Vultr Cloud GPU Accelerated by NVIDIA GB300 NVL72
Harness unprecedented AI reasoning performance through a rack-scale solution combining NVIDIA Blackwell Ultra GPUs and NVIDIA Grace™ CPUs.
Ever-increasing AI demands have created growing compute capacity requirements to support the large context windows essential to multi-step AI reasoning.
NVIDIA GB300 NVL72s combine 72 Blackwell Ultra GPUs and 36 NVIDIA Grace™ CPUs into a single rack-scale, liquid-cooled architecture that makes it simple and affordable to access needed compute capacity and achieve unprecedented acceleration, including as self-service clusters on Vultr Clusters for maximum control.

Built for AI inference, training, agentic AI, and HPC
The NVIDIA GB300 NVL72 delivers up to 50x greater AI factory output compared to a NVIDIA Hopper-generation platform. With up to 30x greater energy efficiency and 15x lower cost per token, it’s well-equipped to support AI agents, and its Blackwell Tensor Core architecture produces high throughput and efficiency for HPC workloads. Vultr’s composable platform makes scaling all those workloads straightforward and cost-effective globally.
Get started with the
world’s largest privately-held cloud
infrastructure company