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

All right. Well, let's get started. Everybody, happy Friday. Thanks for joining us today. My name is Ittai Kidron, and I'm a technology analyst at Oppenheimer. And I have the pleasure of hosting Antonio, the CEO of HPE; and Andrew, who runs IR for HPE. Thanks, guys, for joining us today. Appreciate your time. Before we get started, Antonio and I will have a bit of a fireside chat in the discussion. And of course, before we get started, Andrew would like to say his word as well by the prime directive of the attorneys. But I would like to remind everybody that as much as Antonio and I are going to chitchat here, we would love for you to participate as well, and you have a couple of ways of doing this. [Operator Instructions] And so before Antonio and I get into the details, Andrew, the floor is yours.

Great. Thanks, Ittai. I appreciate you having us here on this morning. So just before we get started, let me read the disclosures really quickly. So you'll hear some forward-looking statements in today's discussions. They are based on risks and assumptions that are described in our annual report on Form 10-K and Form 10-Q. Our actual results could differ materially, and we assume no obligation to update these. More details can be found on our website at investors.hpe.com and our recent Q4 earnings announcement press release dated December 1 of last year. So with that, Ittai, let me turn it back to you. We can start the discussion.

That's great. Friday. We got the short version. That's great. So Antonio, again, thanks for joining us today.

Why don't we start? Clearly, everybody is focused on the macro environment just given everything that we've been through since March last year and given the second and third rounds that are happening. Since you last reported, what can you tell us about your observations of the market, especially perhaps with respect to sales cycles, impact on second, third waves and depending on the regions? And of course, how enterprises -- how do you feel and see enterprises budgeting for '21?

Sure. Well, good morning, Ittai. Good morning, good afternoon, everyone. Thank you for having me today. Happy Friday to everyone. I hope everyone stays safe and well and healthy. So I mean, since we announced our results in early December, we continued to see obviously some uncertainty in the market driven by the resurgence of cases. Obviously, the COVID situation continued to be very concerning out there with a lot of hope and focus on the vaccine. But at the same time, I have to say we are seeing sign of stabilization in the business. That has continued to be the case in the last, I will say, 4, 5 months. And you saw that in our results, particularly in our Q4, where we return to pre-pandemic levels of our revenue at the total company level. But what we see also is a strengthening in certain areas of the demand, driven by what we experienced today. In fact, we are all working remotely, and that has created -- the global pandemic has created some lasting and permanent changes in the way we're going to work, not just today, but going forward. And what that's driving is demand on areas like remote connectivity. Obviously, you got to be connected to participate not just the workforce, but also in the global economy and in this digital economy. Second is everything that's cloud-enabled because it accelerates the deployment of apps and data, whatever you needed. The third element of this is obviously all-around cybersecurity, which obviously, people are very concerned these days. And then anything that preserves capital. And that's where our very strong momentum on GreenLake continues to be there because now people have the opportunity to modernize their estate and then cloudify everything they have, whether it's on-prem or leveraging somebody else's cloud but basically consume it the same way. And last but not least, what I see is a significant push in demand around anything that's around data insights, whether it's machine learning, AI. And we talk about what that enterprise of the future looks like. And 3 years ago, I said that will be edge-centric, cloud-enabled and data-driven. And it's kind of interesting. Unfortunately, pandemic has validated that. We're going to live a much more distributed enterprise, which means the edge becomes even more important than ever. Obviously, cloud is an experience, not a destination, how we bring the cloud to all your apps and data and then everything about the data. So we see that. But even on the core business, I see signs of stabilization. And then obviously, as we think about '21 and beyond, there are going to be new technologies that get deployed that allows us to be a catalyst to strengthen our portfolio.

That's great. Okay. So it sounds like a still choppy, I guess, environment. But to the areas of either priority or investment that enterprises are certainly putting their weight into, right, just given the environment and how do they think about the business position going forward.

And by the way, Ittai, I think the digital transformation is unstoppable. If you talk about digital economy, you are now an IT company no matter what you do. And it's no longer a priority, it's an imperative.

Got it. Got it. That makes sense. All right. So you touched on the several product areas that clearly are part of your business as well. Maybe just from a big picture standpoint before we start getting into each and every one of those, you've kind of laid out a plan where you want to maintain kind of single-digit growth in your business going forward. Can you tell us what businesses do you expect to contribute? How do you expect the business to contribute to this? Which one will be leading in this effort? And which one do you think will be lagging? And how would that potentially change in that 24-month time frame?

Well, if you step back, Ittai, and go back to what I stated and we shared at the Securities Analyst Meeting at the end of October, we have a clear vision for our company, which is to become that edge-to-cloud platform-as-a-service company. We're going to live and [ evolve ] in a decentralized, centralized world. And we talked -- just recently, we talked about the digital transformation, everything starts with connectivity. Secure, seamless connectivity is the first step in everything you do. And so when we think about the edge-to-cloud platform-as-a-service company, we have a 3-pronged approach. Number one, obviously, we've got to stabilize the core business, which is the, call it, the general compute and storage, and transform it into these new cloud-native architectures. And you will see this year an enormous amount of innovation we're going to bring to market in the way we provision a life cycle, a compute or a data service resource. And that will fuel not just the stabilization, but the growth in particular segment of the markets, like what we call the Tier 2, Tier 3 service provider, which is most of them are either born in the cloud or SaaS companies. The telco, which obviously will be a huge demand that we see the demand in deploying 5G, both the core network and at the edge or the new edge. And then obviously, on the storage side, it's all about managing that data gravity in a very software defined and autonomous way. So that's the, what I call the stabilize and transform the compute, which obviously is a big source of our free cash flow. Then the group pockets of growth against that connectivity is Aruba. It's our Aruba Edge Services Platform. And you saw the acquisition of Silver Peak, which strategically allows us to provide that connectivity in a secure way, in an autonomous way for all your edges and all your clouds. So now we're going to have millions of clouds deployed everywhere because anything you deploy, whether the edge in a manufacturing floor, in an autonomous car or in a hospital, is going to be a cloud kind of solution. And you have to connect those clouds to the main cloud, whether it's in the your prem or somebody else's cloud. And that Aruba is a source of significant growth for us because that's an opportunity that we have in front of us to capture a significant amount of share. So if you think about the 2000 to the 2010, we were in the data center networking. In the 2010 to the 2020, we were in the cloud computing. In the 2020 and beyond, we're going to be in the edge computing. And so the Aruba not only will provide the ubiquity and secure connectivity, but also we will incorporate 5G. And also, we will incorporate that edge compute, which we will now deploy in 2021. And then if we think about this massive amount of data we are creating everywhere -- even this webcast, right, we are recording it. By the time we're done, you're going to have a few gigabytes. You're going to store it somewhere and make available. But the point is you have to process that data. This webcast eventually will be translated and the transcript will be automated by the robots, which is an AI kind of solution. That's an example of it. But the backbone for all of these will be high-performance computing. And that's a business we are very bullish about because we have fully integrated IP from the silicon all the way to the application layer with the acquisition of Cray and SGI, an organic business. And as you know, we have been awarded more than $2.5 billion now in business that we have to build and ship over the next 18 months. And that will be a source for growth for us. But then last one, which is in that edge-to-cloud as a service is the as-a-service pivot. Today, $6 billion of the on-prem market is consumer-as-a-service. In 5 years from now, it will be $22 billion. And we have a point of differentiation there in multiple areas. Now first is our software. Our software is all geared to enable these as-a-service experiences. And we've made organic and inorganic acquisition on that front. And that's why GreenLake, in my view, is 2 to 3 years ahead the competition. And second is our HP Financial Services because you need the financing embedded into that experience. It is not a financing business model, it is a services business model that includes financing. So that's where our 3-pronged approach is: stabilize, transform the core; growth in HPC and at the edge; and then the pivot to as-a-service. And that's why we guided, Ittai, today for the 1% to 3% because of the massive inflection points in our portfolio.

That's great. So let's tackle these 3 elements maybe in different order. We'll start with the last one, the GreenLake, which for my view is the most interesting part of your portfolio right here, right now. Clearly, it's in the center of your strategy. And can you talk about the value proposition here? How does this work? I mean is this just like an Amazon? Am I just buying a service and the only difference is that it's Amazon deployed on premise? Is that the idea of this?

Well, at a high level, the way I explained GreenLake is the cloud that comes to you, right? So if you think about the economics of the cloud...

I saw the whole idea of the cloud [indiscernible] goes away from me that I don't want [indiscernible]

But the cloud that comes to you as an experience, right? So here is the interesting thought about it, right? So if you think about the cloud, the cloud is obviously a very efficient model but has -- complexify your estate because the fact of the matter is that you have applications and data everywhere. First of all, data has a gravity and obviously, data is your intellectual property. Number two, today, the cost of compute is no longer the main decision why you go to the cloud. You go to the cloud to ride a wave of innovation at the speed that you couldn't afford to do it yourself. But the higher cost today is the cost of bandwidth. It's the cost of moving data back and forth. I have to say that once you check in the public cloud, with your data, it's like checking in, in Hotel California. You check in, you never check out because it's too expensive to repatriate that data. And so what GreenLake would do, we bring the same cloud experience you will get off premises to the apps and data that they are better suited to be on-prem in a co-location or at the edge because a lot of the applications and data are moving to the edge. In fact, more than 60% of the data is created at the edge. That's why you see that [ explosion ] with the edge and the near edge inference, which we have amazing solutions. But GreenLake brings that experience for all your apps and data, and we manage the entire hybrid estate for you with the same consumption model because if you have an application and data on the public cloud through our managed services with GreenLake, we curate that app and data for you in a very automated way. And so you are no longer in the run part of the business. We extract you from that. We free up capital because you no longer spend the CapEx, you're just paying only for what you consume in an elastic way. And then you repurpose those resources into the innovation side of your business, which ultimately is to extract insight from the data as fast as possible. And that's where GreenLake brings all these elements together. And the first step in that value proposition is software. And people generally don't see us as a software company, obviously, for historical reason. But the whole experience is about the software layers of the GreenLake Central platform with all APIs that we have built and the ability to spin an instance on-premise for that workload that's truly optimized between our software, our infrastructure and our services. And if you have an app, let's say, in AWS or Azure or Google, we actually connect that to your estate, and we manage it for you. And then we only build for you what you consume. Because the problem is when you're in the public cloud, if you don't curate that, you turn it on and you forget And then it's like being on a cab, right, in a taxicab that keeps billing. So we do all of that for the customer.

Got it, interesting. Is there a -- so I want to separate the business part and the technology part when we talk about this. From a business standpoint, are there specific either applications or types of data or types of industries where you say this creates better [ whiff ] having that run in the GreenLake on-premise versus running in a traditional Amazon-like type of a cloud?

Yes. Yes, very good question. I think containerization, virtualization, but anything that's big data-intensive workloads is better suited for this type of solutions because you are managing enormous amount of data. And with our introduction of our runtime software called HPE Ezmeral, we actually saw 3 fundamental problems for our customers. Number one, the ability to run stateful and stateless, meaning legacy and cloud-native applications in the same solution. Second is the fact, automation. You want to reduce operating expenses and remove labor out of the process. We addressed that problem with our automation because of a previous acquisition with BlueData Storage and HP InfoSight that came through the HP Nimble acquisition. And third is all about the data. And with the acquisition of MapR, we integrated that solution into the same runtime, which we are -- now we can manage trillions of data points at massive scale. Great example of that is what we did for Wells Fargo. They have a workload, Splunk, right, since you asked what type of application, that has compliance requirements, a massive amount of data, and we are able to process that data real time at a massive scale. And they only pay for what you consume through GreenLake. So a lot of this is data-intensive workloads, cloud-native workloads but ultimately tied to that data gravity and the ability to do it in a very efficient and elastic way.

Got it. So you've touched on the technology part, maybe we'll shift to the technology. It sounds like -- and of course, I'm simplifying all of this, but you just need a highly sophisticated management software and you're kind of good to go. Like how hard is this? Tell me the investment that went to it, what's proprietary here. I mean Cisco has got a lot of these resources as well. Why can't Cisco come up with something like this or any other hardware company for that matter?

Yes. So we -- you need a lot of software. There is no question about it because it's all about automating things, analyzing things real time and be able to do it in an efficient way, particularly by moving applications back and forth, which obviously microservices and containers are the most efficient way to do it. But at the same time, you have to address the security aspect of this. So we invested a lot in true cloud-native runtime solutions that are basically the future. We address the fundamental data connectivity challenges, which is very hard to do in this environment, particularly in a hybrid world, and then we address the security piece of this. The orchestration brokering, actually, it is something that customers are challenged because the fact of the matter is that the world will be multi-cloud. You cannot lock yourself into one cloud. You want to have choice and flexibility. But to manage this multi-cloud, you need to connect to all these proprietary stacks. And we alleviate all of that for the customer. We provide them the control plan. Basically, we give customers back control, choice and flexibility, and we do that through our platform, which is basically GreenLake Central. And that's all software-driven API, a combination of organic and inorganic acquisitions. And I don't want to spend too much time on the technology aspect of this, but it takes a lot of effort to make this app, I have to tell you.

Got it. Okay, awesome. Maybe we'll talk about the impact of this on your business model. So if I'm paying you as I go, I guess, my -- based on the amount of data and compute that I'm consuming in this platform, what is the trade-off, the positive and the negative to your financial model with the implementation of GreenLake?

Well, I mean, we give a customer flexibility in choice, right? So with GreenLake, obviously, you are on a pay-as-you-go model, which is a utility-based model, inclusive of software and the infrastructure and the services you consume into that. And in some cases, we define a minimum commitment. In some cases, a true elastic model. But general customers like a minimum commitment, and then they go from there and they add and they keep adding. And by the way, we are there for them without them even knowing because we actually can monitor real time the consumption. And then we said, we need to add more storage and we add more storage. Or we need to add more memory and we add more memory or more compute capacity, whatever it is. And in that, the customer, they have the choice to structure the financing underneath 2 ways, either a CapEx leasing or an operating model leasing, depending on what it is. The trade-off for us is that, obviously, if you're in a CapEx leasing, you are able to recognize revenue upfront much faster, particularly on the hardware side. If you're an operating leasing, it gets more distributed over time. But right now, I would say, Ittai, I think it's more like 50-50 at this point in time. And that's why for us, this transition on the revenue is not as hard as particularly on a software, for example, to a SaaS model, where you go through a massive valley and then you come back a few years out. We're going to see a little bit more smoother and linear because of the way customers structure the operating leasing. And that's where we committed that we're going to triple this business in the next 3 years. And we already are on an ARR, which is the annualized revenue run rate of $585 million or so. And we committed to grow that business 30% to 40% on the CAGR because of the differentiation and the value proposition we have in this business.

Got it. Interesting. Is this -- when you think about the overall footprint of your entire storage and server globally, which is clearly massive at this point, is there a percentage in mind, Do you think that much of my footprint longer term should be able to run in this type of our business model?

Yes. I mean if you look at the numbers we committed, it's $4.5 billion. You think it's between 15% and 20% over time. It's just a matter of time because obviously, I call it like a snowball, right? When a snowball comes down the hill, it becomes an avalanche and eventually it gets bigger and bigger and bigger. But remember, it's not just GreenLake as a utility. It's also what goes underneath is also the subscription model to our software. And you will see that everything we do in our compute and storage with that software life cycle management that gets bolted underneath GreenLake will be also a subscription-based model. And even if you don't sell it as a GreenLake, the customer wants whatever -- one of our servers with the software and that they use, then that subscription software will be also recurring for us going forward. So we're going more from perpetualized into also a SaaS model, and that's why the whole architecture comes together.

Got it. Interesting. From a portfolio standpoint, GreenLake as a model is applied to multiple types of products that you're selling. Tell me what products are already consumable in the GreenLake fashion and what's to come? And in what way GreenLake would be different 2, 3 years from now versus what it is today?

Well, all our products are now in GreenLake, whether it's compute or storage, which obviously has been the, what I call the infrastructure-as-a-service piece of this, the IaaS piece of this. We just, in December, introduced what I think is going to be super exciting is HPC-as-a-Service. High-performance computing, generally speaking, has been available to only a few segments of the market, right? Think about government, academia, oil and gas, those -- weather, right? Those are that needs massive amount of computational power. But the problem is the amortization cycle. So the business is getting smaller and smaller and smaller because they want to leverage the latest technology faster. And so we now have done partnerships with co-locations like the Cirrus of the world, the Equinix of the world, where we can put these massive clusters and then only pay for running a specific job. And that's the HPC-as-a-Service, whether it's Apollo or Cray, would be also now under the GreenLake model. And then what we are working now is some interesting solutions where you actually don't need to consume juice per CPU but potentially per core or per memory cycle. And that also gives us even more differentiation. And then last but not least, on the Aruba side, we already have a subscription model, a SaaS model on the cloud. So you can subscribe to Aruba Edge Services Platform where you can provision an access point. Let's say, Ittai now has a different house that needs to go work. We ship you the access point. And with 3 clicks, you get provision, your WiFi. And then we connect you to your Oppenheimer network and then you're good to go. That subscription is a SaaS model into the AESP platform. Today, we already have 65,000 customers under the platform and more than 1.5 million devices. To give a sense, we had roughly 10,000 to 12,000 devices every day to that platform because we design it with IoT in mind. And obviously, security analytics is built into it. But the one -- the next generation for Aruba under GreenLake is going to be what we call the NaaS, the Network-as-a-Service, which is inclusive of the hardware piece of this because Aruba at the core is a software company. They just happen to attach access points and switches. That's where the model is.

Interesting, super interesting. As I think about this platform, I can't help think, let my imagination kind of run a little bit forward on how many services you can launch on top of this platform. There's so many directions you can take within data and AI longer term, right, and other places as well that you can layer on top of this platform going forward. So if we look back in the last 2 years, the acquisitions have been things like Aruba and Silver Peak and Cray. It feels like a lot of the acquisitions going forward could be much more software specific. I mean, of course, all these companies had a software element in them and I'm not trying to discount that. But you could buy businesses that are pure software going forward to enable that on top of the platform. How do we think about GreenLake as a launchpad in the future to completely new software-only businesses?

Yes. Well, I think you just described what our long-term strategy is, is to create a platform that can be consumed as a service, right? And GreenLake is the tip of the sphere of our strategy. Underneath that, what we provide is experiences. And the experience could be a connectivity. It can be a container solution. It can be backup-as-a-service. It can be SAP-as-a-service, right? So we are offering already SAP-as-a-service. Or it can be AI-as-a-service. We already launched an ML Ops as a service. But the point we know -- roughly speaking, there are 100 workloads that enterprises of all sizes uses, and we have narrowed that to 40. And on the 40, we are focused on 17 that can be quickly spin off in the platform and then open it to the ecosystem of partners so they can write new services on that platform. And remember that 70% of our business goes through the channel partners, whether it's traditional distributors or value-added resellers or solution integrators or ISVs. We are actually doing that work through the APIs, integrating them. So there is a way to drag more people into the platform because ultimately, if you are a developer, you don't want to deal with a friction, right? So I just got to the platform. But if the customer says that data needs to stay on-prem for whatever reason, they get the same experience. Today, we offer a lot of the services with things like MongoDB, Cassandra, you name it, that they can be consumed totally on premises. And that's why we believe we have a unique and exciting strategy from the edge to the cloud, which is the architectures customers need in this new world.

Got it. Interesting. Maybe last one on GreenLake. From a regional standpoint, is there a difference in the way this is accepted and used U.S. versus rest of the world?

Not really. The incubation of this business -- and that's why I always have to think about where the innovation happens, right? Actually happened many years ago in an offer we called Flex Capacity. And it was incubated in the Northern Europe. And we saw the potential, and then we brought it on this side of the chart. And then we expanded, and we started adding IP and acquisitions that we did along the way. But it's very consistent. Listen, I cannot share some of the deals, but just today, exactly half an hour ago, we signed a massive deal with GreenLake with a specific entity in Europe. But generally, it happens everywhere. It happens. And what I'm excited about the GreenLake is not just the momentum. It's the size and the number of deals is getting bigger. So the size of the deal now is getting bigger and bigger, and the number of deals is growing better rapidly. And you can see in our orders bookings, right, we go between 30% and 60% every quarter, depending on the seasonality and where we are. But this is going to continue that way. And that's why we felt confident in that guidance.

Got it. Well, maybe we'd take that example. That European customer, don't mention the name, of course, but was that an existing customer of server and storage [indiscernible]

Correct, yes.

And so when you think about the dollars and cents ultimately of them taking hardware from you that they bought, that they installed and had to manage to now moving into GreenLake model, when you do all of that for them, the dollars and cents, is that coming up favorably to you? Like what's the incremental...

Absolutely. Well, let me tell you what -- the benefits for them. First of all, it will be a cloud implementation that we are doing for them under the GreenLake umbrella. It will be multi-country, and it will be for specific type of data they need to manage. In that, we are deploying the full suite of our products, not just in the core data center with the cloud but also the edge with some of our solutions. And remember, when we sell GreenLake, what we have seen, because it is a linear consumption model: a, the hardware margins are better; and b, we attach 100% of our HP Pointnext operational services, which obviously have higher margins. And so -- and by the way, at the same time, we take competitors out because they don't have the opportunity to participate in their model. And then we are no longer in a procurement cycle.

Got it. And the customer saves OpEx. I mean that's...

A lot of dollars because they don't have to invest all the capital upfront, right? So we manage all that for them. And then they focus on what the innovation needs to be around that type of data they need to manage.

Got it. Okay, cool. All right. That's great. Let's move to the switching and edge and start with Aruba, which has been a home run for you guys -- acquisition. Now that's...

That was my first acquisition in 2015.

There you go. So I guess they gave you the CEO position for a reason, I guess. You made this on the right horses. Let's talk about though how Aruba has kind of evolved over time within the HPE framework and -- but also perhaps not just because of technology but also in the context of use cases, in the context of the types of customers you sell to.

Yes. Well, listen, I'm very proud of Aruba for the value that we got. Obviously, it is core to the future of this company because whether you think about edge to cloud, you think about what happened with the pandemic -- and by the way, we're really proud of what we have contributed in the last 12 months with our people and our technology. The fact of the matter, every step on the digital transformation starts with connectivity. And what we see now is a way more distributed enterprise, as I said earlier. And we -- this company has gone through a significant transformation since the 2002 founding, which really, their focus has been what is the most important problem we can solve for customers and how we do it in a modern way that ultimately is all integrated for them. And Aruba, when we acquire that, I actually did a reverse integration. So I took the switching portfolio that HP had, which was another acquisition called ProCurve in 2005 and then we reverse integrated in Aruba. And since then, not only we integrated WiFi and switching with a common operating system in the cloud, and then we also created SD-Branch. And now with Silver Peak, we have also the SD-WAN. And we integrate all sorts of connectivity, including things in the IoT like protocol like ZigBee, Bluetooth and BLE and so forth because now we have a platform we can scale for the IoT era. And as I think about the use cases, think about smart manufacturing, right? So we have to bridge the digital and the physical world. And HPE's unique position into that, also because of our compute capabilities with our edge lines. So we integrate connectivity and compute for that specific use case. When I think about autonomous driving, we just did a very large deal under GreenLake called Zenuity, which now changed the name, it's called Zenzic, but they are developing software for the autonomous vehicles. And when -- I was in one of those vehicles in the test environment that go around -- in 20 minutes, they collect 6 terabytes of data. That data has to be downloaded, okay, when the car comes back in the fleet station. So we implemented our Aruba technology, and then we process that data in a scalable way. Today, they had 40 cars in testing, but they keep adding 10 cars every so often. And so that means the number of storage per gigabyte keep growing. And so they get their scalability and elasticity as they go along the way. But you're bringing everything together for them. But Aruba for us is going to be a source of quite significant growth for the quarters to come and is core to what we see in the market today.

Got it. One of the questions I've gotten from investors is why would Aruba see good momentum here and now where -- work-from-home environment, where -- arguably, who needs to deploy WiFi in an office when nobody is in office? Or there's clearly no sense of urgency in doing that right here, right now. How would you respond to that?

Well, those access points have to be deployed in your houses and it has to be secured, right? So I mean, normally, in an office, think about my office here in the back, right? So that office, obviously, is state-of-the-art, digitally enabled, has all Aruba technology. We don't have phones in that building. It's all Voice over IP. I pick phones either through my PC or my cellphone because it's all connected. There is no phones. But now we need to translate that to your house, right? So in that building, I can drive density with maybe, I don't know, 500 access points. But now I have maybe 1,500 people that I need to send to the homes. And at least if you even have one access point, you need 1,500. There, you need 500. So the growth of this distributed model, right, will continue to expand. By the way, a great example that we won in December, we announced this publicly, is the Pentagon. Pentagon was already an Aruba customer. It used to be a Cisco switching, and now they consolidate and standardize everything with Aruba switching and WiFi. So there's a lot of opportunity. By the way, that building behind me, one of the interesting use cases. You see that glass? The glass has unique characteristics where they change with the temperature and the light. But we are working with several of these companies to integrate the 5G antenna into the glass. Think about the building become your big 5G antenna. But once you go to hit the building, now you offload to the access point because the problem with 5G is $0.05 per gigabit. You go to an access point, it's $0.01 per gigabit. So there's a ton of opportunity use cases that will see exposure in the next few years to come.

Interesting. Well, maybe on that front, I mean, you've talked about Aruba and how the portfolio around it, all the pieces kind of started coming together right at the integration of ProCurve and then Silver Peak and SD-WAN, all of that stuff. Help me understand, when you look at the platform together right here, right now, what's missing, do you think, in the platform? Like, if you and I talked about Aruba and the whole platform 2, 3 years down the road, in what way will it be different than what it is today?

Yes, I think it's not missing. You need to continue to add, I will say. It's more about the security aspects, right? As you add more capabilities, you need to add the security with it. And we think about security built at the core not as an afterthought. One thing is secure the perimeter and one thing is secure your interaction between devices and data of devices and applications. And so whether -- we actually have an AI built into it. Basically, if Ittai comes to the HP office, we give you a guest pass. And obviously, we know what services we can provision to you. But if Ittai started trying to go to certain places once in the office, we actually monitor that. And we can automatically, through our AI technologies, quarantine you because we are monitoring your behavior at the application and data level because of the monitor of the packets that we drive in the networks. And so this is more what we are doing over time. And then obviously, the other one is keep adding more analytics capabilities because at the end, connectivity is all about providing an experience. And so I will say it's not like it's missing. It's the fact that you need to keep adding more and more. But this is why Aruba was conceived as a cloud-native solution. It's a cloud model, by the way, which is easy to add this functionality. By the way, in the GreenLake model, the back-end infrastructure that supports the cloud operations is consistent with our own model. We took the Aruba back-end cloud and we extended to the core business.

Interesting. As I think about the security and AI elements that you seem to be focusing on investing and building over the next couple of years, do those come then as incremental software subscriptions? Is that what...

Yes, correct. It's a modularized mass customization type of model that basically, if you want extra functionality, you can basically subscribe to that. And we keep that packaging very simple because ultimately, it has to be simple to approach and learn it, try it, buy it and eventually last cycle manage. But yes, every time we add more functionality, you subscribe to that if you want that functionality.

Got it. That makes sense. Now you don't really have much about data center switch portfolio. Is there any limitation or handicap in kind of trying to push the Aruba vision through by the virtue of the fact that you can't talk to the data centers, which natively, right, it's not the same thing?

Well, think about the architectures, Ittai. So there is the core data center. There is the edge of the data center, and then there is the campus and branch. And I argue, now there is the micro branch, which is your house, right? So we decided for historical reasons and where our right to play and when is, is to focus on the micro branch, your home now, right, so the branch, the campus and the edge of the data center. And with the introduction of a common operating environment under the Aruba portfolio, which we call it Aruba CX OS, which is a true cloud-native operating system, we extended the switching into the edge of the data center. Think about the top of rack of a solution that sits somewhere. When you get to the data center, right, you need to have bigger capabilities in the core aggregation side, which we have. So think about an architecture in my building. So you have an access point. The access point is connected to a closet in that floor. It falls into the bottom of the building in a core aggregation switching. That core aggregation switching connects to your data center, right? We have all of that in the building. And then now, we have created a core aggregation that we connect to the data center. But once you get into the data center, you need now a massive amount of investments. And we felt at this time was not the right use of our capital because we already have momentum in the other segments of the market. But that's not technical limitation. It's just a matter of time and investment and where.

Okay. And then one other thing is that I remember when you bought Aruba, it was very mid-market focused, let's say. And I remember it was maybe an Investor Day in the New York Stock Exchange. I remember [indiscernible] you talking about the investments we were trying to make to push it more to the enterprise, and there was a [ quarter ] or 2 there where things kind of fell in between the chairs right on the mid-market business, and you had to kind of reinvest again. Where is from a customer standpoint the focus right now? Is it still mid-market? Or do you think you have a very good handle on the enterprise, large enterprises as well?

No, actually, we have momentum in 3 segments of the markets -- well, actually 4. Now enterprise because of the go-to-market that HPE brought to the table, right? And that's where we compete, generally speaking, with Cisco. And there, we compete with Cisco DNA. And there, we win more than our fair share, I will have you. And that's where the opportunity to expand share is. Second is in the commercial space. And we actually made few changes in our go-to-market, particularly in North America, and we see the momentum now in our favor. And then obviously, the mid-market. We generally don't play in SMB because the SMB -- and then again, the SMB is the M, the mid-market, and the S. But the S side, this is where you compete with people like Ubiquity, which is a pure channel model, not really a ton of innovation. It's a prepackaging and so forth. And on the M, we compete mostly with Meraki. And that's where the cloud solutions come in play. So the way I think about that, I think about the competition. But listen, we're getting all traction across the top 3 segments of the market. And then last but not least is the MSPs, which is the telco. And this is where we see significant traction across multiple geos where we actually, in many cases, with [ Y level ], the Aruba solution for the telco. And the telco sells their entire solution, whether it's fixed networks, WiFi, whatever it is, but underneath is Aruba. And a great example of that is what we have done with Telmex, AT&T, Verizon and as well as wet Telefonica in Europe.

Got it. That makes sense. All right. We got one question from the audience from Greg, and he's asking, excluding Silver Peak, you expect edge to increase 4% this year. How should we think about this growth rate versus the overall market? And perhaps maybe you can talk about the WiFi growth versus switching growth, which one is growing faster versus slower?

Yes. What we said is that organically, excluding Silver Peak, we're going to grow this business, and we're going to grow faster than the market. At this point in time, if you look at the previous quarter results, Greg, we did extremely well. Ittai just talked about our Q4 numbers. The problem is the delay factor, understand market share, which we believe we gain between 100 to 200 basis points of market share, that's the point because Cisco obviously had a performance that was not that great with the entire segment going down 16%. The problem is, I don't know how much was down, WiFi or switching, until we report the market share. But we believe we have -- we will gain both WiFi and switching market share. We expect equal contribution because when you go to WiFi 6, generally speaking, there is a need to upgrade your switching because you're doing power over Ethernet. And therefore, you need to operate the switching in the closet to be able to power these access points, which require now more power than the previous switching generation. So that's why we believe equal instantiation. And the WiFi 6 is getting significant traction. And as 5G comes in play, the 2 will coexist because of what I said earlier, right, that the use cases will demand certain characteristics and bandwidth and performance. But at the same time, you don't need persistent [ connectivity ] all the time [ to send ] that you pay enormous amount of money when you can have just an upload and then upload -- offload it to a WiFi.

Got it. Interesting. Maybe last one on this, your ProCurve business has been a long-standing, as you mentioned, business within HPE. I even remember from Carly's days where it was lumped into the corporate from a reporting standpoint. It just tells you how old I am. But help me think about who's pulling who right now, right? And I guess we're trying to get into the attach rate of switching and Aruba. How often is it Aruba pulls a ProCurve switch into the deal versus the other way around?

I think it's -- that was the thesis, Ittai, right? There were 2 components of when we acquired Aruba. One was our go-to-market, right? So Aruba was a very North America-centric company. And in Europe, we had a strong switching business. So in that case was the switching drag in WiFi in Europe. And in North America was the opposite. We had a very strong North America WiFi with Aruba and in this case, was attaching the switching portfolio. But what we learned in the second case is that we needed to have a common architecture because in that context, the ProCurve operating system was designed with different principles than the Aruba experience. And it took us a couple of years to bring it back to the same level. And now you have an integrated experience between WiFi and switching. That's why you see more wins like the Pentagon where customers says, I want the same experience I get on my WiFi. It's simple. I get to get the right policies. By the way, I can manage now heterogeneous environment because our Aruba ClearPass solution as a part of the Aruba Edge actually manages Cisco switches too. And so we use that, in many ways a Trojan horse, to basically modernize and transition everything to Aruba.

Got it. That makes sense. All right. Moving to HPC, which I find to be enigmatic type of a market, right? I mean it's -- not a lot of people know a lot about this space. I mean, obviously, it's very scientific and crazy professors walking around with [ wide closet ], trying to -- just trying to do something going on. Can you help us, first of all, understand the opportunity? How big really is this market? What's your share in it today? And what distinguishes your solutions versus others in the marketplace?

Well, I mean, when you look at the market sizing, inclusive of hardware, software and services is close to $40 billion, right? And we expect this market to grow between 7% and 9% on a CAGR basis. This is a space you need deep, deep expertise and deep, deep technology innovation. And this is a space we are the clear leader, right? Whatever its market share, I think the latest number or so was in excess of 35%. Whether it's the top 500 supercomputers where we are clearly the leader between the Cray and HP, we are more than probably 40 of them. And so when you think about that, what problem we're trying to solve? We're trying to solve the fact that the Moore's Law is coming to an end, the fact that these very data-intensive workloads need a lot of computational capabilities. And there is the mix match of compute and storage and what I call interconnect fabric, a high speed, right, because you're processing, in some cases, billions of transactions. And so when I think about the future of big data, simulation and modeling, AI and machine learning, in particular, deep learning models, you need massive systems. And so that's why we are entering what we call the exascale era, which basically says that we can process 18 zeros after comma. And that means in simple terms, when we deploy one of our exascale system, we will be able to process 1 billion billion, which is a 1 billion square, transactions per second. Now let's bring it to today's reality. If you have those capabilities in the COVID pandemic, right, for the research, it's amazing what we have done to find the vaccine in, what, in 9 months. It's because now they have access to these type of capabilities. We made available our HP systems, patents and talent to the White House and in other entities. But also, we work with specific entities like the University of Huntsville in Alabama, where now they were much more focused not in the vaccine but the therapeutics part of it. And now that we're able to find -- narrow down the number of proteins needed to find a cure. And so this is what we're going to see. But to give a sense, one of these exascale systems is between $600 million and $800 million every time we deploy it. It is -- if you put all the servers stack together, it will be taller than El Capitan Mountain in the Grand Canyon and it will be 2 basketball courts. But our strategy is not just to do that. It's to take these amazing foundation technologies and scale it and democratize it for the enterprise, which we're going to do over the next 2 to 3 years because whether it's silicon or whether it's cloud-native workloads for managing this very AI intensive workloads or whether it's the technology, we're going to make it available on the enterprise systems.

Got it. Interesting. And I guess, GreenLake, that's worked way into this as well, right?

Yes. That's why we introduced HPC-as-a-service offering last month.

Right, got it. So you -- earlier at the beginning of our discussion, you talked about $2.5 billion of business that will be recognized, I guess, over the next 18 months. But why don't you take a step back, what is the sales cycle of this? Like, I can't imagine it's shorter than selling a server or a piece of storage to an enterprise. I would have to think there's far more complexity into it. Definitely overthinking this. Help me think about this.

Well, we won -- we were awarded 5 out of the 6 exascale systems that are going to be built for the United States. This includes the Oak Ridge Laboratory, the Los Alamos, Lawrence Livermore, NERSC, which we're actually -- what we call a half exascale system actually as we speak. And it takes a long time. It takes a long time because you have to bring together a group of people on both sides to understand what problem we're trying to solve. And each of these labs are solving different type of problems. In the case of Lawrence Livermore, right, so we know that it's nuclear physics. It's more about nuclear weapons and all that. If you go to Oak Ridge, you're talking about life science. Those are workloads that require a different type of computational mix. And so the sales cycle is long. And then on the other side, Ittai, is that building these systems takes several months. But then the revenue only is recognized when we turn it on, which means I have to build it, I have to ship it, I have to install it, I have to turn it on, I have to run the workload and the customer had to accept it. And that's where you're going to have this lumpiness in this business. It is a factor of math. I mean last quarter, we delivered a 25% year-over-year growth, right, and 50% growth quarter-over-quarter. And there will be quarters would be not as high. But in aggregate, over the next 18 to 12 months, you're going to see significant growth because of these awarded business we have.

Got it. Interesting. And I guess, over time, as GreenLake work its way into this as well...

In the commercial space, we'll be more interested, right, because in the government space, they run the business on [ things ] that they need to spend.

Got it. So I guess that can move a little bit the peaks and the valleys. But that will take time [indiscernible]

Correct, yes. And remember, when we acquired Cray, it was the same thing as Aruba, right? So Cray was obviously the top of the supercomputer and mostly federal and 2/3 federal, 1/3 commercial. HPE, with our Apollo system and SGI was 2/3 commercial, 1/3 federal. Combine it, you get a little bit better smooth trajectory over time.

Got it. Got it. Okay. So maybe before we wrap it up [indiscernible] respect your time, make sure we do this some time. First of all, maybe could talk quickly on the cost optimization, if you can give us an update on your plans on the cost optimization side, where you stand today? And how do you think about reinvesting some of this versus letting it flow or using some of it for you to buybacks, dividends, things like that?

Yes. So first of all, we are on track to what we committed as a part of our Q2 last year restructure program, which I felt with Tarek, our CFO. And I learned this from prior instantiation, prior restructuring. This was not going to get any better anytime soon. We were a little bit criticized at the time to kind of jump into the changes and aggressive actions, but it paid off for us because I believe you got to be very aggressive when things are not going well in the market to position you well for when the market rebounds. And we believe we are well positioned for that. And so from that commitment, we are right on track to deliver that $1 billion or so in cost savings. And the vast majority will be in 2021 versus 2022. So that's how we communicated this. As I think about capital allocation, as you know, we use a very stringent capital allocation and we think about obviously return to shareholders. We are incredibly committed to our dividends, which obviously, I think is very attractive. As you know, the yield is pretty high because I believe the stock is very, very cheap and discounted at this point in time, understanding we need to show growth at the total company level. But we are unbelievably committed to that. And then we use a dynamic capital allocation for share buyback, what is the right return at the right time while we keep investing in the business because, obviously, it's important we keep investing in the business. From the M&A perspective, I believe I have done now 19 acquisitions since 2015. Those have been very strong for us in terms of our strategy, our pivot in the context for the last 45-minute discussion and very accretive to the company. I mean you think about Aruba as a great example, even Cray for that matter. So we continue to assess what type of M&A we like to do. And obviously, we always talk about the Aruba likes of the model that has a fair price but ultimately brings intellectual property and talent that accelerates our strategy. And as you know, Ittai, today, valuations are a little bit ridiculous in my mind. And timing is everything, as you and I discussed early in the call, right? But that's something we know we have to do. But ultimately, from the shareholder perspective, dividends, strong. We will assess buybacks based on the dynamic situation in the market, where is the right time and then obviously keep driving the growth that ultimately fuels the free cash flow, right? And from the free cash flow perspective, we are totally committed to return to that $1.7 billion, $1.8 billion in 2022, and we feel we are absolutely on the right track to do so.

Awesome. That's great, fantastic. Antonio, thank you very much for your time. This was great.

Thank you, and thanks, everyone, for joining today.

And Andrew, again, and the other HPE team, thank you very much. And everybody, thanks for listening. And if there's any follow-up, feel free to reach out to me or to Andrew. He'll be -- he and I will be more than happy to help out. Thank you very much, everyone, and have a great weekend.

Thank you.

Thank you.

Bye-bye.

Bye-bye.