NetApp, Inc. (NTAP) Earnings Call Transcript & Summary

Good afternoon, ladies and gentlemen, and welcome to the fireside chat with NetApp, Inc. [Operator Instructions] It is now my pleasure to turn the floor over to your host, Mehdi Hosseini. Sir, the floor is yours.

Thank you, operator, and I want to welcome everyone to our fireside chat with NetApp. This is part of our tenth annual technology conference. We're grateful to NetApp's team to have fireside. We have Lance Berger from IR team. I have prepared a set of topics and questions that I will review with Lance. But before we get to that, I am going to read the safe harbor. Today's discussion may include forward-looking statements regarding NetApp's future performance, which are subject to risks and uncertainties. Actual results may differ materially from the statements made today for a variety of reasons, which are described in NetApp's most recent 10-K and 10-Q filed with the SEC and available on NetApp's website. And NetApp disclaims any obligation to update information in their forward-looking statements for any reason. So with that, Lance, I want to thank you for taking the time.

I want to start with some P&L related questions. Obviously, everyone is thinking about higher commodity cost. And if you could just review -- for context, can you help us understand what percentage of your COGS, cost of goods sold, is attributed to memory and specifically DRAM and NAND/SSD.

Sure. Thank you for the question, Mehdi, and thank you for hosting NetApp today. I think when you think about the gross margin implications and the cost of goods sold from a memory perspective, in the first half of fiscal '21, as you know, we saw significantly higher NAND prices year-over-year. That obviously was a headwind to our product margins. As we discussed on our recent earnings call, our current outlook is for NAND costs to be down slightly in the second half of our fiscal '21 half on half. Which equates to as a slight tailwind to product margins in the second half of fiscal '21 versus the first half, obviously. Also, keep in mind that DRAM pricing, as you noted, Mehdi, are up quite a bit for us. And DRAM is a much smaller portion of the bond, but it does offset a portion of the goodness we are currently seeing in NAND. And I think net-net, we expect product margins in Q4 to be consistent with what we saw in both Q2 and in Q3. I think regarding the broader semiconductor supply constraints that you obviously have been talking about as well as the Street. We are proactively working with suppliers and manufacturing partners. We do not expect the supply chain to impact our ability to meet Q4 demand. And as we move into fiscal 2022, you should expect us to prudently raise inventory levels. I think it's the right thing to do given the low-cost of capital and our balance sheet. And I'll just kind of stop there.

Sure. But on average, would it be fair to say that flash SSDs account for maybe 1/3 of the COGS, just on average.

Yes. We haven't broken that out externally for the Street as far as build. But obviously, NAND is the largest component costs within our hardware stack.

Got it. And you highlighted your strategy of increasing inventory into fiscal year '22, perhaps that's in anticipation of higher cost looking into calendar year '22?

Not necessarily. I think what Mike and George were referring to is more about just the broad semiconductor supply constraints that we're seeing across -- I mean, you've seen it in autos. You've seen a lot of different verticals, not necessarily strategically buying ahead on memory pricing from a memory perspective. We've never -- we haven't indulged in that in the past. I can tell you that the vast majority of the inventory that we'll be carrying on the books is simply to make sure that we're not in a situation where we have to build backlog or can't meet demand as we go further into fiscal '22.

Got you. Okay. Now 1 last item on the OpEx. Would it be fair to assume that OpEx could -- our quarterly OpEx could remain below $700 million even if there is acceleration in the revenue growth?

Yes. Let's talk a little bit about margins from the commodity side as well as the OpEx side. As we've discussed in the past, our flash arrays carry higher software content and higher product margins than our traditional spinning disk and hybrid arrays. So the industry's transition to all-flash is a really healthy dynamic for us. I really also encourage you to look at the total gross margin. The mix in our business is changing and is critical for the overall margin structure of the company. Our all-flash arrays not only carry higher software content, but also carry higher dollar maintenance attached. You've seen this dynamic play out in both billings and support revenue growth this year. Additionally, our cloud data services will ultimately be accretive to our overall corporate gross margin. So going directly to your OpEx question. I think Mike and George have laid this out pretty clearly. We are constantly looking for opportunities to reallocate resources to our most strategic growth opportunities, while being prudent and disciplined with our cost structure. That will change, Mehdi. As you know, we haven't given fiscal '22 guidance at this point for the total OpEx. But I would reiterate the commitment that Mike made at our Analyst Day, which is to grow revenue faster than we grow OpEx. And that would hold through in fiscal '22 as well.

Sure. And if I just add my own spin on this, if you're able to continue to improve the mix, higher software, that incremental drop-through could perhaps fund additional SG&A, if needed?

Yes. I think that's the right way to look at it. When you're in a tough economic environment like you've been in the past 12 months, you are going to be much more prudent with how you're thinking about OpEx. We've had significant OpEx savings on -- from the travel piece of our OpEx envelope, and we'll continue to be prudent, but also look for opportunities to invest in the strategic growth. I mean it's no secret that we're leaning into the cloud opportunity, and we'll look for ways within the OpEx stack to fund that.

Okay. Now pivoting to all-flash array. A couple of questions here. George talks about the installed base. We have gone from a low-teen of installed base upgraded -- I'm sorry, let me rephrase it. As of the last quarterly conference call, I think 25% of the installed base has upgraded to all-flash array. Am I getting the number right?

It's actually 27%.

27%. Okay. And it's a significant increase on a year-over-year, or if you were to compare to 2, 3 years ago. Now looking forward, what could happen to accelerate this upgrade?

Sure. So to your point, in Q3, we had 27% of the installed systems under active support contracts, were all-flash. We're -- as you pointed out, we're adding about 100 points per quarter to this figure. I think the key drivers to flash adoption continue to be superior performance and significant data center cost savings, right? So I mean, fundamentally, it's a superior technology. I think our all-flash business will continue to benefit from both -- and to your point, continue to benefit from both upgrading the installed base to refresh cycles and capacity additions of our current flash customers. And then obviously, as we talked about the Run to NetApp plan, obviously, new customer wins within the ecosystem as well. I think in the medium to long term, what really causes an additional inflection in all-flash adoption, NVMe is likely to provide a nice tailwind to all-flash. It's the first storage protocol that's been written specifically for flash arrays and as more applications are written with NVMe interfaces. It should further unlock the value of flash solutions. We're still very nascent in NVMe. I know if you kind of rewind 12 to 18 months, it was a major focus area for Wall Street, and it's kind of -- the discussion around it has died down, but I can tell you, as you kind of look out into the long-term, NVMe will likely be a dominant flash based storage interface between applications and the media.

Yes. Okay. And then just 1 follow-up here. How should I think about the mix of 10 and 8K RPM HDDs and are we getting to the point where some of these storage boxes are running out of maintenance contracts. So that by default, they would have to upgrade to all-flash array. Is that also going to be a factor?

Yes. I wouldn't say by default, but what I would say is that as we get into the replacement cycle for 10 and 8K drives, you are likely to see them move over to all-flash. I think George has highlighted pretty consistently the -- this dynamic within our installed base. And if you kind of look at what happened to the 15K drive market, it's not unreasonable to assume that same dynamic will hold true as we get further into the 10K market.

Can you elaborate on it? How is it different?

I don't know -- we don't necessarily see it as different, to be honest with you. I think you know...

Terrific. Is there anything in the next 12 months that could actually accelerate this upgrade?

Well, I think fundamentally, just a better spending environment, candidly. I think as George has highlighted, I think the dynamics going forward in the forward 12 months, it would -- you'd be hard-pressed to make an argument that they won't be better than the last 12 months.

Maybe perhaps I could be more specific. Could availability of QLC NAND enabled SSD drive a faster upgrade cycle?

It's interesting. Q3, we expanded the breadth of our flash offerings with the introduction of the FAS500f, which is our first QLC technology. It's highly scalable for deployments. As George noted, kind of with unstructured data like medical imaging, EDA, computer age design and manufacturing. I think it's too early to talk about like the margin implications and the supply implications of QLC, it's such a nascent product for us. We feel very good about the flash road map. And as we get further into calendar '21, we'll talk more about QLC.

Got it. Okay. Now moving on to the mature solution products. You no longer break out the mix. But I'm just wondering, should we assume that we have reached a point where mature products are stable? Or is it going to continue to be trending down? Is it like a secular trend and...

Yes. Look, it's probably best to talk about that in the construct of the flash market as well. So let's kind of take a step back. I was -- if you kind of look at the ability to continue to grow the all-flash business. I would say, it's obviously one of our primary strategic goals, not only to grow our all-flash business, but to continue to take share like we've done in the last the last 3 quarters of this current fiscal year. I think one of the most underappreciated dynamics of the storage industry is how concentrated the all-flash market is relative to the traditional spinning disk and hybrid markets. I think within all-flash, the top 5 vendors comprise about 80% of the market, whereas in the traditional HDD market it's significantly more fragmented, right? With those same 5 companies making up less than 45% share. And the reason for this is that only a subset of our peers have been investing aggressively in what we would characterize as a true all-flash portfolio. So to put this in perspective, NetApp has about 5% market share in the disk drive segment, the HDD segment. But upwards of 17% to 18% share in the all-flash market. So as the AFA market continues to outpace the spinning disk segment, it's a very healthy dynamic for us. I think when you talk about the kind of broader trends within spinning disk. We expect there will be a market for high-capacity disk arrays for the foreseeable future. We see benefits for the customers and NetApp in shifting to AFA. But we continue to offer hybrid and spinning disk solutions. It enables us to really have a value conversation with our customers as opposed to trying to force fit specific technology. And I think that kind of -- what I led with is the biggest potential near-term driver is really the overall economic activity coming out of COVID, right? As we kind of get into the back half of calendar '21.

Sure. And especially as 5G wireless infrastructure is -- expands and there is more of a service, that also helps drive edge compute or hopefully, that will make the promise of edge compute materialize. So post-COVID, we could also see edge compute driving or becoming incremental. Would you agree?

I would agree with that. And I would -- what I would say is that anytime data -- important data at the enterprise level is being created, it's a tailwind for enterprise storage, whether it's being created at the edge or it's being created in the cloud or server on-prem. Those are definitely tailwinds for the storage industry.

And in today's -- and I'm going to go back to the all-flash array. When it comes to the edge compute, isn't that another demand driver for AFA?

Well, I mean, I would agree with that in the sense that you need the performance at the edge to really maximize the usefulness of the data, right? And to the extent that all-flash meets that need, I think it's well positioned for sure.

Okay. Because I'm a little bit confused, one of your HDD vendors, Seagate, is aggressively marketing their own objective storage, especially talking of partnership with co-location partners like Equinix and basically saying, yes, edge compute is probably going to drive incremental demand for HDD. And yes, we're going to put these HDD filled boxes at Equinix and lease them out to customers. Kind of couldn't help but think of NetApp a couple of years ago when you first introduced cloud data services. So in that context, it is a little bit counterintuitive to think that edge compute would actually drive demand for HDD. Do you have any thoughts there?

It's interesting. Our view is that the vast majority of the value we add for our customers comes from our software solutions, which we've obviously been improving upon for the last 25 years. We've created a singular OS that cuts across both on-prem and cloud environments, across flash and hybrid, across file block and Object. So we feel better about our portfolio today than at any point in our company's history. And I'll leave it to you to evaluate Seagate's offering. But as George noted, he wouldn't trade portfolios with anyone in the industry. And I think the work that we've done on the cloud side, I'm sure we'll get to that, is -- has been 4 years in the making. And we feel really good about the partnerships and where we sit in that ecosystem.

Sure. And that's actually a good segue into cloud data services. Can you give me an update on the business model, is NetApp still leasing the box? Or are they selling the box?

Yes. Let's take a big step back to talk about how we think about the cloud offering, right? As you know, we are the only third-party solution that's actually embedded directly into the consoles of both Microsoft and GCP. These partners have embedded our ONTAP OS natively into their offerings. And both Azure and GCP sales forces are selling our cloud solution as a first-party service. It's taken a lot of really hard work to get our core technology deployed and certified and integrated into these different cloud partners and data center ecosystems. And the foundation is really starting to show up in our cloud pipeline. I think it's important to realize that we're running the most important workloads on our cloud offerings, applications that run the company's business, whether it be SAP, high-performance computing, mission-critical databases. So these are not workloads that customers would simply lift and shift overnight. It really is a journey for our customers to have the confidence to move these Tier 1 applications off-prem and into the cloud. And we've really been pleased with the pace of the new customer acquisition that has accelerated over this year. And as we think about the majority of the enterprise application deployments on the public clouds, it's still -- it's -- we're still very early days and believe it continues to be an enormous opportunity for us. With regard to the customers' onboarding, dollar-based net retention rates have stayed north of 200%. It's still early days, and those will normalize to kind of, what I would characterize as industry levels between 120 and 130 in the fullness of time, but it's a great start to the offering. I think from a business model perspective, which was actually your initial question. We have a net revenue share agreement with our cloud partners. Revenues recognized in our software maintenance line. Today, we deploy NetApp owned all-flash systems in the hyperscaler data centers. And those systems first show up in our cash flow statement as CapEx and then are depreciated over 3 years in our software maintenance cost of goods sold. I think the great thing that those systems have is expected life of 5 to 6 years, so they become incredibly profitable after year 3. So over time, we would expect more of our solutions to be software only, which I think is kind of the crux of your question. And as this balance moves towards a software-only model, we would fully expect cloud gross margins to be accretive to our corporate average as we scale the business.

Yes. And then during the last earnings conference call, I think, George and Mike were talking about hiring additional sales force for CDS. Help me understand if you're collaborating with major hyperscalers, where -- how do these new sales people help? I was under the impression that you can leverage hyperscalers to find additional customers.

Yes. To be clear, we are definitely leveraging both our existing sales force and channel to drive cloud adoption. And I think on the channel side, obviously, the biggest players are the 3 hyperscalers today. I think that said, we have always been good about leading into market transitions with what Mike would characterize as strategic spend, while also maintaining a disciplined cost base, and obviously, we'll continue to do that. I think if you look at any of the truly successful SaaS companies that we obviously are aspiring to become on the cloud side of our business, right? They have all one thing in common, a true -- and that is a truly differentiated customer success team. So our cloud business is really a land and expand model. And I think to do that well, you need a great customer success team that leverages the initial deployment to cross-sell, upsell and grow capacity. And what I would say is, and I don't know if it necessarily came through loud and clear, but a good portion of the cloud sales resources we're adding really go directly to this customer success motion.

Okay. Got it. And then moving to the topic of untapped. And I may have asked you this earlier, but I'm going to go back, which -- is there any of the older untapped that are coming off of warranty?

I mean in the context of driving an upgrade cycle, I wouldn't be -- I wouldn't necessarily be looking at that. We have a very fluid discussion around upgrade cycles with all of our top customers. And I wouldn't be looking to in the end-of-life of another product to really drive that updated notion. It's just not -- we typically, on the ONTAP side, we do OS upgrades pretty seamlessly every 6 and 12 months. So a fairly -- that's why they're paying us so much for software maintenance, right? For those -- for that goodness that is the ONTAP foundation engine.

Okay. Now I left the most exciting topic for last, Kubernetes, which is like -- we could spend hours on this. So 2 questions. Give us -- actually tell us, inform us what is NetApp's product offerings for Kubernetes? And where are we with the ramp?

Yes. So I think we actually were fairly early to Kubernetes. Obviously, saw it as a key part of the market for us. We have Trident, which has been in the market for some time, which is really a dynamic storage orchestrator for Kubernetes. It basically deploys in Kubernetes clusters and provides orchestration services and really a -- Trident to enable containerized applications to consume the persistent storage for any of our NetApp core storage offerings, right? So that really is improving upon the CSI that's in the market. And the interface and allowing NetApp to attach to the staple Kubernetes environment. So in layman's terms, Trident is simply the interface layer that allows Kubernetes clusters to talk to any -- talk to in-store data with NetApp, right? That was really the first stage of our Kubernetes journey. I think the next major phase is our -- is what we're calling Project Astra, which is a full-blown Kubernetes application data management platform. We'll be -- we obviously haven't gone GA with that yet, but look for announcement coming down the pipe for Astra. And basically, Astra bundles the application, it's unique configuration and application data into a single view for the customer and allows customers to protect it, recover it, scale it, in a really simple and consistent way. I think probably most importantly for the customer in the fullness of time is it really allows customers to easily and quickly move their Kubernetes applications within and across any environment, including the 3 major hyperscalers or their on-prem, hybrid cloud, private cloud.

Sure. One -- this may be a silly question, but when I think about the 3 hyperscalers, it seems to me that each one of them is pushing down their own version of Kubernetes. And I have a hard time reconciling that strategy with third-party product and solution offering.

Well, honestly, I think that's the benefit of something like Project Astra, which allows the customer not to be tethered to the specific hyperscalers that they launched the initial Kubernetes application in. I think that's -- I think that lock in dynamic is really something within the cloud that the customers struggle with. And how do they optimize their spend as well as their performance across the different clouds if they want to do that.

Got it. Okay. And just to be clear, would it be fair to say that NetApp's Kubernetes strategy or Project Astra is more tailored towards enterprise-type customers?

Yes. I think nearly everything we do is dialed into to kind of high-end Tier 1, Tier 2 applications that require the performance characteristics that the ONTAP OS bring to the table. I think you're seeing that in the cloud environments where it is still very early days in migrating Tier 1 applications to the cloud. But as they do that, they're going to want the same performance characteristics that you get on-prem and to mirror that performance here within any of the hyperscalers, let alone within Kubernetes is very challenging for the customer today.

Yes. And just 1 follow-up. How should we think about the storage in the context of Kubernetes?

Well, I think in the context, if you're going to have a full-blown application that you want to leverage the data, right? I mean the application is creating data. And the enterprise really wants to leverage that data. They want to mine that data, they want to understand it. They want to optimize their business around it. So having a seamless interplay between the application and the data layer, it's critical for our customers. It always has been, and I don't see that being any different with Kubernetes.

I see. 6 months ago, one of your competitors, Pure Storage, made an acquisition, and they claim that they now have the technology and know-how to cover different part -- different storage requirement for Kubernetes. Should I assume that with Project Astra, you also have a diversified portfolio of storage. And what would it be? Maybe you can also help us understand the key differentiation between like a stand-alone all-flash array or a storage box that will be used with Kubernetes?

Well, I don't necessarily know that they're all that different. I think if you look at one of the major differences between Portworx and Astra is that, candidly, we didn't pay over $300 million for it. So we -- as we've gone about different verticals in the past, I think what you've seen from NetApp is if we can't innovate internally, we're going to do that. And that's exactly what we've done with Astra. So we feel very good about the application where management platform, I think it's going to compete favorably with the offerings that are currently out there. Obviously, still very early days, given that we haven't gotten GA yet, but very excited about it.

Yes. And any -- would this -- would the GA happen in this calendar year -- before the end of the year?

Well, I'm not going to get in front of Brad and Anthony's team on this one. So we'll to wait for that.

Okay. I figured I'd just take a shot. Okay. Well, this is it for me. Thanks so much for entertaining my questions. I want to thank you again for participating in our annual tech conference. And for the audience, investors, if there's any follow-up questions, feel free to shoot me an e-mail, [email protected]. I want to thank you, everyone, for participating, and wish you a great day. Thank you.

Thank you, ladies and gentlemen. This does conclude today's conference call. You may disconnect your phone lines at this time, and have a wonderful day. Thank you for your participation.