TLDR: CoreWeave has become the first AI cloud provider to deploy NVIDIA’s cutting-edge GB300 NVL72 systems, hosted at Switch’s advanced EVO AI Factories. This deployment, a collaboration with Dell and Vertiv, marks a significant leap in AI infrastructure, offering substantial performance improvements for AI reasoning and agentic workloads.
CoreWeave, an AI Hyperscaler, has achieved a significant milestone by becoming the first AI cloud provider to deploy NVIDIA’s latest GB300 NVL72 systems for its customers. This landmark deployment is taking place at Switch’s state-of-the-art EVO AI Factories, which are designed for advanced data center infrastructure. The collaboration also involves Dell and Vertiv, highlighting a concerted effort to push the boundaries of AI computing.
The NVIDIA GB300 NVL72 platform represents a substantial advancement in AI performance. According to CoreWeave, this new system delivers up to a 10x improvement in user responsiveness, a 5x increase in throughput per watt compared to the previous NVIDIA Hopper architecture, and a remarkable 50x higher inference output for reasoning models. This makes the GB300 NVL72 particularly well-suited for demanding AI reasoning and agentic workloads.
Peter Salanki, Co-Founder and Chief Technology Officer at CoreWeave, emphasized the company’s commitment to advancing AI development, stating, ‘CoreWeave is constantly working to push the boundaries of AI development further, deploying the bleeding-edge cloud capabilities required to train the next generation of AI models.’ He added, ‘We’re proud to be the first to stand up this transformative platform and help innovators prepare for the next exciting wave of AI.’
Switch’s EVO AI Factories are setting new benchmarks for rack-scale infrastructure, providing the necessary environment for such high-density deployments. These factories leverage nearly two decades of thermal engineering leadership, enabling the efficient operation of liquid-cooled, rack-scale systems that integrate 72 of NVIDIA’s new Blackwell Ultra GPUs with 36 Grace CPUs.
This deployment further solidifies CoreWeave’s strategy of being first to market with NVIDIA’s most advanced technology. The company has a history of early adoption, having been among the first to offer NVIDIA H200 GPUs and the first AI cloud provider to make NVIDIA GB200 NVL72 systems generally available. CoreWeave also recently collaborated with NVIDIA and IBM to submit the largest-ever MLPerf® Training v5.0 benchmark, achieving a breakthrough result on the Llama 3.1 405B model.
Also Read:
- Deloitte and NVIDIA Accelerate Enterprise Transformation with Agentic AI and AI Factory Solutions
- Google Cloud’s Q2 Revenue Surges 32% Driven by Strategic OpenAI Infrastructure Partnership
The financial implications of this AI infrastructure buildout are substantial, with CoreWeave planning to spend between $20 billion and $23 billion on capital expenditures in 2025 alone. This investment underscores the high-stakes economics of the AI arms race, where access to cutting-edge hardware is a primary competitive advantage.
