Nutanix CEO Reveals $20 Million On-Prem AI Cluster to Slash Token Costs, Eyes Arm Architecture

Speaking to The Register, Ramaswami detailed how Nutanix's software engineering teams initially adopted external AI coding assistants such as GitHub Copilot and Anthropic's Claude, only to see costs “explode” as usage grew. In response, the company shifted to open-weight models running on its own cluster, eliminating per-token fees. "We are no longer paying on a per-token basis," he said. The move reflects a broader industry trend of organizations matching models and infrastructure to specific workloads rather than defaulting to large language models.

Nutanix also used the briefing to signal progress on porting its stack to the Arm architecture, a shift from its 2024 position where Arm support was seen as a future consideration but not an active project. Ramaswami said Arm compatibility would allow Nutanix to run on lower-cost hardware, addressing customer concerns about high hardware prices and availability. The company is also expanding its hardware compatibility list and supporting external storage devices to ease migrations away from VMware without requiring hardware replacements.

On the AI front, Nutanix updated its Enterprise AI suite with a Model Context Protocol (MCP) gateway, providing identity management and security controls for agents accessing MCP servers. The feature is quickly becoming standard in packaged AI infrastructure stacks.

Financially, Nutanix reported full-year revenue of $2.85 billion, up 12% year-over-year, though net income remained modest at $1.5 million. CFO Rukmini Sivaraman said the company is “focused on delivering sustainable growth and improving profitability.” Ramaswami noted the company added 3,000 new customers during the fiscal year, many choosing Nutanix as a VMware alternative, and predicted the migration trend would continue for another five years.

Why This Matters

Nutanix’s $20 million on-prem AI cluster signals that even large enterprises see token-based AI services as too expensive for widespread internal use, potentially accelerating the shift to open-weight models and on-premises infrastructure.
The move toward Arm architecture could lower hardware costs for Nutanix customers, making hyperconverged infrastructure more accessible in a climate of tight IT budgets.
The company’s continued success winning VMware customers underscores the disruption in the virtualization market, with implications for the broader enterprise IT ecosystem.

Background

Nutanix pioneered hyperconverged infrastructure (HCI), combining storage, compute, and virtualization in a single platform. It has long competed with VMware, which dominates the virtualization market. In late 2023, Broadcom’s acquisition of VMware led to licensing changes and cost increases that drove many customers to explore alternatives. Nutanix has positioned itself as a primary alternative, offering a more flexible and cost-effective platform.

In 2024, Nutanix executives described Arm support as a “future consideration” but not an active development priority. The company’s recent shift reflects a broader industry move toward Arm-based servers, driven by energy efficiency and cost advantages. Meanwhile, the explosion of generative AI has forced many companies to reassess cloud-based AI spending, with on-premises clusters becoming more attractive for specific workloads.

Nutanix’s financial performance has improved steadily, with revenue growing 12% in fiscal 2026, though profitability remains thin. The company’s investment in its own AI infrastructure is a bet on both cost savings and product development.

Key Perspectives

[Nutanix leadership]: Ramaswami argues that owning AI infrastructure gives the company control over costs and flexibility to optimize for its needs. He sees Arm support as another way to lower customer barriers to adoption. [Existing Nutanix customers]: Many are likely to welcome lower hardware costs and expanded compatibility, especially those migrating from VMware. However, some may be cautious about Arm adoption given the architecture’s limited software ecosystem in enterprise environments. [Industry analysts and competitors]: Critics might question the scale of Nutanix’s net income ($1.5 million on $2.85 billion revenue) and whether its growth can be sustained. On AI, analysts note that MCP gateways are becoming commoditized, meaning Nutanix must differentiate elsewhere.

What to Watch

Adoption of Arm-based Nutanix clusters by early customers and performance benchmarks.
Nutanix’s net income trajectory: whether it can improve margins while investing in new architectures.
Further VMware migration announcements from large enterprises, which could drive Nutanix’s customer growth.

ZOTPAPER