Nutanix Releases Enterprise AI 2.8 MCP Gateway

CRN reports that Enterprise AI 2.8 provides centralized controls for AI inference and agentic AI workloads, including visibility into token consumers, model usage, and spending. The report says administrators can apply access policies, set usage caps, and route workloads, although the supplied article text does not specify all routing options.

Blocks & Files reports that Nutanix Private Inference also receives capabilities for high-performance fine-tuning and inference, alongside security and governance improvements. Computer Weekly characterizes Enterprise AI as a Kubernetes-based platform for inference endpoint management across hybrid cloud infrastructure.

Nutanix calls the underlying approach "dual-native," meaning virtual machines and containers are both treated as first-class deployment targets. SiliconANGLE reports that Kubernetes can run on Nutanix's Acropolis Hypervisor or directly on bare metal, depending on the environment selected for a workload. Cornely told the publication, "It is not about getting containers working; it is about how you operate and manage those containers."

The timing distinction matters: Enterprise AI 2.8 is generally available, while NKP 2.19 is announced as an upcoming release. According to Blocks & Files, NKP 2.19 is intended to streamline container management across bare-metal and virtualized environments and include an AI catalog for building and running agentic AI applications.

The release extends Nutanix's recent agentic AI controls. Blocks & Files reports that its Agent Gateway became generally available in July as a control point for agent activity, access policies, and token consumption, while an open-source MCP server was added to Nutanix Cloud Platform earlier in August.

For platform engineers, MCP is increasingly a consequential control boundary because it connects model-driven agents to external tools, enterprise data, and operational systems. Across comparable enterprise deployments, centralized identity, authorization, auditability, and consumption controls are common requirements when agents can make repeated tool calls. Nutanix's new gateway places those functions alongside its existing VM, Kubernetes, and inference management layers, while the practical scope of NKP 2.19's new capabilities will depend on its final availability and implementation details.