Nutanix has launched Nutanix Agentic AI, a software stack it positions as a way for enterprises to build and operate “AI factories” for agent-based applications across hybrid environments.

The product integrates with NVIDIA AI Enterprise at the Agent Builder layer and orchestrates configurations aligned with NVIDIA-certified AI factory designs. It targets organisations expecting large numbers of AI agents, high concurrency, and frequent changes to services and workflows.

The launch reflects a broader shift in enterprise AI spending away from experimentation and model training towards production use. Companies are now grappling with the operational demands of running many AI services at once, along with governance requirements for access, data handling, and security controls.

Nutanix framed the challenge as infrastructure complexity rather than model selection, pointing to issues such as shared resource access, policy enforcement, compliance, and sovereignty requirements when deploying agentic workloads at scale.

“Contrary to AI infrastructure for model training that was optimised to run ‘one big job,’ production Agentic AI infrastructure needs to handle scale and high rates of change for thousands of AI services, agents, and concurrent users and developers. Nutanix Agentic AI extends our AHV hypervisor, Flow Virtual Networking, Nutanix Kubernetes Platform, and Nutanix Enterprise AI to deliver a cloud operating model to enterprise AI factories, enabling infrastructure and platform teams to simply build, operate, and govern AI factories, while providing Agentic AI developers with the performance and rich set of models and AI platform services they need,” said Thomas Cornely, Executive Vice President of Product Management, Nutanix.

Stack components

Nutanix Agentic AI combines infrastructure orchestration, security tooling, and platform services for data scientists and developers building agent-based applications. Nutanix described it as an AI platform service layer and a model service that sits alongside a Kubernetes environment.

A key element is Nutanix Enterprise AI 2.6, which includes an AI Gateway service for policy control across cloud-hosted and private large language models. The release also adds support for the Model Context Protocol server and fine-tuning, which Nutanix said improves how agents connect to enterprise tools and data sources.

The software adds support for NVIDIA’s Nemotron family of open-source models, datasets, and training tools, aimed at giving developers more choice for agentic systems that can interact with tools and complete multi-step tasks.

Kubernetes focus

Nutanix is extending its CNCF-compliant Nutanix Kubernetes Platform with a catalogue of open-source AI tools, including notebooks, vector databases, MLOps workflow engines, and agentic frameworks.

The Kubernetes layer integrates with NVIDIA AI Enterprise, allowing developers to deploy NVIDIA NIM microservices, including Nemotron, from within the environment.

Infrastructure changes

On the infrastructure side, an early access version of NVIDIA topology-aware AHV enhances the AHV hypervisor for GPU-dense servers by automating the allocation of physical resources to virtual machines.

Nutanix has also updated Flow Virtual Networking to support offloading the network dataplane to NVIDIA BlueField, reducing CPU and memory use on hosts while maintaining network performance.

The stack includes storage and data services, with Nutanix Unified Storage aligned to the NVIDIA AI Data Platform reference design. Nutanix said it delivers linearly scalable read and write performance for large numbers of GPU clients, and highlighted a high-capacity tier for KV Cache offloading. It also supports S3 over RDMA and NFS over RDMA for low-latency data access.

Partner hardware

Customers can deploy NVIDIA-certified AI factories on hardware from Cisco, Dell, and Supermicro, with Nutanix and NVIDIA jointly validating supported configurations.

Nutanix is also working with NVIDIA on integration with the NVIDIA Agent Toolkit, including the NVIDIA OpenShell open-source runtime. It described the work as part of a foundation for autonomous agents in enterprise settings.

Industry views

Steve McDowell, Chief Analyst at NAND Research, said the combined approach could reduce operational hurdles for enterprise AI teams.

“Nutanix’s Agentic AI stack removes much of the infrastructure friction that can slow down enterprise AI projects. By bringing the layers together-from Models-as-a-Service at the top, to an AI platform built on a standardised Kubernetes distribution, down to GPU-aware hypervisors and DPU-accelerated networking-organisations get a more coherent AI stack, enabling AI factories that deliver strong performance and security while driving down the cost per token,” said McDowell.

NVIDIA also linked the shift to security and scale requirements when running large numbers of agents.

“Agentic AI requires high-performance infrastructure that can securely manage thousands of agents at enterprise scale. Nutanix’s integration of NVIDIA Agent Toolkit and open Nemotron models gives enterprises a foundation for building and operating efficient AI factories – and we’re working together to scale autonomous AI in the enterprise as AI agents continue to evolve,” said Justin Boitano.

Nutanix said the overall solution includes products that are generally available or in early access, with additional components expected to become available soon.