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

Hello, everybody. Tim Long here, along with my colleague, George Wang. Thank you for joining us for this fireside chat with Arista Networks. We're fortunate to have Ita Brennan with us. Many of you probably know Ita has been CFO there for a while. So we're just going to dive right into Q&A.

So Ita, let's start with maybe some more high-level type ones, and then we'll get into some of the product and market questions. So first of all, I think the market obviously had a very favorable reaction to the Analyst Day and some of the guidance, particularly for 2022, that 30% revenue growth number. Could you talk a little bit about how broad-based -- obviously, we're seeing cloud strength across the board from all of the companies and a little bit on the enterprise as well. But from an Arista standpoint, can you talk about the real drivers behind that nice uptick? And what do you see as risks and opportunities, risks to this number and opportunities to exceed it?

Yes. No. Look, I think we've seen some good strength of demand across kind of all parts of the business over the last while. I mean enterprise -- we showed this at the Analyst Day, the enterprise and kind of including campus has been growing very nicely, very steadily with like a 30% CAGR over the last 5 years, growing new logos, new opportunities. So that piece of the business has been performing very consistently over time. And we see a good pipeline heading into 2022 for that piece of the business as well. We've talked about the campus piece and that we would double our $200 million going to $400 million. So that's a big contributor to what we see for 2022. The providers business has returned to growth, and we expect that to continue to contribute. And I think on the cloud, I mean, cloud had been somewhat muted for the last while. We've now come through kind of new product qualifications. These customers are kind of laying out their investment plans, particularly with the new 400-gig cycle ahead of us. And together with the kind of some of the extended lead times, we're getting some better visibility into what they're planning to do as well, right? So I think when you think about the demand picture, it's healthy across the board and with an uptick in cloud as you'd expect, as we kind of head into this new cycle coming off of a period where they hadn't been investing with kind of the normal spend that you'll see from that piece of the business. What's constraining all of that, honestly, is the supply side of things. We've taken a view, obviously, with the 30%, kind of assessing everything that we know, risks and opportunities around that. But really, supply is probably shaping that 2022 view at this point even more than demand. We'll just have to continue to manage that as we go.

Okay. Great. That's helpful. And maybe just looking at a little bit the 15% longer-term CAGR. Yes, it seems like the enterprise campus side, there's a lot of room there given just the low amount of market share that you have. What do you think kind of the cloud cycle impact will be? I mean you've lived through some of these big boom periods and then real lulls for that cloud vertical at a time where -- you guys don't report orders. But obviously, Cisco and Juniper 2 quarters in a row, triple-digit order growth. That's not real and sustainable. So what's the -- what drives that 15%? What are the kind of assumptions that -- like what happens with the cloud vertical, which I think you said is running closer to the 40% of your mix these days?

Yes. Look, I think the cloud cycle, when you think about kind of cloud spending, we do think maybe it's a little bit more cyclical going forward. If you think back Q4 '19, early 2020, that was the first time we had seen any cyclicality in that cloud business, right? Prior to that, it had been a long investment cycle, adding footprint and other things and you hadn't really seen any pause or so as you went through product cycles previously. So that was the first time that we saw that kind of decline. I think Q4 '19 was the first time as an industry, we saw a decline in cloud network spending and then obviously was somewhat muted for some time period. So I think our view now is we have to at least prepare and think about the business as a business where that vertical is obviously healthy. They're investing for all of the different growth drivers that they have, AI, machine learning, et cetera, but that it may not always be just up and to the right. Over time, it will grow, but maybe there's some cyclicality to that. And we're kind of building that into that outlook that we've talked about, which says the enterprise piece of the business can maybe be a more consistent grower over time and then maybe cloud layers on top of that with some cyclicality. Now obviously, we don't have a perfect base view of what that looks like. But it does seem that, that might be kind of how this evolves. Now we have some good products. There's new product cycles in the future coming. So I think overall, the spend is positive but again, maybe a little bit more cyclical than what we've seen from them kind of pre Q4 '19. So our outlook is trying to encompass that and say that -- periods of more aggressive growth from them. Maybe their spend is slower in other periods. But when you look at it over time, you're still growing in the kind of mid-teens CAGR.

Okay. Great. Great. And maybe the last one from the Analyst Day, then we'll get into the verticals. The margin outlook is kind of comparable similar to where you are now. You guys have really exceeded on the margin front. The gross margin has been incredibly resilient and stable, as has op margins. So how do we think about kind of this very strong growth yet no operating margin leverage? Is this going to be continuing to invest in other vertical or tangential businesses, and that's -- that level of investment is needed for this 15% growth? Or what else is going on, on the leverage side?

Yes. I think, look, we have to have some movement on the gross margin line in the near term, right? As you think about it, we've been operating kind of at the upper end of that range for the last couple of quarters. I think now as you think about some of the supply chain impacts and the cloud mix maybe shifting towards cloud a little bit in that time frame, that will put us kind of more firmly in the range of that 63 to 65 and depending on the quarter, maybe even towards the bottom end of that range. So that's part of kind of -- there's definitely some impact there, I think, in the near term. On top of that, when we think about investing in the business, I mean, I think about R&D and sales and marketing and really the drivers on the R&D side. We would like to accelerate hiring into the R&D organization particularly on the software side. We see lots of opportunities that can drive revenue growth in the near term and in the longer term. Whether we'll achieve kind of the hiring that we'd like to have there or not, we'll have to see. Same thing on the sales and marketing side, we've been adding 36% to the sales and marketing organization. We'd like to continue to do that, making some investments around channel and channel management at the same time. So I think our desire, for sure, would be we grow that top line at 30%, could we grow OpEx at 30%. We haven't done that historically. We'll grow OpEx somewhere around low 20% range in 2021. We hope to do better than that in 2022, but we'll see how it plays out. But I think we're certainly giving the business the right to go and make those investments because we do think it drives kind of that future growth from that future scale of the business.

Sure. Great. Great. Maybe if we can switch over to the cloud vertical, which I think recently has been your largest segment. Kind of on the customer front, maybe a 2-parter. One, obviously, you disclose annually, you've got 2 pretty large customers in there. And there was a point where -- particularly Facebook, I know there was revenue recognition with deferred revenues and whatnot but certainly had a step down. And even Microsoft was kind of flattening out and they kind of really went the other direction. So just curious what you think has happened with your large customers. Is it just they're growing so much or you're taking share back or something like that? And then second, curious how you think Arista is doing with some of the other hyperscalers as well as getting deeper into the Tier 2, Tier 3 cloud, which I'm sure has a pretty long tail to it.

Yes. Look, I think it's always hard to call share in these accounts, right? I mean we kind of look at it as where are we -- as we go through these product cycles, where are we earning our fair share, winning use cases, et cetera. And I think this is the third, fourth cycle probably since I joined Arista even. It's been great to see kind of the team execute so well through these cycles and come out the other side with maintaining and expanding kind of where we play in those networks. And I think we've done that again successfully here at 400-gig, and kudos to Anshul and the team for driving that, right? So the exact percentages, we haven't talked yet about where these guys will be for 2021. But I think we're very happy with kind of the demand profile that we see from those customers in front of us as we head into 2022. And that's broader as well to -- I think they dwarf kind of the other cloud titan customers. But those customers are large customers in their own right if it weren't for the fact that they're being compared to the 2 larger guys, but those other customers are growing as well and are investing. And that's cascading into that tier -- the specialty cloud part of the business as well. I think that's a competitive space. I think people are very focused on improving their cost and their power consumption across these footprints. So as we move into these new cycles, we are seeing strong interest not only from the big guys but also kind of across the rest of that business as well.

Ita, just as a follow-up to the cloud vertical. Can you talk about the kind of lead time and the kind of preordering for customers? So you guys talk about kind of full visibility kind of stretched out over a year versus kind of 2 quarters in the past. So can you kind of give more color on what's the latest you guys are seeing just in terms of the customer ordering and the kind of lead time front?

Yes. I mean if you think back to kind of pre all of this, COVID pre supply, the dynamic -- the longest kind of lead time item we had kind of from a manufacturing perspective was the IC and the chips. And that was -- we thought that was a very long kind of 13 to 23 weeks kind of lead time depending on the component, right? And we offered those key components for these customers. And therefore, the visibility from an order and demand side was much shorter, right, and could be much shorter because you were able to respond to upside. It was generally upside, respond to upside pretty quickly. If you fast forward to today, I mean, lead times are now 50-plus weeks on the component side. You have to -- we are buffering. Obviously, you've seen that in our financials on the raw materials side. But you have to be much more thoughtful and these customers are being much more thoughtful about what do those long-term forecasts look like. And we're working very closely with them to try to figure out what should the deployment schedules look like. Because I think you have to stay focused on kind of deployments in this environment and really figuring out what can you commit to and what time frames and then does that meet their needs and what are their priorities in those time frames. So that's been a big change in terms of visibility because it has to be just driven by the supply of -- the current supply environment.

And with Arista being a leader in 100-gig, as the world shifts to 400-gig, can you kind of talk about Arista's competitive advantage versus other competitors as sort of we begin the deployment for 400-gig?

Yes. I mean, look, I think it's just continued execution, not only -- if you think about it in a -- take a hyperscale piece of the business. It's just this collaboration over long-term and technology road map is the first piece of that, right? I mean we're working now on kind of the next generation and the generation after that with these customers. That's where Andy and Anshul and those guys are spending a lot of their time. It's not so much on this cycle anymore but on kind of the next product cycle, the cycle beyond that and what that architecture needs to look like. So I think it starts there. And then it all comes back to just executing, bringing those products to market, quality, hitting those time scales, et cetera, and being responsive to and capable from a technical perspective and responding to the needs of what these guys are driving as their priorities, right? And being able to do that consistently through cycles, I think, is very valuable to the customers. And that's what earns you the right to kind of have those conversations and be more collaborative.

Sorry, making sure I'm not on mute. Ita, just one last follow-up on cloud and George's question there. Obviously, you guys, as you said, did great in 100-gig. And some of your competitors that did not try to use these as dislocation points and try to insert themselves more. So understanding you've got incumbency, you've got execution and good collaboration and embedded base with some of these customers, what is the risk for market share moves or pricing moves or things like that as the transition kind of ramps to volume from your standpoint?

Yes. I mean, look, there's always been multiple sources for these customers. They want to have multiple sources. That's not new, and that's not different. And again, performance is everything, right? Execution is everything, right? And that's not a new dynamic, right? That's been the case for a long time. So I think it really, again, comes back to that very same principle of [ if you are a ] continued kind of history of execution with these customers, then you earn your right for their business. And it's a technology-driven decision and it's a technology capability. And that fits very well for us, right? So again, it's always hard to talk about share in the individual customers. But the exposure to new use cases and being able to continue to drive that business, I mean that's kind of a result of your execution and the execution of the technology team. Pricing is obviously an element. They're very savvy customers. They understand cost structures, et cetera. And again, that's not new either, right? So we continue to execute against that. It's a big piece of business for any supplier. So you have to be responsible and reasonable in terms of how you think about pricing. And that's been pretty consistent, too, over time. So I don't know that we've seen anything different in that -- from that perspective.

Okay. Great. Great. If we move over to enterprise campus part, as you mentioned, doubling next year in revenues. That's -- obviously, you guys have taken big share with that. Curious, just it's a newer business for you and the sales force. So how do you think about visibility here? I'd say traditionally, this vertical probably has better visibility than what a cloud vertical would have. So just curious what you look at, I mean, customer counts, win rates, at bats. How do you get confidence in both the near-term 200 and the longer-term 400 and the longer-term $750 million?

I think there is some extension of visibility here, too, right? They still have to -- this part of the business still has to respond to the same lead time extension that we talked about for cloud, right? So there is some -- there's definitely been some increase in visibility there, too, right? Beyond that, it's more of a traditional pipeline -- enterprise pipeline type analysis and business where you can see the opportunities that are -- that people are pursuing and getting a view as to when kind of those opportunities will close, et cetera, right? But the extended lead times are also feeding into this in terms of increased visibility as well. So I think it's traditional -- think about traditional enterprise pipeline with win -- good win rates, good traction. And then on top of that, you've got a little bit longer window as well just because of the extended lead times.

Okay. Curious about the -- kind of the product portfolio here. You guys kind of brought it out over time. But wireless seems to have become -- wireless LAN have become more important. And you guys seem a little under-indexed there. So can you talk about the importance of your wireless offering? And particularly, what you guys can do to maybe beef that up and help win those enterprise deals where it might be wireless-led compared to wireline or even data center tie led?

Yes. I think maybe I'll start, and then Liz can chime in as well on the product side. I mean I think it's just -- there's definitely a matter of emphasis here, right? For us, we will always look at these opportunities as coming from the switching side to the wireless, right? And -- but you shouldn't forget that kind of that wireless offering is still part of -- CloudVision is still managed by CloudVision, et cetera. So what we see a lot is we're engaged in an opportunity. Sometimes it's a data center customer, sometimes it's not. But it is starting with kind of the wired switching platform and then the customer has an appreciation of CloudVision. And then -- so we will sell, upsell, if you like, the wireless piece and that functionality of CloudVision across the two as part of that sale, right? So that's our starting point. It's just our natural starting point given kind of the -- that we start with the network. Liz, I don't know if you want to add anything on the product side?

Sure. I think when you look at the large enterprise customers, which is what our sales force interacts with, many of them that have experienced the quality of EOS, they want that same quality in other parts of their network. And these network engineers that we're working with in the data center space or if we've had other opportunities with them, they're the same teams in many times that are making the decisions for the campus deployments. And so to be able to take that -- the quality that they found in EOS and the management structure in CloudVision and extend that into the campus is kind of the natural progression. You think of the way that campus network is laid out. It's really just another leaf off of your data center. And so having the WiFi component is necessary because it is a part of the campus. And we have done a great job of bringing in kind of the mojo quality of experience into our CloudVision WiFi. And customers have really valued that by having it being drawn from the same CloudVision that they use to manage the rest of their network. So it's that holistic orchestration and automation. It's not just the data center and not just the WiFi but the entire enterprise network.

Okay. Maybe one more follow-up here on kind of the campus side. Ita, could you talk a little bit about, kind of following on what Liz said there, how much of this growth that you're seeing is kind of new to Arista business? And how much is -- you partnered on the data center side, and that's pulling you in. So just curious how much of this growth is more or less bundled or led through incumbency and how much is kind of new wins?

Yes. I mean I think we were surprised to see a fair amount of, I would say, campus led kind of deals with these customers. I mean it's still targeting kind of that larger enterprise account, right? But we are definitely seeing large enterprises that are serviced by the incumbents, are looking for that quality that Liz talked about, like a high-quality kind of solution. And their choices have been limited, right? So they're interested in looking at Arista. So there's definitely a set of customers who have had an experience, maybe don't like the fulfillment pricing models of some of the newer products from the incumbent and then come to look for kind of a replacement solution for that. So there has been a fair amount of that, probably more than we would have expected kind of when we started on this journey. But at the same time, the natural motion, I think, is -- for a big piece of this business is also coming from customers that know us already, right, and are just saying, okay, now that -- I understand the whole quality story. I've experienced it. I understand the automation. I use CloudVision, this is a very natural extension in that situation as well when their campus kind of comes up for renewal. But there has been -- I mean there are a lot of large enterprises where we don't solution the data center today who are interested in the overall Arista solution. And we'll start maybe with the campus because that's the next thing they want to go and make some investments in.

Ita, I want to ask about service provider vertical. As Arista gets more traction in the service provider space, can you kind of briefly discuss your strategy around increasing the telco penetration? And do you see a similar opportunity for share gains in the service provider versus the campus side?

Yes. I think service provider has been interesting as a vertical for us. We saw some really good traction in certain use cases when we brought kind of the 7500 switch router platform to market back in '17 and '18, and we saw some really good growth and good wins with good, large telco names, right? But the use cases were specific, right? We didn't -- it was difficult to get broader than those use cases because we didn't have kind of the legacy routing stack, et cetera. So if you fast forward to today from that, I think we have -- firstly, we've built kind of a more, I would say, complete new routing stack for the cloud and for the hyperscale customers that's been leveraged in enterprise, and it's been leveraged by some service providers, new protocols, new features. We've also kind of spent some time on specific service provider-only features that we think apply across the board. And we've done some work specifically for some service provider customers as well on those features. I think Anshul would say we have a credible seat at the table now with that kind of combined stack. We would still need to find customer partners and work and probably do some more work specific to them on their feature set if we were to do more of these broader use cases. So I think we're continuing to kind of add to our capabilities. We're partnering with customers and solving specific problems for them. And then we can be deployed more broadly in their footprint. But it's going to be a continued effort, right? We need to find those partners and we need to continue to build kind of that base. So it's not yet at the point where you could say a service provider will consistently grow the same way maybe as enterprise and campus has been. It's more opportunity driven. But I think we're in a stronger position to take some more of those opportunities. But we still have more to do there, for sure, to make that more of a consistent grower over time.

That makes sense. Just a quick follow-up on the traction related to the campus and service provider. Obviously, you touched earlier just in terms of the operating expense ramping investments as you increase the channel and kind of S&M, R&D. Can you briefly talk more on just kind of implication of the campus and service provider growth strategy related to the operating [ expense ]?

Yes. Look, I think the reason why service provider is so intriguing to us is just that there's a ton of leverage and there should be a ton of leverage on the technology and the product side, right? I mean the cloud-led kind of development that we're doing on routing and other capabilities are absolutely applicable to the service provider as well, right? And there -- and that's why it's so intriguing to us to figure out how do you really help them leverage that and make that work, right, because that's very powerful from a business model perspective as well, right? Sales and marketing-wise, the service provider piece is -- it's maybe a little bit more intensive than cloud, but it's still a large customer technology-led interaction, right? So it's not that different from that piece of the business. On the campus side, obviously, it's more enterprise. A lot of our kind of sales additions are added kind of sales and sales engineering head count is targeted towards that enterprise and campus piece of the business, right? That's where most of that growth is going. We have started kind of our channel journey, if you like, again, not for that large enterprise sale, which is what's driving the business today, but more for later when you think about mid-market and other parts of that market. And we need to be strategic around how we think about adding partners -- adding channel partners, et cetera. So that's gradual. So I don't know that there's a huge step function in go-to-market associated with some of these kind of adjacent expansions. But we do need to be consistently investing and consistently executing to keep that traction going.

Maybe one more, if we could. Obviously, Arista over the years has done a good job going from kind of data center switching to routing to campus switching. Could you talk a little bit about the next multiple years when you think about other technologies, be it optical or storage or SD-WAN? What do you think some of the other good opportunities for Arista are going to be to grow into?

Yes. I mean I think, first of all, there's a lot in front of us, right? I mean this is a big TAM, and we're still relatively underpenetrated in a lot of this market. I think one of the things Arista is good at is kind of focusing on what's in front of you and making sure you're executing well against that and then taking incremental steps around that to kind of drive that next phase. So I think visibility, monitoring, security of the network, these are things that kind of we've taken some step into already, and we'll continue to do that, right? And we're very conscious that there needs to be adjacency and synergies when you do this and that it should leverage that EOS capability that's pretty unique, right? When you think about data lake and some of the early steps there, it's all leveraging that very unique way that EOS captures network state and the visibility that you get from that architecture, which is pretty unique in comparison to what any of the competitors can do. So it's kind of -- I think you'll see us do stuff that kind of continues to leverage that. There's a data set there that's very valuable. We're now marrying kind of some of the big switch monitoring data sets in there. We're leveraging some of the Awake sensor data sets in there. We will look to add other data sets to that. And you've got a very unique view then of the network that's captured in a very unique way and then leveraging that to grow other pieces of the business. I think that's kind of the strategy as you go forward. So I think incremental adjacent and leveraging kind of that unique architecture is probably the way to think about it.

Okay. Great. That's helpful, Ita. I think we're up against the end of the time here. So thank you so much for joining us. I appreciate all the conversation here. Thank you, everyone, on the line for joining. Enjoy the rest of the conference. And Ita, look forward to seeing you in person out there soon.

Yes, hopefully. Thanks for having us. Thanks very much.

Okay. All right.

Bye-bye.

Thank you.