Nscale Brings Vera Rubin AI Racks Into Global Cloud Deployment

Nscale has switched on its first Vera Rubin racks. The company announced on November 18, 2026 that its NVIDIA Vera Rubin NVL72 rack-scale AI systems were online, starting a global rollout expected to continue over the coming months. The deployment combines Dell PowerEdge XE9812 systems, factory-integrated liquid cooling, high-speed networking, and a fleet operations model designed to move each rack into service without treating operations as an afterthought.
Each NVL72 rack contains 72 Rubin GPUs and 36 Vera CPUs, plus memory and high-speed connections designed to operate as one rack-scale supercomputer. The Dell PowerEdge XE9812 systems add capacity for customers training and serving large models, including workloads that reason through multistep problems.
Nscale identifies large distributed training, long-context reasoning, reinforcement learning, agentic inference, and low-latency services processing high token volumes as target workloads. NVIDIA claims the stated configuration can deliver up to 10 times more tokens per megawatt than GB200 NVL72 — a bold efficiency comparison, because apparently the data center still wants a seat at the power bill negotiations.
One rack, two networking layers
The systems divide networking duties between the rack and the wider cluster. NVLink provides the scale-up fabric inside each rack, connecting the GPUs so they can work as a tightly coupled system, while a RoCE Ethernet network connects racks into larger clusters.
RoCE, short for RDMA over Converged Ethernet, supplies high-throughput, low-latency communication for distributed AI traffic. Nscale’s Ethernet fabric also includes high-performance switching, congestion management, telemetry, and automation on a standards-based foundation.
The design aims to keep GPUs working on parallel jobs instead of waiting for memory, networking, or operational systems to catch up. That matters for training and inference workloads spread across many racks, where a slow connection can turn expensive hardware into an elaborate waiting room.
Factory integration moves work offsite
Nscale says it is the first in the industry to deploy Dell’s fully integrated, liquid-cooled L11 networking rack. Built on Dell’s IR9148 platform, the rack combines eight NVIDIA SN6600-LD switches with shuffle boxes, power shelves, an integrated rack controller, and an in-rack coolant distribution unit.
Dell integrates and tests the populated rack at the factory, removing the need for onsite network-rack assembly. The approach is meant to reduce deployment work and help Nscale bring high-performance computing infrastructure into production faster.
“Dell PowerRack integrates and validates compute, networking, power and cooling in the factory, reducing the work required onsite and helping Nscale bring its high performance compute infrastructure into production faster,” said Ihab Tarazi, CTO of Compute and Networking in Dell Technologies’ Infrastructure Solutions Group.
Dell’s system overview describes the XE9812 as the Vera Rubin successor to its XE9712. The system uses direct liquid cooling and NVLink-interconnected GPUs in a rack-scale design, keeping compute, cooling, and connectivity aligned inside the same deployment unit.
Racks enter the cloud operating model
The hardware enters Nscale’s cloud operating model as soon as it comes online. Fleet Operations manages enrollment, configuration, validation, burn-in, network integration, maintenance, repair, and retirement across the rack fleet.
Control Center applies software, firmware, network, and security baselines before systems move through validation. That operating model gives Nscale a way to manage the rollout as a repeatable fleet process rather than a sequence of isolated hardware installations.
The first systems are now online, but the larger story is the rollout model. Nscale is pairing rack-scale AI hardware with factory-tested networking and centralized fleet management, aiming to turn Vera Rubin capacity into a global cloud service over the coming months.
Based on
- Nscale Activates Its First Vera Rubin NVL72 Racks for Global AI Cloud — unite.ai
- NVIDIA Expands Vera CPU Platform to Accelerate Scientific Discovery | Markets Insider — markets.businessinsider.com
- AI Infrastructure: GPU, Cloud & Compute | VentureBeat — venturebeat.com
- Nvidia Archive for November 2026 – Page 1 | The Verge — theverge.com
- This week in the big AI data center buildout. | The Verge — theverge.com




