Nutanix, Inc. (NTNX) Earnings Call Transcript & Summary

All right. I think we're about to -- ready to get started here. Rajiv, Rukmini welcome to the GS Communacopia Technology Conference. Thank you both for being with us today.

Great to be here. Thank you.

Thank you.

Well, Rajiv, let's start with a big picture question. Where would you like to see Nutanix in the next 5 years?

Yes. I think it's -- for the last many years, since we were founded as a start-up way back 15 years ago, we've been a pioneer and innovator of HCI, but always been a challenger in the market, playing up against and winning our fair share against large incumbents whether they be the legacy storage vendors, whether they be companies like VMware and competing against public cloud. And where we've come today and where we're going is that we've now built out what I call a full infrastructure platform. A platform that can run all kinds of applications, existing applications, the new modern applications, the AI applications being built in the future. And they can be run anywhere and the data associated with these applications can also be managed anywhere: on-premises, edges, public clouds. And the opportunity I see for Nutanix over the next 5 years is to become a de facto platform in this area, a leader in this area that companies of all shapes and sizes and organizations trust as a trusted innovation and long-term partners that they can run their businesses on and as a leader. So some of the proof points I'd just say that we are pretty happy about is, for example, over the last couple of years, we've been named as a leader in Gartner's distributed hybrid infrastructure Magic Quadrant. And what gives me pride is the fact that we are the smallest company to be in that leadership quadrant up against people like AWS and Microsoft, big companies that are out there, and Oracle. So our vision, again, aspirations, I should say, are really to become the de facto platform for companies -- for applications and data.

That's a great jumping off point. But before we dig into some of those pieces, I wanted to get -- talk a little bit about your background, right? You spent time at Cisco, Broadcom, VMware before landing at Nutanix. Can you tell us a little bit about what you took away from those experiences and how that's informed your leadership of Nutanix?

Yes. I've learned a fair amount that I think have proven to be quite helpful for my current job. I've learnt the market dynamics in the space that we operate in quite a bit. I've learned what it takes to innovate in this market, what kind of products need to be built and how to look at what's happening in terms of the future and anticipate that. Also, I think the businesses that I ran and many of those companies are much bigger than what I'm running today, right? And so I've learned out of scale. And a lot of these cases, we've had to go scale businesses that -- to multiple billions of dollars. And I think that notion of what it takes to scale these organizations and execute, I think, is what have brought to the table as well with Nutanix. So it's been a good journey. It's been 5 years now almost, and I'm looking forward for the next 5.

That's great. Now Rajiv, the debate between private and public cloud has been ongoing for years. So from your perspective, what are the enduring advantages of private cloud infrastructure? And why do you believe this remains a resilient and strategic opportunity in today's landscape?

Yes. And if anything, I would just say over the last 5 years, the pendulum has swung back and forth on this. It used to be -- 5 years ago, a lot of CIOs were talking about just going all in on the public cloud. And I think the ones that did go in realized that it's not for everything, right? There's a play for the public cloud. Clearly, it's great for getting new workloads up and running quickly, new apps up and running quickly. It's like a drug. It's very easy to consume and get it done, but then it's very expensive. It's a very expensive habit. So what happens today is that everybody realizes that the private cloud, if it's properly engineered, can be much more cost-effective. But beyond that, there's a lot of gravity on data. There's, for example, concerns on sovereignty, data locality, privacy, security and real-time actions. And now we're getting into this AI world. And same thing happens with AI, it's all about the data. And so where is that data, where is the data being generated, where is it going to be consumed and you're going to need to run your workloads right there and a lot of data is being generated and consumed in data centers and in the edges and manufacturing sites, and if you're an oil and gas company, exploration sites, etc. So there's a lot of action happening on-premises and at the edge as well. So which is why I do believe that this is actually an enduring opportunity for what we call hybrid cloud, which is, it's not one or the other, it's really both. And it's about being able to manage these applications and data wherever they may be and create that infrastructure that can be flexible. And that's what we are aiming to do at Nutanix.

And why is it that you think that most of the AI workloads today are running on public cloud? Do you inevitably see a situation where some of that gets repatriated back to on-prem?

Well, actually, I don't think there's a huge amount of AI workloads. Right now, a lot of action has been on training and building these models. And that's naturally a place -- public cloud and large compute centers are where you build these models, you get them trained. And where I think it gets much more distributed is when it comes to inferencing and a slew of new applications are being built that do AI inferencing. Now if you look at the market, so far, it's been all about training. And then on top of that, there's been software companies that are using some of these AI models, these AI native companies to go -- starting to deliver services. And that also starts to happen in a public cloud. But then as enterprises and organizations start actually using AI to meet their needs on a daily basis, they're going to find again that -- and we're already seeing this, that they have to do this where the data is. And the data, again, is a lot of it is on-prem, a lot of it is in the edges. And that's why I think AI like every other application will also be a hybrid application.

Very interesting, and I want to come back to that as pertains to GPT-in-a-Box, but I want to pull Rukmini into the conversation. Rukmini, Nutanix recently reported 4Q results. It sounds like there may have been a bit of confusion around how to interpret ARR, NRR given some of the larger, more complex transactions that you guys are now doing on top of the change in methodology that you guys introduced. So maybe you could clarify for the audience with the dynamics at play here?

Yes. So the change in methodology that you are referencing is we proactively decided to align our ARR recognition with when customers are provided with licenses. And so this is purely a timing adjustment. It's a proactive thing that we're doing. It's -- like a start of a new fiscal year seems like a good time to do it. And if you look, we also provided a schedule for investors at the back of our investor deck where you could see both the new and the old methodology, very clearly for the last 8 quarters, so that folks had all the information. What we're also reminding folks is that going forward, all of our ARR and associated metrics, NRR, like you mentioned, are going to be based off of the new updated methodology. So the starting point for Q1 that folks should think about is the ending ARR for the most recent quarter under the new methodology, right, although that this is a transition quarter, if you will. And I would say that the reason we do this is because, again, it's aligned with when customers have provided the licenses and it's more reflective of just say how industry practice, but it's purely a matter of timing, right, in terms of the [ actuals backed ] itself.

Very good. And maybe moving on to kind of fiscal '26 guidance more broadly. Can you walk us through some of the puts and takes that inform the initial guide? How do you balance some of the long-term secular tailwinds versus some of the near-term dynamics you're contending with: Fed uncertainty, duration headwinds, deal cycle elongation?

Correct. So I think a few things. Maybe I'll start first with what do we see as our long-term growth drivers. And that obviously also will continue in fiscal year '26. So the first one is our continued deployment of our HCI architecture, as Rajiv said, which was our pioneering architecture into legacy environment. So the vast majority, we believe, of a private environments are still running a legacy, what they call a legacy 3-tier architecture: separate compute, storage and networking. And so there's a big opportunity for us to go in there and displace that with our modern architecture. That's our bread and butter. We've been doing it for years, and we intend to continue to drive that. One tailwind to that growth driver is what is happening in the market from a competitive perspective, which gives us an opportunity or a multiyear opportunity really to accelerate that move. So think of that as 1A and 1B of our growth drivers. Another one I would bring up here is what Rajiv said earlier in terms of the world becoming more hybrid, he talked about the distributor hybrid infrastructure that Gartner puts out and so on. And we think that's still in early innings, and we have more value to add there. The third driver, I would say, is cloud native and modern applications. which is addressed by our Nutanix Kubernetes Platform, NKP, which is relatively small and newer for us, but it's a very mature product and platform that we acquired a company about 18 months or so ago that had been existence for a while. So the technology is quite mature and now we're able to take it to market. We've had a good year with NKP in fiscal year '25, and we expect that to grow going forward. And that's a secular win because we can now bring to our platform the more modern and cloud-native workloads. And then the last growth driver, I'll say, is around our partnerships. So we've announced a slew of partnerships, and those should be driving more leverage and more incremental top line for us. So those are kind of all the things that are the growth drivers. As you said, there are things that also give us some caution in terms of overall uncertainty in the critical geopolitical environment, U.S. Fed uncertainty because of just how they're buying, changes of personnel and so on, that we've also had to bake into the guide and then contract duration. So our revenue, I think, as you know, is not fully ratable. There's an upfront component as well that is that is affected by contract duration, and we expect duration to blow down slightly year-on-year. And so that, those 2 will be a bit of an offsetting effect. But that's all baked into the guide that we gave for fiscal year '26, which is about 15% growth at the midpoint.

Very helpful color. And for Rajiv, one dynamic that's unique about the partnerships that Rukmini is referencing here is when you think about Dell and Pure Storage, right? Nutanix is now supporting external storage. You already announced 2 early wins with PowerFlex, pretty exciting, considering it's only been in the market for a couple of months. So can you talk about what informed this decision to really start supporting external storage? And what are the advantages to this approach?

Yes. I mean I think for the longest time, our whole approach was we were going to displace external storage with our HCI platform. And that still continues to be a big motion for us. Now if you look at the progress that we've achieved against that, in this enterprise market over the last 15 years, I think HCI has gotten to be maybe 20% to 25% of the overall market. The remaining 75% or 80% is still very much on traditional storage. Now we will continue to eat into that over time, and that's still very much a core motion. And in the past, by the way, there was really no opportunity or no reason to support external storage because, first, the external storage folks viewed us as a competitive threat. And second, there was really not much interest in customers wanting to displace a VMware hypervisor for somebody else. So we didn't have this opportunity in the past. But what's changed now is the fact that VMware now has raised prices and therefore, customers are looking for an alternative. And they're also pushing HCI with their VMware Cloud Foundation, which is putting the external storage folks also at risk and they need to partner with somebody else. And that's why we've seen this sort of interesting situation in the market where we have customers coming to us as well as the storage partners coming to us saying, please support, and let's work this, and that was the genesis of us starting out with Dell and then now Pure Storage. And I do expect that we'll do more of these over time. So the rationale for us as we get into this market now is that we can now insert into the broader market with a subset of our offering in a much easier way, without requiring customers to make architectural changes. They can continue to deploy the same architecture that they were deploying, use the same hardware that they were working on. And just overlay out software instead of the current software that they have. So it's a much easier migration path for the customer than requiring them to migrate to HCI. So I look at this and say, "Okay, now we can go up to that remaining 75% and we can get in the door faster in some of those accounts than we could if we just only had HCI." So that's what drove the decision.

Right? So now you can wedge into the 3-tier environments. So it's kind of a 2-step motion, right?

Exactly. And over time, we'll try to earn or keep with the rest of HCI.

Yes. And what's your confidence rate that inevitably, when these come due for hardware refreshes that you can start to roll those over into full stack HCI deployments?

Well, not all of them will roll over. Some portions -- some subset of those, we would be able to roll over into a full stack. Others will continue on external storage. And now we have the ability to go figure out and manage across both of these environments. And we also have the ability to tier our solution as well from a pricing perspective in terms of a full stack and maybe a subset of the full stack. And it just gives us a lot more weapons in our arsenal to go out there and compete.

And Rukmini, I think one question that's always on investors' minds is, now that you're supporting external storage, how do the economics compare relative to a full stack HCI?

Yes. So as Rajiv said, this is not the full stack in terms of the capability. At the same time, we are providing an option to customers who are looking for an alternative at an important time for them. And so if you think of sort of the bookends on one end is what our full stack offering is offered at. And then on the other end, is what the main competition, right, in this case, Broadcom's VMware used to have available, which they no longer do, by the way, in terms of the stand-alone hypervisor solution, which they used to have along with some other capabilities. So those are the 2 bookends, Matt. And for us right now, I think we have room in terms of where we price around that. And so we are trying to balance, and its early days, I should say, because it's just -- as you said, the solution has just been in the market for a couple of months here. So what we're trying to balance is where there is value for a customer to adopt this solution, but also giving us flexibility and making it a smaller hurdle for them to switch to the full stack when there is an opportunity like down the road when they are depreciating their hardware to switch to HCI. So we're balancing a few things there as we think about pricing and it's early days. So we're also discovering what the market will bear.

I just want to make one point on that, which is sometimes when people think about this compute offering, they think of it, it's a hypervisor. It's actually much more than a hypervisor. Because if you look at our stack today, we have the compute virtualization. But then we have a network virtualization. We have the operations. We have Kubernetes. And yes, we also have our own storage. Now effectively, the way you should think about this external storage offering is it's got everything in there except for the built in storage, right? So it is a full stack of compute, network operations, Kubernetes, all of that is in that stack. It's much more than a hypervisor. And then the customer has the choice of whether to use HCI storage or external storage, and as Rukmini pointed out, there's a big pricing umbrella that we can go work under to make this happen.

Does that come all bundled? Or can they adopt NKP down the line and the...

Yes. We always want to provide our customers flexibility, to use what you need, not force it upon the customer, not force them to buy stuff they don't want. Because for us, it's all about the customer success and then adoption -- they have to adopt it and they have to be happy with that, and then they have to renew and expand with that product. So that's a motion that we try. So we are very focused on what they want and how they can adopt it and be successful.

Very helpful. And I want to drill in on the new customer momentum. It's been very strong for Nutanix for the last several quarters. You've been investing quite a bit in channel enablement. You've also announced several OEM partnerships. We touched on a few of them. How would you rank the drivers of this accelerated customer momentum? And how durable do you believe this is?

Yes. I think, first of all, I think for us, I would say the first part of it is the fact that we have, as an organization, focused on going and landing new customers from a GTM motion, from a go-to-market motion in terms of customer programs, partner programs, our own sales force and incentives for our sellers. So we've done a lot of that work ourselves. The second, I would say our channel has certainly contributed, right? We have a channel-first model. And we have incentives into the channel partners. Many of these channel partners have these as existing customers that may be new to the Nutanix family, and we welcome those customers coming in. And the third, OEMs, especially Cisco, I would say, has been a good contributor of new logos for us over the past entire year as this is the second year of Cisco with us. So I think all 3 have contributed quite a bit. And of course, there's -- on top of all that is a general notion of many customers wanting to migrate away from Broadcom VMware. So that just provides an overall umbrella under which we are executing.

Right. Very helpful.

I'll add one thing there. As we said on the call that we expect to continue to add mid- to high 3-digit new logos on a quarterly basis. So more or less sustain the number of new logo additions. Now the growth has been really good in fiscal year '25. So that growth rate is going to be hard to sustain off of what is now a tougher compare. But as Rajiv said, there's lots of reasons why we want to continue to add customers on top...

I mean I'll just give you a perspective on that. I mean, so today, Nutanix has about 30,000 customers. The addressable universe out there is about 200,000 customers. So there's still a lot of companies that we can go try and land.

Great. Nutanix's product velocity has expanded the scope from infrastructure modernization to really hybrid multi-cloud enablement. How are customers now embracing this broader portfolio in practice? And what are some of the most compelling ways they're adopting more of the platform?

Yes. I think the -- one of the main things we've seen is that the number of use cases within the customer environment has gone up substantially. So we used to just be, okay, on-prem HCI, and that was our motion. That's what -- that was the use case that we were selling. Now today, we are selling a much broader platform, right? They can use us with external storage, they can run their modern applications with Kubernetes and I'm very happy with how that part of the business has grown over the last year. It's still small, but we're doing well there, and we expect to continue growing that portion of the business, where they can use this on-prem as well as on top of public clouds. And we've seen some initial -- I mean, we've had that in the market for quite a while with AWS and Azure. They're bringing Google to market. We've actually seen over the last year, some initial success in landing a customer first on the public cloud. And again, a lot of this is driven by the fact that these were VMware customers in the public cloud, and they're looking for an option. And migration from VMware to Nutanix in the public cloud can be done fairly easily very quickly. And so we've actually seen a handful of customers, I can't call it a trend yet, but a good set of customers actually start their journey with Nutanix now in the public cloud. So you can see that this all this widens aperture of what we can go offer a customer in terms of meeting their needs and meeting their needs for applications and being able to run them in many different places.

So you're starting to see more lands with NC2 as the anchor product. Is that on the back of some of these deals that you struck with AWS around the migration?

Yes, that's absolutely correct. Yes. In fact, that AWS expansion partnership is a very critical part of this.

And it's still ongoing?

Yes, ongoing. Yes.

And then, Rukmini for you, how early are we in the multiproduct attach opportunity, right? You guys have a ton of new products to sell as a part of the broader platform. So when do you expect this to be a more meaningful contributor to NRR beyond kind of the workload expansion opportunity?

Yes. So to be clear, it does contribute today. I think we can do more. I think, is the -- and so you're right to ask the question. And so what we've done over the last, call it, 12 months or so is to hire more portfolio specialist sellers, those who are experts in areas like NDB or NKP, which are Kubernetes platform in order to assist our main core sellers to drive more of this portfolio attach, because our core sellers and the technical sellers who work with them, certainly understand the platform really well and what customers are looking for, but they're not as deep experts in this particular product area. So the idea is to create really a partnership between the 2 teams so they can drive more portfolio attach. We think we can do more really across the portfolio, but we think NKP is showing some encouraging signs. Like I said, fiscal year '25 was a good year for NKP. Still small, to be clear. It's a very small portion of the overall base, but should be growing nicely. So as we think about '26, so there's some investments we've made. Now those portfolio sellers are ramping and should start to be productive here over time to help drive more expand with our customers.

And I'm curious, Rajiv, for you, like when you think about the NKP opportunity, is this more migrations of existing VMs on to containers? Or are you -- is this really positioned as more of, hey, we're going after net new apps because this is -- they're all being developed through a containerized environment. So can you peel back that onion for us a little bit?

It's a combination, but more oriented towards the latter, which is net new applications because there's a ton of new applications being built. And people, when they build new applications today, run them on Kubernetes typically. And every AI application that we're talking about is also going to be running on Kubernetes in some form or fashion. Therefore, a lot of this is in front of us as new applications are being built. And yes, at the same time that the new applications are being built, some portion of the existing application footprint will be refactored into containers. And so we don't distinguish necessarily between those 2, but our view is to provide customers with that flexibility, which is you can run your existing applications, but at the same time, you're going to have a mix of existing and new applications and we can run both of them on the same platform. And you don't have to choose a priority, do I do this or that? You can mix some match.

Very helpful. And maybe we move on to the big topic of conversation, which is VMware. You guys have been quite vocal about the opportunity around the VMware disruption, but also, I think, very deliberate in messaging this as a gradual sort of multiyear opportunity. So maybe remind us why this is the case? And how far along you think we are in this opportunity?

Yes, I think it's -- we've already said that this is a large, multiyear opportunity. And many analysts have pointed out that some subset of VMware customers will leave over a period of a few years. The numbers are very -- Gartner and others have said 30-plus percent of customers will leave. And clearly, we're seeing that. We're seeing newer customers migrate. We've added 2,700-plus customers over the last year. And we're seeing that increased base of customers, and almost all of them were using VMware before. Now the reason this takes time is because infrastructure is fairly sticky. To migrate infrastructure from one platform to the other, in many cases, first, there's the timing of the VMware licenses, number one. Number two, for HCI at least, you have to replace, in many cases, existing hardware. And so when a customer buys hardware, they typically depreciated over a period of 3 to 5 years. And so they have to -- you have to time the refreshes with the depreciation cycles. So that's the second aspect. Third is migration requires some work. Of course, we try to automate the actual process as much as we can, but it still requires the customer to have a focused project and migrate and put in some effort to go and migrate everything and learn a new platform, whatever that may be. And the fourth is just I would just say inertia. There's a lot of inertia in this business. Customers have a lot of priorities. And so all of these, I think, contributed to the fact that migration is surely happening, but it happens at a gradual pace. And to characterize where we are, I would say we are in our second innings of a baseball game.

Very, very early then. I think about the timing of renewals piece, like in '22, the deal was announced; in '23, it was closed, right? So you kind of look at '25, '26 like a lot of these customers are on 3-year ELA. So I mean how big is the kind of pool or addressable pool over the next kind of 12 to 18 months when you think about this opportunity?

Yes. I mean, look, the -- I think this is -- again, this gets changed out over time. By the way, some customers who renewed once. I've already renewed with Broadcom. In fact, what Broadcom said is the vast majority of their target customer base is now on VCF subscription. Okay. And the question is, again, as they come up for renewals and some of them will come up for renewals now, some of them renewed maybe last year. So there's kind of a continuously coming out pool of revenue. It's not -- I don't think there's like a peak event where everybody renews, but it's a gradual sort of renewal event. Now again, they have to plan for these migrations in advance. So if there's a renewal event, they have to give themselves time ahead of that to make sure they can plan and execute the migration. Depending on the size of the environment, for small customers, like we had a law office, for example, migrated a couple of data centers, maybe tens of servers, and they were able to do that within a matter of a couple of weeks. For medium to -- medium-sized enterprises from anywhere from 25,000 cores to 100,000 cores, we've done many of these migrations in under a year. For the very large customers with millions of cores of compute, that might take 2 to 3 years to migrate, and so they have to plan -- start planning and start executing ahead of the migration time. And some of them start. Some of them have completed. Some of them haven't started. In fact, many of them haven't started. And so that's still says that some of these guys are going to actually renew again for the second time, and then again, we might get in there as a second vendor. We've seen that in many cases where they bring us in as a second vendor, and then we win a portion of their estate and then we expand over time.

Right. Rukmini, from your side, I mean what have you seen from kind of a pricing perspective out there as it pertains to kind of VMware? I mean are you seeing Broadcom get more or less aggressive on kind of the pricing side as you guys go after some of these larger enterprise deals?

I would say things have generally played out more or less as we had expected it to be because I think Broadcom playbook with their acquisitions is fairly well understood. What is different about VMware is they have a much larger customer base than perhaps some of their earlier acquisitions. And to Rajiv's point, it can be quite sticky. In general, they tend to acquire businesses that are sticky for a good reason. And as we know, they've made some changes to their pricing to make it more expensive going from just using hypervisor to the full platform, which is what they have done. Now for us, we have always been comparable in terms of pricing in our full stack compared to theirs, just to be clear. We're not here in the market to be some sort of discount offering. We add a lot of value to customers. Our TCO, total cost of ownership savings, are meaningful to customers. And so we want to price in a way that is commensurate with the value that we're adding to customers. So that's always been our approach, and continues to be our approach. Now of course, do we have a view around volume-based pricing, right? So if there's a customer with, like Rajiv said, on the largest of the large, 1 million-plus scores, we'll take that into account as we think about pricing and price it lower than someone who's maybe a much smaller deployment. So naturally, we'd make adjustments like that. But our intention is to be competitive in the market because of the value we bring. And we'll also like pick our spots where we want to go and win here, right? And I think they've also -- Broadcom, for example, has also been quite clear even publicly about where they're focused in terms of which accounts they care about, which are the largest ones and so on. And so we are focused, as Rajiv said on making sure that our value proposition is more around -- we are a long-term partner. Our Net Promoter Score has been 90 for over a decade. We are innovating in hybrid cloud. We're innovating in Kubernetes. We're innovating in AI. We are the partner you want for the long term. And then pricing is one of the components versus sort of trying to win purely on price. So it's really all of the above.

Understood. I want to shift the conversation to margins. Nutanix has done a commendable job growing into its cost base with free cash margins now in the high 20s. Your guide suggests OpEx will run a bit hotter in '26 versus the past couple of years. I guess what are the primary investment initiatives for this year? And how would you frame the drivers of leverage in the years ahead?

Right. So our margin -- operating margin on a non-GAAP basis for fiscal year '26, the guide at the midpoint was a touch higher than what we did in '25, which I believe is what you're referring to. And in terms of the investments that we're driving -- for us, the #1 priority is driving growth because we see a large opportunity ahead of us, and we want to make sure that everything we're doing is in favor of getting that growth to be as high as it possibly can, also knowing that we have to drive leverage to the bottom line and drive improving operating margins because I will tell you that we're not at our long-term operating margin. We know that we can do a lot better than where we are. So in terms of investments that we've talked about, like I said earlier, portfolio specialists, for example, like those are on the go-to-market side. A few other investments on the go-to-market side, I'll call out. One is we've increased the number of reps that we have out there in the field because, again, large opportunity. It's not a huge increase, but there has been an increase in our reps as we enter the new fiscal year and then the people associated with those reps. So they'll have sales engineers, they might have other kind of supporting functions to help provide the support for that rep. So those are some of the areas in go-to-market. Another one I would highlight is around marketing where we have had to dial up the volume on our presence in the market because we are smaller, as Rajiv said, like from a lot of other companies in that Magic Quadrant. And so we have invested in branding, marketing, getting our -- getting awareness, higher awareness in the market about who we are and why we're the right partner for companies. There are some examples of investments on the go-to-market side, but also investing in R&D. So making our core platform continuing to build on the core platform. We're very proud of how long -- how far that core platform has come and it's how enterprise ready it is. So we'll keep investing there. Other areas of investment in the R&D side are Kubernetes that we've talked about a lot here on the AI side, external storage support has been another investment. So we think these are all things that are directly linked to our growth drivers and where we can see a return. So when we look at '26, there is a bunch of OpEx that's coming in, that is run rate of people that we hired towards second half of fiscal year '25 that are all coming into '26. And then any incremental ones, we'll be very prudent about where we choose to invest. Now I think to finish off the answer here to your question, what are the overall drivers of leverage? I said we're not at our long-term margins. So the first one I'd say is our mix of renewals as a percent of revenue or billings will continue to grow over time. And as long as we can continue to transact those renewals more efficiently than we do land and expand, that should help drive leverage to the bottom line, and we've demonstrated that over the last few years, and we expect that to continue. Another driver is the productivity of our sales reps. And so when I say productivity, I mean specifically land and expand productivity, so incremental business that they're bringing in. We've seen a nice improvement in that over the last few years. We're not yet at what we would characterize as industry benchmark levels of ACV productivity, so we intend to continue to drive that with our sellers. And then the third piece is I'll call sort of good hygiene in terms of any incremental investment, it's really thought through, it has a real ROI, how can we leverage AI across some of the things we're doing. So more around just disciplined execution is a third bucket of things that we believe will continue to drive margin expansion over time.

Fantastic. And with 2 minutes left, Rajiv, I want to close out with a question on AI. You've been in the market with GPT-in-a-Box for over 2 years now. There's a big debate around the ROI of use cases right now, especially around AI. So where are you seeing the most success today? And how close do you think we are to sort of enterprise AI maturity to the extent that it can start benefiting Nutanix's business?

Yes. I think it's still pretty early days for us. We've seen a good set of customers trying out the product, some of them are getting to production with smaller applications. These applications tend to be largely around things that we've already seen out in the market, right, optimizing support all kinds of document summarization initiatives or call summarization, for example, I mean, if you have salespeople talking to clients and delivering insights and checking for compliance; fraud detection for money laundering, for example, or in retail stores. So these are the kind of, I think, use cases that are emerging. I think we've seen, like I said, the first set of use cases, I think customers starting to take the simple ones into production. These are relatively small clusters, well contained with reasonable ROI. Support is the obvious one where they can get good ROI. And I think on these things, the key driver that we've seen is data, right? Where is the data are located? And do they need to secure the data and run it in a private base securely. That's where we see this. I still think we are in the very early stages of this. And I think over the next few years, we're going to see a much more of, I think, inference use cases starting to be deployed in volume. I think where people are still trying to figure out exactly the ROI that they can get as they optimize staff or optimize workflows -- entire workflows and I do think there's a lot of optimization to be done. And as those start coming into play, I think you're going to see inferencing applications take off.

Excellent. Well, with that, we are almost exactly out of time. So thank you so much, Rajiv and Rukmini for joining us today, and appreciate it. Looking forward to having you next year.

Thank you very much, Matthew.

Thank you.