Arista Networks, Inc. (ANET) Earnings Call Transcript & Summary

Great. Good afternoon, everyone. Thanks for joining us at the UBS Global TMT Conference. My name is David Vogt. I'm the enterprise, hardware and networking analyst here. And thanks again for joining in person [indiscernible]. And so with me today, we're excited to have John McCool, Arista's Chief Platform Officer. And in the room somewhere is Liz Stine from Investor Relations. Do we need to read it over? I think we're good or all good, perfect. So the first one, I have [indiscernible].

So why don't we just jump off. I think many of you in this room have probably met John or heard John present in a different forum. But I do know from e-mails that I've gotten over the last couple of weeks, they're definitely are familiar with sort of your role in your purview. So maybe I think it would help us if you could spend in a quick minute or 2 kind of discussing what falls under your sort of title, your role, your purview just [indiscernible] level set the conversation?

Sure. So Chief Platform Officer at Arista. I have responsibility for our new product hardware development as well as our manufacturing and supply chain and the latter half will be quite busy this year.

Great. So why don't we start there? Okay. So many of the meetings that we've had today and yesterday and even over the last week [indiscernible] is slow being a backseat more of the normal cyclicality and macro drivers of the business. But given sort of the stuff that you've had over the last 18 to 24 months, given your position with hyperscalers and enterprise, what you've seen by vertical end product? I know a lot of your products are effectively you can be used in multiple iterations and different customer verticals for sure. But what are you seeing today maybe relative to 6 months ago or 90 days ago in terms of supply chain situation?

In terms of supply chain situation. Yes. So I think over the course of COVID, the supply chain situation has evolved from factories closing, to get workers [indiscernible] or to control orders who focus around semiconductors and the semiconductor mismatch between supply and demand related to capacity, where it's particularly acute is in power semiconductors. So these tend to be on older processor nodes, which really lacks some investment coming into COVID. And then the demand for these products is very broad based, not just networking or even IT, there is consumer, et cetera. That's really been in the last half of the year in 2022 is a key constraint to go into power supplies. And in fact, they can be used across multiple of our products though not necessarily segmented by cloud or enterprise. They're very generic devices across the boards.

How does that get resolved? These are [indiscernible] nodes. There's no incentive for anyone to make any investments in this technology. I would imagine the same dynamics that have caused -- besides COVID, that have caused a lot of the shortages will persist as we come out of this economic cycle. So as a company, how do you plan for this? I know you've been in the broker markets buying product at elevated prices. And you've done some reconfiguration, I think on some of your products. But over the [indiscernible], how do you navigate this potential? I don't want to see a persistent challenge, but it's definitely going to be something that I think will probably crop up from time to time that maybe display disruption?

Yes. I think there's been many areas of supply chain during COVID, which showed us that there's an opportunity for more focus and attention on multi-sourcing or regional areas. We've gotten very clear about where things are sourced from, from our suppliers and making bets around that. We've extended purchase commitments largely to deal with these kind of situations as well. And I think that we're working very closely with the suppliers in the way that we haven't thought about.

Should we read anything [indiscernible] in the most recent period? Does that suggest that maybe supply has done a little bit better? You've been managing it a little bit more effectively.

I think as we went into 2022, we saw the challenges ahead and really we're very focused on placing those bets and now we're starting to see some of them played out. I wouldn't read really into that.

Perfect. And to your point, just maybe just a final point right now. If I look -- I think you said publicly and you've written about this, like 6,000 parts or 6,000 vendors that you work with. When you look at what's happening in other parts of maybe the tech industry with lockdowns and potential civil unrest, have you been thinking about maybe ways to further diversify your exposure in parts of world that may have potential trouble in the future? There's discussions about obviously new fabs being built in the U.S. I don't know it's not kind of core to what you view right now. But have you kind of thought about maybe diverse sort of your exposure geographically over the long term?

Yes, I think in terms of direct fulfillment, effectively, what we buy and make, we're very cognizant of where our factories are and what products are getting built in the various factories. As you go multiple levels down the supply chain, it gets more challenging. In fact, you start to see -- you may have 2 suppliers, but they're all [indiscernible] some resident or chemical that's coming from the same location. We're drilling down much further than we ever have in the past.

Got it. Okay. And so along those lines, maybe just turning to the big picture, just a ton of discussion and market demand durability by vertical, maybe just starting with your sort of bread and butter over the years, the hyperscalers, there's a lot of discussion that maybe workload growth is slowing to some degree, capital intensity kind of grows more or less in line, I guess, with workload migration is kind of the way we're thinking about it. I know you have reasonably good visibility in the next 3, 4 quarters. How do you think about your relationship, though, from a visibility perspective post COVID? I guess the question is you've been working, I know, very intimately with your customers, you're sharing road maps, sharing designs, does that change, going forward does it revert back?

I think there's 2 pieces. We've always had very good insight into technology direction, technology wins and co-design even before COVID. So directionally, where things were going. And I think that continues when this supply chain situation starts to resolve or lead times come in. With the extension of lead times, though, procurement and deployment decisions had to be extended, right? Does that go back to normal? I guess one thing I'd go back to the semiconductor side, what people say it's easing. This time for semiconductor is the worst case, the most difficult one used to be coming 4 weeks. Today, the worst case one is 70 weeks. I don't think we see that snapping back to those 24 weeks for a long time. And especially in the cloud segment, they're much more cognizant of the semiconductor lead times and planful around that. So I think we'll see some easing, but I'm not sure that brings snap back to that over a period.

I mean, obviously, it sounds like it's a reasonable assumption to assume that even if lead times come in from 70 weeks to 52 weeks or somewhere a little bit tighter, the pro-development work plus the lead time should give you reasonable visibility in hyperscalers...

Into some planning and direction for quarters -- it's tough to say.

Is a reasonable way to think about it?

I would say that there's a level of intimacy that's required to participate in the cloud business that needs an engineering level of engagement that will persist.

Got it. Okay. And then maybe on the enterprise side, a little bit of a different market, like different go-to sales motion. It's a new -- not newer, but it's not the original market that Arista sort of was known for. What do those conversations look like today from an enterprise perspective? Are the conversations changing in terms of where the decisions are being made? [indiscernible] a lot of companies that we've done calls with that it's basically a decision at this point. Has it reached that point in terms of your conversations with your customers? Or is it still on a project base by project base basis?

Yes. We're very involved in the project base by project base decisions. We've had a small historical presence in the data center 5 years ago, organically, we did routing [indiscernible] in data center. We built the campus use case, network visibility. So we've, in the last 5 years, expanded the number of use cases that we can go into an enterprise and truly present ourselves as an alternate end-to-end from the incumbent. So when we go in, we're very much into a new account, decisions had been made around the project. They're looking for another supplier. It's less about refresh activity, which could be pull off essentially because you're not going to refresh this year being right on the budget.

So it's not like you're going in and you have this existing campus footprint and there's a refresh opportunity. You're basically...

We're looking for those opportunities and yes, times have changed.

Right. Got it. So against that backdrop, does that mean from a -- I know [indiscernible] a lot of people. But from a company perspective, some of the larger competitors in your space talk about how they are fighting back to keep share in some of the verticals? And you've taken a lot of share in some key verticals. But given that you're more of a rip replaced story, let's say, enterprise, does that mean in a sort of a challenging economic climate where project base by project base is decided that theoretically should be able to continue to take market share?

That's our objective to become a share gainer. We've given the sales teams more opportunities to go after projects 4 years ago, they went for the data center refresh and if that wasn't up for refresh, you talked to another account. Now we'll have a campus opportunity, network visibility story, maybe even [indiscernible] away to the security competency. So there's many more touch points in an enterprise that we can play against.

I mean where are you seeing more of these touch points being successful? Campus has been successful [indiscernible] enterprise data center very successful?

The core pain point that we touch is my network is hard to operate. It's difficult. I can't upgrade security. It's difficult for me to patch. It's just difficult. That's where our team is focused, and that may present itself a campus opportunity or a data center opportunity, and we're fairly agnostic about that. Just looking for a way to continue to either grow wallet share within the company we're in, we're bringing in a new customer.

That was one of my next questions. You mentioned wallet share. Is there a way to look at sort of your relationships with your customers from an attach point perspective? So for example, this company historically was an enterprise data center customer. Now it's a campus and enterprise data center customer. Do you think about your business that way? And does that decrease or increase the stickiness and decrease some of the cost of maintaining and/or maintenance of that account and drive better profitability?

Absolutely. I know our sales team look at opportunities and wallet share within the accounts and big bets that they place. I don't think we can correlate with the stickiness. We're just trying to gain share either by grabbing another customer or increasing the wallet share. Some of the large customers still have significant footprints of switches and routers that are an opportunity for that to move forward.

And then maybe just finally on enterprise. It certainly sounded as if maybe at your Analyst Day, there was a little bit of hesitation in terms of maybe demand for campus on the enterprise side. Is that a reflection of maybe a lot of the initial wins were legacy sort of touch points for Arista customers that knew the product well, knew what you bring to the table and now maybe the next leg of growth is a little bit [indiscernible] tougher, but maybe it's a little bit slower in terms of onboarding effect?

Actually, we feel like we were pretty par on target with the demand. It was the ability of the supply chain to really drive that revenue perspective.

Got it. So does that mean backlog has been building in [indiscernible]?

Interest is -- that's about backlog, but interest in general and I think opportunity we see pretty strong.

Got it. Okay. Sticking to campus, over the last couple of years, it's a small market for you, but some of the bigger competitors have been aggressively raising price. Has that been an umbrella that you can price under? Or are you using that as a leverage point in addition to your technological position to gain share or?

Yes, I think we get questions about either pricing or supply as a leverage point. I mean, we're very focused on the operational simplicity as the key win and the sticky win in the long term. So not so much on the cost of ownership, right, operational complexity, reducing cost of service calls, automation that all those components.

So the total cost of ownership matters than sort of that initial price. So conversely, if pricing kind of softens as we go through '23 and, let's say, calendar '24, plus of an issue for you because you're not sell or competing directly on price?

Yes. We're focused as a solution that's better.

Got it. Okay. And so when you think about the 800-pound gorilla in the room in campus, for data that I've seen, 8, 10 quarters really haven't seen much in the way of growth in that market. Now there's some debate on growth because of backlog drawdown and other sort of metrics, but absent the backlog drawdown, it seems that their position is relatively stable and different share as the market grows. Are you -- how do you think about their competitive position vis-a-vis you in terms of you're small now, but as you get bigger, is there a risk that this competitor becomes a little bit more impressive right? We've seen it in the past where companies have gotten a couple of points of market share in campus with 3 points of market share, and maybe Cisco has been a little bit more aggressive on pricing with those customers to retain and/or win back. How do you think about that over the long term?

I know you're selling on solutions, but I think we've seen -- I mean we're up against [indiscernible] extremely competitive and very focused on maintaining their share. We see the data center piece you see that new cloud where they at one point had dominant share. So I think we're used to that dynamic. I don't think we would see or expect any changes from the very competitive position that we've seen in the past.

And you mentioned cloud, have you seen them aggressive on enterprise data center recently? It looks like they're losing share, at least in the data that we've checked looked a relatively difficult quarter. You just had an incredibly strong quarter. I just love to kind of get your perspective.

Again, I think it's more solution-based and people looking at sort of the value of having cloud fusion and being able to automate those environments and not having a separate network for campus, a separate network from routing and a separate network data center that requires 3 different operating systems. That's been the selling point.

Okay. And then sticking with enterprise data center, obviously, it's been incredibly strong. We have this conversation with investors all the time in terms of what's the long-term investment cycle for this market? For example, I'll use UBS as an example, 1/3 of our assets are [indiscernible] and 1/3 is our own data center and 1/3 in the cloud. That's migrating slightly more towards probably public cloud, [indiscernible] we've been public about that. In your experience and from your view, how do you see the enterprise data center market evolving vis-a-vis, let's say, public cloud? Is this a market that can grow reasonably faster than GDP? Or can it go faster public cloud becomes significantly more expensive, we're hearing a lot of pushback of public cloud is maybe not the right solution for everyone anymore, [indiscernible] I'd just love to get your kind of perspective.

No. I think that that's an interesting point. I do think people are measuring now the expense of the public cloud, but there's still an advantage of ease of use and deployment and then thinking about [indiscernible] that you've also talked to customers who think about strategic workload that needs to run on-prem. And I guess, as your 1/3 might fit into that category, which this can never go because this has to be on site for data domain versus whatever you have. So it's tough to say how that changed, but I think we're at a point where people are thinking in terms of -- they've already rationalized their portfolio, mail into the cloud, right? So I think we're where people are thinking about the balance of their investments going forward.

And do you think the big delineating factors not [indiscernible] is cost? And the reason I ask is those companies that we talk to understand the value of the cloud. But what they didn't understand maybe years ago [indiscernible] how costs continue to escalate and now the bigger part of your [indiscernible], you're running at P&L. And so when we have these conversations with some Fortune 500 companies is that I don't want to move more to Azure. I don't want to move more to AWS because I have a P&L to hit. So does that maybe lend some ability to maybe enterprise data center or some of these as-a-service offerings that some hardware companies offer, maybe price a little bit more aggressively, offset the growth?

It's hard to say just from a cost of deployment, you also have geographical diversity. You have the ease of moving and migrating workload. So if you zoom back, I think it's less clear about those cost benefit trade-offs. And it's tough to say.

Right. Okay. So just I'm trying to figure out what [indiscernible] is the long-term investment cycle for public versus enterprise? I know you're well positioned on both sides of the ledger, but we hear from a lot of investors who are very public cloud centric...

Right, it's one or the other.

This is a 20-plus-percent capital entity business enterprise is a much lower business.

And we've tried to -- if we look at kind of the solutions, especially as we go forward, focus on solutioning in the enterprise or hybrid environment, multi-cloud, on-prem, multiple off-prem clouds and trying to make that operational model simple to do all of that.

But you're agnostic. I guess maybe just turning to agnostic. And I know financials that aren't your focused not going to ask your models or anything of that nature. But when we look at this -- when we look at sort of the different verticals, typically, hyperscalers are a little bit more aggressive on price of gross margin given the nature of the relationship, enterprise a little bit better. But when you think about different growth rates going forward, to your point, you're agnostic about technology platforms, but are you agnostic long term, where you want your -- what your business will look like -- is there -- maybe not financial is not the right way to look at it. Is there a risk being overly dependent upon [indiscernible] because they tend to be more cyclical throughout enterprise? [indiscernible] then there's a digestion period.

I think you have to look at service provider, cloud enterprise as they are separate markets, and they have their own care about their own solutions and architectures that you need to fulfill. I think that on the cloud side, there's been just phenomenal growth over multiple years. We do think there's an element of cyclicality potentially that we've never seen a cycle. We've seen some flatness in 2019, 2020, but they do tend to follow the technology roles. So we would -- but not being in cloud would be incredibly hard as an infrastructure provider today, a networking provider because you get scale, they're really driving the architectures effectively. I think that when Arista started, the premise was, everyone is going to build my cloud. And now we're seeing enterprise customers building leaf and spine architectures, driving a level of automation that you wouldn't have even thought possible 5 to 7 years ago. So they're clearly the thought leaders today around networking.

Since you just mentioned leaf and spine and digestion period, how are you thinking about your planning in terms of your road map from a technology perspective? How do you think about the risk of digestion, right? It's a cyclical business you just mentioned. Do we -- Arista continue its investment road map through the cycle? So if you think, let's say, '23, '24, '25, '26 [indiscernible] I guess you have a 5-year plan out there? Maybe '25, does it change on let's say there's a shortfall in demand next year. Does it change how you think about where you're investing, what your research dollar priorities are going forward?

Yes. I think if you're here in the infrastructure business and networking, the dynamics follow Moore's Law, right? Things get better, faster for similar costs. And we made a bet on merchant silicon because we saw the economies of scale being in the silicon business. So being able to go faster, move that model to be pretty efficient in delivering platform, so we expect the continuation of that going toward. Next-generation platforms are continuing -- it's just what you do business in this market.

And since you mentioned merchant silicon, obviously, that's been an advantage, I think, this year, having a very tight relationship with Broadcom. What are you seeing in terms of new entrants in silicon? And is it something that would interest you in your merchant silicon company effectively, right, all you use? Is there anything down the road that maybe changed that decision design new products, new platforms?

We stepped back from that. We're even happier with that decision, and I think we were at the company inception is just incredibly expensive to build silicon, many of the basic building blocks are used in other markets besides networking, the substrate design, the packaging, SerDes, all the technologies that go into it. And just as a number, people talk about out of the data center of the networking component being 10% to 15%. So storage and compute and all consumer electronics really driving these investments that benefit networking, it's tough to go the other way around. It's just not volume.

So I mean, so does that -- do you think it gives you a competitive advantage vis-a-vis I know one of your big competitors went down [indiscernible] they went down their own silicon route to some degree. Does that give you sort of, in your view, a cost advantage going forward, flexibility or [indiscernible]?

I think flexibility and consistency a road map, it definitely has supported us. And our competitor, in some ways, there's a different positioning of that silicon, but it's always been focused on captive silicon internally even back since its inception.

Got it. Back at the Analyst Day, I think there was some discussion about expanding your addressable markets. And one of them was the AI spine. Can you -- that was a little bit new for us. Next time when you guys talk about it, can you kind of talk about what that market entails and how you think after share in that particular new addressable market?

Yes, I was thinking more about that. I know we would talk about that. And it's been a long time since something that wasn't a computer has connected to the network, right? We had servers connecting to desktops and then we had laptops, probably the phones, the first thing that was not a computer that got connected to the network and drove a new market, right? So in some ways, we're now connecting dedicated AI devices to the network. Those clusters are built out as a separate system. So most reminds us of high-performance computing, where you had a cluster that was dedicated to computing something, it then connected to the broader network. So it's new TAM, it's new opportunity. It's probably not very well sized by the analysts. But we see it happening in the cloud networks today that we're invested in. And we believe also this is back to where the clouds are important that, that will start to take place in other segments like the enterprise.

I know it's early days, but what that architecture look like? Obviously, this is new for us. So how would you be positioned from a technology perspective?

One thing we see that's been important in that market is the intelligence seems to scale with the size of development connected to the network. So having very large mavericks that are single tier, if you connect a lot of devices is important. The second thing is the very expensive devices with memory. So any stalling, you're not utilizing perhaps that. So keeping the AI delay stead is a very important metric. And typically, people didn't believe you could do that with Ethernet. So our [indiscernible] suggestion control and some of the others that we put into our products, we believe that we can make Ethernet work very effectively at 400 gig. And it also drives just higher speeds into those devices themselves.

And have you seen any sort of use cases right now or trials that give you encouragement that there's a real commercial opportunity in the intermediate term?

With the AI cluster piece that you talked about, we talked about a number of use cases, probably the most near-term realizable.

How would you frame that 2023? And would it have sort of a similar -- I know I'm not getting any specific numbers, but when you think about everything that Arista has done, it's all at similar kind of profitability profile. Is that how you would think about these types of businesses, where...

Like our typical network business.

Typical network business, yes. Got it. Okay. And then obviously, sizing, you said it's impossible early days analysts can give this. Is it a market that you think has a handful of winners, 4 or 5 players that are attracting the economic rents from this market kind of like where we're today in some markets where if I look at hyperscaler type of data centers, it's only a handful of players that really compete at scale? Or is this going to be -- you're going to see a ton of smaller players that can maybe proliferate this market because it's diverse enough, right, is going to be, I think?

That's a really good question. I think there'll be some overlay in systems and data centers today that people have large data centers, including enterprise or something like that. Will there be specialty AI providers? I don't know. But it will be an interesting [indiscernible] it's such an early [indiscernible] how it's going to play.

Yes, the reason why I ask is if you find these inflection points in these markets, where maybe it's a winner take most in some cases that it's a hyper-growth market with a fairly big TAM. You have a new entrant that comes out of [indiscernible] or an existing player that has a much smaller presence that you maybe didn't quite fully comprehend the scale and scope of what they bring to the table and change the step function the dynamic of their business, and that's what we now are trying to figure out, helpful. Can we -- and this may be a little bit different. SD-WAN. For years, it sounded as if you were to be blunt, most interested in the SD-WAN market. I think maybe if that's a fair characterization, like, this is great. You talked a lot about it at the Analyst Day. What's changed? What's changed in the company's perspective? How are you thinking about the opportunity, whether it's a go-to-market or technology perspective, kind of expand on that?

So maybe I'll just step back and talk about the application a little bit. And of all things in the network, I think this area has gotten the most names over time. So how will I connect my enterprise to [indiscernible] at the edge, right. It could be -- I'm running a bunch of supermarkets. I have to connect all the stores together or I'm an insurance company, have a number of branches, et cetera, et cetera, et cetera. And back in the '90s, it was about integrated services. I had to have both video, security, all in one box, and that was the technology [indiscernible], right? Then we went into WAN optimization because we needed a ton of bandwidth and T1 was slow and you had to do something cool, right? And then came SD-WAN where a lot of features and functionality about running over multiple networks and getting [indiscernible] doing that kind of thing, right, saving some costs. So we would have been the 15th [indiscernible]. We also built up some core technologies that connecting the edge is important to have, like or just our routing stack in the enterprise and POE devices and edge devices are already part of the portfolio. So now can we do something cool that would differentiate us. And I think the teams got some really good ideas about a solution that would offer some different technologies and applications to sell that edge connection that we're really excited. So it's a combination of having the technology as a basis that we feel fundamentally very good about already and seeing some inflection point in the market that we can take advantage of.

And so to your point though, there's 12, 15 and 16 players in traditional SD-WAN, when you have a conversation with a customer, a potential customer, what solution or what pain point then does this solution directly address that's maybe not being addressed currently in place?

Yes. So I mean, consistent with Arista, it's always been in the enterprise network, building on those networks in a more consistent fashion with EOS, a single operating image software and the management stack runs across all. I don't think you'd see any surprises from us in this area. It will be consistent with everything else we've done. If you're running the campus, you'll be able to run this, not a different operating system, very consistent.

Understood. And then we talked about over time, campus was a new opportunity. This new SD-WAN approach could be a new opportunity. AI spine can be new opportunity. What else out there are you looking at today from a road map perspective that you think you don't have the core capabilities or competency today. But if I think out 3, 5, 10 meters, the market is moving in this direction. And given our position, we can be there, and we need to be there.

Yes. I think the one thing that we've talked about is we're excited about what we've done in the management stack with CloudVision. We announced CloudVision as a service. And then we've announced the Data Lake. So this is an extension of our management architecture, being able to take multiple information of the streams from all the devices and have a large data repository that we can federate with other folks either trying to consume that data or add information into that database. So that's a technology area that it's something that we're very focused on as a team. And with our Awake acquisition, being able to make good security decisions around the information that's going through the networks, not endpoint security, but what's flowing through the pipes? Is it right all and making some sense of that is very interesting to us to see that directionally.

So does that mean that you need to pursue incremental security asset purchases, acquisitions? How does the week has a dedicated sort of -- it's always a dedicated problem, right? It's not a complete security solution per se, doesn't have endpoint, right? But from a security perspective, where are you today in terms of what you need to the point where you want to be at?

I would say -- and just given the TAM that we laid out at the Analyst Day is very sizable, right? So how do we get more of that? And what functionality strength is to differentiate and bring that above? So Awake wasn't really about being a security player about getting the most insight from my network that we're in a unique position to be able to extract that, right? So I don't think we think we have to go into other tangential market segments or different buying centers and really focused on that network buying center, but differentiating the product and extending it like we've done with these other use cases. 5 to 10 years, maybe there'll be other use cases that we see where the network can be adapted to connect. I don't think it's 5 years ago, we would have dreamed up the AI use case kind of knew what was out there, maybe that would happen. I think people are talking about it, but at the sale that is starting to [indiscernible].

Obviously, you guys did very focus on smaller technical as your capability. Can you walk through how you think about adding on new technology capability? You just talked about all these different use cases, [indiscernible] when you look at it outside or you look at a technology that drives decision or [indiscernible] and obviously, there's some economic financial consideration. Is there anything that maybe would create sort of a deviation from that? It was a larger asset. I mean everything to date in the last several years has been...

Sure, acknowledge...

Very small sitting in...

It's just difficult to find something that would be complementary, not dilutive that we pick model. There has been such a thing. I don't think we would be necessarily against it that's difficult.

And can I ask maybe in a different way? If I think back in 2020, 2021, it seems like there was a plethora of small deals done across the industry, not just you, but Cisco and Juniper, people are fairly aggressive, and it seems at least from the data that we track moderated to some degree. Is that a reflection of the economic climate, maybe private companies have unrealistic expectations in terms of what they want? Or is it a question of that maybe there's just no assets that make sense for your use case or [indiscernible] use case? It seems like the M&A market has been much more subdued piece of the envelope...

It's tough for me to say outside risk I mean we've been very focused on the assets we've acquired, how we use the cultural fit is important. And if you also think about our value proposition around EOS and the software stack has to be consistent with that. So they won't necessarily switching and routing. We've been able to go [indiscernible] different operating system is not really connected to the network, but what we did with wireless, big switch, we already had visibility into the network, but being able to make that more usable and consumable was an important thing to add on to the solution.

Remind me how much big switch sold to customers now?

They had kind of 2 aspects, but I had one that was more of a fabric base kind of software-defined networking, but they had pivoted to more of a network visibility [indiscernible] and we've built that into our operating system to extend that to kind of switch [indiscernible]. Yes, it's a full solution for network visibility. So it's a full solution with an ELS. But ELS by the extended use by third-party, but it works [indiscernible].

Understood. I mean, do you have much off box of Arista software sales?

We do have some, but it's not the phrase.

So basically, you get the full complement of solutions and you get the full functionality in the best use case, [indiscernible] customers to risk the software it makes it a competitive advantage. And then maybe one final point along those points, we've heard from a variety of consumers that, that sort of solution has helped maybe swing pendulum back away from White Box to an Arista branded vendor in the data center. Is there any truth in that are you seeing that sort of customer decision?

I think, again, you can kind of parse this by cloud and enterprise. On the enterprise side, there really isn't a viable third-party operating system that makes it consumable.

On the hyperscalers?

On the hyperscalers, I think, a component of this also in the supply chain and being aggressively invest has helped in this environment on the OEM side.

So it sounds like you don't think let's say it like not getting well. I want to just talk about specific customers, but that your solution is not a good reason why maybe it's been a little bit better than, say, like Facebook's own operating system vis-a-vis with White Box.

I would say the situation on the cloud side is pretty static. I think that the people who have chosen to use White Box were vertically integrated still remains some that are 100%, some are mixed changes.

But there's still an opportunity for you to take share in some of these hyperscalers predominantly buybacks over the years.

I don't know, we continue to focus on White Box as a competitor.

Got it. All right. I think we're out of time. John, thank you for your time.

Thanks for the time.

Thank you for joining. Liz, thank you for joining. And if anyone has any questions, please reach out, and we're happy to help. Thank you.