By CoreWeave Inc. (CRWV) stock gained on Wednesday after the AI cloud-computing company said it had brought up a multi-rack Nvidia Vera Rubin NVL72 cluster on CoreWeave Cloud.
The setup connects hundreds of Nvidia Rubin graphics processing units (GPUs) into a single system, allowing customers to run training and inference workloads across multiple racks.
CoreWeave said the infrastructure is designed to provide additional capacity for larger AI models, more demanding inference workloads and reinforcement learning at scale.
The stock had gained as much as 4% before trimming the advance to around 3% after the US Federal Reserve raised interest rates to 4% in a unanimous decision and signalled further rate increases.
CoreWeave connects hundreds of Nvidia Rubin GPUs
A single Nvidia Vera Rubin NVL72 rack combines 72 Rubin GPUs with 36 Vera CPUs, along with Nvidia networking and data-processing technology.
CoreWeave said multiple racks can be connected through Nvidia Spectrum-X Ethernet to operate as one scale-out cluster.
The company said it was the first AI cloud provider to validate and bring up a Vera Rubin NVL72.
Chen Goldberg, Executive Vice President of Product and Engineering at CoreWeave, said the multi-rack configuration connects hundreds of Rubin GPUs as a single scale-out cluster.
Goldberg said the system provides customers developing agentic AI with greater scale and faster iteration.
CoreWeave said its infrastructure automates rack setup, firmware updates, validation, power and cooling processes. It also tests complete racks before they are moved into production.
However, Bernstein said in a September 14 report that CoreWeave could be among the companies most exposed if artificial intelligence firms reduce spending on model training.
The research firm said CoreWeave’s heavy exposure to AI infrastructure could leave the company particularly vulnerable to a shift away from large-scale training workloads.
Bernstein also pointed to CoreWeave’s concentration in rural and lower-tier data-centre markets as a potential risk if AI demand increasingly favours metro-focused operators.
Bernstein maintained its Sell rating on CoreWeave while raising its price target to $74.
AI Object Storage gets new features
Alongside the new compute infrastructure, CoreWeave introduced several additions to its AI Object Storage platform.
Cross-region write acceleration allows AI workloads to write data in the region where they are running while CoreWeave copies the data to another region in the background.
The company said this can reduce pauses for customers conducting training workloads across multiple regions.
CoreWeave also introduced an Archive tier designed as a lower-cost option for customers retaining data for longer periods.
The company said the tier has no retrieval fee, early-deletion fee or fee for reading data from the Archive tier.
Its Local Object Transport Accelerator, or LOTA, can cache data closer to AI workloads.
CoreWeave said the technology reduces read latency by eight times compared with reading from a traditional storage cluster.
The company also said its expanded AI Object Storage system delivers reads at local NVMe speeds while reducing latency compared with traditional storage clusters.
CoreWeave adds lease agreement as Fed weighs
The company also received a boost from a commercial development involving Blockfusion USA’s subsidiary North East Data.
The subsidiary converted a previously non-binding letter of intent into a definitive 15-year anchor lease with CoreWeave for capacity at its Niagara Falls, New York campus.
The agreement includes two five-year renewal options and a companion expansion agreement.
CoreWeave’s stock later pared some of its gains after the Federal Reserve raised its interest rate to 4% in a unanimous vote.
The central bank said inflation remained elevated and that the policy action would support a more timely return to its 2% objective.
The post CoreWeave stock gains as Nvidia Rubin cluster boosts AI cloud capacity appeared first on Invezz
