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

All right. Good morning, everyone. Thanks for joining. We are live here on -- with mighty John McCool, who I suspect most of you have met in prior sessions -- actually, in person, just as John has been part of the Arista crew for many years now. And previously, we tracked with him at Cisco. One of the things that I love about John is that he has great perspective. He's been in many different parts of the IT industry, not only in networking, but on the broader industry as well. And at the moment, he happens to be just the person that we want to speak to, John, as you do a lot of work on the supply chain.

So if you can give us a bit of up-to-date work on how you are navigating and what we should expect from here. But before we get to that, I am hopeful that you can start us off with a bit of a bigger-picture question. You have been through a bunch of spending cycles in the past, John, whether it's here or memory or at Cisco. And just wondering if you can help put into perspective how this one compares to some of the prior up cycles that you've seen in spending.

That's a great question. And I think when we look at some of those previous cycles, they've always tended to be focused on maybe a segment or vertical. I think what's very different about this cycle is certainly you see the spend increasing in the networking segment but across all the sectors of IT, compute storage, but it's even broader than that, right? You're seeing spending increase on kind of the consumer side of IT. People buying laptops, deploying home networks. You're seeing changes in the industry around electronic going to automobiles. You're seeing push to the edge and edge computing. So I think a lot of the things that have been teed up around transitions somehow were accelerated by this pandemic. People thinking how did we get through the other side and how do things change. I think a lot of folks are replacing bets across a very broad set of technologies and verticals. And that's, to me, one of the core differences. It's so broad-based and multi dimensioned, and I haven't seen anything like that before.

Is it -- one of the questions that we talk about with investors a lot is the durability of the spending strength. And I think there's one school of thought that says, hey, the world shifted, and suddenly, whether it's telcos or enterprises or webscalers, traffic was shifted. And now companies are spending in conception to that, but that's not super durable. We'll get back to the old growth rates on the other side of the supply challenges. But there's also another school of thought that says, well, we've actually just hit the accelerator button on a lot of digital transformation that a lot of us will stay remote and continue to do all the things that we've done remote in the past, plus a whole bunch of new things. And so can you help sort of gauge for us like how durable this IT spending uptick feels to you at this point?

Yes. And probably more towards the latter , but I think that there's probably a mix of both. I think you have a set of people that are more planning, given the longer time horizons. They see with lead times, et cetera, and just doing things that they would have been doing but thinking more through the timing of that. But I also think you are seeing some underlying, more dramatic changes towards the future, I think, which will have a longer-term effect than just this pandemic period. Certainly, in our industry, the transition to cloud has been kind of doubled down, if you will, in this cycle. And I think that's been going on for a long time, but I think people are embracing that. As people are becoming more reliant, the concept of hybrid cloud and how they architect and build their data centers, both internally but leveraging multiple cloud providers, has really taken hold. You're starting to see more of the transition and investment around 5G. And then in the data center, the infamous transition, the 400 gig -- or 400 gig, 200 gig, 100 gig, as we like to call it, has really come into play as part of this. And I think they're all going to continue post-pandemic time if we ever do get out of this and get to some return to normal or a new version of normal, as we know it.

Well, we were joking before, but I'll say it again, there will be a time [indiscernible] conference, John, and I look forward to it. I look forward to that day.

Excellent.

How do you think the pandemic has affected the calculus of companies as they consider kind of public cloud versus private cloud?

I think part of the calculus is just backing up is I might not have people on-premise to operate my infrastructure. I have to become more resilient. I can't place all my bets in one place, one data center, one geography. Pre-pandemic, we thought about just some resiliency of I have something in Asia. I have something in North America. I'm good. With the pandemic, we've really seen that disruption can occur in any location, and you have to be agile and being able to perform in this environment. So clearly, it places more focus on building out in the cloud. But there are decisions folks have made to operate internal data centers or infrastructure on their premise, and they're more concerned about operating them from the cloud. It's even tougher. It's always been tough to find people to build and run networks. This has placed even more a burden on finding people that can operate networks at scale. So the operational cost and complexity, ease of deployment, all of those are getting emphasized more in this environment than we've seen in the past.

Well, we've had some CIOs say to us, well, it's very much about the public cloud now, and we want to make sure that we get there for some of the reasons that you've just suggested. But we've also had a few that have said, well, now, I've actually gotten a little better sense that if I put too much in the public cloud, it's going to cost me an arm and a leg, so I've got to maybe not, in some cases, pull things back from the public cloud. But in most cases, it's just been a little bit more sensible about what they can do in a colo facility, perhaps, versus what they can do with AWS or Azure.

That's right. And certainly, there are some applications that have to be on-premise, IoT type of applications, things that are very specific to the business. The networks have to run in the campus effectively to make that happen.

How has the pandemic affected your enterprise trajectory, and that is both on the data center side, but also in your burgeoning campus switching line?

From a business perspective or -- kind of give me a little bit more on what you're looking.

I guess, well, certainly from a business perspective, like how has demand been on the enterprise side? But I guess one of the things that we all worry about on this side is that, at some point, those initial customers that were desperate to bring you into the campus market from success in the data center side. Well, you'll have filled up that demand. And suddenly, you'll find yourself really trying hard to compete very aggressively against Cisco, not that you've done already. But has the pandemic made it easier or more difficult to gain that incremental customer?

Sure. I think for folks that are listening, I think the one thing that is very different about the enterprise -- and when I -- for networking. And specifically, we play at the high end of the enterprise. Think about Fortune 2000 as opposed to sort of mid-market. It's been a very biased market in some way, if you compare it to storage. You might have a leader that has 35% share and somebody next, #2, is 25% share. There's been this perception that to make a network work, it has to be a single vendor with the same approach and the same architecture end to end, and that's been reinforced over several decades. And it's not dissimilar to when we came into cloud. And that's led to kind of a normal market share from the leader. And we've been able to go first into the harvest place, which is the data center, where you have needs for the most resilience in your network infrastructure and uptime and been able to service that market and grow well beyond anybody had done prior against the incumbent. And now we're taking on the campus. And as we've gone through that -- first of all, there's an enormous amount of market share still in front of us, so I would say we are definitely at the early innings here in networking in the enterprise. But as we build out more places in the cloud, as we like to call them, but you think about routing, data center, campus, WiFi, customers are seeing us as a viable end-to-end alternate source. And they're seeing, as we go in, there doesn't have to be a monolithic environment with a single vendor and a single vintage from that vendor. It can actually be multidisciplinary by taking these cloud approaches and managing much like the cloud providers do.

So I guess, status -- the pandemic -- you've been competing against Cisco for many years, and the pandemic, per se, hasn't really changed that too much. Is that...

It hasn't changed it too much, but I think the pandemic forces people to sit back and have some introspection. And as opposed to just day-to-day, I'm doing this every day. I'm going to get up. I'm going to fill out another PO. I'm going to replace this box. Especially at the beginning of the pandemic, I think there was a step back to like what does this mean and how should we be doing things. And kind of I think that punctuation, if you will, might have opened up some more interest in talking to us or maybe it was just more peers have gone to other places and companies and have familiarity with Arista and understand how they can bring us in. But it's certainly been -- I think we've seen some changes, very positive, in the enterprise market for Arista.

Okay. Well, that's good. We know that you've got that $200 million target out there. It doesn't sound like that's in much jeopardy, and it seems like there's a decent amount of growth that we should think about behind that.

Absolutely.

Okay. That's great. Let's do talk about the supply chain because you, more than most of the companies in the group, have seemingly sort of navigated that reasonably successfully, both not just nice revenue upside to where we all were thinking, but also able to translate that to earnings. And not everybody was able to do both, John. So give us a little bit of a forward-looking sense -- or how you were able to make that happen and kind of what that means for the second half of the year.

Yes. I mean, I think the team has worked very hard in the last 18, 19 months since the pandemic broke open, kind of viewing this maybe initially as something that's happening in China but then realizing this was going to be worldwide and really stepping up to what we had to do. I think on our last call, we were looking at this is not over yet, right? The challenges are going to continue into the future, and they're really threefold, right? You have sort of this issue around COVID and the ability for people to dependably come to work. And that might be because people are quarantined or they're ill, or there might be some government order that restricts their travel. And that continues to happen, and we see that happening with this Delta wave rolling out. Then there's this more specific piece around the semiconductor capacity and lead time, which is exacerbated by this cross-segment demand from all industries for electronics. So that's kind of the environment. In terms of...

Is there a third one that you were going to mention?

Yes. Just kind of backing up a little bit, I think if you talk about the inventory piece that you asked about specifically, a lot of that -- most of that is just related to the extension of lead times from our sub-suppliers. So you've seen this crazy extension of lead times in the semiconductor components, and we've had to react to that, starting to plan our business for what that looks like, lead times of 52, 53, 54 weeks from now, so what will people be buying and consuming this time next year.

Do you -- I mean, I guess, one of the things that you've done is not just inventory additions to the balance sheet, but your purchase commitments are up quite a bit as well.

Right.

And I guess, what are the pluses and the minuses of such large purchase commitments? I mean, on the -- I guess, on the one hand, yes, if things go well, then the supply is there. But are there pricing elements of those extended purchase commitments that we should watch out for on the margin side?

They're kind of independent components in some aspects, right? The extended lead times. If you're going to be in the business next year, and you believe you're going to be in the business next year, and lead times are 52 weeks, you have to place an order. So in some ways, we're reacting to the marketplace that we're all in with these extended purchase commitments. We're fortunate that we are in the new product cycle in many of these technologies with the R3 Series. So it gives you confidence -- more confidence to make a bet a year out because the products themselves have a longer life cycle ahead of them. So that's one aspect of it. The second part of your question is, definitely, we're in an environment where there's increasing cost, even down at the level of copper and gold and resin and the things that make up chips and devices and printed circuit boards. And that's kind of independent of the purchase commitments, so just basically the environment that we're dealing with.

Okay. Yes. That makes sense. I guess, are you able to pass along higher costs to your customers? Or maybe, in some cases, some items, yes; on some items, no? It's been a long time. I mean, there's been this type of constraint in the networking community, and I'm not sure that the customers are necessarily all that set up for these types of pricing negotiations.

Right. And I want to just reiterate, it's broader than the networking industry.

Of course. Of course.

Base-level commodities that are going in and kind of moving up in the pricing of semiconductors and following. So it is a different environment that I think we've operated in for a long time. In terms of passing along these increases, right, there's many levers that we have that we can do something around expedites or changing discounts or price increases. I think what we've been clear about, and maybe this caused some confusion, is we're not looking at this disruption in any way to increase our margin profile. We stay committed to the 63%, 65% range that Ita and Jayshree have talked about operating the business at. And what we'll do is doubling down on our cost-reduction activity and then passing along increases that we feel are more sustainable in pricing that will kind of get through this COVID period but look to be more of an ongoing cost structure, and that gets passed along to the customers.

Okay. How far into this do you think we are, John? And are we -- what are the signals that we should look for, for the supply chain constraints easing?

I think that's probably the hardest question you're going to ask me this morning. I kind of come back to the fact that this whole thing started because people were getting sick and couldn't come to work, and that causes a disruption in supply. So I think there's people that kind of can predict the return to a more normal basis than me, but I think we're far from over that. The vaccination rates are starting to pick up in the rest of the world but far behind where we've been in the U.S. So that has some while before we see that transition, and I don't know how long that goes on. But new patterns are starting to emerge on how you deliver freight and getting it baked into the system. And new capacity is starting to come online or emerging. So I think we're signaling that we're prepared to continue to operate in a similar environment clearly through 2022 and, if necessary, to 2023.

It doesn't sound like things are necessarily getting incrementally worse at the moment. If the freight patterns are solidifying and new capacities coming online, that seems like it's a little bit of a -- incremental positive.

I would say on the people-related pieces, either folks have figured out how to do social distancing and operate their factories and have learned the procedures and things that you have to do to deal with COVID on one hand. I think the thing that's changed in the back half of last year was the emergence of the semiconductor constraints which has as much to do with demand as it has to do with capacity and particularly on some of the older process nodes that may be more constrained.

Yes, which, I guess, for 400 would be less of a concern for you all, John, right?

Well, there's a lot of things that go into making a 400-gig switch. Some of them are the very new stuff, and some of them are little tiny old stuff. So you have to worry about all the ingredients to deliver...

The chip that controls the fan.

Right, right. There's all those little bits and pieces have to come together at the same time to deliver it to a customer.

All right. Well, I guess on -- without putting words in your mouth, I guess I would say that it sounds like some parts are getting better, and some parts may not be yet, but at least let's not worry too much about things getting that much worse, barring change in COVID patterns or another port shutdown or whatever.

Yes. I think this is the environment we're in, and we're in it for a little while.

Okay. All right. What does it look like, John, when the supply chain constraints -- is this something that happens slowly? Or does it just unwind all at once? I mean, I feel like our lead times in January went from like, I don't know, 1.5 quarters, 2 quarters or whatever to like a year and like a month.

Right.

And so I would imagine if they unwind that quickly, that would cause some commotion in the supply chain. I just want to be ready for that.

Yes. I think there's many moving parts on these constraints, and maybe they unwind in different pieces, maybe different market segments. Demand starts to ease up. Automotive or consumer, who knows which one of them, starts to ease up and free up some capacity. So things might change at different times. I think people will hold on to the longer lead times longer, and you'll see more ability to expedite and to bring in devices and obtain devices that right now might be harder, so it's difficult to predict. But I think you see more of a phasing over a few quarters of some easing up of different segments of this as a first indicator.

All right. Fingers crossed. Let's move on to talking about software a little bit. One of the frustrations for me in covering Arista is that I feel like very few people actually buy products, primarily on the box, right? It's just been a software company since its inception 15 years ago or whatever.

Correct.

So I'm hesitant to say, well, let's talk about software. What I really mean is let's talk about software-only sales.

For sure.

And I guess that is something that is becoming a greater focus across the industry, including for Arista. Where are you in thinking about how your customers want to consume the software that you've written? Is it still in those chunk purchases? Or is there more demand for an as-a-service type model than there once was?

Yes. I think if you're delivering the service as a service, that makes a lot of sense. And you've seen us kind of extend the software portfolio. We have a core EOS, which operates on-premise, on the device, on the switch, or if you're using the device as a router, it sits there. We have on EOS a license, if you want to extend it routing features, and that's offered as a license. Then we have -- starting with the CloudVision WiFi that was part of the Mojo acquisition. We have a cloud-based WiFi network that we sell as a service; CloudVision as a service itself, our homegrown management stack for operating the wired network, if you will, that ran on-premise now available as a service; and then Awake. So we're broadening the portfolio for more subscription services. But in terms of EOS, our view is that the people wanted to procure that along with their hardware, and the expectation is that would run. And if you have a maintenance contract, you'd be able to get new revisions of EOS and download it, either for new features or functionality or service and security patches that might come along. We don't see that model changing to a subscription model. It's very clear to us that customer preference there is if they're buying a device and running it in-house, their expectation is they get a license to do that, and they're happy to move to a maintenance contract to get upgrades. But we're not really adding value to delivering that service once we deliver it on-premise. Unlike what we're doing with CloudVision WiFi, we're actually operating a service. It's operating in the cloud. there are certain costs associated with that, and people are willing and happy to pay a subscription service. That makes sense to you?

Yes. No. That makes total sense. I'm thinking about the analogy of trying to buy a PC and then paying monthly for Windows.

That's all right.

Okay. One of the things that you've hinted at before, you mentioned 5G. And John, service provider is a really tricky thing, as you know better than most. And Arista had its -- had some energy a few years ago going into that market and if it's petered out for a bit. And I guess I'm wondering is this finally the moment for Arista in service provider? Or is it going to continue to be hand-to-hand combat?

The -- I'm not sure there's a moment in service provider. We got out of the gate with some really strong wins. And based on the premise that you can take a switch, it has large routing tables, you can start to deploy that as a router, I would classify them looking backwards in hindsight more of greenfield opportunities that maybe didn't pay much homage to the legacy infrastructure and those deployments and not highly replicable. We've also said and I think because we haven't seen that demonstrated in the numbers that we're continuing to invest in service provider. We see this as kind of the third leg of the stool, and the team has gone and invested significantly on features that tie in this next-generation routing technology into existing MPLS and segment routing-based cores and built out a number of use cases and have continued to extend that, along with the most recent announcement that we made.

Sadly, I think we have to hold it there, unless there are a couple last points that you would like to make, hey, you, investors don't really understand this, and it really irritates me on the conference call that nobody ever asked about that.

Well, I think just since we have a few seconds left, I'll just keep it pretty high level. But I think that we have gone from specifically cloud customer, cloud focused, taking those principles into enterprise and service provider and extending them with new use cases around the campus or in the light of the routing announcement to the edge, which I think is going to be very relevant as people start to think about edge computing and low-latency connectivity into premises in the future.

Well, I look forward to ongoing discussions about that, including at the upcoming Analyst Day. Hopefully, that will be more than virtual. We'll see. John, thanks very much for your time. As ever, a stirring discussion. And Charles, I know you're looking out there, so thank you for your efforts as well for making this happen. Okay. Thank you, everyone. We'll talk soon. Bye.

Thanks Jeff. Bye-bye.