The Challenges of Building AI Factories

Despite the promise, building and operating AI factories is complex. Organisations face hurdles like fragmented infrastructure, a very fast rate of change of hardware and software stacks (including models), heterogeneous GPU and CPU generations, and the need for scalable, secure environments. Add to that the need to manage demands from a variety of sources within the organisation, all with their varying levels of security and performance needs and levels of separation from other entities. These challenges can slow deployment, increase costs, and introduce risk—especially when managing complex and sensitive workloads like agentic AI.

Nutanix: A Foundational Software Platform for AI Factories

Nutanix plays a critical role in enabling AI factories by providing a secure, scalable, and efficient software platform that enables organisations to easily create shared inference services using the Nutanix Enterprise AI (NAI) solution to meet the diverse AI needs of the organisation, providing freedom of choice of models as well as underlying hardware.

In addition, the Nutanix Kubernetes Platform (NKP) solution provides a production-grade Kubernetes® base without the complexity of managing each individual component of the stack that otherwise makes running Kubernetes difficult.  The Nutanix platform complements the AI factory solutions offered by various hardware vendors based on NVIDIA Enterprise Reference Architectures.

Added Challenges for Government and Security-Conscious Organisations

For government agencies and security-sensitive enterprises, the bar is even higher. Data privacy regulations, compliance with standards like FIPS encryption, and the need for sovereign control over infrastructure and data create additional layers of complexity. According to a 2025 US Government Accountability Office (GAO) report, 10 out of 12 federal agencies cited policy and compliance as major barriers to AI adoption.

To meet these challenges, NVIDIA introduced the AI Factory for Government reference design—a full-stack AI architecture tailored for deploying AI workloads in high-assurance environments—provides a blueprint for meeting these challenges. Built on the NVIDIA Blackwell architecture, it integrates NVIDIA-Certified Systems, NVIDIA Networking, and NVIDIA AI Enterprise software, along with government-ready containers and open models like NVIDIA Nemotron. 

NVIDIA’s government-ready containers have been tested to run reliably on NKP and interoperate with NAI. This provides seamless deployment of the AI Factory for Government using NKP as the Kubernetes base, and NAI to build the shared inference service to operationalise the NVIDIA software components.

Integrated AI Factory Solution Overview

The joint solution provides consistent performance, simplified operations, and leverages hardened configurations, FIPS encryption, and continuous vulnerability monitoring, making it ideal for regulated environments. NKP can run on virtual machines - often preferred for multi-tenant use cases - or on bare metal servers where there isn't a need for a virtualisation layer. Earlier this year, Nutanix announced that NKP will ship Ubuntu Pro as an additional option with NKP Pro and Ultimate licenses for a validated, well-integrated bare metal Kubernetes stack on which to run your AI workloads.

Together with leading hardware OEMs, NVIDIA, and Nutanix deliver complete AI factory solutions for enterprises and governments. These solutions incorporate NVIDIA AI architecture for accelerated computing, NKP for simplifying the Kubernetes layer, and NAI for scalable inference management, enabling organisations to scale AI workloads across on-premises and hybrid cloud environments. 

The enhanced security and encryption technology included in the AI Factory for Government reference design spans all the components of the joint solution, including NKP.  The flexible design supports future growth without model or deployment mode lock-in. It adapts easily to evolving use cases, from law enforcement and healthcare to financial management and national security.

SUPPORTING QUOTES

Justin Boitano, vice president, enterprise AI products, NVIDIA.

"Enterprises and government agencies are seeking secure, scalable AI architectures to streamline operations and solve challenges that demand both flexibility and compliance. Together, NVIDIA and Nutanix are enabling organisations to confidently deploy AI factories that accelerate innovation while meeting the highest security and regulatory standards."

Jeremy Foster, Senior Vice President and General Manager, Cisco Compute, Cisco.

“With Nutanix NKP and NAI, we’re helping customers bring Cisco Secure AI Factory with NVIDIA to life - delivering secure-by-default AI deployments that run on Cisco’s GPU dense servers with NVIDIA AI infrastructure and software. By simplifying Kubernetes and AI lifecycle management, Cisco and Nutanix are helping organisations move from pilot to production faster, and accelerate ROI on their AI investments.”

Thomas Cornely, SVP of Product Management, Nutanix

“We’re excited to work with NVIDIA and hardware leaders to deliver a fully integrated, hardened AI factory solution tailored for government needs. With built-in FIPS-compliant encryption across every layer, we’re helping organisations protect sensitive data while accelerating AI adoption. By leveraging NKP and NAI, customers gain robust governance features—like fine-grained access controls, auditing, API token management, and full infrastructure visibility—enabling a fast time to first token, cost control, and the robust security for mission-critical AI workloads.”