Nutanix invests $20 million in on-prem AI cluster to curb LLM token costs and target one-year ROI

Facing high hardware costs and exploding token expenses from services like Copilot and Claude, Nutanix allocated $20 million to construct an on-premises AI cluster that runs open-weight models. The platform incorporates a Model Context Protocol gateway to manage identity and data access, aligning it with industry standards. CEO Rajiv Ramaswami said the cluster will eliminate per-token fees and allow selective use of frontier models on external resources when needed.

He also noted that expanding support for Arm architecture should enable customers to run Nutanix software on cheaper servers and facilitate migration away from VMware without hardware replacement. In its latest financial briefing, Nutanix posted Q4 revenue of $757 million, a 16 percent year-over-year increase, and full-year revenue of $2.85 billion, with net income of $1.5 million. The firm added 3,000 new customers during the fiscal year, many of whom chose Nutanix as an alternative to VMware.