Hewlett Packard Enterprise Company (HPE) Earnings Call Transcript & Summary

Okay. Folks, I think we're going to get started. Thanks, everyone, for attending the second day of our 2024 Global Technology Conference. My name is Ruplu Bhattacharya, and I'm a Director with the IT hardware and electronics manufacturing services equity research team. Today, I'm filling in for Wamsi Mohan, who is the senior analyst. And we're going to kick off the session with HP Enterprise. We have the team here, and we have CFO, Marie Myers. So Marie, it's great to see you again. Marie has been with HP Enterprise since January of this year, but she has lots of experience with HP. In fact, she was with the other branch of HP since 2002. She was appointed CFO of HP Inc. in 2021. And I think even before HP, you were with Compaq.

Yes.

So she's got lots of industry experience, so we hope to have a great discussion today. So Marie, thanks for joining us today.

Thanks very much, Ruplu. So before we get going, I've got to read out this comment. So my remarks may contain forward-looking statements, so please refer to our SEC filings, including our most recent Form 10-Q for a discussion of the risk factors that relate to our business. That's one out of the way.

So Marie, we've got lots of things to talk about. But maybe I'll just start with you've had experience with both branches now, HP Inc. and HP Enterprise. What made you switch to this branch of HP?

Yes. No, thanks very much, Ruplu, for that question. Well, first of all, I've always admired the transformation that Antonio and the team were driving at HPE. And really, I'm excited. What drew me to HP was the opportunity to be part of the AI inflection. I could sort of see from the other side in terms of where the business was going with AI PCs. And the opportunity here is just enormous in terms of where the market's at. And I feel HPE is just incredibly well positioned. And in fact, Ruplu, in my first job out of grad school was in the Server and Storage division of Compaq back in the ProLiant, x86 days. So it felt like I was coming home. But yes, look, I'm really excited. We just had earnings yesterday, and I think we came off a phenomenal quarter. We had a trifecta. We beat revenue. We beat cash flow. We beat EPS. So as a CFO, it sort of doesn't get any better than that, Ruplu. So excited to be here.

Yes. No. I mean you had a really good quarter and report yesterday. And we're going to talk about a lot of different things, and we're going to spend some time on AI servers. But let me ask you about an announcement that you had even before the earnings call, which is on the H3C divestiture. And what -- can you just remind us what exactly are you selling H3C now in tranches? And how does that affect the value that you hope to realize for H3C?

Yes. No, thanks for that. So just to clarify right up front, so basically, nothing has changed in terms of the value. The construct of the deal remains absolutely the same. So it's still the $3.5 billion. But what we've done is really clarify the timing with Unisplendour. So if you look at the 8-K we filed, you can see we've broken the payment schedule up into 2 key tranches. The first is the 2.1, which will come in the August to October time frame, and then the clock starts for the remaining 1.4, which will come 16 months later. So 8-K and the agreement with Unisplendour was really an opportunity to get much more clear about the timing of the payments. So with that, we retain the 19%, obviously, which is what we updated in the guidance model yesterday.

Got it. And can you just talk about HP's overall strategy for China? I mean how do you see that market?

Nothing has really changed. I mean in terms of the agreement with H3C, we still have a sort of vice versa agreement in terms of our stuff, their stuff, in China and outside of China. So nothing changed there, Ruplu.

Okay. Got it. The other thing I want to talk about before we get into the details of the earnings call yesterday is the Juniper acquisition. So can you remind us like what's the rationale for that acquisition? And how are you thinking about the revenue and cost synergies related to that acquisition?

Sure. Maybe I'll just sort of clarify some comments I think Antonio made yesterday, which is we're very excited about Juniper, one of the reasons I wanted to join. Secondly, the deal still remains on track for end of calendar year, beginning of next fiscal year. And then obviously, I mean, I think the business with Juniper and HP is -- brings together an amazing portfolio. Juniper, if you think about their strengths, it's really much more in sort of the service provider, the data center space, which is obviously super important now because of where AI is going and then sort of like AI-enabled native networking. And then if you look at our portfolio, it was really focused on sort of campus like big Home Depots, et cetera, and then also on SASE. So if you put this portfolio together, you now get breadth and reach that we never had. And I think the data center piece is super important. It is, as Antonio has said, I think, commented many times, probably the piece of the whole AI ecosystem that's least understood. So I think it's going to be the right place at the right time as you think about this whole AI sort of system and servers, et cetera, and where -- how things are evolving. In terms of cost synergies, revenue synergies, the deal was predicated on $450 million of cost synergies, which we still believe is the right number and about a $500 million cost structure to achieve those synergies. So that's, in essence, the structure of the deal. More to come as we get closer to closing it out.

And one question we've received from clients related to this is, are there any restructuring charges? Or how should we think about cash flow accretion from that deal?

Yes. As I mentioned, about $500 million in charges and then in terms of free cash flow, obviously, probably slightly accretive, a real one.

Another thing that happened was you have restructured your segments. So can you talk about the rationale for that? And how are we thinking about the business in terms of different segments?

Yes, no, absolutely. Maybe I'll just start out with a philosophy as the CFO. One of the first things I did when I joined was we sat down with Antonio and look at this through a different lens. And I use the word simplification. I'm a big fan of having really clear, easy-to-understand messages for our investors. And so the first move we made was to actually put all the server business together into one segment, which I think today one it makes it easier for you guys. You can get out there and compare us quite easily to other folks in the business, and it's a direct apples-to-apples compare. So hopefully, that's made your lives a little bit easier. I think secondly, in terms of just the business itself, it is one big segment. And I think AI has made the segmentation somewhat different. The way we look at the market today is you've got sort of 3 big blocks. You have the model builders that are out there today. You have the sovereigns, and then you have the enterprises. And I think today with the stack that we've got and the way we put that sort of server portfolio back together, it enables you to understand the business in that context. So a lot more transparent and a lot clearer. And the rest of the segments, frankly, Ruplu, haven't made any changes. But obviously, simplification to me is paramount.

Got it. That makes sense. So now let's start talking about each of the segments. So let's start with the Server segment. You had a phenomenal report yesterday. The thing that struck me as unique was the commentary on enterprise AI. We've heard your competitors, Dell and Super Micro, they talk about hyperscale, they talk about second-tier CSPs. But the focus that you've had on enterprise AI seems to be unique. So can you talk about what you're doing with enterprise? What are you seeing? Are you actually seeing enterprises getting into AI?

I'd say the answer simply is yes. And I think what we saw this quarter is that we saw green shoots in the enterprise. And I think what was interesting was it was diverse. It was companies that you would imagine in industries like automotive through to health care. There wasn't one -- one of the things we're looking for is there one sector, is there one vertical, for example, that's taking off. But what we saw was it was sort of, interestingly, across the board financial services, et cetera. So I think one genuine interest in different verticals was noticeable. I think secondly, it's across the globe. And Antonio had returned from -- he'd been in Europe. I think he talked about this on the call yesterday, and a lot of conversations in terms of just interest coming out of different parts of the world. So I think to me, early days, I might add. I think Antonio said about 15% of our cumulative order book on AI is enterprise. So that's a good sort of summary, I think, in terms of just the level of interest. But I would just say, look, it's early days. Use cases, still starting to emerge. I mean we're seeing enterprises experiment with about probably 5 to 6 top use cases, everything from software through to service. They seem to be the areas of marketing that companies are focused on. But this is an area that HP, I think, plays. This is why one of the reasons I joined because, honestly, this is the stronghold of HPE, the Enterprise. And I think that's where we have a portfolio that is uniquely differentiated.

No, I agree. And 15% of $4.6 billion, I mean, that's...

It's not a small number.

Yes. That's not a small number. All right. So maybe the next thing I want to talk about is AI server margins because I think that's a focus for a lot of clients. Can you talk about how you see the trend in AI server margins, both on a sequential as well as on a year-over-year basis?

Now I'm pleased you asked that question. It's obviously attracted a lot of interest in the last couple of weeks. So look, overall, as you know, we report our Server segment from an operating perspective together. So the margins are all sort of rolling up in there. We guided to 11%, and we hit 11%. That's the outlook for the year. First, I'd like to clarify a point just on margins. So when you look at it from a year-on-year perspective, and I think we all have forgotten last year, about a year ago, we were coming out of the sort of backlog peaks commodities. We're in a different place. So it's important to remember that server margins about a year ago, we're hitting a big peak. So if you do it year-on-year, you got to take that bridge into account because that backlog was priced with a pretty sweet commodity profile. Now if you look sequentially, our margins have been, I think, Ruplu, very positive in terms of maintaining the bottom end of our range. If you think about the puts and takes in there, obviously, from a headwinds perspective, we've seen commodities now. We're in a much more inflationary environment. And then secondly, we've had a strong mix of AI service. And as you've heard from many of our competitors, it is a competitive market. But we're out there competing, disciplined in terms of both cost and price, and that's sort of like our mantra. Now is that business competitive? I'm going to be the first to say, yes, it is, but you got to be disciplined about it. In terms of some of the sort of the tailwinds that we're seeing, we've got Gen11, which is our sort of traditional Server business. It's migrating very well. The AUP is there. We're getting that benefit. And then secondly, we've talked about cost discipline. That's something that is also very crucial to our portfolio. And we're seeing the benefits of the actions we're taking on OpEx. So all that together gives me confidence in our operating margins of 11%. But certainly, servers are competitive.

So I want to ask a couple of more questions on AI server margins, and these are questions that you've gotten from clients. So if we -- you had a strong quarter. You took a full year guide, both for revenues as well as for EPS. But for -- on the EPS side, it looks like you're getting some benefit on the OI&E line as well as there's a part of H3C that is still in the P&L. So is there -- I mean how should -- in that context, does it mean that AI server margins are dilutive? So how should one think about that?

Yes. No, thanks for that. Look, so let me just unpack the guide for everybody. So we did guide up $1.85 to $1.95. And what we did was pass through the portion of the $0.03 of H3C that was related to that remaining 19% stake. And that was really tied to the fact of the time, which we said was about 16 months. So we thought it was prudent to put it back in the guide. So that's sort of step one. In terms of our Q2 performance, as you correctly said, we did outperform on OI&E. We did have a one -- and I caveated a onetime benefit, which we said would be nonrepeatable. Henceforth, we adjusted the outlook on our OI&E guide. So yes, we had the onetime benefit there. And then in terms of just the revenue, we did actually guide up revenue by an additional percent. But actually, in my comments, and I think folks might have missed this, we pointed to confidence towards the higher end of the range we do. So hey, too early to call yet, but look, I'm kind of giving you guys a breadcrumb. But yes, we're leaning in and saying, "Hey, we're looking towards the higher end of the range." So being prudent, Ruplu. I think I've already been in the role 5 months, actually maybe less than that. We've just been through a guide down, so trying to be prudent. We've still got sequential improvements in networking that we expect to see modest, I might add, and some improvement in the Hybrid Cloud segment. So all of that together is what gives me confidence to put that guide out in terms of $1.85 to $1.95.

Makes sense. As you mentioned, the AI server market today is very competitive, and people are trying to gain customers, gain market share. So in that context, I mean, do you have a baseline margin below, which you won't go? I mean there are some investors who are concerned that margins can be 0 or margins can go negative. So how do you think about how aggressive you want to be on margins in this land grab, so to speak, for AI server customers?

What I'd say, Ruplu, we've got a framework, and we use that to run the business with. And I think you've seen the results. I mean, to me, nothing sure is better. Actions speak louder than words. So you've seen our results, Ruplu. We use a very well-established framework in terms of deal structure, and we evaluate each deal, deal by deal. And you've seen the kinds of customers we're dealing with. I think in the announcement -- we made some strategic announcements about relationships that we've formed, and that's the kind of customer base we like to work with.

Maybe, Marie, I want to ask you about the relationship with Microsoft. Can you talk about like how long the relationship has this been? And what are the type of workloads that you're helping Microsoft with?

Yes. No, very, very pleased to be working with Microsoft. And as we said in the announcement yesterday, HP is helping Microsoft extend the Azure AI platform to customers like OpenAI. And I think that's just evidence of the quality of our portfolio, the fact that customers like that like to work with us. And frankly, a lot of that is based and predicated on our skills in liquid cooling, which I think is also another point that came up on the call yesterday. Liquid cooling is a skill that HPE has had for decades. And it goes back to the original sort of acquisitions that we did in some of the heavy compute space. And those decades of experience are now going to be very, very beneficial to customers that we talked about like Microsoft. But moreover, as the chips come on board with Blackwell, maybe it's not quite understood that they're going to be completely in need of liquid cooling. And I think that's where HPE has more than 300 patents and decades of experience.

So it seems to me that liquid cooling will become much more relevant going forward because all the new accelerators that are coming out, they need to have liquid cooling to maintain the energy, their temperature requirements. So when we think about margins, again, I'm coming back to margins because it's a focus for clients. Do you think that gives you the opportunity, having this expertise in liquid cooling? Can you charge more for that? Can margins be better for liquid cool racks? And how do you think about margin progression over time?

I'm really glad you asked that question because actually, this quarter, for folks who might not have noticed, we actually added services into our AI systems disclosures, and there was a reason for that. And the reason was exactly what you said. Services right now is sitting in our sort of backlog to a large extent because as those data centers come online, customers need that handholding or white-glove treatment to be able to actually stand up to the centers. And a big part of that is going to be the expertise and the know-how around that liquid cooling. And obviously, services, as you know, have a very nice margin profile. So we see that as obviously one of the important profit pools that is going to emerge, plus we have IP in this space. As you know, intellectual property is super important. We have more than 300 patents. So all of that together over time should be very accretive to the business and help us in terms of managing our margin structure. So that's how we're thinking about it, which really also ties to the fact of why we decided to include services that we could start to give you some transparency.

No, that's great. Marie, help me understand the relationship of AI and GreenLake. Is one helping the other? Like are you seeing more GreenLake business because there is demand for AI? And is GreenLake helping customers get into AI? And how did you -- what was the impact on GreenLake revenues and ARR from AI? If you can give us your thoughts on how these things work together?

Yes. So AAR, first of all, very happy it was up 39% year-on-year. And actually, one of the biggest drivers of that was actually AI, and it was actually the GreenLake. So what we like to think is we see GreenLake as a really unique differentiator because it's a sort of easy way for customers to get into enterprise AI. And we've actually seen some of our bigger customers lean in on the GreenLake option. So we like to think as enterprises become much more interested in generative AI that we have this unique differentiated offer with GreenLake that will allow them to get in because it takes the headache away, frankly. They don't have to deal with all the setup, the maintenance of the data center. We take that on. We take on the management of that, and it really allows them to have the sort of like the best of both worlds because if you think about it, if you're a customer, you want to be able to manage your data, you want to have that on-prem capability, and that's certainly what we can provide.

That makes sense. Marie, Antonio made a statement last night, which I found interesting. He said that the pipeline for AI servers is multiple times the backlog. So given that, is it reasonable for us to think that sequentially, you should see AI server revenue, order and backlog growth in this next quarter?

Look, I'd say from an outlook perspective, we're very excited about the fact that we're seeing such strong demand and interest. And as I mentioned, we see the green shoots in the enterprise. Now I'd say, look, from a guide perspective and an outlook perspective, everything is in the guide. That's why we raised revenue. You saw us there, we took revenue [ 1% ], but that reflects some of the enthusiasm that Antonio had yesterday. And similarly, I think we've captured that in terms of both our outlook on EPS and cash flow. So for the moment, we definitely see the opportunities, and that's why we're rolling through the increased revenue.

I want to ask you a couple of more questions on the Server segment. Let's talk about supercomputers, the Cray supercomputers. I mean how do you see revenues from that flowing in, in this year?

Actually, we -- obviously, supercompute is -- I'd say the way to think about supercompute is really the sort of work that's going on with a lot of sovereigns. I mean if you equate how supercompute is being used in many different countries around the world, it is actually the sort of first step into sovereign. And you think about what a lot of these countries are doing, they're using that capability to start to enable models. We do have a really large deal for -- in the U.S., it's called El Capitan. And actually, we're expecting to see that in the back half of the year in Q3 and Q4. It's one of the bigger items that's driving some of our AI system revenue in the back half of the year. So these deals, you got to think about, they're lumpy, too. Obviously, they're big, large transactions. They're complex. They're hard to stand up, and it requires a lot of technical acumen and sometimes all the complexity around liquid cooling. So different type of transaction to traditional server. Henceforth, given the size and the scale, these deals tend to be much more lumpy. I always say to folks that's like building an oil refinery. People -- I'm from Texas, so I apologize. But in Texas, and you talk to people is like, hey, yes, I'm in the business of building oil refineries. Well, I'm like I'm in the business of building big, large data centers. It's a whole different mindset, but still they're just capital-intensive, and also they're quite complex to do. So that's a different way to think about it, and supercompute is getting you into sovereign.

Makes sense. The other thing is we've talked about AI servers. What about general-purpose servers? How has demand been for the Gen11 servers? I think you raised your AUPs. Is there more scope to raise prices?

You keep pushing me, Ruplu. But I'd say, look, first of all, we've been successful in passing through the increases in commodities. And frankly, it's shown up on Gen11 with better AUPs. And we're very pleased with the performance of Gen11 so far to date. Definitely, it's on track and perhaps even slightly ahead of track. And actually, you see it in the unit numbers in traditional servers. Traditional servers were actually up double digits year-on-year and single digits from a unit perspective. So we're seeing unit volumes. You're seeing the incremental AUP. So overall, very pleased with the performance even of traditional servers. And I think, Ruplu, that sort of opens up -- we're starting to see that potentially customers are also using some of the CPU capacity for inferencing. I think that's not well understood yet, but that's something we're going to stay very close to understanding.

And in terms of partners, is NVIDIA the main partner today? Or do you have any plans to expand beyond them over time?

When it comes to gen AI, our primary partner, I think as we've been very clear about is NVIDIA. And in fact, for those of you who aren't coming, we're having our Discover Conference here in a couple of weeks, and we actually are having an IR Summit, Ruplu. I'm not sure if you -- are you coming?

I will.

Yes. You better be there. And actually, Jensen is going to be on stage with Antonio. I think Antonio is giving the first keynote at the Sphere, and Mr. Jensen is going to be there for quite some time. So I'd encourage you to go. We're going to have all our assets. You're going to be able to see liquid cooling. I mean I think Antonio is bringing what he calls his favorite plumbing with him. So I mean -- and it's really -- I think once you kind of see it, like a great thing if anybody wants to come to Houston, let me know. I'm happy to give you a tour about data center because once you see the liquid cooling capability, it becomes very real. So that's one of the things we're looking forward to at Discover. We're going to have all our sort of servers there with all the different sort of aspects so you can appreciate just the size of the portfolio and moreover, the technical differentiation. So very excited about that.

Got it. So it looks like HP has a lot of great future in AI servers. We've got lots of opportunity. I also want to talk about some of the other segments real quick.

I didn't think you thought -- I thought you've forgotten the rest...

Well, let's talk about Intelligent Edge. How are you seeing revenues and margin progression in that segment?

Look, I think we've been clear about it. Intelligent Edge has been transitioning. And we see -- we saw Q2, I might add, really probably as the trough in that business. And you look at it, you can look at the numbers. I think they speak for themselves. The good news is channel inventory remained in good shape through the transition. I think we were very prudent about managing our channel inventory. So the inventory in the channel, I think, is relatively in decent shape, still some pockets in Asia that we're working through. But what I would say is that we do see a modest sequential improvement in revenue, and that's really driven by public sector-led education. So that's the signal that we see in the back half of the year. In terms of margins as well, we've taken some actions around operating cost structure, really addressing the impact to the business. So we expect to see also some sequential improvement in our operating margins in edge in the back half of the year as we see those actions start to play out, Ruplu.

How about Hybrid Cloud? How should we think about revenues and margins?

Yes, Hybrid Cloud, frankly, is in transition. This business will be probably hit a low point in terms of our operating margin. And that's because we have been investing in the GreenLake platform that we talked about. We expect that some of that investment will start to taper off in the back half of the year. We're also in the midst of a very significant transition on storage, both in terms of go-to-market, block file and electro. We're really excited about electro. We see a great pipeline there. So we do expect to see margins come back into the mid-single-digit range in the back half of the year. That's operating margins as that business starts to -- we see the dividends pay off in terms of both better expense management and also the transition and storage. I think that storage transition to me is absolutely crucial for our company.

Great. In the time that we have left, I want to ask you some financial questions. Talk to us about how you see your -- HP's inventory trending, cash conversion cycle and expectation for free cash flow for this year.

Yes. The free cash flow is still very much in line with the guide that I gave last quarter. We expect at least $1.9 billion for the year. Obviously, we had a strong Q2. And just want to bear -- sort of clarify to folks that was really driven by just more timing of payments. From a working capital perspective, we do expect to end the year relatively neutral. So I want folks to understand that in terms of Q2 to Q3 and how that cash flow walk would look like because typically, for HPE and even for HPQ, cash flow comes in the back half of the year. We're a company -- both HPs are driven by enormous cash flow in Q3 and Q4. Still expect that step up, perhaps more moderated from Q2 to Q3. So that's how I think about that. In terms of inventory, we did hit a high note on the balance sheet this quarter. And the way to think about that is most of that increase, more than about 1.2 came from servers. And we're really positioning. We talked about El Capitan. A lot of those big deals are still sitting in inventory as we realize the revenue in the back half of the year. We shipped those deals, then move out of inventory. So we expect to see inventory come down somewhat in the back half of the year. I would add that we like to do strategic buys, where it makes economic sense. Frankly, it shows up in the P&L. As a CFO, we'd be doing a strategic buy, and it has the right economic profile. We want to be able to maintain that flexibility to do strategic buys. So that's another reason that sort of drove up the inventory, Ruplu, I want to clarify.

That makes sense. Antonio made an interesting comment. He said that OpEx in first half of '24 was less than first half of '23. How should we think about OpEx for the second half of '24?

Now as a CFO, I like OpEx to go down. Let me just put it that way. So I'm very focused on OpEx management. I think I made a career out of it. So look, we're going to have a onetime increase in Q3, just as we go through some marketing events. But then I expect that to come down in Q4 and to end year-on-year lower, positioning ourselves nicely going into '25. So to me, discipline around cost and price. And then ultimately, driving structural cost reductions is super important in a business like this.

Just to clarify on the free cash flow guidance, does that include any payments related to H3C?

No. We don't have any of the payments for H3C or divestiture that we did. So yes, all that -- once all that rolls in, obviously, once we get into the quarters, we'll put that in our updates. But at this point in time, none of that's in the cash flow guide.

And maybe in the 2 minutes or so we have left, how about investments? How are you thinking about CapEx? How are you thinking about R&D spend? I mean what are some of the areas? How should we think about that dollar amount of spend trending over time?

Well, typically, it's pretty seasonal. I think if you look at our -- the way we pace CapEx throughout the year, you typically see a seasonal uptick in Q4. Nothing really newsworthy in that, Ruplu. Most of the CapEx is tied up in the HPFS business. And typically, we've made the right prudent investments in R&D, IT infrastructure, et cetera.

Okay. Again, I want to end the session with another question on AI servers because that's -- clients want to hear about that. So again, this is a very competitive market. Are you being prudent in the sense that are you walking away from any deals? Like -- or is there -- are you trying to be accommodative for everybody? So just your thoughts on are there segments that you don't want to be part of?

I think Antonio described this really well. There are 3 main segments. There's the model builders, the sovereigns and the enterprise. And look, we're participating in all those segments. As we gave you the disclosure around our Enterprise segment, I think I've talked today about supercompute and sovereign. And I think we updated you in the call yesterday around the relationships we've got in the model builders. So I think we're playing in all the segments. It's a matter of how you want to play, Ruplu. So I think the mantra is here. This is a very exciting market. It's an exciting time. We intend to be part of it. And we've got our own strategy around how we wanted to take in those markets and play.

And on sovereign AI, do you think that there are unique qualities that HP has to tackle that demand from sovereign entities?

Oh, absolutely. In fact, we're doing -- I mean for folks for clarification purposes, we're delivering sovereign AI today. Just to be clear, we've got supercompute in the U.K., supercompute in several different countries that we are delivering that is being used for AI purposes. And we don't see that sort of -- let's just say, we don't see countries stepping away from that opportunity. So I think HPE is uniquely positioned for sovereign AI. And clearly, it's everything we've talked about today, including liquid cooling. All of that capability comes home in these types of deals.

Okay. I think we covered a lot of different things. So Marie, thanks again for joining us.