Nutanix and Supermicro plan joint AI infrastructure solution for private inference

Nutanix, a hybrid-cloud and AI platform provider, and Super Micro Computer (NASDAQ: SMCI) announced a jointly engineered AI infrastructure solution powered by Nutanix Agentic AI to help enterprises curb runaway AI costs link. The offering combines the Nutanix Cloud Platform with Supermicro servers, eight NVIDIA Blackwell Ultra GPUs per node, and NVIDIA Spectrum-X networking, aiming to reduce custom configuration and integration complexity.

“Collaborating with Nutanix, we want to accelerate time-to-online for enterprise AI,” said Matthew Thauberger, Chief Revenue Officer, Supermicro.

“Combining Supermicro’s engineered servers with Nutanix’s full-stack software will deliver a pre-validated, rack-scale solution that bypasses traditional lead-time delays. It will help give organisations predictable access to high-density GPU compute while removing the operational complexity of scaling AI.”

“Moving AI from experimentation into production remains a challenge for many enterprises, with infrastructure complexity and hardware procurement often slowing progress. Our expanded collaboration with Supermicro is a direct response to this barrier,” said Thomas Cornely, Executive Vice President, Product Management, Nutanix. “The Supermicro systems, engineered with our Nutanix Agentic AI full stack software, will help enterprises simplify how AI infrastructure is deployed, operated, and secured, and scaled, enabling them to simply deploy private inference and control the cost per token as AI adoption scales.”

Supermicro with Nutanix Cloud Platform: Reducing the Complexity of Enterprise AI Infrastructure

• The engineered solution from Supermicro and Nutanix is intended to provide a hardware foundation that helps eliminate custom-order delays by using the Supermicro server systems which houses eight NVIDIA Blackwell Ultra GPUs per node.
• The resilient single-tenant control plane is expected to feature a dedicated, resilient Nutanix management cluster (NX-8170-G10) to protect against management plane failure without disrupting GPU execution.
• For high-performance storage scaling, an integrated NUS cluster (NX-8150-G10) is optimised for high-bandwidth I/O requirements across AI data ingestion, checkpointing, and cache warming.
• NVIDIA Spectrum-X Ethernet networking combines NVIDIA Spectrum-4 Ethernet switches (SN5610/SN5600) with NVIDIA ConnectX-7/8 SuperNICs for high-performance, predictable GPU-to-GPU scale-out connectivity and NVIDIA BlueField-3 DPUs for infrastructure offloading and in-band management, complementing NVIDIA’s scale-in networking architecture.

“As organisations across the Middle East and Africa accelerate their AI initiatives, the focus is shifting from experimentation to delivering measurable business value from AI in production. However, the cost of inference, access to GPU infrastructure and the complexity of integrating multiple technologies can slow this progress. Our expanded collaboration with Supermicro is designed to address these challenges by simplifying the deployment and management of private AI infrastructure, giving enterprises greater control over their data, security and AI operating costs. By combining validated server systems with Nutanix’s full-stack AI software, we aim to help organisations across the region scale AI workloads more efficiently, reduce deployment complexity and build a more predictable foundation for long-term AI adoption,” said Sush Kajaria, EMEA Alliance Sales Director, Nutanix.

“The AI imperative is shifting from simply building raw compute capacity to optimising the cost and efficiency of every inference delivered. Hardware, memory, networking, model optimisation, and orchestration all need to be managed as a system if organisations are going to scale inference efficiently and translate AI investment into real business outcomes,” said Matt Kimball, Vice President & Principal Analyst, Moor Insights & Strategy.

Availability The Supermicro systems powered by Nutanix Agentic AI will be available early 2027 from Supermicro and its channel partners.