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

Thanks, everyone -- wow, it just got louder. I'm Jon Zauderer from Citi. Thank you all for joining us today. I'm really excited that Marie Myers is here from HPE to talk about the company. But I want to remind everyone a couple of things. One, this is just for Citi clients. No press or media should be in the room. So if you are, please leave. And the other is I wanted to read this statement, this -- I'm reading on behalf of Marie, is her remarks may contain forward-looking statements, so please refer to the HP Enterprise's SEC filings, including their most recent Form 10-Q, for a discussion of the risk factors that relate to their business. Now walking into the room, one thing that struck me immediately. So yesterday afternoon about this time, I hosted the CFO at Keysight, and I realized you both have the same grandparents. Right?

Myers.

I mean, because I've been doing this way too long. But if we invite in Broadcom, HPQ, I'm probably missing 1 or 2 others, we'd have the whole original HP family here, so -- and Agilent, sorry. And Agilent, that's what I was thinking. So let's keep it in the family. And we're going to keep this very interactive. So I have some questions, but certainly, raise your hand. We have a mic if anyone wants to ask any questions. But let me just start first by asking Marie just to kind of give us an overview of the business because you are in different areas and the business has grown by M&A. So just talk about how do you -- for the investors who maybe don't know the HP Enterprise story as well. What are the different pieces, moving parts?

Yes. No, sure, absolutely. First of all, good afternoon. It's a pleasure to be here. For those of you who might have missed it, we actually announced earnings less than 24 hours ago. So I was doing earnings yesterday and I'm here in New York today. So in terms of just our results, first of all, we had a really strong quarter. We hit the high end of our revenue range. So we're up double digits, 10% on a year-on-year basis. EPS above consensus at $0.50. And also hit our cash flow estimates at $669 million. So overall, really pleased with the quarter. And in terms of the business, I'll walk you through and sort of unpack the different segments of the company. But maybe I'll start out with just a general macro comment for -- just sort of get a sense of what things look like. I'd say overall that right now, we're staring at an IT sort of spend environment that's definitely improving. And actually, what we saw on a quarter-on-quarter basis was definitely sequential growth across all sort of key pillars of our business. But what we're seeing is it's still uneven. So some geographies are definitely ahead of others. And what you typically see is that the Americas region is probably leading the pack in terms of just being -- leading a bit of that recovery. So the rest of the world still sort of catching up. But overall, definitely, I think the headline is an improving IT environment and definitely an improving macro as it pertains to the businesses we're in. From our business perspective, we sort of operate what we have is three key segments. First segment is our Server business, which includes everything from what we call AI servers through traditional servers. Then we have a hybrid cloud business, which includes our GreenLake platform and our storage business. Then we have our Edge -- Intelligent Edge, which is our networking business. And then we have our leasing business, which is the HPEFS portfolio. So they're sort of like the four key pillars of the business. And the leasing business really enables much of what goes on in those different categories. I'm happy to walk you through it as I'm sure you'll ask me questions as we go on.

No, no. That was a great overview. And I guess we have to start with AI because everything else is secondary there. But when -- we get the question a lot about the AI opportunity. So can we talk about what does AI mean to HPE? Where is your opportunity there?

Personally, I always like to look at any business through the lens of a customer. And from an AI perspective, I would say that we sort of look at this broadly in three distinct categories. We have what we call our Tier 1, Tier 2 players, which include CSPs. Secondly, our sovereign sort of customers. And then thirdly, the enterprise AI. Now I would say, as a company, we -- actually in our last quarter, we hit about $1.3 billion in AI server revenue. We saw a mix of customers in that revenue base. So we had a combination of the Tier 1s, Tier 2s, sovereigns and enterprise. And so for us, it was the highest-ever revenue shipment. We are up about 40% Q-on-Q. It's about 30% of our server business. So it's becoming definitely more meaningful and very important to us. And we actually exited the quarter with a very healthy backlog. So overall, the business is definitely growing and important. And I think I mentioned in my earnings guide yesterday that we expect to see some level of modest sequential growth in our AI server business going forward. But as we think about it from a customer perspective, certainly, what we're focused on is both profitable deals. Because in the -- what I've described the Tier 1, Tier 2 space, it's very competitive. And we see lots of competitive deals come across our desk, but we're focused on driving profitable growth, so we're trying to be disciplined on cost and price in those deal segments. Now as we sort of move into other segments, which include the sovereign and the enterprise AI, in the sovereign space, for those of you who don't know us, HPE has had a long history of doing business with sovereigns. We have been, I'd say, the backbone of many different countries sort of computing for decades, actually, in fact. We just recently announced a very important strategic deal in Japan with the AIST. That's a great example of sort of the compute support that we give to many countries. And many countries are starting to explore how they can create sovereign, enable sovereign capabilities so that they can sort of fuel their own country's growth, whether it's in important work that they're doing in national security; or secondly, how do they even sort of enable local startups to really sort of get ahead in this space. So we're participating in the sovereign space, and we see that as clearly a major pillar of strength going forward. And then finally on enterprise AI. And I'm getting tons of questions on enterprise AI. What I'd say is that today, honestly, there isn't a customer that I don't meet that's not talking AI. And as a CFO, I get out there and I meet with our customers and SIs. Everyone is experimenting. And in fact, about 80% of the enterprises are either in some sort of early adoption stage or have a proof of concept. But it's still early days. And so what we're doing in the enterprise side, and I know we're going to get to this later, I'll talk a little bit about some of the offerings that we're coming out with to make it easy for companies to really get going. But enterprise AI is clearly where we have a right to win, and it's certainly an area that we look forward to because we see that the margins in enterprise AI are going to look a lot more like traditional servers going forward. So still a new market, but one that we're really, really excited about. And I'd say, overall, what AI does is that it opens up the TAM. And clearly, this is a great point in time for a company like ourselves because we're participating in a market that has a much more expansive TAM due to AI.

It's sort of funny because I always get investors say to me, figure out when enterprise AI is going to really start to move needle and call me 3 minutes before. And I'll call your 4.

Okay. Thank you.

Talk a little bit -- you said you have a right to win. Now as part of that is also -- because again, you've been at the table with sovereign funds and enterprises globally. But also, you have a good legacy history in high-performance computing. Is this -- like how does HPE think about high-performance computing? Is it an -- is AI -- that's just a natural extension of what we're doing? Or this is a completely new opportunity for us, but this gives us the skill set we need to succeed?

Probably a bit of both, but more the latter. I'll give you a really good example of one of the strengths and one of the pillars of strength that I think we've developed, and that's in the area of direct liquid cooling. Now clearly, as the future generations of GPUs get released, they're going to become highly dependent on this capability. And the good news for us as a company, we have decades of experience, and in fact, I think 300 patents or so, in this space. And the expertise that is required to develop direct liquid cooling capability is very deep and very technical. And so I think the good news for us is, because of or the experience we've had and the legacy history of super-compute, we're able to draw on that experience and really be able to enable customers, who I think, sovereigns and enterprises, are really going to need that help. The other thing, which is a bit of an unknown-appreciated fact is we actually have facilities and manufacturing locations that do this. And so we're actually having an AI Day in October in Chippewa Falls, where we're going to really highlight this strength and capability which really comes from our legacy heritage that you referred to. So I think that's just one example of where I think we have a huge opportunity and a right to win, and a pillar of strength, frankly.

I like that. I already made a note.

You made a note that you're coming.

Yes, we're coming. Chippewa Falls in October. Why not? The -- well, let me phrase it this way. Back in June, I met with one of your U.S. competitors, and we talked a lot about AI servers. And the argument they made was that they felt, that because of the complexity of these servers, and the fact that NVIDIA is moving from kind of the traditional server 2-year cadence to a 1-year cadence, is putting tremendous pressure on resources and engineering and testing and validation, et cetera. And as a result, they thought the bigger U.S. OEMs might have an opportunity versus the ODMs just given that scale. And I haven't been to Taiwan so I can't give you the rebuttal to that. So I'd love to hear from an HP standpoint, how do you think about your competitive advantage versus the ODMs from -- moving from traditional server to AI server? And how do you think your market share could go over time?

Yes. No, look, it's a great question. I believe we probably have three core pillars of strength. The one we just spoke about, which is I think that experience in direct liquid cooling, which frankly took decades to really mature. So I think that's one key strength. Another that we have, which is a strength actually, and I credit Antonio to the vision here. He started GreenLake back in 2019. We now have a consumption model and a capability that also what we're seeing is that customers like that option. And in fact, we just announced actually at our Discover event, the launching of what we call our private cloud AI offering, which essentially the way to think about that is jump-starting enterprise AI. We've announced four offerings, very simple, small, medium, large, extra large, so you kind of scale the capacity. But what it really does, it enables an enterprise to literally plug and play. And we've grown so far is to actually even load up NIMs. For example, in the life sciences area, if you're a life sciences company, you can basically plug in and go straight to work on a life sciences model. So we've shrunk the time to get started on AI, and we've made it easy for an enterprise to get up and get going and to consume it in a different form. And I think that's -- these are just 2 examples of -- actually 3 examples of the differentiators that I think we bring to bear versus some of the other folks that you spoke about. And I might add that, on private cloud AI, people it's a bit of a tongue twister because the abbreviation is PCAI. So you don't want to say AIPC, but that's a twist. But nevertheless, I think it is a real differentiator. We've just announced it. We've got hundreds of customers who've called in wanting it and we literally went live with autos 2 days ago. So really excited about it, and I think that's a great example of what we can do that others simply can't do.

I like it. And by the way, there's a HP rep out in the hall, if anybody wants to get in on the order take in. You talked earlier about enterprise customers and customers that you're talking to. Can you talk to us a little bit about what are the type of customers you're working with on AI? What are their use cases? Is there any customer concentration?

Sure, absolutely. And firstly, I'm a user of GenAI myself. But look, there isn't a customer that I don't talk to that is not thinking or experimenting with GenAI. And certainly, I would say that the use cases are sort of broad and deep, and companies are at various levels of the spectrum. I think there's one common factor though, which is the companies that are really leaning into GenAI, typically, it's a conversation that's being driven at the C-suite. Either by the CEO, it's a mandate where the CEO sees GenAI as a transformative opportunity and that person is really advocating to use GenAI as a catalyst to drive transformation in the company. Whether it's real efficiency or productivity improvements, that individual, is think of being the sort of the champion for GenAI. Then there are other companies that are really trying to get a toehold and figuring out what is the right use case. But what I've seen in terms of use cases is everything from sort of customer service agents. It's a very ripe area for applying GenAI, as is R&D. I think a lot of companies have a lot of software engineers, haven't necessarily gotten to use all the capabilities. That's another one. One other area is marketing. I have a really good friend of mine who's a CMO who specializes actually in helping companies really drive efficiency in content creation, language translation. It's an area actually in GenAI that is super easy, it's quick to start, and you can make enormous gains in marketing fairly quickly. And then functionally, there's just lots of use cases starting to emerge, I'll be honest, even in finance. I was chatting with a CFO of a large insurance company the other day, and this CFO had literally enabled a language model to every single one of his finance employees and was driving transformation in finance with everything from forecasting accuracy through to better modeling, sort of leveraging the abilities of these language models. So the spectrum is pretty broad. I wouldn't say today there is a sort of standard off-the-shelf example for a company. But what you see is there's a wide spectrum. The ones who are leaning in or ahead tend to have a C-suite mandate. Others are quickly following. And is a wide spectrum, but the use cases tend to be centered around probably 4 to 5 key categories. And I'm using it myself in my own department. In fact, we're going to use it in IR. So watch out for our next earnings call. So hopefully, we've got our GenAI up and going for our earnings release.

All right. Well, I'm going to ask you one question just about AI versus non-AI in the sense of, when you think about the business, are there differences between? Whether it's in buying patterns or deal sizes or types of customers? Any sort of -- as you look at the business, what are the differences you see between traditional -- or AI and non-AI?

I think it simply comes back to the customers again. So as I mentioned, it's sort of like the broad categories of customers, Tier 1, Tier 2, sovereigns and then enterprise, obviously, the size of deals scale significantly. So that's one obvious example. But I wouldn't say, look -- and I'm not sure if we're going to talk about this later, just in terms of our traditional -- I don't want to lose sight about traditional server business because it's very healthy. In fact, I would say that our demand in our traditional is -- we're not seeing cannibalization. That's one question I often get asked, are we seeing cannibalization? At this stage, not. But certainly, the traditional business is healthy. The margins are healthy as well. We've seen increases in orders. We're in the midst of a generation inflection in traditional servers. And all of that's playing out very well in terms of our margin structure. So good news is, while AI is going on, we're managing a gen shift inside servers and that's going well. It's a very rich configuration. It's driving higher AUPs in that business as well. So good business as well while the AI business is carrying us.

All right. So just a follow-on question, then I'm going to open it up to the room. If anyone has a question to ask, you can raise your hand, and we'll get you the mic. But just specific to AI -- I'll let you have a drink.

Yes. Thank you.

Take a drink. So you get excited about this AI stuff...

It's my lingering cough.

But the question I ask -- I wanted to ask you was, just thinking about last quarter, the margin differential or the margin profile between traditional and AI servers, was there any notable change in the margins you saw last quarter? Or was it just mix that you saw?

No, no. Thanks, first of all, for asking that question. It's certainly been an area of focus. So like let me just sort of unpack it for you. First of all, this quarter, from a year-on-year basis. If you look what drove gross margins, it really was the mix of our networking revenue. So if you remember a year ago, we're at a very different point in the networking business. So obviously, with that revenue adjustment that took place as the industry went through a correction, it affected the gross margin, as you would imagine. Secondly, from a sequential basis, what impacted gross margin was the actual mix of AI revenue. And I think I mentioned right when we started today that we're very proud of the fact, actually, we shipped $1.3 billion. We converted significant AI revenue in the quarter. We exited the quarter with a backlog of $3.4 billion. But what that meant is that [indiscernible] percent of our servers were actually AI. So that is what impacted gross margins. But let me unpack it a little further for you here as well because I think this is where you're going. From an OI perspective, from a margin quarter-on-quarter, nothing changed. If you looked at the incremental server revenue that we shipped in AI, the margins were basically sort of flattish quarter-on-quarter. In fact, our operating margin was 10.8%. We expect in Q4 to be back at 11%. Our guidance range is 11% to 13%. So expect, even though we'll have some slight sequential improvement in AI revenue, that our traditional server segment -- sorry, our server OP margin will be back at 11%. So I just want to kind of put that in sight and let folks know that traditional server margins are healthy. We're getting a lot of questions about that. I just want to clarify that for folks. And then honestly, enterprise AI, we do expect that, that is going to be a lot more like the margin structure of traditional sort of servers, albeit still an early market yet.

Okay. I'm going to open it up to the room. If anyone had any questions, raise your hand and we'll get you a mic. Right here in front. I shouldn't sit in the middle...

Can I just ask you to build off of what you just said about the next quarter and margins kind of trending a little bit higher next quarter? How -- over the next, call it, 2 to 3 years, how do you expect AI margin, the trajectory of AI margins, to go?

I think the key here is the adoption cycle of enterprise AI. I think as I mentioned just a moment ago, we do expect that enterprise AI margins will look a lot more like margins for traditional servers. Obviously, it's still in early days new market. But I think that's going to be closely linked to how things play out over the next few years. And really, one of the things we're very focused on right now, which is why we announced at our Discover event, new products in the PCAI space, is we want to help enterprises get that time to market and adoption -- and reduce that adoption cycle. So I think what's going to be important for us is really that adoption cycle that we see and the success that we start to see emerging around AI. Good news. Like I said, we have hundreds of customers that we're interested. We're seeing a lot of interest in PoCs. But certainly, I think that's the biggest hurdle for enterprise AI today.

Any other questions in the room? Somebody is nodding no. Oh, wait, somebody back there?

Back here. I think you said modest sequential growth for AI revenues over the next few quarters...

I'd stick with Q4, sir. I'm just giving to the end of '24 guide. So from Q3 to Q4, just for clarification, expect modest sequential growth in AI revenue in server.

Was there a follow-up, or that was it?

[indiscernible]

I'm just going to comment towards the end of '24 because we're going to come to our end of fiscal year here shortly, and then we'll give '25 guidance as we close out the fiscal year.

Fair enough. On the right here. Your left, my right.

Awesome. I have two. I was just wondering if you could provide any context around what percent of the backlog, in rough terms, whether it's just like a minority or close to a majority, have some sort of direct liquid cooling attached today. And when you think about moving from Hopper to Blackwell. If a Blackwell server costs $100, for the customer to buy it, what do you think that direct liquid cooling attach could be for HPE? Is it like $5, $10 of direct liquid cooling attach?

No, no, great questions. In fact, great precursors for the AI Day that we're going to have our very shortly. So we haven't disclosed any sort of facts and figures around direct liquid cooling. But we will certainly give you more insight, if you can attend our AI Day, around the capabilities we have and how we see our unique advantages in that space. So very excited because I think, to your point, those chips are going to require that capability. And the good news is we have it in spades. And the good news is we have manufacturing capacity both here and in Europe, that enables that deployment, and plus the services attach around it. But while we're on the pipeline, maybe I'll just add a couple of comments because I know we're getting questions about pipeline. Look, I'd just say, I think we said it on the call, that overall, our current pipeline for AI service is multiples of our backlog. And what I'm really excited about is if I look at enterprise, the pipeline there has more than doubled in our backlog. So that gives me confidence as we sort of discuss the AI inflection. Today, what we're seeing is enterprise is sort of remaining steady in the low to mid-teens in terms of its participation in our cumulative backlog as well. So that's also another, I think, important data point that gives you the insight into how enterprise AI is really going to sort of impact AI service going forward. So all in all, comments about direct liquid cooling, more to come when we see you in Chippewa Falls.

Okay. I'm not going to get a sneak peek here...

No, you're not going to get a peek here. You got to...

I was trying.

Nice try though.

I'm going to bring it back up here. And then unless somebody had a question, I'm going to -- we'll circle back to the audience.

Sounds good.

So one of the things I want to note is that HP Enterprise is a multifaceted story. You started by saying there's 4 businesses and 4 like main pillars to the story. And unfortunately, part of the challenge in the investment community, you get pigeonholed into one aspect. It's like every time I meet with Pure Storage, people go, what's happening with that hyperscaler win? They don't care what else is happening at the company. And it's unfortunate. Or you meet Seagate here, how's the HAMR qualification coming along? And nobody seems to pay attention to what else is happening. So let's pay attention to what else is happening at HPE outside of AI. And we'll start with hybrid cloud, because it's a big part of the business, and storage. What are some of the trends there? What are some of the things you're seeing on that side of the business?

No, thank you. Thanks for the question. So on hybrid cloud, I would say this last quarter, we definitely saw the impact of some major product and model shifts that we're having. So I think we hit over 5% on our operating profit, so a significant sequential improvement, and we expect that to continue into the sort of same range as we go into Q4. And what's happening in hybrid cloud, a couple of things. First of all, GreenLake. And I think Antonio, I know, is very passionate. He started GreenLake back in 2019. We have in the high 30,000 in terms of customers, so we're starting to see more and more customers turn to the GreenLake platform both in terms of as a model in terms of consumption. AAR I think is a great indicator and barometer of success there. We hit $1.7 billion in the quarter. We're right inside our range of 35% to 45% CAGR over 3 years. So overall we're shifting the model. And then if you unpack it further, one of the areas in a hybrid cloud we're really focused on is storage. And thank you for bringing up Pure Storage. I think it's a great segue. This quarter, we're -- actually this year, we're in the middle of a very significant transition in storage, both in terms of both product. We're moving to our Alletra MP platform. And then in terms of our model, we're actually shifting to a much more software-defined storage. What that means essentially is that we're going to be deferring a lot more about revenue over time. So that's important to understand, and we'll give you more insight to that as we guide the end of the year. But I think what we're pleased about is that so far to date, we've seen orders sequentially increase significantly. We're seeing that, that product transition is resonating in the market. We've also gone through a massive sales force transformation, where we're moving to a much more specialized sales force because we realized that, that type of storage sales requires a lot more handholding with the enterprise. So the good news is we're in the transition. The good news is we're still in the transition, so it's going to take time. But we're starting to see early signs of that transition, and it's positive so far to date. So I think that is a good one. And then finally, hybrid cloud is where we launched the private cloud offering that I mentioned on AI as well. So a lot going on in hybrid cloud, but more to come.

Okay. But I'm going to move on to intelligent networking, but just I'll poll the room one time if there's any questions around storage or hybrid cloud.

Or around network...

Going once? Okay. Intelligent network has been a great story. It's been a great story, and it gets better. So there's been some M&A or pending, whatever you want to call it, but there's things happening there.

There's a lot going on in there.

There's a lot going on. And it's funny because I saw one of your competitors in networking yesterday, and they were like, "Oh, the customers are all confused now, what's going to be the strategy? So we're going to pick it all off here." So again, everyone is entitled to their opinion. What's your opinion? What's going on in networking? What are the key drivers? You're moving up market, up margin. Talk about how HP thinks about networking today and in the future.

That's a long story, by the way. Let me start with the -- first of all, in terms of the quarter, I'll anchor it back. I mean, we definitely are through the trough. I think that's the question we're getting from a lot of folks. I think you've seen that from some of our industry peers. The good news is that we've moved through that trough of networking. If you look at our sequential revenue, we're up 3%. And we're seeing what I would say the bright spots of the early signs of the recovery, both geographically and in certain segments. And from a geographical perspective, certainly, as you would typically expect, it's the Americas region where we're starting to see those early shoots in the sort of the early stage of the recovery. In addition to that, what we've seen from a business perspective, it's really in WLAN, services, SASE this quarter, which we saw some of those green shoots of recovery. I might add as well that what gives me confidence is also just the state of our channel inventory. And today, our channel inventory, if I look at it right now, it's in a healthy shape, the sell-out continues to be decent. So all of that together gives me confidence that we are definitely on the path to the recovery. I would just predicate and say it's a modest recovery, though. And as we look forward from Q3 to Q4, we do expect to see some modest improvement in the business going forward as well. So overall, networking sort of in a state of recovery. But I'd say what gives me a lot of time, a lot of focus going forward, is obviously as we get into closing the Juniper transaction. And as of yesterday, for those of you who might missed it, pleased to say we actually closed on our stake of H3C, our 30% stake. We received $2.1 billion in cash. It's in our bank. So we're really happy about that. And our Juniper transaction is still very much on track to close, as per our prior guide, in terms of end of calendar '24, beginning of calendar '25. So we're going to look like a fairly different company once we double down on networking. And what that means is that our gross and operating margins will have a significant, dramatic impact from being very -- with a large concentration of networking in those numbers. And what that means is that, from an operating profit perspective, more than 50% of our operating profit going forward once we close the transaction is actually going to come from networking. So I assume going forward, we might spend more time on networking than AI. I'm not sure, but we'll see.

When is that Analyst Day going to be?

It's coming. Yet to be announced. It's coming. It's going to be a record-attended one, I can tell, based on the questions we're having. But yes, we will be doing some type of I'll just say networking event. I see Shannon getting worried about what we're going to do there. But we'll do some type of networking event after we close the transaction.

I like this. Okay. And it's funny because you already took away one of my questions about HC3. Well 2 days ago, so what do you expect from me? And of course, the other one, which I'm wondering now, is just even relevant, is I was going to ask you about margins, particularly given rising input costs. But as I've stood around this conference the last 2 days, input cost sounds like they're not rising anymore, at least on the chip side. So how do you think about margins either -- because again, there's so many moving parts here. I don't know how you do this. But as you think about all the moving parts, how do you think about margins generally? I mean, you've guided only for the October quarter. But just how do you think about the puts and takes around margins?

Look, I mean, from a commodity cost perspective, we certainly -- we're managing through inflationary cost environments. And I think as a company, we've got a good history of managing through those cycles. We all lived through it on COVID. We know how to do it. And there are the headwinds and tailwinds in the various businesses, so you've got to be able to figure out the right pricing strategy. And also the other thing is, frankly, take advantage of strategic buys where it makes economic sense. And I think I made those comments on the call yesterday. We'll continue to be prudent about it, but where it makes economic sense, we'll lean in and do a strategic buy to ensure we get the right sort of economic price structure out there as well.

Okay. One for the audience, and I come back. Oh, we got somebody. Okay. I love when they help me out because...

Can you just talk a little bit more about the Juniper transaction? You're adding quite a bit of leverage. I know you're targeting reducing leverage kind of post-transaction. Can you just kind of talk about how that looks, how much is debt repayment versus EBITDA growth? And then once you get to that leverage target, kind of how do you envision the financial policy on a go-forward basis?

Sounds like they are all the right questions to ask on the networking event. So we still haven't closed Juniper yet, and I'd say nothing's really changed from what we said at the onset when we closed the transaction. We do expect to manage our leverage down. I think we said we would want to manage down sort of post '26. But frankly, outside of what we said back when we consummated the transaction, all the type of insights that you've asked for, we'll give you that once we close and then once we have that networking event. So stay tuned is what I would say.

I was going to ask a follow-up question just to talk a little bit about with your CFO hat on. I don't know if you have another hat, but you're doing a great job at the whole thing. But on the CFO hat side, remind us about inventory trends, cash conversion cycles, free cash flow expectations. Like how do you think about the operations of the business?

A lot. So no, let me walk you through how to think about inventory, sort of cash flow, and what we think about working capital and the like. So from an inventory perspective, we were up sequentially around $350 million. That was driven really by two primary factors. The first was obviously the build plan for AI revenue conversion. So as we ship the AI revenue, which I mentioned, we do expect to see a modest sequential increase from Q3 to Q4, you'll see that depart out of inventory. And then secondly, we also want to be opportunistic for strategic buys where it makes economic sense. So you're seeing that impact our inventory. So as a result of that, from a working capital perspective, certainly, it was a use of cash in Q3. And then we expect to convert that, as I mentioned, in Q4. As those AI systems move out the door, we see working capital being a source of cash in Q4. Henceforth, we expect to -- as I mentioned yesterday, we guided to $1.9 billion in free cash flow for the year. As we're sort of getting down to the last quarter, we just sort of mentioned that we really expect to hit $1.9 billion. We've only got a couple of months left to go here in the fiscal year. So that's how we're thinking about it. And candidly speaking, one thing that I think is clear from the AI business, that there is such a large momentum in these deals. You have to build up and build large, frankly, large complex structures, whether you're servicing sovereigns or you're servicing some of the Tier 1, Tier 2. So you have to build inventory, you have to do an installation. A customer has to go through an acceptance period. So that all takes time, and that's why you see some of those inflections in inventory. So just bear that in mind as you look through our inventory numbers. However, we do expect to see a sequential reduction in our inventory going from Q3 to Q4, I might add.

Okay. I have one last question, but I'm going to take one more from the audience because I'm being blinded by these lights. So I feel like I'm going to confess to something. Do we have any questions in the audience?

Are we expecting to deploy the $2.1 billion of H3C proceeds? Will they be used to fund the Juniper transaction? And when do you expect to realize, I think, the remaining balance of the stake sale on H3C?

Yes, no. To your second question, I think we said it'd take about 18 months or so to realize the second stake, which we still got about 19% left in H3C. And the $2.1 billion, pleased to say it's sitting in our cash, in our bank accounts right now today. So that's a good thing. And obviously, we'll be giving more disclosure around proceeds and how we assume to use those proceeds as the transaction date closes. So still got to wait a little while here. We expect, as I said earlier, the transaction won't close till end of calendar year '24, beginning of '25. But obviously, just really pleased we got the cash in the door earlier than we had originally potentially expected. So super good news. And the fact it came on earnings day. I thought that was a really good thing, actually.

Okay. We're good? Okay. Last question. Let's close strong here. Why should investors buy HPE?

That's a great question. Well, look, first of all, I joined HPE myself from HPQ only a few months ago. And 1 of the 2 main reasons why I joined. First of all, and I think hopefully, this resonates with investors. I think we're at the inflection point of an amazing transformational TAM opportunity with AI. And I believe that HPE is incredibly well positioned to participate in that TAM. And frankly, we have a right to win, as we talked about throughout our conference today, particularly in areas like sovereign and enterprise AI. And so excited to be part of that. And I think from an investor perspective, we're a great company for you to be considering. Secondly, we have a very transformative M&A transaction right in front of us. And in itself, that is going to provide, I think, a significant sort of transformation to the financial structure of the company. And really, as we think about it, more than -- like I said earlier, more than 50% of the operating profit is going to come for networking. So two, what I think, great building blocks to be thinking about going forward. And then finally, I'd say from a capital allocation framework perspective, we -- our framework hasn't changed. We have a very balanced framework and it relies on four main tenets. The first is we'll pursue growth and invest in growth where we get value. And obviously, that Juniper transaction sort of checks that box. Secondly, we'll continue to obviously pay dividends. Thirdly, buy back our shares. And then finally, we want to maintain our investment-grade credit rating. We have a very important financial services business that is actually a very strategic asset, particularly as companies look to adopt AI. The HPEFS team, we've seen tremendous growth in their volumes specifically around AI. So it's so important for us to continue to maintain that investment-grade credit rating. So overall, I think that we are a great investment. The time is right. As I said earlier, the transaction still is expected to close in the time frame I mentioned earlier, and we're in the early days of the AI adoption and inflection cycle. So I think we're incredibly well positioned. Delighted to be here and very happy that I joined as the CFO right at this, I think, strategic point in time.

I was going to say moving from HPQ to HPE, you could have just said, "I just decided that PCAI is better than AIPC." But I don't want to make the other guys feel bad. Well, Marie, thank you very much for your time. Thank you all for joining and enjoy the rest of the conference. Thank you.

Thank you. Thank you. Thanks very much.