Nutanix CEO Ramaswami On AI, AMD GPUs, And Why Partners Need To Move Faster

Nutanix CEO Rajiv Ramaswami tells CRN how the company has expanded beyond hyperconverged infrastructure into hybrid cloud, AI inferencing, external storage and partner-led services opportunities to build a platform that provides customers and partners unequaled infrastructure flexibility and choice.

Anyone who thinks of Nutanix strictly as a developer of hyperconverged infrastructure should have a discussion with Nutanix CEO Rajiv Ramaswami.

Nutanix has actually evolved from its HCI roots into a cloud software infrastructure platform for virtual machines, containers, AI applications, public clouds and external storage, Ramaswami told CRN in an exclusive conversation and as part of a short press conference after the company’s fiscal 2027 financial report.

Ramaswami told CRN the company’s core mission is letting customers run any app and manage data anywhere, from on-premises to Amazon Web Services, Microsoft Azure, Google Cloud and GPU-focused neoclouds.

“We offer customers a lot of flexibility and choice in a single platform,” he said. “They can run their VMs [virtual machines] or containers equally well. They can run on-prem or in the public cloud equally well. Licensing is flexible. You buy a license, you can use it anywhere, and you can run whatever you want.”

That run-anywhere mantra extends to AI, where Ramaswami said Nutanix provides choice to help customers reduce costs.

“AI on Nutanix is all about landing customer applications on the Nutanix platform, and this is where our full agentic experience comes into play,” he said. “Today, if you look at the different layers in the stack, our software stack at the silicon layer runs on Nvidia GPUs but will soon be running on AMD GPUs as well.”

Ramaswami also told CRN that AMD’s recent $250 million investment in Nutanix will help Nutanix support AMD GPUs.

“It’s a joint development effort,” he said. “We work together as [AMD develops] GPUs because we will be supporting those GPUs. We get access to them. And we also get to provide that as a choice to our customers.”

For channel partners, Ramaswami said Nutanix’s broader portfolio creates more opportunities around modernization, containers, AI, cloud, migrations and mitigating component shortages.

“We are not a services company,” he said. “We provide some level of services, but we’re very happy to have our channel partners take on the services piece completely, and from a channel partner’s perspective, this is where they can add value to their customers and value for themselves.”

There’s a lot going on at Nutanix. To learn more, read CRN’s conversation with Ramaswami.

Define Nutanix.

Nutanix is a cloud software infrastructure platform company that provides a full software stack on top of which companies of all shapes and sizes run their existing business-class applications and their modern applications, including their AI applications, anywhere, and manage all the data. The key thing is any app, all the data, managing it anywhere. That’s what we do today as a platform. If you look at the platform at this point compared to where we were, most folks know us as founding the HCI market. We continue to be the leaders in that category. But we’ve evolved this company from being an on-prem HCI company to extend our platform to run on top of the three major clouds: AWS, Azure and Google. We now support a variety of external storage, so we don’t only have HCI. We now have a full cloud stack with compute, storage, networking, operations, automation, management, etc. And storage becomes one of the options: HCI or external storage.

We also have a platform today which allows customers to run modern applications. One of our value propositions is that customers can mix and match containerized applications with virtual machine applications. And finally, with our more recent product portfolio announcements, we have a platform for running agentic AI applications. So we’re one platform today that runs all these applications from traditional virtual machine applications to AI applications wherever customers like to run them.

You talked about how Nutanix’s definition has evolved. How do you think the definition will change over the next couple of years?

It’s a mix of all these. Five years from now, I think a good majority of the applications will be more containerized and modern applications including AI. Nutanix naturally becomes a platform for doing that. There will be increasingly autonomous applications where agents are doing more of the work compared to today. So the nature of the applications will change, and the platform will continue to evolve as and when that happens.

You said Nutanix’s software stack runs on all three major hyperscale cloud providers. Do the hyperscalers run the Nutanix software natively?

The software runs natively on bare metal. All three of these hyperscalers provide bare metal on top of which customers can run the Nutanix stack.

Are the hyperscalers taking advantage of that as well on their own technology?

We have a good partnership with all of them. We have an especially good partnership with AWS, where we also do a lot of joint go-to-market work. I’ll give you one example from our recent conference. State Street is a large global bank, a good company. They were on stage talking about their use case. They run 3,000-plus virtual desktop instances on Microsoft Azure on our platform. They’re also looking to run and migrate a lot of the compute workloads using our platform, our software on top of AWS. So they operate in an environment where they have Azure, AWS and on-prem data center, and they’re using Nutanix in all those.

Who would you now say is Nutanix’s primary competition? Many of your traditional competitors are a lot smaller compared to Nutanix.

If you look at Gartner’s Magic Quadrant, what they call distributed hybrid cloud, hybrid infrastructure, we are in the Leaders quadrant, and along with other big, massive players [including] the public cloud players. For the public cloud players, we both collaborate and compete because they compete for workloads without us. If there’s a workload on Nutanix and it’s run in the public cloud, then they get it. So they are competitors and collaborators.

Of course, we have Broadcom on the one side. They’re the established legacy provider. And we compete more and more against Red Hat.

What competitive advantages does Nutanix have?

I think one of the first things, aside from the technology itself, is our customer NPS [Net Promoter Score]. We take care of our customers. We build long-term partnerships. We are proud that our NPS has remained at 90-plus, which is the very best in the industry, even as we have continued to scale. The second differentiation is we offer customers a lot of flexibility and choice in a single platform. They can run their VMs [virtual machines] or containers equally well. They can run on-prem or in the public cloud equally well. Licensing is flexible. You buy a license, you can use it anywhere, and you can run whatever you want. A lot of flexibility and choice. Third is we try to make things really simple and easy. Typically, Nutanix platforms can be operated by a very small number of people very easily, with a lot of automation. Very simple to use, very robust, very resilient. Consumers like simplicity in terms of operating complex enterprise environments, so we provide that.

How far has Nutanix gone in terms of developing a platform for agentic AI for its own use or for customers?

I think about AI in three different facets. One is customer applications, AI applications running on the Nutanix platform. We call it AI on Nutanix. The second is AI capabilities inside of our own products to make them more automated and easier to use. That’s AI in Nutanix. The last is our own internal use of AI inside the company for efficiency and productivity. We call it AI at Nutanix.

AI on Nutanix is all about landing customer applications on the Nutanix platform, and this is where our full agentic experience comes into play. Today, if you look at the different layers in the stack, our software stack at the silicon layer runs on Nvidia GPUs but will soon be running on AMD GPUs as well. AMD made a strategic investment of $250 million in Nutanix so that we can jointly support their solutions and take them to market.

What was behind that investment?

They made that announcement because they see the value in us providing a solution based on AMD to the enterprise market and really liked the idea of providing our customers choice. That’s at the silicon layer. Our full stack runs on top of GPUs, and we essentially provide the stack to run inferencing applications. We effectively provide inference endpoints. Customers can choose to run any open model they like on this platform. They can pick their model—an LLM or a smaller language model—and run it, and we deliver a turnkey inferencing endpoint out of the stack so their application simply consumes it.

If you look at how an agentic application works, there is an application that typically runs on CPUs inside an enterprise data center. This may be an agent, or it may be an application that needs to consume inference, and this inference can be consumed from a frontier model like, for example, Claude or OpenAI. Increasingly, there’s a lot of focus on cost, and people are also trying to build out their own inferencing solutions with open models because they are more cost-effective and more predictable. We have a gateway solution that sits between the application and the models they are accessing to ensure that we provide visibility on costs and who has access to it and providing some controls there on both usage and access to these models. We also optimize what applications are using what models. Not everything needs a frontier model. Customers will use frontier models when they need to, but they’ll also use open models to reduce costs. Token costs are one of the biggest issues in the use of AI. We try and help customers reduce token costs.

The second thing we provide is the actual inference endpoint itself. That’s a full stack for delivering inference on top of standard GPU-based hardware. That stack can be running inside enterprise data centers, but it can also be running on any neoclouds with GPU access. Customers want to consume GPUs wherever they’re available. GPUs are more available in neoclouds today. You can run our stack on neoclouds and then consume the inferencing capabilities through that.

What does Nutanix get from the AMD funding? Are you getting early access to the GPUs or technical support?

It’s a joint development effort. We work together as [AMD develops] GPUs because we will be supporting those GPUs. We get access to them. And we also get to provide that as a choice to our customers. At this point, Nvidia of course is by far the market leader, but customers want alternatives, and the fact that we also have a solution with AMD is a good alternative for them. … Part of this AMD funding also supports our joint marketing efforts.

Does Nutanix get first access to those AMD GPUs?

I can’t comment on whether it’s first access, but we get early access.

And when will the AMD GPUs be ready?

We expect by the end of the calendar year to have our first solutions with AMD GPUs.

Does Nutanix still sell hardware?

Nutanix does not sell hardware or carry any hardware on our books. We do have what’s called an appliance made by Supermicro. Customers can choose to use that server if they’d like, but they can also choose to get servers from Dell or HPE or Lenovo or Cisco or any server—you name it. We are agnostic from a hardware perspective.

Has the shortage of components, particularly server and storage components, impacted Nutanix, even though you don’t sell hardware?

Even though we don’t sell hardware, we are impacted by it because customers need hardware to run our software. If you look at our business, there’s a ‘land’ component, which is getting new customers on [Nutanix], then an ‘expand’ component where existing customers expand their use, and a ‘renewal’ component where existing customers who already bought a solution renew after their subscription ends. The renewal is not impacted by hardware. That’s a reasonable chunk of our business that doesn’t get impacted as customers continue to use our product. It’s on the land and expand side that we get impacted by rising hardware costs, and it depends on the overall customer budget.

What is Nutanix doing about it?

First, we are providing customers as much flexibility as possible across server configurations and server platforms so they can choose whatever is available and easiest with the lowest cost. Second, we also work on the public cloud, and increasingly, in some cases, public cloud hardware is starting to be more cost-effective than buying a server on-premises. That’s a fairly recent trend, and we’ve seen customers actually make use of public cloud now rather than just buy servers and run our software on it. That may be the easier, cheaper option, right now. Third, and this is where our platform expansion has come in, we support a range of external storage components. A lot of our customers, and we add about 500 to 1,000 new customers every quarter, are looking to migrate [to Nutanix], and in a lot of cases, our platform, our software, runs and supports many of their existing storage arrays. We support Dell PowerFlex, Dell PowerStore, Everpure, and will soon support NetApp. So as long as they have those storage arrays, we can run on their existing servers and replace the software, so they don’t need to buy any hardware to deploy us. That’s a significant benefit during a time where hardware costs have gone through the roof.

What percentage of Nutanix deployments are going with external storage versus Nutanix’s own storage offerings?

Keep in mind that this is relatively new, and it’s still a small percentage overall, but growing very fast. For example, we’ve had Everpure now for two quarters. We've had Dell PowerFlex for most of the year, but that’s a relatively small footprint, but Everpure is fairly broadly deployed. We just got Dell PowerStore up and running. Now we have NetApp up and running. Those are broad platforms, and we saw significant uptick in Q4 with those. … We expect very rapid growth.

How does the external storage versus Nutanix storage decision impact channel partners?

We all think of ourselves now as a cloud platform where storage becomes an option. You have compute, you have networking. You have all the cloud management. You have the Kubernetes version. You have the AI pieces, and you have storage. And storage can be either HCI or external storage, and that depends on the situation and the use case and the customer landscape. Channel partners are very familiar with external storage. They’re very familiar with our platform. So to me, this is a very natural thing. They should just be considering what works well for the customer.

Does Nutanix have to certify the hardware platforms on which the software is run?

Let’s distinguish between HCI and external storage. For HCI, we certify a broad range of platforms across these OEM vendors. We test our software to make sure it’s certified and works, and the customer can order those servers from Dell, HPE, any of these vendors. When it comes to external storage, it’s far less stringent. Our customers can run our software on any server as long as we support its external storage array.

As the company expands its offerings, looking at AMD GPUs and inferencing and so on, how has that changed your relationship with channel partners? Are you having to do more in terms of certification of partners?

There’s a much broader opportunity with Nutanix. It’s not just selling Nutanix for HCI use cases. It’s selling Nutanix for a broad set of use cases that are very central to customers. They want to modernize their infrastructure, run modern applications on containers, work on AI, go to the public cloud. These are all significant top-of-mind use cases for customers for which Nutanix can be a solution. Partners need to get on board, understand the capabilities of the platform, and be trained technically. We would also love for them to become service partners when it comes to migration services. Every significant, complicated migration requires services. We are not a services company. We provide some level of services, but we’re very happy to have our channel partners take on the services piece completely, and from a channel partner’s perspective, this is where they can add value to their customers and value for themselves.

How have channel partners changed their business in response to the increased move by Nutanix to the public cloud?

The public cloud is not new. Channel partners need to embrace it, and we again go through our channel every time, whether it’s public cloud or on-prem. It’s the same license, the same stack we sell. Customers can deploy it on-prem or deploy it in the public cloud. We don’t limit them. So our channel partners are part of the equation. What I would encourage channel partners to do, though, is to proactively have those conversations with customers because the customer is going to find public cloud as an option, and they’re going to do it with or without the involvement of the channel partner. So I would urge them to get up to speed in terms of what’s possible in the public cloud, how they can help their customers, especially in this world where you know buying on-prem hardware is difficult.

What are your strategic priorities for the rest of the year?

It continues to be addressing our growth opportunities across all of these pillars that I described: modernizing infrastructure, running modern applications, AI, helping customers go to the cloud. Those are all priorities focused on driving growth. As part of that, in terms of the tools we have, we’ll have more of a focus on external storage, on public cloud, and on containers and AI. In terms of go-to-market, the ecosystem is getting broader for us. We now have significant OEM partners who are taking our solutions to market, including Cisco, Dell and Lenovo. We’d like to get more leverage through our channel friends here. We would like the channel to be a big force multiplier for us in terms of going after the broad market, to enable the channel to do more of the work up front and also get more of the rewards as a result. Our channel programs [provide] partners with the appropriate incentives to go in there, help us land new customers, expand the customer base, and represent our full portfolio of products. We like to get leverage through channel partners, so that continues to be one of our priorities. And on the product side, as we move forward we will support a broader range of external storage platforms. We will continue to enhance our Kubernetes platform. AI is innovating at a rapid pace at this point, and the market is changing dynamically. We’ll keep up with that and make sure we’re on the forefront when it comes to AI inferencing.