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

All right, welcome everybody. I'm Meta Marshall. I cover the networking space here at Morgan Stanley. We're delighted to have Ita Brennan here today, CFO of Arista Networks. I'm going to start with a brief disclosure. For any important disclosures, check morganstanley.com/researchdisclosures or talk to your sales representative. So Ita, welcome.

Thank you.

I wanted to start by diving into there's been this debate over the past couple of years about your relevance was capped at 100 gig and you wouldn't find as much relevance to your cloud customers at 200 and 400 gig? I think clearly, we've seen that the opposite is true. Can you just go into whether there's been a shift in how partners are -- how your customers are looking to partner with you and how this upgrade cycle is a little bit different?

Yes. I think, look, it all goes back to we've been partners with these companies for many product generations at this stage and for a very long time. And the good thing about it is it's all engineering-led. These are technical buyers. Their engagement is at an engineering level. Whether we're solving problems for 2 generations out or we're solving problems for current problems in the network today, there's this great collaboration at the engineering level that really drives that recurring stickiness in the business. And there's lots of changes from architecture, decisions perspective, again forward-looking and current, that we engage with these customers on and help contribute to those decisions and those discussions. And that's part of that collaboration that Anshul talks about with these customers, and making sure that we're bringing our expertise to the table to help not only can it be part of the product solution but obviously be part of determining what that roadmap looks like, what that architecture looks like going forward. I think that's the key to being able to anticipate some of the changes that are happening from a technology perspective, maybe lead some of those changes, and be able to participate in a meaningful way on a recurring cycle basis. So for us 400 gig wasn't that different. A lot of the discussions around 400 gig were happening pre-100 gig shipping in volume. It's that long in terms of the lead-in. Now you're talking about 800 gig and beyond at that level in those conversations. So I think that's key is that collaboration from a technical perspective, the sharing of ideas, the trust that's between those engineering organizations is a big part of this collaboration over time.

Got it. When people hear about Sonic or they hear about different initiatives, they often think like, oh, they're trying to internalize certain functions. And I guess what -- how do you see some of those efforts by some of your web-scale customers to internalize versus where that creates even more opportunity for you guys?

Yes. I think all of their focus all the time is to make it better and make their operations scale. They're growing, they're still growing, they're still expanding what they do. So the focus is really on making it more efficient, making it scale better. So we've always been very supportive of that, whether it was switch routing platforms, whether it was some of their own internal work that they wanted to do. Our approach has always been to support that, because really what it's driving to is a more efficient network that lets them scale and do more. And as long as -- so we've found over time repeatedly that supporting those internal efforts, helping accelerate them in some cases, support them in other cases, just exposes us to even more. And because of the nature of their business and how that business is growing, there's always something new for us to go do or new use case for us to go solve in that network. So that approach has worked very well. I think it's a compliment to Anshul and Jayshree in how they think about the customer. But really when you support the customer in these efforts, it's just -- every time it's exposed something new.

Got it. To speak of something new or expanding opportunities, you clearly have a fair amount of deferred revenue right now that's tied to some of these projects. Can you just help level set? Is that multiple projects? Is that some of the expansion that you're talking about? Just how we should think about some of that deferred revenue coming off over the next year or being recognized?

Yes. It's pretty typical when we have like a new product -- set of products. We saw this at 100 gig. We're seeing it again now at 400 gig. Just as you deliver these platforms at scale, early in that cycle that there would be some acceptance associated with that to give them an opportunity to really deploy it in their environment. It's very hard to pre call some of that just because you don't have environments of that scale to test in. So it's pretty typical for us to build a deferred revenue balance at this time, a product deferred revenue balance. We've said it's largely cloud related, at this point. It's multiple customers. It's not just a particular use case. I think what you'll see happen is those will cycle within that balance. So we're not -- we haven't assumed any decline in the balance year over year when we think about the 30% growth case, but it will cycle. There will be things that will be accepted and then some new dollars will go into that as we go through the year.

Got it. Maybe moving away from the super-large cloud guys for right now. Your tier 2 and specialty clear -- specialty cloud providers revenue accelerated last year as well. What are you seeing in maybe a change in behavior in these customers in terms of building their own networks versus leveraging cloud, just are you seeing a desire to control their own networks or just anything on their behavior?

Yes. We come through a period, if you think back to end of '19 and to the beginning of 2020 where a lot of those players were rationalizing what they were going to do themselves versus what they would put in the cloud. And so we saw some volatility honestly in our business with them in that timeframe. But now having right-sized what they want to control for themselves and found where they're competitive -- where that brings them some competitive advantage, we are seeing those footprints grow. And we saw that last year that piece of business grew nicely last year. They will also look to leverage the newer products and new technology, so we'd expect to see some continued growth around that. So that's specialty cloud customers will tend to be customers who reach a point where they really start to evaluate whether what they want to control themselves, cost structure wise what's better to control and have themselves versus what's in the public cloud. And I think we see that pattern of you get to a certain scale and size where you start to really think about that and then you resolve that and then grow the pieces that you maintain for yourself.

Got it. Does their buying patterns really differ from what we would see with other customer sets? Is it more project based? Just anything.

It's technology -- I think there's some linkage to technology for sure. They will want to follow or be on par with what the hyperscale customers are doing from a technology perspective. It's a little different maybe to enterprise. They'll probably be more aggressive around following the technology than maybe what you would see from an enterprise customer.

Got it. Maybe also turning to the service provider side. This has maybe been an opportunity that's taken longer to develop than maybe you wanted to, but it's still managed to be quite a good business for you guys. What do you see as a catalyst there for more faster changes to their networks? Is it 5G? Are there core network upgrades that we can think to as creating opportunity for you guys?

Yes. That has been very much a use case where you can find a fit to like a telco cloud-like architecture, we do very well. And we've been growing, where we're deployed, and good customer acceptance then it grows. It's really how do you expand the parts of the network that can be serviced with that cloud architecture over time because we haven't gone back and built every feature that would require for you to go compete for the existing infrastructure. So it's really what can drive more cloud-like architectures for them. I think 5G ultimately will do that. There'll be more -- initially it's at the front end of the design, but once it gets to aggregation and managing the aggregated traffic from that, that's likely to be more of a telco cloud-like architecture and that's where we can compete and so hopefully that helps. But it's really the more that their networks want to look more like a cloud network work, I think that gives us a real opportunity to do something compelling for them. It's just that's taken a long time and we're still looking for those insertion points. So I think that's the go forward is more can you find customers that are ready to make a move to that more cloud-based architecture. And when they are, I think then we can compete well for that business.

Got it. Moving on to enterprise, you've done quite well here and are now expanding the offerings into campus. Just where are you seeing the -- I think there's a fair amount of runway for you guys with existing customers, but just how do you think about growing out that sales channel or expanding the campus business? And is there a revenue threshold that we should think of when it's like, okay, this is no longer just attached to existing customers?

Yes. Even today, I think, Jayshree commented on the call, in the Q4 new logos for campus, there were probably 50% that were not existing Arista customers. So there is -- we are seeing still in that large enterprise part of the market which is probably close to 50% of the overall campus market, but accounts where we didn't have a data center footprint before. So we are seeing both of those drivers. Still a direct sales motion to get to the customer, but there's an equal desire I think on the part of large enterprises that's been very much a single vendor sale previously. So having somebody else bring a quality solution there, we do see good interest in evaluating Arista in those cases as well, even if they're not already deployed in the data center. And some of those will have data center footprint, some of them won't. Obviously, if they do have a data center footprint, it's a good cross-sell. But in any case, a large enterprise will have some meaningful campus business that we would pursue in a direct sales model.

And is there a point where you would expand the sales model to be less direct or just how do we think about that developing?

Yes. And we're doing that. We're working on doing that, again, not to solve for the near-term goals but to start to build some momentum for the mid-market part of that overall market. And that will take some time. There's no doubt that we will need to find partners, build a business with those partners, have them be successful. The more we expand our enterprise business which we're doing, the easier it'll become to adopt partners for that campus business, but it's still going to take time. So we're fortunate that we have this large enterprise TAM to go chase first. But at the same time, we need to be making some progress on that indirect as well.

I think one of the questions that I have been running through my own mind is the buyer for the data center being different and maybe more sophisticated than your traditional person managing a campus network. And so is the value proposition of EOS and the simplicity still the same to a campus buyer or just how does that sales messaging differ from data center to campus?

I think at some point that was true. It's become less true. I think the more that you have different devices different entry points to the network, more complexity around this whole hybrid work, use IoT, use some buzzwords. But it's like the more complexity there is in that campus network, the more people working hybrid models, et cetera, the more important it becomes to just stay in the business. I think you are seeing a real interest again in technology around that, whether it's just the management of that, the security of that, the visibility to that. So I think you're right. If you think about -- go back in time, we weren't that interested in the campus market, honestly, because it was very static. It had the technology it had and there wasn't a lot of churn and there wasn't a lot of new technology. But now there's a lot of new investment happening around the campus and there's also a refresh happening in the technology. So it's a good time to be looking to insert into that campus network.

Okay, perfect. Are there areas where either your resellers are -- I'm trying to get a sense of security has been an area obviously that a lot of networking guys have gotten into and you guys have had select purchases but have largely chosen to partner. Just what has informed that decision to partner more versus have that in-house?

Yes. Again, it comes back to I guess the trade-off between focus and really being able to drive the differentiation. So when you think about the Awake acquisition, for example, that was a -- it was a first step to say, we would like to have more security expertise in the team because we think securing the network is becoming more and more important and more complex. So what -- I think we just announced last week our first integration between the Awake product set and switching. But there's real synergy there between the data that EOS collects and monitors and the Awake sensors and being able to marry the 2 of those as part of the switch and then sell that as part of the switch solution. So we took our very first step and it's the first step towards integrating those 2. But when you think about security, that's how we think about security is saying, okay, we have a unique perspective on the network, we have a unique access to data on the network. How do you leverage that and secure what's happening within the network, as opposed to trying to do things that are out of our domain expertise when we have this TAM that we can go and address. And so we'll stay focused on doing things that are focused around the network itself.

Okay. I'll give you a break from the technical questions for a second and dive into financials. Obviously, Arista has had phenomenal operating margins over the years, but just in terms of your guidance this year of operating margins of 38% or just investments you would like to make, just how do you see areas in which you want to invest for growth over the next couple of years?

Yes. I think, look, on the sales side, we've fallen nicely into this motion of adding roughly 30% plus or minus to the sales team, and we feel like there's a good absorption of that and a good return on that. On the software side and the R&D side, it's all about people and talent, and we would go as fast there as we can. That was -- we didn't do as well there as we would have liked last year just because it was a difficult year for hiring, et cetera. We hope to do better this year. And that's just because there's so much work that you could do that that's tied to revenue that you definitely want to acquire those resources as fast as you can. Can we grow that at the same rate as the top line? We'll see, but there's definitely intent and that would be the right thing to do if we could do it. I think the other factor this year we've talked about gross margin a little bit, but I think gross margins will be pressured a little bit and that'll obviously impact the operating margin as well. And so we'll see where we come out. But I think we do have a serious intent to go drive hiring on those 2 areas.

Got it. One element of that gross margin is just maybe some of your larger customers purchasing, but the other, now we can turn to everybody's favorite topic of supply chain. Clearly, this has been an area of disruption that you guys have managed better than a lot, but just what are you seeing just in terms of is it stabilizing, getting better, just any update there?

Yes. I think it's hard -- it's always hard to baseline things like this. I think what we said on the call is we don't feel like it got necessarily much better or worse. I think it was reasonably stable. You still have -- you're still in an environment where you have surprises happening, and for us to really get some momentum behind truly driving higher volumes and shortening lead times, et cetera, et cetera, we're going to have to -- those surprises are going to have to -- the number of surprises have to come down. You have to start to really feel like you're getting your arms around more and more of the components. We've been broadening out our focus -- direct focus, if you like. We used to manage key components. Now we're managing a broader set of components still through the CMs but forming contact directly with the end customer. We've obviously made some expanded purchase commitments just to help secure your position in line. That helps for sure. But I think we're not yet at the point where you'd start to say you're really seeing some significant improvements and the level of surprises is coming down, et cetera. We now think that's probably a 2023 dynamic best as we can see now.

Got it. You just alluded to this inventory build that you've put into place having meaningfully more inventory than you have. Could you just give a sense of was that all just a genius foresight? Just what were the signals and what is the biggest component of some of that inventory?

Yes. So when you think about the purchase commitments, it's not that I have the inventory today. I wish I had but we don't. So it's really -- it's more taking your -- putting out your commitments for the future in the hope that by being in the chain, you'll have better access to inventory. It still won't be perfect. You still can have surprises, but at least you've got your consecutive positions in line and that hopefully would help with the supply. The inventory itself it's up a little bit. We did double-down on some of the key components early on and you saw a little bit of growth in the raw materials and stuff because of that. But a lot of this is really more making confirmed commitments to the future periods so that hopefully you get that inventory when you get there. And again, the expectation is that it still won't be perfect. But it helps. It helps you maybe be in a better position as to supply, it should help with expedite fees and other things hopefully, and the cost of it, at this stage, and because we're early in a product life cycle is really tying up cash. Well, ultimately, we'll end up having to have some cash to cover some of those commitments. But that seems like a good trade-off to us versus being 12-plus months away from any -- being able to have any upside if you don't have those commitments in place.

Got it. Maybe added here versus the end of just any Ukraine-Russia impact since it's tightening various things around the world.

Yes. There's nothing significant directly for us. We don't have a huge exposure there. Are there second and third order effects to the supply chain? So I think that's still something that will have to work its way through the system. So I think there's nothing specific that we know of today, but it is one more shock to the a system that didn't really need any more shocks right now. I think so we'll have to see how if there's anything else that comes out of that.

Got it. Clearly investors can see where the source of the 30% growth comes from this year with just a lot of product upgrades or upgrade cycles kicking off. But just what gives you confidence in the longer-term double-digit revenue targets that you laid out at the Investor Day?

Yes. We talked a little bit about the different building blocks. And I think you've got the enterprise business which really is a new logo share gain story, and I think we have some good momentum there. So we need to continue to execute against that. So we think that business is a -- it's been growing roughly 30% CAGR for the last 5 years. I think we can continue to execute there. There's lots to do in front of us around that business. I think cloud we've established that we can go through repeated cycles and get our fair share of that business. What their spend will do is something that's a little harder to predict if it will be a straight line up into the right or will it have some cyclicality to it. I think with that something that will evolve over time. Right now they seem to have an urgent need to deploy equipment and that will -- with the supply constraints and everything else, that will last for a while. And then is there some volatility to it after that, we'll see. But I think we get to a world where when cloud is strong, you're going to grow very healthily. The rest of the business hopefully supports growth in a period where cloud is maybe more muted and you have that model over time which is you have a good, steady CAGR -- double-digit CAGR growth over time.

Got it. I want to open it up to questions if there's any from the audience. All right, perfect. I have more questions. So if we turn to the future, clearly, EOS and the modularity of it and the ease of use of it has been the differentiator for Arista for now the last 10 years. When you're seeing either your cloud customers push you for new features or enterprises push you for new features, what are those that'll be the next generation of differentiators? Is it AI? Is it increased security tie-ins? Just how do you see differentiation developing?

Yes. I think it all does build off of the EOS. And whether it's pushing the boundaries on ethernet and AI and being able to do more and more that comes from just pushing the envelope and development of those high-capacity systems over time. That's leveraging EOS plus silicon plus other things. So we'll continue to do that. That's bread and butter to be a good supplier to the larger cloud customers, et cetera, that should continue to do that. I think on the enterprise side, there's more to do around just helping them manage the network, making that simpler, easier, more reliable across their deployments, so more with CloudVision, more monitoring, we've talked about more security of the network with leveraging Awake. There's a lot to do there still and that'll get pulled in conjunction with customers over time. But yes, I think we have an important role that we can play in just helping make that infrastructure more robust and easier to manage for enterprise customers, which is what they want, right? They really want to be more efficient service providers to their businesses and just more robust, and it's becoming more and more important to companies to be able to do that. So I think there's a lot we could do there with the tie-in of Awake and some of the other visibility capabilities that we've been adding.

Got it. Another area where you guys have alluded to is potential entry into other web-scale customers. And just what is that evaluation process or entry or opportunity set for you guys that you see with some of the other web-scale guys that you don't do major business with today? I understand you do business with everybody, but...

Yes. I think the concept that Anshul has developed over time of make-and-buy is very interesting in this context. And can you -- can a customer be better served, if you like, with that combination, what does a branded supplier like Arista bring to the table, and then in conjunction to having multiple sources of supply, which is what they're obviously trying to achieve, right? And that model has worked very well. Yes, I think there's some openness to that with other customers. The problem is it's really hard to get to. So again, it's not -- I think about this as something that's absolutely worth exploring and engaging in, but it's not a simple thing to do and it'll take time even if you were to be successful. So I think it's something that's I think for us to worry about for now as opposed to anybody else. And if we were ever able to make progress, I think we could -- we would talk about it some more. But it's intriguing. It's just really, really hard to do.

Got it. Perfect. Well, Ita, thank you so much for being here today and appreciate seeing everybody in person again.

Yes. Thanks for having us.