Nutanix Inc. today announced a trio of major initiatives aimed at expanding its footprint across cloud-native applications, enterprise artificial intelligence workloads and mission-critical infrastructure.

The announcements, made at its .NEXT 2025 conference in Washington, D.C., include the general availability of the Cloud-Native AOS operating system, deeper integration with Nvidia Corp.’s artificial intelligence software and and a partnership with Pure Storage Inc. to deliver integrated infrastructure for high-performance environments.

With the announcements, Nutanix aims to address some pain points in modern information technology environments. By detaching AOS from its traditional hypervisor dependency, Nutanix hopes to bridge a gap in cloud-native storage options that often lack enterprise-grade features.

The integration with Nvidia’s AI tools positions Nutanix as a middle layer that connects raw infrastructure with AI model deployment and management. The Pure Storage partnership aims to offer a more open and flexible alternative to VMware Inc.-centric stacks.

The company is betting that enterprise IT teams want unified platforms that can manage workloads across VMs, containers and AI frameworks without sacrificing control, performance or data mobility.

“Our vision is that we want customers to be able to run anything, anywhere,” said Lee Caswell (pictured), senior vice president of product and solutions marketing at Nutanix. “This is about running any containerized applications and the latest gen AI models, whether on external storage or not.”

Bridging Kubernetes and storage

Cloud Native AOS is a version of Nutanix’s core AOS storage software that runs in Kubernetes environments, allowing it to be deployed across public cloud and bare-metal infrastructure without the need for a hypervisor. Nutanix said its goal is to extend enterprise-grade data services, such as integrated disaster recovery and application mobility, to containerized applications regardless of where they run.

The product addresses a growing gap in hybrid multicloud architectures: managing persistent data across diverse Kubernetes environments. Though container orchestration is common, stateful workloads require robust data protection, replication and migration capabilities typically found in virtualized infrastructure.

Cloud-Native AOS extends these services natively into cloud-hosted Kubernetes services such as Amazon Web Services Inc.’s Elastic Kubernetes Service as well as bare-metal environments, simplifying data management and disaster recovery for containerized workloads. It allows developers to use Kubernetes application program interfaces to provision, manage and secure data across environments.

It’s the first tangible outcome of the Project Beacon initiative announced two years ago, Caswell noted. “It’s a containerized AOS that was architected from the start to be independent of any hypervisor, which gives us the flexibility to extend deeper into public cloud and further into the edge,” he said.

Initial early access is now available for Amazon EKS, with broader availability planned by the end of 2025 for on-premises, bare-metal Kubernetes clusters. Targeted use cases include backup across cloud availability zones, workload repatriation from public cloud to private infrastructure and unified storage management across hybrid deployments.

Caswell said the ability to provide production-grade data services like snapshotting and disaster recovery in Kubernetes environments without a hypervisor is “a material change” that enterprise computing consistent “whether you run in the data center, in the cloud or at the edge.”

Agentic AI with Nvidia

The latest release of the Nutanix Enterprise AI platform now features tighter integration with Nvidia AI Enterprise software, including Nvidia Inference Microservices and the Neural Modules framework. The company said customers can develop and run agentic AI workloads — AI systems that use multiple models and tools to plan, reason and act autonomously — across on-premises and cloud infrastructure.

Nutanix said the release positions NAI as a unified control plane for AI agents spanning multicloud environments.

The update includes a shared model service architecture intended to reduce resource consumption by allowing different applications to reuse a single set of large language model endpoints. Nutanix said the approach helps conserve graphics processing unit, central processing unit, memory and storage resources while simplifying deployment and scaling.

Nvidia integration supports a range of agentic capabilities:

• Secure, shared LLM endpoints for reuse across multiple applications.
• Support for Nvidia Llama Nemotron open reasoning models, NeMo Retriever, NeMo Guardrails and function-calling for external data integration.
• Guardrail models to enforce content filters, topic controls and prevent prompt injection or jailbreaks.
• Tighter coupling with the Nutanix Cloud Platform and Unified Storage for structured and unstructured data access using technologies like Nvidia GPUDirect Storage.

Applications can be deployed on Nutanix’s hyperconverged infrastructure, bare-metal servers and Kubernetes distributions certified by the Cloud Native Computing Foundation.

“Many of our customers are excited about AI but don’t know how to get started,” Caswell said. “We’re helping them by integrating models for guardrails, reranking and embedding directly into enterprise workflows so they can deliver production-grade LLMs using their private data.”

He added that customers don’t necessarily want to train models themselves but need a reliable path to deploying them: “They are looking for a fast path to deploying optimized models for AI applications lwith a choice of LLMs, GPUs and infrastructure,” he said.

VMware customer enticements

The Pure Storage partnership will combine Nutanix Cloud Infrastructure, powered by the Nutanix AHV hypervisor and Flow networking and security, with Pure’s FlashArray connected by the high-speed Non-Volatile Memory Express/TCP storage interface.

The announcement is intended to take advantage of the turmoil in the virtualization market that followed Broadcom Inc.’s acquisition of VMware Inc. and subsequent changes in licensing and support termss. Gartner Inc. last year predicted that “cost pressures are driving enterprise-scale VMware customers to migrate half their virtual workloads by 2028.” Nutanix and Pure aim to capitalize on this trend by offering an alternative to VMware-based stacks.

Key elements of the planned offering include support for compute- and storage-intensive applications such as AI and data analytics, integration of Nutanix’s Flow microsegmentation and disaster recovery orchestration with Pure’s data encryption and SafeMode immutable snapshots and support for servers from Cisco Systems Inc., Dell Technologies Inc., Hewlett Packard Enterprise Co., Lenovo Group Ltd. and Super Micro Computer Inc. Cisco said it will include Nutanix in its FlashStack converged infrastructure platform.

“This is a material change in our architecture,” Caswell said. “We’re extending Nutanix value to customers with large storage estates without giving up the hyperconverged value proposition. It’s a complementary play for customers who want storage managed independently but still want application-centric control.”

Caswell framed the new model as addressing customers who are hesitant about transitioning away from VMware. Early access is planned for mid-2025, with general availability targeted for year-end.

Photo: SiliconANGLE