Nutanix expands AI infrastructure platform to control costs as agentic workloads explode

Nutanix Tackles Agentic AI Infrastructure Complexity at .NEXT 2026

Nutanix rolled out significant expansions to its agentic AI infrastructure platform at its annual .NEXT 2026 conference in Chicago, addressing the mounting challenges enterprises face as AI workloads transition from experimental pilots to production-scale deployments 1

1

. The company introduced two critical additions: an AI gateway within Nutanix Enterprise AI that governs which agents access which models and at what cost, and Service Provider Central, which enables providers to build multi-tenant GPU clouds and sell AI service catalogs including GPU-as-a-service and Kubernetes-as-a-service to enterprises facing silicon shortages 1

1

.

CEO Rajiv Ramaswami emphasized that the enterprise computing stack is undergoing its most consequential transformation in decades, with Nutanix positioning itself as the platform where both today's applications and tomorrow's AI workloads run 4

4

. "We truly want to be the platform company where all applications run," Ramaswami told theCUBE, describing how agentic infrastructure has shifted from an experimental workload to the organizing logic of every new application 4

4

.

Cost Per Token Crisis Drives Infrastructure Modernization

The economics of agentic AI are forcing enterprises to fundamentally rethink where they run inferencing workloads. A single user action in an agentic workflow can trigger hundreds of downstream agent calls, each consuming tokens at scale and driving up costs 1

1

. Dan Ciruli, vice president and general manager of cloud-native at Nutanix, explained that customers now face complex tradeoffs: "Do we call an API where we're going to pay per token? Do we use some infrastructure at a service provider where we're paying for time, but then we get to generate all the tokens? Or does it make economic sense to buy some hardware, run it on-prem and now we're just buying electricity?" 1

1

.

This cost control challenge is giving rise to an entirely new discipline: AI FinOps. Nutanix highlighted the growing importance of usage metering and cost per token visibility to prevent AI initiatives from becoming budget liabilities 5

5

. "As you use more and more models, you run into challenges around tracking usage and managing cost," Ramaswami noted 5

5

. For Wynn North America, cost efficiency has become a major factor in evaluating whether to maintain cloud-based AI deployments or invest in on-premises infrastructure 5

5

.

AI Governance and Shadow AI Emerge as Critical Concerns

As agent proliferation outpaces visibility into how AI workloads consume resources, unmanaged AI deployments—what Nutanix calls shadow AI—represent a growing operational risk 3

3

. "As agents sprawl, models and tools need to be controlled and governed," explained Anindo Sengupta, vice president of product management at Nutanix 1

1

. The company's AI gateway addresses this by integrating AI governance directly into the core platform rather than treating it as a separate infrastructure layer 3

3

.

Regulated industries face particularly stringent requirements. Dan Regalado, CIO of Wynn North America, emphasized that "our gaming data cannot leave the state. Data security and data residency are non-negotiable for every one of our resorts" 5

5

. This regulatory reality is shaping deployment decisions, with organizations reassessing whether production workloads belong in public cloud or hybrid environments where governance and control can be maintained 5

5

.

Data Sovereignty and Hybrid Cloud Drive Platform Choices

Data sovereignty has emerged as a major growth vector for Nutanix, with governments worldwide building sovereign AI clouds to keep data and economic value within national borders 4

4

. "The whole move towards sovereignty is here to stay," Ramaswami said, noting that government initiatives to finance these buildouts are creating a direct pipeline of anchor customers 4

4

. Ramaswami predicts that agentic AI will be a "true hybrid application" for most enterprises, with workloads distributed across public cloud, private cloud, edge locations, and neo-clouds—a new class of service providers offering specialized AI services 2

2

.

The flexibility to run workloads across multiple environments matters because proximity to data, real-time inferencing requirements, and regulatory constraints all influence where AI applications can execute 2

2

. Stephen Hall, vice president of infrastructure and operations at BlueCross BlueShield of Tennessee, underscored this reality: "Infrastructure leaders need adaptability because the industry will keep evolving" 5

5

.

GPU Utilization and Storage Become Strategic Priorities

Nutanix's platform evolution centers on making GPU resources work harder by eliminating idle compute that inflates cost per token. "GPUs sitting idle is bad, because think about it—on the one hand you're spending more and more tokens, and if you're going to need to buy more and more GPUs to go and use that, it's not efficient," Ramaswami explained 4

4

. The same virtualization logic that improved CPU utilization a decade ago now applies directly to GPU workloads 4

4

.

Underpinning these capabilities is Nutanix Kubernetes Platform Metal, described as the only dual-native platform supporting any combination of VMs, virtualized Kubernetes, and bare metal Kubernetes from a single control plane 1

1

. The company also announced a partnership with NetApp to integrate NetApp Intelligent Data Infrastructure with Nutanix Cloud Platform, addressing the reality that storage has become the last line of defense in the AI era 3

3

. "When you think about the cybersecurity angle, storage becomes the last line of defense for customers. It's not a matter of if, it's a matter of when," said Sandeep Singh, senior vice president and general manager of enterprise storage at NetApp 3

3

.

Ecosystem Expansion Signals Platform Maturity

More than 100 partners sponsored .NEXT 2026, spanning major cloud, server, storage, and chip providers—a signal that reflects the network effect of a genuine platform rather than a product line 4

4

. A strategic partnership with AMD, in which AMD committed up to $250 million in investment and joint engineering to co-develop an open agentic AI platform, reinforced that the ecosystem is hardening into something structural 4

4

. "The value of a platform is directly tied to the ecosystem around it," Ramaswami said 4

4

.

For organizations with limited IT staff, reducing operational complexity through unified platforms has become critical. Josh Hostetler, lead platform engineer at Tire Rack, described his team's reality: "I'm on a platform engineering team of three people. Our goal was to reduce administrative burden without adding another tech stack or more engineers" 5

5

. As Ramaswami framed it, the shift from prompting to delegating autonomous agents means AI must now be treated like core infrastructure: "This is now about your competitive edge" 5

5

.