NVIDIA AI Enterprise can be run on Amazon Web Services (AWS), Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure (OCI), Alibaba Cloud, and Tencent Cloud. There are four methods for deploying NVIDAI AI Enterprise in the cloud, the following table summarizes:
Instance Type |
CSPs |
NVIDIA Virtual Machine Image (VMI) |
License Included |
|---|---|---|---|
| Standard | AWS, Azure, Google Cloud, OCI, Alibaba, Tencent, ByteDance | No | No |
| NVIDIA Private Offer VMIs | AWS, Azure, Google Cloud, OCI | Yes | Yes |
| NVIDIA AI Enterprise VMIs | AWS, Azure, Google Cloud, OCI | Yes | Yes |
| NVIDIA GPU-Optimized VMIs | AWS, Azure, Google Cloud, Alibaba | Yes | No |
| Managed Kubernetes | AWS, Azure, Google Cloud | No | No |
| Marketplace Red Hat OpenShift | AWS, Azure, Google Cloud | No | No |
NVIDIA VMIs contain key technologies and software preinstalled from NVIDIA for rapid deployment, management, and scaling of AI workloads. Whereas, standard cloud instances may not include the proven supported NVIDIA AI Enterprise configurations, such as RHEL or Ubuntu based Operating Systems Certified for NVIDIA AI Enterprise, the NVIDIA driver, and the NVIDIA Container Toolkit.
Please refer the NVIDIA AI Enterprise Supported Cloud Services Product Support Matrix for additional information
Regardless of which supported instance you choose, a license for NVIDIA AI Enterprise is required and may not be included in the provisioned instance automatically. In most cases, you are required to bring-your-own-license (BYOL), except for on-demand “pay-as-you-go” Enterprise VMIs and Private Offers, which include a separate license process. Detailed information for licensing NVIDIA AI Enterprise in the cloud can be found in the Cloud Deployment section of the NVIDIA AI Enterprise Packaging, Pricing, and Licensing Guide.
The upcoming sections provide further details for each Cloud Service Provider (CSP).