NVIDIA NIM (NVIDIA Inference Microservices) are pre-built, optimized containers specifically engineered to run on NVIDIA GPUs for high-performance inference tasks. When an administrator selects a NIM model for an endpoint within Nutanix Enterprise AI, the platform recognizes this inherent dependency. The "Use GPU" checkbox is therefore automatically selected and locked to enforce this mandatory hardware requirement. This ensures the model is deployed with the necessary resources to function correctly and prevents misconfiguration that would lead to deployment failure or non-functional endpoints.

A. The locked state is a direct requirement of the selected model type (NVIDIA NIM), not a separate, broader cluster-level policy.

B. This is incorrect; Nutanix Enterprise AI supports various models, and some can be configured to run on CPUs without requiring GPUs.

C. The checkbox is locked upon model selection, which occurs during the configuration process, not after the system has validated the final settings.

1. Nutanix Enterprise AI User Guide v1.1: In the "Creating an Endpoint" section, the guide specifies the behavior of the user interface during model selection. It states, "When you select a model that requires a GPU, the Use GPU checkbox is selected and cannot be cleared." This directly confirms that the model's inherent requirements dictate the UI behavior. (Reference: Nutanix Enterprise AI User Guide, Chapter: "Model and Endpoint Management", Section: "Creating an Endpoint").

2. NVIDIA NIM Documentation: NVIDIA's official documentation for NIM clarifies that these microservices are "built on the foundation of the NVIDIA CUDA platform" and are "optimized to run on NVIDIA GPUs." This establishes the fundamental dependency of NIMs on GPU hardware. (Reference: NVIDIA NIM official product documentation, "NIM Overview" section).