Ecosystem Integration

The NVIDIA AI Enterprise Infrastructure software suite has an extensive list of supported integrations with Kubernetes-based container orchestration platforms, whether bare-metal or virtualized. Customers can take advantage of the capabilities it offers in a number of ways, depending on their use case and point in their path to production AI

• The infrastructure software can be used with upstream Kubernetes, and NVIDIA produces deployment and sizing guidance for getting the most optimal utilization. This gives customers a straightforward path to standing up an AI Factory for test, development, and non-production applications.

• Enterprises can make use of this software in production when they purchase commercially supported solutions from enterprise management software companies. These solutions have supported integrations with the components of the NVIDIA AI Enterprise suite, and customers can choose to use which ones to use based on their needs.

For a platform engineering team, managing a specialized AI factory feels exactly like managing a standard Kubernetes cluster. Administrators interact with the primary Kubernetes control plane to deploy AI software containers, allocate GPUs, and auto-scale inference models using the exact same declarative YAML manifests and Helm charts they already use. By translating specialized hardware requirements into standard cloud-native APIs, the software bridges the gap, allowing enterprises to scale complex AI infrastructure using the operational discipline and visibility tools they already possess.