakout, Density, ServiceabilityNVIDIA B200 vs GB200: HGX vs Rack-Scale, and What Changes to DeployHot Aisle vs Cold Aisle Containment at GPU DensityNVIDIA GB300 NVL72 Explained: Specs, Power, and What It Takes to DeployData Center Migration for AI Infrastructure: A Practical Field GuideGoogle TPU vs NVIDIA GPU: What It Means for Your AI Infrastructure BuildData Center Structured Cabling Standards: TIA-942, TIA-606-C, BICSIData Center Rack-and-Stack Services for GPU Builds: What's IncludedHow to Choose a Data Center Liquid Cooling CompanyWho Deploys GB200 / GB300 NVL72 Infrastructure?How Long Does GPU Cluster Deployment Take?GPU Data Center Deployment in Texas: Who Does It and How to HireGPU Commissioning & Acceptance: What to Demand Before You Sign OffSite Readiness Before the GPUs Arrive: Power, Cooling, Floor, PathwaysRail-Optimized vs Fat-Tree: The Field Wiring Plan, Port by PortIn-House vs. Outsourced GPU Deployment: How to DecideGPU Rack Assembly: What Drives the CostSpectrum-X vs InfiniBand: What's Different for the Cable PlantNVLink Spine Cartridge & Copper Backplane Handling: Field ProcedureGB300 NVL72 Deployment: Power, Cooling, and the Cable PlantGB200 vs GB300 NVL72: What Changes for DeploymentNCCL Bandwidth Validation: Proving a GPU Fabric Before ProductionGPU Rack Receiving, Staging & Lift Plan: Moving ~1,360 kg Racks Without DamageOverhead Busway Installation for 100kW+ GPU Rack DropsThermal Burn-In for GPU Clusters: Duration, Watch Items, Pass/FailStructured Cabling QA/QC for GPU Racks: Bend Radius, Slack, Torque, DressingOTDR & Insertion/Return-Loss Testing for GPU Cluster FiberLiquid-Cooling Loop Commissioning for GB200 & GB300 NVL72: Flush, Fill, Leak-Test, Accept