Inside Dell & CoreWeave's World First Vera Rubin Deployment

Dell Technologies has delivered the world’s first operational NVIDIA Vera Rubin NVL72 server rack to AI cloud provider CoreWeave.

The move marks an early deployment of NVIDIA’s next-generation AI infrastructure platform ahead of broader availability in the second half of 2026.

The fully operational rack passed comprehensive diagnostic testing over the weekend, arriving at a time when demand for AI compute capacity is accelerating across AI labs and hyperscale operators.

Built around 72 NVIDIA Rubin GPUs and 36 NVIDIA Vera CPUs, the liquid-cooled system delivers up to 3.6 exaFLOPS of performance for trillion-parameter AI models.

This deployment places CoreWeave among the first organisations to gain access to the rack-scale platform as NVIDIA moves Vera Rubin into full production and expands manufacturing through its global partner ecosystem.

Michael Dell, Founder and CEO of Dell Technologies, says: “The world’s first NVIDIA Vera Rubin NVL72 server rack is here.

“We’re thrilled to deliver the first working, liquid-cooled Dell Technologies PowerEdge XE9812 for CoreWeave – built for the next era of AI infrastructure.”

CoreWeave gains ground in the AI infrastructure race

The shipment provides an early example of how AI cloud providers are scaling data centre infrastructure to support increasingly complex AI workloads.

NVIDIA describes Vera Rubin as its most extensive POD-scale platform to date.

The architecture combines five purpose-built racks that operate as a single AI supercomputer designed for agentic AI workloads.

The Vera Rubin NVL72 platform integrates NVIDIA Rubin GPUs, NVIDIA Vera CPUs, networking, storage and security technologies into a unified rack-scale system.

According to NVIDIA, the platform delivers 10 times the agent throughput at scale compared with the previous-generation NVIDIA Grace Blackwell platform.

CoreWeave’s early adoption also reflects growing competition among AI infrastructure providers seeking access to the latest accelerated computing platforms.

The company is already one of NVIDIA’s closest cloud partners and has been expanding its data centre footprint to support AI training and inference workloads.

Alongside CoreWeave, NVIDIA says cloud providers including Firmus, GMI Cloud, IBM Cloud, IREN, Lambda, Microsoft Azure, Nebius, Nscale, SpaceXAI and Vultr are adopting NVIDIA Confidential Computing capabilities within the Vera Rubin platform.

NVIDIA ramps global production

The CoreWeave deployment coincides with NVIDIA’s announcement that the Vera Rubin platform is now ramping into full production.

Its manufacturing is being supported by a broad ecosystem of server makers and infrastructure partners. Dell Technologies, HPE, Lenovo and Supermicro are among the major system builders producing Vera Rubin-based systems, alongside suppliers including ASUS, Foxconn, GIGABYTE, Inventec, MSI, NetApp, Nutanix, Quanta Cloud Technology and Wistron.

NVIDIA says hundreds of supply chain partners across more than 350 factories in 30 countries are supporting the production ramp, including 150 partners in Taiwan alone.

Jensen Huang, founder and CEO of NVIDIA, says: “Agentic AI is a new kind of workload. One prompt can launch a thousand-step journey of reasoning, retrieval, tool use and response generation.

“Vera Rubin was built for this moment – an AI factory engine that delivers intelligence at scale, with the performance, efficiency and security needed to power the next industrial revolution.”

The production milestone demonstrates how infrastructure vendors are moving from prototype and announcement stages into large-scale deployment of AI factory architectures.

Networking, security and AI factory design

Beyond compute performance, NVIDIA establishes Vera Rubin as a platform for building large AI factories containing hundreds of thousands or even millions of GPUs.

A key part of that strategy is the introduction of NVIDIA Spectrum-X Ethernet Photonics, which the company describes as the first co-packaged optics-based switching platform using 200Gb/s SerDes technology.

NVIDIA says the networking platform delivers five times better power efficiency, five times longer AI uptime and faster deployment compared with systems relying on traditional transceivers.

CoreWeave, Lambda and Oracle Cloud Infrastructure are among the first adopters of the networking technology.

The Vera Rubin platform also incorporates NVIDIA BlueField-4 data processing units, supporting software-defined networking at speeds of up to 800Gb/s. The technology is designed to improve tenant isolation and strengthen operational control across large multi-tenant AI environments.

With the first operational Vera Rubin NVL72 rack now running at CoreWeave, the next phase of AI data centre buildouts is already moving from roadmap to deployment.