Nutanix announced at its .NEXT conference on April 7 a pair of additions to its Agentic AI infrastructure stack designed to help neocloud providers sell managed AI services rather than raw GPU compute. The two new capabilities, a multi-tenant AI management portal built on Service Provider Central and an Enterprise AI Gateway for governing agent access to models, target a specific gap: neocloud providers that have GPU capacity but lack the operational layer to sell higher-margin services on top of it, according to the Nutanix press release.
From GPU Rental to Managed Services
The neocloud market has grown rapidly by offering on-demand access to GPUs, primarily for AI training workloads from a small number of large customers. Nutanix’s expansion addresses the next phase: scaling inference and running agentic AI applications in production for many enterprise tenants simultaneously.
The multi-tenant portal enables service providers to offer a catalog that includes GPU-as-a-Service, Kubernetes-as-a-Service, model-as-a-Service, and agent orchestration, all while maintaining tenant isolation and granular resource management. Previously, most neocloud operators were limited to the first tier, according to the Nutanix announcement.
Thomas Cornely, Executive Vice President of Product Management at Nutanix, said in the release that the platform is “designed to enable neocloud providers to run diverse, governed AI services across compute, AI platforms, and agentic AI applications, all from a single pane of glass.”
The AI Gateway: Cost and Governance
The Enterprise AI Gateway introduces a control plane that governs which agents access which models and at what cost. Anindo Sengupta, Nutanix VP of Product Management, told SiliconANGLE that the gateway is “really around cost and governance. As agents sprawl, models and tools need to be controlled and governed.”
The cost angle matters. Dan Ciruli, VP and General Manager of Cloud-Native at Nutanix, noted in the same interview that a single user action in an agentic workflow can trigger hundreds of downstream agent calls, each consuming tokens at scale. He described the emerging discipline of AI FinOps as a direct consequence: enterprises will increasingly choose where to run inference based on per-token economics, not just availability.
Infrastructure Under the Hood
Both announcements sit on top of Nutanix Kubernetes Platform Metal, which the company describes as a dual-native platform supporting VMs, virtualized Kubernetes, and bare-metal Kubernetes from a single control plane. It ships with CN-AOS (an enterprise storage layer) and a catalog of open-source AI projects announced at Nvidia GTC, according to SiliconANGLE.
Competitive Positioning
Nutanix is positioning against AWS Bedrock AgentCore and Azure’s agent infrastructure with a provider-first pitch: rather than competing directly for enterprise end-users, Nutanix enables the neocloud operators that serve those enterprises. The multi-tenant capability also supports sovereign AI deployments, giving enterprise users control over data, infrastructure, and AI operations within their jurisdiction.
General availability for the new capabilities is targeted for H2 2026. Early access is available to existing partners now.