Nutanix and Netris partner on multitenant GPU cloud stack
Nutanix (NASDAQ: NTNX) and network automation specialist Netris have announced a strategic collaboration to deliver a jointly engineered multitenant infrastructure stack targeting neoclouds and enterprises running GPU-accelerated AI workloads. Announced at NVIDIA GTC in Berlin, the combined solution pairs the Nutanix Cloud Platform with Netris's Network Automation, Abstraction, and Multitenancy (NAAM) platform. A tech preview is planned for early 2027; no pricing or commercial terms were disclosed.
The collaboration is aimed squarely at operators trying to monetise large GPU clusters as a service. The core engineering challenge the two companies are addressing is tenant isolation: giving multiple customers access to shared GPU infrastructure while enforcing hard, hardware-level separation at every layer of the stack, from virtual machines through to the physical switching fabric. The joint solution will also expose self-service cloud networking constructs, including VPCs, elastic IPs, NAT and L4 load balancers, accessible via web console, REST API and Terraform.
What the stack delivers
The combined platform is designed to automate network fabric configuration in lockstep with compute provisioning, so that as a neocloud operator onboards a new tenant, the appropriate network segmentation is configured automatically rather than requiring manual intervention. Netris claims 70-plus live deployments of its NAAM platform across AI cluster operators, and the company is backed by Andreessen Horowitz.
Ketan Shah, Vice President of Product Management at Nutanix, said the partnership addresses a genuine complexity problem for GPU cloud operators. "Building the cloud operating stack required to monetise GPU clusters at scale is remarkably complex," he said. "By pairing Nutanix's Agentic AI solution with the Netris NAAM platform, we will provide AI cloud operators with an out-of-the-box solution to accelerate their journey from metals to margins."
Alex Saroyan, CEO and co-founder of Netris, framed the collaboration as a synchronisation problem. "An AI cloud needs two layers working as one," he said, describing compute orchestration and network automation as requiring continuous alignment to deliver reliable multitenant services.
Market context
The neocloud segment, specialist GPU-as-a-Service providers such as CoreWeave, Lambda Labs and a growing number of regional operators, has expanded rapidly on the back of enterprise demand for AI inference and training capacity. These operators compete with hyperscaler managed AI services but differentiate on dedicated capacity, data-residency controls and support for NVIDIA's partner programmes.
NVIDIA's Cloud Partner programme carries its own technical requirements for multitenancy and isolation, and the Nutanix-Netris stack is explicitly positioned to help operators satisfy those requirements. This is a commercially significant angle: qualification under NVIDIA's partner framework can open access to preferred GPU allocation and co-sell motions.
The broader infrastructure management market for AI clouds is contested. Competitors range from established fabric-management vendors through to hyperscaler-native tooling and open-source orchestration projects. The Nutanix and Netris approach is differentiated by its emphasis on hardware-enforced isolation at the switching layer, rather than relying on software-only network virtualisation, which analyst commentary included in the release suggests will be an increasingly important differentiator as inference workloads become more security-sensitive.
The tech preview timeline of early 2027 leaves meaningful execution risk. Buyers evaluating the stack will want to see performance benchmarks across East-West GPU cluster traffic, demonstrated tenant onboarding times, and clarity on which physical switch vendors are supported within the NAAM automation layer before committing to the platform for production deployments.