Nutanix Expands Platforms for Agentic AI and Multicloud Operations

At its .NEXT conference the week of April 7, 2026, Nutanix unveiled an expanded set of capabilities across the Nutanix Cloud Platform aimed at enabling agentic AI and hybrid/multicloud operations. Announcements center on an extended Agentic AI stack with multitenant and service-provider features, a bare-metal Kubernetes extension called NKP Metal, deeper Kubernetes support, and enhanced management controls for cost, governance and tenant isolation.

Enterprises building AI factories need platforms that combine scalable GPU resources, orchestration for agent-driven workloads, and operational controls for cost and data sovereignty. Nutanix positions itself as an independent software vendor that can run across OEM hardware (Nvidia, AMD, Intel) and provide unified orchestration across edge, private cloud and public cloud targets.

Nutanix described new multitenant/service-provider capabilities designed to enable so-called “neoclouds” — providers offering on-demand AI infrastructure and services such as GPUs-as-a-service and Kubernetes-as-a-service while preserving tenant isolation. Lee Caswell, SVP of product and solutions marketing, framed the strategy as “one platform and one experience only.” Nutanix also emphasized operational controls: tracking token usage in large-language-model environments to avoid cost surprises (“important as customers look at how they deploy AI without getting surprised about the costs”). NKP Metal extends the Nutanix Kubernetes Platform to run K8s workloads directly on bare metal, targeting performance-sensitive AI training and edge deployments that rely on dense GPU configurations.

The changes affect architectural choices for enterprise AI. NKP Metal lowers barriers to use bare-metal GPU pools under Kubernetes orchestration—valuable for training and latency-sensitive inference—but shifts operational responsibilities around provisioning and firmware management. Multitenancy and provider-grade controls simplify offering shared GPU/K8s services while addressing governance, cost management and data sovereignty — common blockers for enterprise AI rollouts. The token-usage emphasis signals that observability and cost-control tooling will be required alongside model orchestration.

product availability and supported hardware lists for NKP Metal, integration depth with Nvidia agent tooling, specific controls for token/meters and chargeback, and initial customer/service-provider pilots of the neocloud capabilities.