Key Highlights CoreWeave (CRWV) shares jumped approximately 3% during Wednesday’s premarket session following the announcement of its multi-rack Nvidia Vera Rubin NVL72 cluster deployment. Th
Shares of [[LINK_START_1]]CoreWeave (CRWV)[[LINK_END_1]] advanced roughly 3% in Wednesday’s premarket session after the AI infrastructure provider revealed its successful deployment of a multi-rack Nvidia Vera Rubin NVL72 cluster within CoreWeave Cloud.
CoreWeave, Inc. Class A Common Stock, CRWV
This achievement builds upon CoreWeave’s June breakthrough when it first brought a standalone Vera Rubin NVL72 rack into production. The latest announcement represents a substantial expansion of that initial success.
The recently implemented multi-rack configuration links hundreds of Rubin GPUs into one unified scale-out infrastructure. This represents a substantial increase in computational capacity for enterprises operating demanding AI applications.
Each Vera Rubin NVL72 rack incorporates 72 Rubin GPUs paired with 36 Vera CPUs, featuring Nvidia NVLink 6 technology, ConnectX-9 SuperNICs, and BlueField-4 DPUs. The multi-rack configuration integrates several of these units through Nvidia’s Spectrum-X Ethernet infrastructure.
The ConnectX-9 SuperNICs enable 1.6 Tb/s scale-out bandwidth per GPU through multiplane, multirail architectures. The infrastructure can accommodate approximately 128,000 GPUs per rail within a non-blocking network fabric.
The architecture’s modular nature enables additional racks to be integrated without requiring complete fabric reconfiguration. This scalability is crucial for organizations experiencing rapid growth.
Chen Goldberg, EVP of product and engineering at CoreWeave, noted that the multi-rack Vera Rubin infrastructure provides organizations developing agentic AI with “greater scale, faster iteration, and higher productivity as models and agents continuously learn and improve.”
The infrastructure is engineered to manage both model training and inference operations at enterprise scale. Agentic AI applications, which demand continuous model learning and autonomous decision-making, require substantial computational resources.
In conjunction with the cluster announcement, [[LINK_START_3]]CoreWeave[[LINK_END_3]] released two additional capabilities for its AI Object Storage platform.
The first capability enables cross-region write acceleration, allowing data to be written with local latency while background replication occurs to remote locations. The second is an Archive storage tier offering economical storage without retrieval charges, early deletion penalties, or read fees.
CoreWeave’s Local Object Transport Accelerator (LOTA) technology delivers read performance at local NVMe speeds. According to the company, this reduces latency by as much as 8x versus traditional storage cluster reads.
LOTA delivers throughput of up to 7 GB/s per GPU, CoreWeave reports. This performance is achieved through managed caching deployed on each CoreWeave Kubernetes Service node.
The Archive tier serves as an economical solution for infrequently accessed data that must remain within CoreWeave’s platform.
Nvidia shares rose approximately 1% on Wednesday in response to CoreWeave’s announcement. The partnership between both companies remains under close industry observation, with Nvidia’s hardware forming the foundation of CoreWeave’s cloud platform.
CoreWeave emphasized that it remains the exclusive AI cloud provider to have successfully validated and implemented the Vera Rubin NVL72 architecture across multiple racks.
The post CoreWeave (CRWV) Stock Gains 3% as Company Launches Nvidia’s Cutting-Edge AI Infrastructure appeared first on Blockonomi.
