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

Great. Thanks, everybody, for joining us this afternoon. I said it in the last session, I'm going to say it again this session whoever joined me for the whole day, we kept the best companies to the end. It's a strategy, by the way, to attract the investors to stay to the end. So I'm revealing my secrets here. But I'm very pleased to host Arista for the next session. And we have Ita Brennan, CFO; and we have Charles Yager, Director of Product and Investor's Advocacy. I prepared a list of questions, as always. You do see me now with a new headset. I'm a smooth operator because of voice quality issues, but at least you hear me well. If the visuals are not great, at least you hear me well now.

So with no further ado, I'm going to start with the first question, which is kind of a high-level question. Ita, last year, all my questions to you were about slowdown in cloud spending. This year, all my questions are about the increase in cloud spending. So can you talk about the cycles in cloud? And give us also the historical perspective, kind of what -- and it's your experience, right? It's not about cloud spending in general, but rather your experience with cloud customers. What are the drivers? What's happening? What's the outlook, et cetera?

Yes. No, for sure. If you think about cloud over, I would say, pre back half of 2019, I mean, cloud had been just a continuously growing, expanding piece of the networking space, right? And you had these new businesses, you had expansion into new geographies, expansion into new use cases, et cetera. So I think just generically, if you were to go back and look at the numbers, you'll see that cloud with spending and continue to grow spending very aggressively through that time period. And as a result, for Arista, we also benefited from that, right, with our combination of some very strong spending in 2017/'18 around this 100-gig cycle and the switch routing convergence that we saw in that time frame. And then I think for the first time, both at an industry level and again, as a result for us, you saw some falloff in that spending in Q4 of '19 and then continuing kind of at a more muted level into 2020. And that obviously impacted our business, but I think with hindsight, when you look at the industry numbers, it was broader than that. It was very much a digestion of their spending. So we were -- you think about how has our perspective changed around cloud over the last couple of quarters. I think we had -- we felt that business had stabilized. We were starting to see some recovery in spending. We have some new products and new product cycles. So we were in qualification and some pilot builds around those new products. So you've seen kind of a gradual improvement in confidence. We've kind of gone from being stable to feeling like it would contribute to now you're seeing some healthy momentum again in that business. And that's somewhat product cycle driven. I think it's driven by the strength in their businesses through COVID all kind of aligning to what we can see now as being a healthier spend cycle for them. And I think we're very happy with kind of our position as we've come through just from a product and from a new technology perspective as well, right? I think it does leave us thinking that this piece of the business will always, in our minds, be a growth over time piece of the business. Maybe it's a little bit more cyclical going forward than what we had seen prior to that Q4 '19. I think that's something we have to kind of acknowledge and figure out how the rest of the business kind of builds in around that. But it is great to see kind of some recovery in that part of the business and that we've -- the team has done a really nice job of executing against the new products for them as well.

Got it. Has the visibility improved with cloud? Meaning if you go back, and again, I'm not asking just about the quarter or the near term, I'm asking also historically. When you go back to the kind of visibility you had 2 years ago and then last year and now this year, where do you stand?

Yes. I mean we've always had this idea of 1 to 2 quarters of visibility, maybe a quarter of kind of order level visibility and then a couple of quarters of project-level activity-based visibility. I think that's improved right now, I mean, honestly, partly because of some of the supply issues that the industry is facing. And if you're dealing with a 52-week lead time on chips, maybe even longer, then you have to try to get some more visibility on the demand side. And these customers have been trying to provide that, right? It's not that they're -- it's, perfect, discretely plotted, but they are giving direction in terms of what they think they're going to do and what needs to be deployed in kind of roughly what time frames, right? So I think we have -- we've benefited, if you like, from the supply environment just in terms of being able to get that a little bit longer range view, and they understand that we need that in order to be able to respond. So I think that has improved in terms of just what they're willing to share so that we can react to what they need for deployments.

So that leads me to 2 questions. One is, in times of constraints, component constraints or any constraints in the industry, we see a phenomenon called double ordering, where companies order more than what they need just because they can't get everything. How do you manage it? How do you make -- it's natural. If someone they know -- if a client thinks they can only get 80%, they'll order 1.2x what they need and they'll get 80%, that will take them to what they need, right? So how do you manage it? And what do you think is happening in the industry now about this topic or around the this topic?

Yes. I mean I think for us, our approach to this has always been -- we've had not quite -- maybe not quite to this extent, but we've had similar supply constraints in the past. And I think our approach is always to focus on deployment, right? Worry less maybe about just bookings and demand signals and worry more about deployment planning with the customer. Because when you do that, I think you get a truer view of their priorities and you plan according to how you're going to actually deploy products. I think that's a healthier way to think about it. We're fortunate that we have some large customers who make up a reasonable piece of the business. So if you spend time with them, adopting that approach, providing some visibility to what you're doing to satisfy their demand on the supply side and then I think you can get to a truer view of what really the priorities are and what actually needs to be deployed and when. I think that's how we think about it and what we use to drive our thinking around the business.

And the supply chain issues, when do you think you're going to see an easing of the situation?

Yes. I mean it's certainly tight, and we think it extends well into next year. Beyond that, we'll have to see. I mean obviously, capacity will get added to the system, but it's going to take some time.

And how do you manage it? What are the areas of tightness?

Yes. I mean we're focused obviously on the key components. And I think we were fortunate that we came into 2020 with probably the best inventory, lowest inventory position metrics from a from a finance perspective. But unfortunately, when you ran into COVID, that was a problem, right? So we had started to kind of take some more aggressive positions around supply and some key components that we buffer in response to COVID, which I think has helped us a little bit as we've gone through this cycle. And we are going to continue to buffer those key components. We're at the beginnings of product cycles, which helps, right, because then you've got a longer window to consume those, right? So we are key components. We will buffer and we'll be -- for new products, we'll be more aggressive, and we'll increase buffers there. Then there's just the stuff that's going to show up, right, in the supply chain. It's commodities, it's smaller components that somebody's supply chain decommits to them, they decommit to us, you have to try to rationalize that. So then there's just a lot of blocking and tackling around the other pieces of the supply chain and the surprises. And you still got COVID impacting manufacturing and manufacturing capacity in different places, and you have to manage those. And I think the team has -- we know how to do that now, right? There's a process around that now, but you still have to respond to kind of these things as and when they happen and just keep driving, keep resolving them one at a time, right? So there is a fair amount of focus on just the execution and the response to things as they come up in addition to just buffering the key components.

I want to go back and maybe last question for this topic, it's about the cloud vertical. In the past, we published -- and I don't know even how much you disclose the names or not, but there are 2 big customers, right? You have Microsoft and there's Facebook. I want to talk about the other customers, not about Microsoft and Facebook because I think we fairly understand what they do in kind of the ups and downs in their cycles. Can you talk about trends with the cloud in general outside of these 2 customers? Where do you see strength? What about Tier 2 clouds? What is driving the demand over there?

Yes. I mean I think when you think about the other cloud tightened pieces of the business, I mean, we've obviously have -- we've talked, I think, fairly openly about having 2 of the other 3, if you like, in that bucket. And those businesses have been growing nicely. They grew well last year. It's on a size that's much smaller obviously than the 2 that have been greater than 10% customers. But I think we continue to execute well with both of them. We're well positioned with both of them. And then it just comes down to what's their strategy around how much do they spend on their own infrastructure and then how much do they spend kind of internally versus maybe with branded and other vendors, right? I think that's something that will evolve over time, whether there's greater investment in on-premise data centers or how do they consider the kind of make-versus-buy decisions. But that's really what drives, I think, significant progress with those customers. Other than that, we've been executing very nicely, and it's been growing kind of at a reasonable rate over time.

Got it. I want to talk about -- switch gears to enterprise data centers. What are the trends with the enterprise data centers? First of all, have we seen a decline last year with COVID? And are we seeing a recovery? Is this the cycle that is happening? Or can you talk about the big drivers with the enterprise DC market?

Yes. So for us, it's been more -- I think it's a share gain, new logo, new win type environment with the enterprise, right? It's -- we came through COVID, probably Q1 '20 was where we saw some disruption while everybody tried to figure out exactly what was happening. But then it came back reasonably well for the rest of the year. So we didn't necessarily see a dip in the discerning the data center enterprise piece of the business, right? We saw continued kind of wins and continued kind of growth in that business through the rest of the year. So I don't know that we see this kind of reduction and then kind of rebound. It's been more consistent, right? And I think that's just because for us, it's more about winning new opportunities and new logos that it is necessary about some macro trends in the business. And we're targeting these larger enterprise customers who tended, I think, to come through that and be more focused on kind of investing through that period than maybe some of the smaller enterprise customers that were more impacted by COVID.

And do you have the same kind of visibility or the same kind of optimism for the near term with enterprise as you have with cloud?

I think enterprise has been growing very nicely for us over the last couple of years, and I think we believe we can continue to drive that, right? I think there's a good pipeline of opportunities there, and the team is doing a nice job of executing against that. So I think that's -- cloud, I think, was more significant to us because we're coming off kind of a period of uncertainty, if you like. I think the enterprise, it's more just kind of continuing to add some sales resources and have those resources be deployed and execute against kind of that really share gains in that piece of the business.

Got it. When I ask you questions about 400-gig before you or Anshul or anyone in the company, the answer was keep your expectations low. It's going to start slow. It's going to ramp gradually. 1.5 years past, 1 year past, what's the experience now with 400-gig?

Well, I think that's still somewhat true. I think I like the way Anshul kind of talks about this as it's 100, 200, 400, right? It's not -- trying to separate those based on port speeds at this point is hard, right, because it's all leveraging the same kind of underlying silicon technology. And it's really -- you need to look at all 3 kind of combined, right? So I think that's what we're seeing. Customers are at different phases depending just on their architecture and their roadmap. I think 400-gig proper is still a DCI -- primarily a DCI use case starting to drive some dollars kind of in the second half of the year. I think that's very consistent with what we've been saying all along. And then you're seeing deployments of 200-gig for certain customers and then continued 100-gig deployments as well just across the customer set.

Got it. One of the opportunities you highlighted in the past was campus switching. And I want to start from the basics. It's a big opportunity. How do you address the market, meaning, what is the innovation or what is the disruption that you bring to the market? Because in data center switching, the disruption was very clear and obvious and didn't have a response at the beginning from the big guys.

Charles, would you like to take a shot at the differentiation disruption?

Yes. I think that cloud architecture and technology hit the data center first, and it was really all about scaling the bandwidth and reducing the operational costs. And so we learned a lot by feeding this technology to the cloud titans, and we put a lot of that technology into CloudVision, which is a management automation tool to manage it. And so that actually worked quite well in the enterprise data centers, and they took advantage of that technology. But the cloud technology is not really restricted to the data center, it expands into the campus. And so you want automated campus deployments. You want to be able to lower the operational expense in the campus. And the fact that it's a single operating system and a single management tool for your data center, for your campus and your wide area network is a fundamental principle of cloud technology. And trying to just simplify that and unify that is really something that hasn't happened in the campus yet.

So Charles, I want to go back to the basics because it's very clear where the need was in the data center environment. In the cloud -- in the campus environment, it's not even clear if there is a need, meaning the biggest driver -- bandwidth is not a driver in a campus environment because there is -- in a sense that there is enough bandwidth, there is more bandwidth than you can process with each computer. It's especially in the campus where a lot of the processing is done in the cloud. So bandwidth is less of a driver, and it's about maybe workforce growth, hiring of people, which is not a big driver. So I'm starting from the basic question. What are the use cases where your innovation is needed, where you can actually disrupt the market? What are the entry points for you to the market? Through what angle are you entering the campus market?

Yes. I think that lowering the operational cost will never go out of fashion.

That's true.

It is a cost that companies incur. Improving the reliability through better management tools will never go out of fashion. So that's what the cloud architecture has done. I mean the cloud titans really innovated this architecture. It's not limited to the data center, it applies to the campus. But you can even take the campus further by saying with this new world of COVID, people are going to be very mobile. And some people may go to work part of the week and they work at home part of the week. And so you want to be able to identify those people and say, "Are you in the engineering team? Are you in the finance team?" And then you have access control over that. Some person could be sitting in the same conference room as you and yet have different access to different applications. And so CloudVision can help you achieve that in an automated way and also being proactive about avoiding any problems. I mean, let's face it, there can be problems in a campus network, and you want to avoid those problems. So just making the IT administrator's job simpler and easier through automation and through the management tool, CloudVision, and having that tightly integrated, the campus and the data center, with one management tool simplifies the overall process.

Maybe thinking about it differently, what is the profile of customers? I mean campus is a big market. It's a giant market. It's almost about $10 billion, give or take, right, campus switching. What are the -- what's your target market? Within that big market, do you address the entire market? Or do you address submarkets of it? And I'm not looking for the number, I'm more looking for the profile of target customers.

Yes. I mean I think the initial target market for us is the large enterprise, it's the larger enterprise. And that's for a couple of reasons. I mean, obviously, that's where we play in the data center. In a lot of cases, that's where we have kind of the established familiarity, use cases, et cetera, already. And it's also a part of the market where there's been a dominant player, like a significantly dominant player. There really hasn't been an alternative for customers that customers were willing to adopt in that part of the market, right? So I think there's a real need for a competitor that is accepted as having the same high level of quality, capabilities, reliability, known trusted vendor that can compete in that part of the market. So that's -- initially, that's the target, is to go after the larger enterprise piece of the campus. That's where the connectivity from a customer perspective and from a product perspective between data center and campus can be more important, and it's an area where we think there's a real opportunity and a real need on the customer side to have a viable, high-quality alternative vendor.

Got it. Is your -- speak about the completeness of your product. And I'm not talking about campus switching, I'm talking about your campus offering. Speak about the completeness of your product, meaning you also have a WiFi play. You have a campus switch play. Talk about the product portfolio in total. And are there parts that are missing? What are the parts of this portfolio that are doing well versus others? I just want to understand the holistic approach of selling to the campus environment.

Yes. I mean I think we believe now we have all of the elements that make a compelling campus solution, again, targeting that kind of large enterprise customer. So all the way through from campus aggregation, the [ seat ] switches, WiFi, again, managing it all centrally with CloudVision, which is a very -- which is an important part of that. I mean obviously, we acquired Mojo, integrating Mojo into CloudVision have been able to kind of drive that from CloudVision was an important part of that story. We're going to continue to add features and capabilities like the macro segmentation that Ken talked about a while back, things that customers are going to bring that are important in terms of how they evolve this -- the new campus, if you like, and helping to manage some of those channels to provide more visibility. Awake can be a part of that as well where you provide better visibility to what's happening in those campus environments, where you've got a lot of different entry points and a lot of different tools functioning in your network and being able to observe kind of what's happening in your network and develop some expectations for what's things that are happening that shouldn't be happening in your network, if you like. So the combination of all of that, I think, gives us a good end-to-end solution. But we'll still partner with -- partners for other pieces, right, whether it's with security partners for some of the perimeter, VPN-type partnerships, so there's other partnerships that we will have around the edge of that. But for the -- what we consider to be the -- our job, which is the network end-to-end, including the WiFi and providing that visibility and managing -- helping to manage that footprint, I think we have -- we believe we have a good suite of products to address that today.

Got it. One last aspect of campus switching is the go-to-market. You started as a data center company. It's kind of a different channel versus campus or not, unless you address the same customers, basically. Can you talk about your go-to-market efforts, your sales channels? And what kind of adaptation did you have to do in order to address the campus environment?

Yes. I think there's 2 pieces to it, right? One is the campus sale is very adjacent to the data center enterprise sale, right? So everything that we do to add resources and to broaden our coverage model for the data center enterprise sale, and we've been doing that and we'll continue to do that. It's directly leverageable to the campus, right? When the sales team is selling to a data center customer, they have every opportunity to sell the campus as well and you can leverage kind of the experience across to the campus. So we'll -- every time we continue to grow that, we'll continue to grow the campus capability as well, right, it's like a land and expand almost from that data center. At the same time, we do believe that over time, you're going to need to have a broader channel sales effort as well to continue growth in that campus business. So we have started to, and we're probably into our second year now, of a campus-targeted channel program, targeting particular partners, strategically aligning with some partners and then building channel programs to benefit them and to help kind of expand that campus. So I think we're seeing some progress there, but it's early, right? And that's not something we need for the metrics we have today in front of us today, but we know that we'll need that later to help continue to grow that business. So we are investing in that now, and it will -- it has to be targeted and it has to be incremental just because of the structure of the channel kind of means that we have to find those partners and we have to align ourselves with strategic with particular partners. But that is in train, and we'll continue to invest there as well.

Got it. I got a question from investors. They want me to go back and ask about cloud and risk of current cloud customers adopting white boxes.

Yes. I mean there really hasn't -- I think if you look at this particular product cycle and the evolution, I think we don't believe that there's been any fundamental shift towards white box as part of the cycle, right? And that's -- I think that comes back to some of Anshul's comments about how it's not as obvious and compelling that, that's kind of what should happen or needs to happen or that that's where the biggest advantage is for the customers over time. It's a lot less obvious. I think when you start to look at the technologies and you look at the analysis with these companies than it is maybe when we sit in these forums, we think about how that's just obviously how it's going to progress over time. So I think we need to continue to execute. We need to continue to drive roadmaps with customers. We need to continue with these development partnerships effectively with these customers. I think if you do that and you execute well on that, we don't believe it's obvious that you continue to see some expansion of the white box in these businesses. So I think there's real value to these customers and having a partnership with something like this.

Got it. I want to switch gears to routing. In the few minutes that are left, I want to talk about your disruption to the routing market. What do you bring to the market that is so different than the existing offerings? Same question I asked you about campus so I want to ask about routing.

Yes. I mean maybe I'll kick it off, Charles, and you can chime in as well. But I think the real disruption, I think, to routing is all around using merchant silicon and leveraging that merchant silicon with new software protocols that just drive a much denser, high-powered, lower cost routing alternative. And obviously, the first place, again, that this gets deployed and where we've seen big success with this is with the cloud, right, and with those cloud customers. And you can -- we have developed, I think, some very compelling solutions for the hyperscale customers that are leverageable into the enterprise and have been leveraged into the enterprise around routing and just low-cost routing. And this is a combination of switch router, common hardware platform with new modern routing protocols is very powerful, right, and is being deployed by those customers in multiple different use cases. So I think that's the disruption. The disruption is taking really leveraging merchant silicon with the software, the right software protocols to really change the cost structure of some of those routing use cases and allow for much more interconnectivity between data centers, et cetera, which is super important to these customers.

Yes. And the only thing I would add is that we're making a switch that does routing. And the competition has very legacy type of routers, very legacy type of switching, but they don't really combine the two. And by combining the two, we're really reducing the amount of inventory for the customer needs and simplifying the operation by using the same management and operation -- network operating system for both routing and switching. And that has worked well for the cloud, and it will work well for the enterprise and the service providers as well.

So that leads me to my last question due to lack of time, which is -- I spoke once with a market player, and he said, "Even if Arista is very, very, very successful, you will never see it in the numbers because it's just a license fee on top of a switch." So because you're so price disruptive, the opportunity -- and again, according to him, but he's external to the company, the opportunity is relatively small in dollar terms in the areas where you're successful. So the question is, going back to you with the same thing, is it really just a license fee on top of a switch? And if that's the case, is there a dollar opportunity there?

Yes. I mean there's nothing wrong with a license fee on top of a switch and expanded TAM, right, and that's effectively what that is, right? You're increasing the addressable market. For sure, there was a cannibalization of that original router TAM just because of the size and scale and cost ASP per port in that market. But I mean we weren't participating in that anyway, right? So for us, it's absolutely an opportunity, right, because even if the -- even if you're compressing the ASP versus the historical, that wasn't addressable for us before, right? It's a whole new business that's addressable, and it's accretive because of the routing license. But it's just added business and added TAM where we can compete very much on our turns, right? With our legacy technology, our heritage technology, you're right, that drives that. So I think it's -- yes, it's very much an opportunity for us.

Got it.

It's not the same as it was before from a routing player perspective, but for us, it was all upside, right?

Got it. Great. Unfortunately, we ran out of time. So thank you, Ita and Charles. Thanks very much for your very open answers. And if I do get any additional questions from investors, I'll make sure I'll pass it on -- pass them on to the company for further discussion. Thank you so much.

Thanks very much for having us. Thank you.

Take care.

Bye-bye.