CoreWeave announced that it has become the first AI cloud provider to bring up and fully validate NVIDIA Vera Rubin NVL72 on its cloud platform, marking a significant milestone in the deployment of NVIDIA’s next-generation AI infrastructure. The achievement extends CoreWeave’s support for NVIDIA hardware and includes rigorous system-level validation of the entire rack-scale architecture.
The announcement comes as agentic AI workloads drive new infrastructure demands. As AI models expand to trillions of parameters, context windows stretch into millions of tokens, and persistent reasoning becomes commonplace, inference performance has become a critical factor in determining how quickly AI companies can scale and innovate.
NVIDIA Vera Rubin NVL72 features 72 NVIDIA Rubin GPUs and 36 NVIDIA Vera CPUs per rack, connected through a 260 TB/s sixth-generation NVIDIA NVLink fabric. According to NVIDIA, the platform delivers up to 10 times better inference per watt, requires up to one-fourth fewer GPUs, and reduces cost per million tokens by up to one-tenth compared with NVIDIA Blackwell systems.
To support production-scale deployment of Vera Rubin, CoreWeave introduced several purpose-built innovations through its CoreWeave Mission Control platform. These include Valvey, a software-defined liquid cooling system that enables real-time monitoring and control of cooling infrastructure at the rack level, and Racky, a unified rack control appliance that aggregates power, cooling, and environmental sensors into a standardized management platform.
The company also highlighted its multi-rail, multi-plane networking architecture, which supports both NVIDIA Quantum-X800 InfiniBand and NVIDIA Spectrum-X Ethernet with RDMA over Converged Ethernet. CoreWeave said the architecture delivers 1.6 Tb/s of backend bandwidth per GPU and can scale to hundreds of thousands of GPUs within a two-tier network design.
In addition, CoreWeave is incorporating NVIDIA BlueField-4 DPUs to support secure multi-tenant AI cloud operations, enabling accelerated infrastructure services, lower latency, improved data access, and stronger tenant isolation.
The deployment was developed in collaboration with key infrastructure partners. Dell Technologies supplied the platform architecture through its PowerEdge XE9812 servers, while Micron provided 7600 SSDs, which CoreWeave said represent one of the first liquid-cooled NVMe storage solutions deployed at rack scale.
CoreWeave noted that the company has established a track record of high-performance AI infrastructure, citing record-setting MLPerf benchmark results, Platinum rankings in both SemiAnalysis ClusterMAX 1.0 and 2.0, and a top ranking for inference speed and price-performance for Moonshot AI’s Kimi K2 in independent benchmarking by Artificial Analysis.
KEY QUOTES:
“Our research depends on infrastructure that’s both powerful and reliable, and CoreWeave has delivered on this as we’ve scaled across NVIDIA Hopper and Blackwell. Their ability to deliver highly performant clusters with full cluster observability and a support team that engages deeply on hard problems gives us the confidence to partner with them on Vera Rubin. We are excited about the efficiency gains at rack scale translating into faster training runs and shorter iteration cycles for our researchers.”
Craig Falls, Head of Quantitative Research, Jane Street
“The agentic era demands a fundamentally different approach to infrastructure, one that keeps pace with workloads that reason continuously, scale unpredictably, and operate in production around the clock. What separates infrastructure that performs in a lab from infrastructure that performs in production is the depth of engineering underneath it. With patent-pending innovations like Valvey and Racky, CoreWeave has done the full-stack orchestration work to enable Vera Rubin to perform the way it was designed to, not just in a lab, but at production scale for the world’s most demanding AI teams.”
Chen Goldberg, Executive Vice President of Product & Engineering, CoreWeave
“Vera Rubin is the most capable AI platform NVIDIA has ever built. CoreWeave has consistently been at the frontier of deploying each new generation of NVIDIA architecture at scale, and their full-stack, end-to-end approach to Vera Rubin, from cooling to orchestration, is how the world’s most ambitious AI teams will push the next AI frontier.”
Ian Buck, Vice President of Hyperscale and High-Performance Computing (HPC), NVIDIA
“Dell Technologies and CoreWeave share a commitment to delivering innovation that performs at the frontier of what AI demands. The PowerEdge XE9812 was engineered for exactly this kind of density and precision. Working with CoreWeave to bring up the first NVIDIA Vera Rubin NVL72 rack is a direct validation of what enterprise-grade hardware can do when it’s paired with the right operational expertise.”
Michael Dell, Chairman and Chief Executive Officer, Dell Technologies
