TLDR
- CoreWeave stock rose around 3% in premarket trading Wednesday after announcing deployment of a multi-rack Nvidia Vera Rubin NVL72 cluster on CoreWeave Cloud.
- The cluster connects hundreds of Rubin GPUs into a single scale-out unit built for agentic AI workloads.
- CoreWeave claims to be the first AI cloud provider to validate and bring up a Vera Rubin NVL72.
- The company also launched two new AI Object Storage features: cross-region write acceleration and a new Archive tier.
- Nvidia (NVDA) edged up around 1% on the news.
CoreWeave (CRWV) stock climbed roughly 3% in premarket trading on Wednesday after the company announced it had deployed a multi-rack Nvidia Vera Rubin NVL72 cluster on CoreWeave Cloud.
CoreWeave, Inc. Class A Common Stock, CRWV
The move follows CoreWeave’s earlier milestone in June, when it became the first AI cloud provider to bring a single Vera Rubin NVL72 rack online. Wednesday’s announcement takes that a step further.
The new multi-rack setup connects hundreds of Rubin GPUs into a single scale-out cluster. That’s a meaningful jump in compute density for customers running large AI workloads.
A single Vera Rubin NVL72 rack combines 72 Rubin GPUs with 36 Vera CPUs, Nvidia NVLink 6, ConnectX-9 SuperNICs, and BlueField-4 DPUs. The multi-rack version stitches multiple such racks together using Nvidia’s Spectrum-X Ethernet networking.
The ConnectX-9 SuperNICs deliver 1.6 Tb/s of scale-out connectivity per GPU across multiplane, multirail paths. The setup can support roughly 128,000 GPUs per rail in a non-blocking fabric.
The modular design means more racks can be added without redesigning the fabric each time. That kind of flexibility matters when customers are scaling fast.
Chen Goldberg, EVP of product and engineering at CoreWeave, said the multi-rack Vera Rubin cluster gives customers building agentic AI “greater scale, faster iteration, and higher productivity as models and agents continuously learn and improve.”
The cluster is designed to handle both training and inference jobs at scale. Agentic AI workloads, which require models to continuously learn and act, put heavy demands on compute infrastructure.
New Storage Features Also Launched
Alongside the cluster news, CoreWeave introduced two new features for its AI Object Storage product.
The first is cross-region write acceleration, which lets data be written at local latency while being replicated to a remote region in the background. The second is a new Archive tier that offers lower-cost storage with no retrieval, early deletion, or reading fees.
CoreWeave’s Local Object Transport Accelerator (LOTA) delivers reads at local NVMe speeds. The company says it reduces latency by up to 8x compared to reading from a traditional storage cluster.
Storage Performance Numbers
LOTA provides up to 7 GB/s of throughput per GPU, according to CoreWeave. It achieves this through managed caching on each CoreWeave Kubernetes Service node.
The Archive tier is positioned as a cost-efficient option for data that doesn’t need to be accessed frequently but still needs to stay within the CoreWeave ecosystem.
Nvidia stock edged up about 1% on Wednesday alongside the CoreWeave announcement. The two companies have a closely watched partnership, with Nvidia hardware sitting at the core of CoreWeave’s cloud infrastructure.
CoreWeave said it continues to be the only AI cloud provider to have validated and deployed the Vera Rubin NVL72 architecture at multi-rack scale.
Stop guessing and start investing with confidence. KnockoutStocks gives you the AI insights, market intelligence, and stock research you need to spot opportunities, cut through the noise, and make smarter investment decisions — all in one powerful platform.
Sign up today and get 50% OFF full access to our premium stock picks.
Simply use coupon code SPECIAL50 at checkout to claim your exclusive discount.
