Nutanix Expands Enterprise Infrastructure for Production Agentic AI
Nutanix has announced enhancements to the Nutanix Cloud Platform (NCP), including the general availability of Nutanix Enterprise AI (NAI) 2.8 and the upcoming Nutanix Kubernetes Platform (NKP) 2.19. The company says the updates are designed to address a growing challenge for enterprises moving AI from experimentation into production: running AI workloads alongside applications and data that remain distributed across virtualised and containerised environments.
Rather than requiring organisations to create separate infrastructure silos or rearchitect existing workloads, Nutanix is taking a dual-native approach that supports virtual machines and containers through a common operating and governance model. Thomas Cornely, Executive Vice President, Product Management, Nutanix, said enterprises should not have to rebuild existing systems to adopt AI. He said the company’s approach is designed to allow organisations to run AI alongside existing applications and data while maintaining consistent operations and governance.
NAI 2.8 provides a centralised platform for AI inference and agentic AI, with new capabilities aimed at controlling how AI agents interact with enterprise applications and data. A key addition is the generally available Model Context Protocol (MCP) Gateway, which provides a controlled interface for AI agents connecting to tools and data through MCP. Nutanix has also introduced an MCP Server for NCP to support development of agentic applications with access to infrastructure managed through the platform.
NAI 2.8 also introduces enhanced private inference and fine-tuning capabilities. These include multi-GPU inference using tensor parallelism, parameter-efficient fine-tuning and speculative decoding, which Nutanix says can accelerate LLM token generation by up to 2.5 times. Security is another focus, with granular identity and access management, custom roles and least-privilege controls intended to reduce the risk posed by rogue AI models and agents. The platform also supports air-gapped NVIDIA NIM deployments.
The upcoming NKP 2.19 is designed to simplify Kubernetes operations across bare-metal and virtualised environments. The release will introduce NKP Metal, aimed at automating operating system, firmware and container deployment on bare-metal Kubernetes infrastructure. NKP on Nutanix AHV will continue to support virtualised environments, with integration with Nutanix Flow providing network-level isolation for AI agents.
An AI Applications Catalog will also provide curated deployment options for technologies including Kubeflow, Milvus and Slurm. NKP has achieved Cloud Native Computing Foundation certification for Kubernetes AI conformance, supporting Nutanix’s positioning around standardised infrastructure for enterprise AI workloads.
The announcement comes as organisations across the Middle East and Africa increasingly move AI from proof-of-concept projects into mission-critical environments. Mohammad Abulhouf, Vice President and GM, Middle East & Africa, Nutanix, said enterprises in the region need to scale AI without separating it from the applications and data that underpin their operations.
The company is also expanding its ecosystem through its Powered by Nutanix: Verified Services programme and Service Provider Central, aimed at helping partners develop and monetise managed infrastructure, cloud-native and AI services. For enterprises, the broader message is straightforward: production AI increasingly requires not just powerful models, but infrastructure capable of keeping AI close to enterprise data while maintaining security, governance, flexibility and control.
