As enterprises move from experimenting with generative AI to operationalising it at scale, the underlying infrastructure is becoming a central challenge. The rapid evolution of large language models (LLMs) over the past three years has driven unprecedented demand for compute, storage and networking architectures that can support growing model sizes, higher token volumes and increasingly complex AI pipelines.
Against this backdrop, Nutanix and NVIDIA are positioning an integrated AI operating environment aimed at simplifying how enterprises deploy and manage AI infrastructure built on accelerated computing platforms.
From experimentation to AI factories
The initial wave of publicly available LLMs demonstrated how token-based prediction could be applied to tasks such as text generation, image creation, summarisation and translation. As organisations began moving these capabilities into production, infrastructure requirements expanded rapidly—placing pressure on performance, scalability and power efficiency.
NVIDIA’s AI infrastructure has emerged as a foundational layer for processing the tokens that power generative AI workloads. However, as models grow and workloads diversify, enterprises are increasingly confronted with the operational complexity of stitching together compute, storage, virtualisation, containers and AI software stacks.
Nutanix said its response has been to develop an integrated AI operating environment designed to help enterprises assemble and operate AI infrastructure more efficiently, without treating each layer as a separate project.
An integrated operating model for AI workloads
At the core of Nutanix’s approach is its Acropolis Operating System (AOS) for storage and the AHV hypervisor for virtualisation. These are complemented by additional platform components, including the Nutanix Kubernetes Platform (NKP), Nutanix Enterprise AI (NAI), Nutanix Unified Storage (NUS) and Nutanix Database (NDB).
For bare-metal Kubernetes deployments, Nutanix also works with Canonical to offer Ubuntu Pro, aligning operating system support with NVIDIA’s AI software ecosystem. The objective, according to Nutanix, is to provide enterprises with a single, integrated environment that spans infrastructure, orchestration and AI services.
The company describes this integrated AI operating environment as a turnkey system for running what are increasingly referred to as “AI factories”—production environments built on NVIDIA AI infrastructure and NVIDIA AI Enterprise software, including NVIDIA NIM microservices. The emphasis is on reducing the time and complexity required to move from infrastructure delivery to active AI workload execution.
Preparing for next-generation acceleration platforms
Looking ahead, Nutanix said it is working closely with NVIDIA to support upcoming acceleration technologies, including the NVIDIA Rubin platform. Planned support includes NVIDIA Vera Arm-based CPUs and NVIDIA Rubin GPUs across bare-metal and virtualised environments using NKP and AHV.
Nutanix also indicated support for NVIDIA’s Inference Context Memory Storage Platform, enabled by NVIDIA BlueField-4 storage processing, as well as NVIDIA Spectrum-X Ethernet Photonics switch systems to address data centre-scale AI networking requirements.
These developments reflect a broader industry shift toward tightly integrated compute, networking and storage architectures optimised specifically for AI workloads, rather than general-purpose infrastructure.
Open source and ecosystem alignment
Nutanix emphasised that its integrated AI operating environment continues to align with its open-source strategy. This includes support for NVIDIA’s open-permissible models and NIM microservices, allowing enterprises to adopt newer LLM capabilities without being locked into proprietary stacks.
By aligning with NVIDIA’s expanding AI software and hardware ecosystem, Nutanix is aiming to position its platform as an abstraction layer that shields enterprises from some of the complexity of rapid infrastructure change—while still allowing access to the latest acceleration technologies.
Infrastructure as the enabler of AI scale
As generative AI moves deeper into enterprise operations, infrastructure decisions are increasingly shaping what organisations can realistically deploy, scale and sustain. Rather than focusing solely on models, enterprises are now evaluating how quickly they can provision environments, control costs and maintain operational consistency across AI workloads.
The Nutanix–NVIDIA alignment reflects this shift: positioning AI infrastructure not as a collection of discrete components, but as a cohesive operating environment optimised for token generation, inference and training at scale. For enterprises building long-term AI strategies, such integration may prove as critical as model choice itself.