Welcome to the Third Quarter 2022 Arista Networks Financial Results Earnings Conference Call. During the call, all participants will be in a listen-only mode. After the presentation, we will conduct a question and answer session. Instructions will be provided at that time. If at any time during the conference you need to reach an operator, please press the star followed by zero. As a reminder, this conference is being recorded and will be available for replay from the investor relations section at the Arista website following this call. Ms. Liz Stine, Arista's Director of Investor Relations, you may begin.
Thank you, operator. Good afternoon, everyone, and thank you for joining us. With me on today's call are Jayshree Ullal, Arista Networks President and Chief Executive Officer, and Ita Brennan, Arista's Chief Financial Officer. This afternoon, Arista Networks issued a press release announcing the results for its fiscal third quarter ending September thirtieth, 2022. If you would like a copy of the release, you can access it online at our website.
During the course of this conference call, Arista Networks management will make forward-looking statements, including those relating to our financial outlook for the fourth quarter of the 2022 fiscal year, longer-term financial outlooks for 2022 and beyond, our total addressable market and strategy for addressing these market opportunities, supply chain constraints, component costs, manufacturing capacity, inventory purchases, and inflationary pressures on our business, the potential impact of COVID-19, extended lead times, product innovations, and the benefits of acquisitions, which are subject to the risks and uncertainties that we discuss in detail in our documents filed with the SEC, specifically in our most recent Form 10-Q and Form 10-K, and which could cause actual results to differ materially from those anticipated by these statements. These forward-looking statements apply as of today, and you should not rely on them as representing our views in the future.
We undertake no obligation to update these statements after this call. Also, please note that certain financial measures we use on this call are expressed on a non-GAAP basis and have been adjusted to exclude certain charges. We have provided reconciliations of these non-GAAP financial measures to GAAP financial measures in our earnings press release. With that, I will turn the call over to Jayshree.
Thank you, Liz. Thank you everyone for joining us this afternoon for our third quarter 2022 earnings call. We delivered record revenues of $1.17 billion for the quarter with a non-GAAP earnings per share of $1.25. Services and software support renewals contributed approximately 16.3% of the revenue. Our non-GAAP gross margins of 61.2% was pressured by escalated costs due to supply chain as well as a substantially higher Cloud Titan mix. We expect some of these trends to continue into 2023. Cloud Titan was our largest vertical, followed by Enterprise and then Cloud Specialty Providers, Financial, and finally Service Providers. International contribution was at 17% with the Americas at 83%, once again reflecting stronger cloud customer influence.
Our Q3 2022 results reinforce Arista's customer relevance in both Cloud Titan and Specialty Cloud Providers. I would like to invite Anshul Sadana, our Chief Operating Officer and Cloud Tsar, to shed some light on this.
Thank you, Jayshree. Our partnership with the Cloud Titans and specialty providers keeps getting stronger. While supply chain got most of the attention for the last two years, we have achieved several technical milestones with our customers during this period, taking our products from labs to pilot to high-volume deployments. We are successfully deployed in more use cases than before, including the leaf-spine cluster designs, data center-wide spines, data center interconnects, backbone, WAN, and edge. Our customers have a deep appreciation of our expertise and execution. The Titans, starting with Microsoft and Meta, are continuing with their 400 Gbps journey. They're using our latest fixed and buffer modular 400 Gbps switches in these new deployments. We are continuing to expand our use cases with the other Titans too. While they are relatively smaller, these partnerships are healthy and continuing to grow as well.
The specialty cloud providers are continuing to grow too. They have very similar leaf-spine designs, but with smaller clusters. They too have the same DCI and regional designs, just fewer ports. We are in a competitive industry, and our customers like some diversity. However, we are very well positioned to maintain and grow our share given our quality, execution, and strong partnerships. Back to you, Jayshree.
Thanks, Anshul. We are experiencing one of our best ever Cloud Titan growth and revenue this year since IPO. Therefore, we expect north of 45% contribution in 2022 from this category with a diversified product portfolio and use cases, as you heard from Anshul. Speaking of new products, Arista introduced the next phase of routing this quarter in Q3. Arista's cloud-grade routing platform, powered by Extensible Operating System EOS. And our network data lake, NetDL, is built on a consistent architectural foundation for multiple routing use cases such as peering, content delivery networks, cloud connect, enterprise edge, mobile and metro edge, as well as carrier core. We have successfully transformed legacy routers to modern routing.
For our customers securing data in transit, Arista's innovative TunnelSec technology provides inline encryption at wire speed from 10 Gbps all the way up to 400 Gbps, eliminating the performance bottleneck of traditional encryption. Arista's latest R3 Series routing portfolio continues to deliver for our adjacent market sector that we began six years ago. As I mentioned previously, our customer momentum continues. Our million-dollar logos have doubled in the last three years, and Arista's market share in 100, 200, and 400 Gbps ports climbs, placing us at a number one market leader position, according to many industry analysts. Let me illustrate with a few customer wins this year. The recurring theme to remember is that Arista is diversifying its business with many use cases such as data center, campus, routing, cloud, and edge-based, all on cloud principles.
Our first enterprise win highlights the ever-growing strength of a single solution for diverse use cases in the data center, in the campus, and in data center interconnect routing. Based on industry standards and monitoring capabilities with Arista's DANZ Monitoring Fabric, or DMF, Arista was uniquely positioned to provide a single EOS binary code version using CloudVision and our NetDL stack. The next is an International Americas win in the financial sector, providing secure data center solutions. Arista's competitive advantage was twofold. First, our single EOS stack provided reassurance that they could handle security patches, bug patches, new software code upgrades with zero downtime, much better than our peers in the industry. Also, we can do much better automation. Traditionally, it used to take customers several months, if not years, to go live with a single data center with their incumbent vendor.
Arista's programmatic APIs, continuous integration validation with DevOps integration was deployed across multiple data centers within just a few days with no human intervention or errors, offering the lowest total cost of ownership. Our third win demonstrates our strength in campus in the healthcare sector. This customer was looking to upgrade its legacy infrastructure in multiple hospital locations and moving more to a cloud-operated model. Arista provided that cloud-managed network offering across the data center campus and cognitive Wi-Fi and also WAN edge and internet edge. The single pane of glass offers real-time network as a service with visibility, health checks, troubleshooting, and provisioning to the client, all the way to campus and to the cloud infrastructure. The next international win shows our continuing presence in the public sector.
This customer wanted to move to 100 Gb Ethernet using Arista's spine and spine platforms to connect into incumbent DWDM edge and load balancers for massive route scale. Arista's advantage was the lowest software security vulnerabilities over a nine-year period compared to alternatives that were significantly higher. This, by the way, also applied to Arista's cognitive Wi-Fi with zero vulnerabilities. Our final win emphasizes Arista's ability to provide encryption as a part of a secure routing solution. This was an international tier two service provider, whereby Arista delivered full line rate data in transit for the customer's remote sites. This bespoke design encompasses layer three virtual private network, VPN, over VXLAN routing solutions with macro segmentation based on access lists and direct flow rules, all managed by CloudVision.
Overall, you can see that Arista continues to gain customer relevance and market share in building our vision for a client to cloud network. The promise of not only delivering on superior technology, but the ability to help our customers realize the operational benefits via world-class quality, support, services is a real compelling advantage. We look forward to sharing more of our vision and strategy at our Analyst Day this Thursday, November 3rd, 2022. Now, I will turn over to Ita for our financial specifics.
Thanks, Jayshree, and good afternoon. This analysis of our Q3 results and our guidance for Q4 2022 is based on non-GAAP and excludes all non-cash stock-based compensation impacts, certain acquisition-related charges, and other non-recurring items. A full reconciliation of our selected GAAP to non-GAAP results is provided in our earnings release. Total revenues in Q3 were $1.177 billion, up 57.2% year-over-year and well above the upper end of our guidance range of $1.025 billion-$1.075 billion. Overall demand for our products remains healthy across all areas of the business. Supply challenges continued throughout the quarter, with ongoing supplier decommits constraining shipments and requiring higher cost broker purchases and expedite fees.
Services and subscription software contributed approximately 16.3% of revenue in the third quarter, down from 17.6% in Q2. This largely reflected accelerated growth in product revenues, while services and software continued to grow on a more consistent basis. International revenues for the quarter came in at $199.1 million, or 17% of total revenue, down from 20% in the second quarter. This reflected strength in U.S. revenues in the period, and particularly with our larger cloud customers. Overall gross margin in Q3 was 61.2%, just above the midpoint of our guidance range of 60%-62%. As previously discussed, our current lower gross margin ranges reflect a healthy cloud mix and higher levels of broker component sourcing and expedite fees.
Operating expenses for the period were mostly flat to last quarter at $227.7 million or 19.3% of revenue. R&D spending came in at $150.1 million, or 12.8% of revenue, up from last quarter at $148 million. This primarily reflected increased headcount costs in the period, with slightly lower new product introduction costs. Sales and marketing expense was $62.8 million, or 5.3% of revenue, and G&A costs were $14.7 million, or 1.3% of revenue, both mostly consistent with last quarter. Our operating income for the quarter was $492.1 million or 41.8% of revenue.
Other income and expense for the quarter was a favorable $6.1 million, and our effective tax rate was approximately 21.3%. This resulted in net income for the quarter of $391.9 million, or 33.3% of revenue. Our diluted share number was 314.4 million shares, resulting in a diluted earnings per share number for the quarter of $1.25, up approximately 69% from the prior year. Now turning to the balance sheet. Cash, cash equivalents, and investments ended the quarter at approximately $2.98 billion. We repurchased $47.6 million of our common stock during the third quarter at an average price of approximately $99 per share.
As a reminder, we've now repurchased approximately $740 million or 7 million shares against our October 2021 billion-dollar board authorization. The actual timing and amount of future repurchases will be dependent on market and business conditions, business requirements, stock price, acquisition opportunities, and other factors. Now turning to our operating cash performance for the third quarter. We generated $134.1 million of cash from operations in the quarter, reflecting strong earnings performance, somewhat offset by increased working capital investments. Increases in inventory and other assets are mainly driven by receipt of components for future shipments, including shipments delayed due to supplier decommits. This trend should reverse once overall supply conditions for these decommitted components improve. DSO came in at 51 days, flat to last quarter. Inventory turns were 1.7 times, down from 1.9 times in the prior quarter.
Inventory increased to $1.1 billion in the quarter, up from $852.8 million in the prior period, reflecting higher component and peripherals inventory and an increase in switch-related finished goods in transit. Our purchase commitment number for the quarter was $4.3 billion, down from $4.5 billion in Q2. These multi-year purchase commitments reflect overall strength in demand and the current long lead time supply environment. As a reminder, we continue to prioritize newer early life cycle products for inclusion in these strategies in order to help mitigate the risk of excess or obsolescence. Our total deferred revenue balance was $941 million, down from $1 billion in Q2. The majority of the deferred revenue balance is services related and directly linked to the timing and term of service contracts, which can vary on a quarter-by-quarter basis.
Approximately $165 million of the balance, down from $228 million last quarter, represents product deferred revenue, largely related to customer-specific acceptance clauses for new products with our larger customers. Accounts payable days were 56 days, down from 63 days in Q2, reflecting the timing of inventory receipts and payments. Capital expenditures for the quarter were $10.4 million. Now turning to our outlook for the fourth quarter. We came into the year calling for 30% year-over-year revenue growth, somewhat balanced across our market sectors and heavily constrained by supply. We now expect revenue growth for the year of approximately 45% at the midpoint of our Q4 guidance, with a healthy contribution coming from our Cloud Titan customers in what has remained a stubbornly constrained supply environment.
Our Cloud Titan customers are now expected to account for approximately 45% of our revenue for the year, with major contributions from Meta and Microsoft. Demand from our enterprise and provider businesses has also been strong, exceeding our original expectations from a demand perspective, but with revenue somewhat constrained by supply. We expect gross margin pressure to continue in the quarter, with some continued need for broker purchases and expedite fees in response to ad hoc supplier decommits, combined with a healthy cloud mix. Even allowing for the lower gross margins, we expect healthy operating margins for the quarter, again demonstrating the resiliency of the business model, with significant bottom line flow-through from increased revenue scale. With all of this as a backdrop, our guidance for the fourth quarter, which is based on non-GAAP results and excludes any non-cash stock-based compensation impacts and other non-recurring items, is as follows.
Revenues of approximately $1.175 billion-$1.2 billion. Gross margin of approximately 60%-62%. Operating margin of approximately 40%. Our effective tax rate is expected to be 21.5% with diluted shares on a post-split basis of approximately 316 million shares. I will now turn the call back to Liz. Liz?
Thank you, Ita. We will now move to the Q&A portion of the Arista earnings call. To allow for greater participation, I'd like to request that everyone please limit themselves to a single question. Thank you for your understanding. Operator, take it away.
We will now begin the Q&A portion of the Arista earnings call. In order to ask a question during this time, simply press star then the number one on your telephone keypad. If you would like to withdraw your question, again, press the star one. We ask that you pick up your handset before asking your question in order to ensure optimal sound quality. Your first question today comes from the line of Tim Long with Barclays. Your line is now open.
Thank you. I was hoping you could just talk a little bit about kind of the enterprise and campus strategy. Obviously, you guys have been growing pretty well there. If you could kinda update the, you know, the $400 million bogey for the year. You know, you've mentioned,
Oh, we lost him. Tim, are you there?
The rest of that business as we're looking for next year, and
Can you hear audio? Tim, you are
$200 million
You're sounding muffled.
Just more in the-
I think I'll interpret your question, Tim, even though I'd only heard half of it, which is what is your status on campus and enterprise strategy, and how are you doing to the $400 million target? Obviously, we'll get back to you when we've completed the year. At this point, we would say that the strength of our campus business from an orders and perspective is very strong, and we still expect to hit $400 million. We are very supply constrained, so we're not yet sure if we will hit the $400 million from a revenue and shipment perspective.
I'm sorry, if you could just touch on the wireless LAN contribution. It sounds like that's doing a little bit better now. Sorry about that, I was muffled a little.
We are very pleased with our wireless LAN contribution. If you think back, we acquired Mojo Networks in late 2018, early 2019, and it was a very small single-digit business. We expect the wireless LAN contribution to become a triple-digit business next year.
Thank you, Tim.
Okay, thank you.
Thanks, Tim.
Your next question comes from the line of Alex Henderson with Needham. Your line is now open.
Great. Thank you very much. Spectacular results for the quarter, for the year, spectacular results last year. I guess the question really boils down to, you know, are we going to see a reversion to more normalized growth? If that happens, does that bring the margins back in to the 37% range? Have we set ourselves up against extremely difficult comps as we go into 2023? How do we think about the mechanics of where we are? Alternatively, clearly, supply constraints persist. You sound more constrained than you had earlier, and that would imply to me that at least for the first half of next year or even longer, you might still be in the strong growth trajectory.
Can you just address how we should be thinking about the environment as we move forward? Thanks.
Alex, thank you for the kind words, and I'll tee it up and certainly let Ita speak to some of the details you are asking about. First of all, I just wanna say I'm very proud of the team for executing. As you know, we came into the year expecting 30% growth, and here we are telling you it's gonna be 45%. A huge contribution from Anshul and the team, not only on the Cloud Titans, but also on the enterprise and rest of the cloud business, as we call it, led by Krishnamurthy and Ashwin. A lot to be proud of, especially as the numbers get larger. As you think about the numbers getting larger, it's going to be difficult to sustain 45% growth every year.
I'd love to have it, but as you know, Arista is a volatile business, and you have to think of us across a three- to five-year CAGR, not just on an annual basis. I still think we have a large TAM, we're gonna do very well. But we, you know, with the looming uncertainty of recession and CapEx spend, et cetera, it's definitely difficult to predict beyond a year. Having said that, I think, as I said before, we're having one of our best ever years in 2022. Even though the comps will be difficult, we feel confident that if supply chain gives us some relief, we can do well in 2023 as well. Ita, would you like to add anything?
Yeah. I think, Alex, if you go back to the business model, you know, obviously the gross margin kind of variability that we see that's kinda inherent to the business model is to do with the mix of the business, right? That kind of works very nicely when you see accelerated cloud. You see that flow through to the operating margin line, even though it might be impacting the gross margin in that time period, right? You know, as we talk next Thursday, we'll talk some more about kinda what the growth rates look like for next year and kinda how that flows through the model. There's no doubt that when we accelerate growth like this, it's hard for us to accelerate spending at the same rate, and you never really kinda catch up for that, right?
It's one to be proud of.
Thanks so much.
Thanks, Alex.
Your next question comes from the line of Paul Silverstein with Cowen. Your line is now open.
I guess I shouldn't ask you what everybody wants to know, which is what's your outlook for next year?
Give it a couple more days, Paul. Just a couple more days.
All right. I had to ask it. Everybody else is asking. It sure doesn't sound like from the numbers that macro had much, if any, impact. F5 last week did identify macro as a meaningful driver of a very disappointing outlook. Any thoughts you can share in terms of what you're seeing? Is it affecting one iota customer decisions as to your products and solutions? Any thoughts would be appreciated.
Look, you know, I don't think of Arista as a bellwether for macro, so we let the pundits and economists speak about that. I think it's fair to say that data center spend has been very strong, and we're confident of our near-term strength both in Cloud Titans, Enterprise, and Campus. If there's a place that we see some softening at all geographically, I would point to Europe. They've had the effect of the post-pandemic, the energy crisis, the war, the Brexit. It's just a whole lot of things going on there. Our numbers are small, and even there, I would say it's more than enough to execute better. Overall, macro is not yet an issue for us. We'll keep a vigilant eye on it, but so far, so good.
Jayshree, just to be clear, you're not seeing elongated sales cycles. You're not seeing cutbacks, deferrals, or cancellations to any appreciable extent in orders, especially from Enterprise. I appreciate that you're small, but your Enterprise is not nothing. It's a growing business. Correct me if I'm wrong, but historically, in times of economic duress, most organizations have declined to change vendors or put off changes in their infrastructure that would result in bringing in a new vendor like yourself.
Right. I don't think Arista is any more a new vendor. I think we are the best-of-breed vendor. What we see, especially in networking, unlike maybe.
Non-incumbent. Non-incumbent.
Yeah, fair point. We're not a legacy vendor. That's fair. I would say, if anything, Paul, just to answer your question more directly, unlike server or storage, networking is a very, very strategic spend. There's a long qualification cycle that goes in. There's proof of concept labs. There's intense testing for scale. We're all struggling with lead time, so switching vendors doesn't give you any particular advantage. If you're looking for best of breed and best operational advantages, our customers would prefer to wait for us. Of course, we don't wanna test their patience. We're gonna do everything we can to execute. You're seeing that in our purchase commitments, but we gotta execute even better.
All right. Appreciate the response. Thank you.
Thank you.
Thanks, Paul.
Your next question comes from the line of Simon Leopold with Raymond James. Your line is now open.
Thanks for taking the question. I wanted to get a little bit of help understanding how to think about the purchase order commitments. Just looking at the prior 10-Q filing, you had a 12-month commitment of, like, $2.9 billion, and you mentioned today that the overall commitment came down a little bit to $4.3 billion. But what I'm struggling with is if we just sort of try to back into implied revenue, it's well above what most of the street is modeling for the next year. And there are a number of ways we could think about it. You could either write off inventory, you could hold higher inventory levels, or certainly you could have much higher revenue than we expect. I'm wondering if you could give us some advice on how to interpret these purchase order commitments.
Thank you.
Yeah, Simon, I think the, obviously, the goal is to make the purchase commitments that we need to support the business, but it is multi-year, right? We are thinking about this kind of longer term than just, you know, the year that's ahead, right? That's why you're seeing the numbers kind of, you know, not map perfectly to whatever your revenue assumption is for just 2023, right? We'll, again, we'll talk about 2023 and the growth rates when we get to the Analyst Day. Then maybe you can ask this question again, and we'll try to provide some more guidance. It really is about being multi-year, and some of these products are early-stage products where we believe we can, you know, we can put a, so that there's a limited risk in terms of, you know, E&O in reserves and really you're trying up cash, but that's kind of a, that's worth it in terms of being able to have the products and have the components in place when we need them.
Thank you.
Thank you, Simon.
Your next question comes from the line of Meta Marshall with Morgan Stanley. Your line is now open.
Great. Thanks. Ita and Jayshree, you know, maybe if you could just comment on whether some of the upside that you're seeing from the Cloud Titans this year is due to kind of greater ability to serve them and get them more product than they wanted or if you're actually getting kind of stronger orders than expected and just kind of on the deferred revenue being recognized if there are projects that are kind of changing in timing that led to kind of the stronger than expected results. Thanks.
Thanks, Meta . So, you know, I think it's fair to say, as Anshul has often said, that our relationship and partnership with the Cloud Titans continues to be strong and couldn't be better, especially with Microsoft and Meta, who we fully expect to be greater than 10% concentration. I wish we could ship them more, and I think if they were hearing this call, they would be saying, "I wish you could ship us more too." I don't think it is a case of shipping them more. I think it's a case of demand from them and a very strategic partnership at the engineering level, at the network design level, at the operational level, at the procurement level. Actually, Anshul, I'd like you to, since you're dealing with this daily, say some more.
Absolutely. Meta, the demand from these customers is just strong. Yes, there's some timing of revenue and deferred and so on, but that's not what's leading into this trend. The raw demand from these customers continues to be strong. They wanna do more build-outs.
If you look at how many data centers and regions they want to build out, it's very strong. If you look at their DCI ads with 400 Gb and 100 km type of distances, building big regions, that's very, very strong. They love our execution. As Jayshree mentioned, if we could get them more product, they would happily take it as well. All in all, a phenomenal outcome compared to where we were two years ago, but in a very healthy cycle with our cloud customers.
Matt, on the deferred, I mean, it's a nothing for the year and outlook, right? If you look at it year-to-date, the reason the deferred is not a contributor at all at this point.
Great. Thanks.
Thanks.
Your next question comes from the line of Ben Bollin with Cleveland Research. Your line is now open.
Good afternoon, everyone. Thanks for taking the question. I was hoping we could talk a little bit about just the broader supply and demand balance. I'm sure we'd all, you included, would love to know when this ends. I'm curious how you're thinking about it evolving, you know, from a supply perspective, when you think it, you know, maybe gets directionally better or when it's really better. Also curious how it's influencing the duration of visibility you're seeing from customers. That has gotten longer. Is it still getting longer? Is it starting to shrink? What are you seeing there? That's it for me. Thanks.
Okay. Hey, Ben. I you asked really two questions. What's the direction of our visibility? Is it getting longer? Generally right now, because of our lead times, our visibility is approximately six months to a year. We've never had that kind of visibility, particularly from our cloud providers. Directionally, it's not getting longer, but staying about the same is what I tell you. Then what are we seeing on supply chain, and when is it gonna get better? Gosh, one of the reasons that Ita, Anshul, and myself made such a concerted effort, we got board approval for very large purchase commitments, multi-year, despite a smaller revenue, is exactly not to be in the spot we are, which is we wanted to get all our components.
Sadly, we have almost all the components except a handful, and you can't build a system without the last few components. We do have still some severe shortages. As I said before, we expect this to go into 2023, and I think our long lead times will persist in the industry for networking, especially, for the next several months and for 2023. We don't see a lot of relief in sight until we can get all the components. Being short of a few components still means we can't build a system. Expect us to really only see improvement a year from now.
Thanks, Jayshree.
Thanks, Ben.
Your next question comes from the line of Matt Niknam with Deutsche Bank. Your line is now open.
Hey, thank you so much for taking the question. On the guide for the non-GAAP op income margin, I think you did 42-ish% this quarter. I know the guide talks about a nearly 200 basis point step down despite stable growth margin. I'm just wondering if you can talk about where you see opportunity for incremental investment. Then if I could just sneak in one follow-up on the deferred revenue question. It looked like deferred revenue dipped about $92 million sequentially. I'm just wondering what's baked into the revenue outlook for 4Q from a deferred perspective. Thanks.
Yeah. On the deferred, the product deferred decline in Q3 was about $60 million. There's a piece of it that's just services, service contract timing and stuff in there. Really the thing that's kind of directly linked to the revenue is about $60 million. For Q4 in the guide, we're assuming no deferred, right? No deferred impact, right? That's that question. The other question was, o h, the expensing.
Yeah, the expenses and operating margin. I think, look, we were flat pretty much on OpEx in Q3. You know, some of that is timing. You know, that wasn't the original plan, obviously. Some of that will recover and some of as we head into Q4, we'll see, you know, some sales commissions and other things at the back end of the year tend to be a little higher as well, right? It's kind of a combination of maybe some timing and then just the normal kind of Q4 dynamics.
Thank you very much.
Thank you.
Thanks, Matt. Welcome to your first call.
Your next question comes from the line of Fahad Najam with Loop Capital. Your line is now open.
Thank you for taking my question. Jayshree, I think at a recent investor conference, you said that you think the industry can grow double-digit revenue. If I recall, you said you were connecting clouds or data centers. I wanna further understand your response to a previous question that if the supply chain improved next year, you would do well. Well versus what? That double-digit outlook that you had earlier, is that what you're thinking? Just can you help us understand?
Yes, Fahad, I'll try. I think in terms of demand, we fully expect to grow double digits next year, and I hope for years to come, unless there's some real macro issues. Specific to next year, again, we will guide specifically at Analyst Day. I think, if supply chain were to be relieved, we would grow double digits. If supply chain continues to be constrained, we'll still grow double digits. Either way, I'm seeing us do well next year.
I have one follow-up.
The question of how well.
What's the risk of your cloud customers having some level of inventory build up? I understand that there is a significant constraint on networking equipment, but there has been evidence of i nventory mismatches in the server and storage space as some other cloud titans. What's the risk that there is maybe?
Yeah.
a mismatch
Very little, Fahad. I think it's important to understand that, nobody's been in a position to have any kind of inventory on networking. Anshul, you wanna add to that?
Fahad, our customers will consume everything we can ship. We are not worried about inventory build-outs anytime soon.
Appreciate it. Thank you.
Thanks, Fahad.
Your next question comes from the line of Ittai Kidron with Oppenheimer. Your line is now open.
Thanks ladies, and great quarter. I guess a couple of questions. Jayshree, you talked about your activity with the cloud guys and, you know, how the cloud guys do want diversity in suppliers. Of course, there's always timing in projects where you kinda run ahead or below where you probably should be, just given the timing of projects. Do you get a sense right now on if you're punching above or below your weight kinda in, and I'm calling weight whatever the cloud guys really think you should be at long term from a share standpoint?
Hey, welcome back, Ittai. We haven't chatted in a long time. That's a really good question. You had to ask a tough one. I do think we're punching at our weight. In terms of demand, even above our weight. In terms of actually supplying, maybe below our weight. By the way, I'm only talking about the cloud guys, not myself. Anshul, you wanna add to that?
Well, I'm not here to ref a boxing match, but in terms of supply and overall demand, Ittai, customers like to diversify. I think most customers have already set their plans. We're not expecting any big dramatic changes from here on. We talked about for a couple of years now that customers truly like our product. They like our execution. The partnership is healthy. Even worldwide, because of claims from many companies on how they'll be ahead, and they'll beat us, and all of that, in the end, we've done really well. If someone takes a share in one of the roles, we take share in other roles, too.
As a result, because of the expansion in WAN, and edge, and routing, and other use cases, we've done well, and our expectation is we'll actually continue to do well going forward as well.
Got it. Okay. Maybe as a follow-up, Jayshree, I mean, you did talk about the weakness in Europe, and it's obvious that there should be weakness over there, but your business over there is down on a year-over-year basis. I guess I'm kinda wondering, the traction and the share gains that you're having in the US, one would think that even in the current environment, if you're able to get share gains in Europe as well, you'd still be able to eke growth out of there. Maybe not 30%-40%, but perhaps a 10%-15%. Help me understand, what. Maybe you could go a layer deeper into what's going on in Europe.
Do you think that the current environment, maybe going back to a previous question, Paul's question, in this environment, perhaps customers are just less likely to consider change? They'll just stick with what they've got, and maybe that's a hindering element.
I was answering the question, Ittai, more on is there a recession. In a broad sense, if at all, we're generally seeing very strong demand worldwide. If at all, we are seeing some softness and weakness, it would be in parts of Europe because of all that they're going through. Specific to Arista, I see no reason we shouldn't be growing everywhere, including Europe, but our rate of growth in Europe may be slower for exactly the reasons you mentioned, that their overall economy. If you look at the GDP, and if you look at the top three countries, you know, Germany, U.K., and France, none of them are really substantially investing. I think it's because we have small numbers, we can do fine.
We should be able to still grow double digits, but that is the only sign of recession we're so far seeing.
Got it. Very good. Good luck. Thanks.
By the way, Ittai, just to correct you, we are growing year-over-year in Europe as well.
Yeah.
Okay. All right.
I think the geo numbers, Ittai, have a lot of cloud influence in them, so it depends. When you strip that out, the kind of in-region business is growing. Yeah.
Okay. Well, maybe, Ittai, on that point, it was, did the U.S. grow? Can you tell us how much the U.S. grew without the Titans?
Off the top of my head, no. Yeah.
You have to do it by your end.
Yeah. Yeah. We'll show you some stuff at your end for sure.
All right.
On the verticals of the business.
Appreciate it. Good luck. Thanks.
Thank you.
Your next question comes from the line of David Vogt with UBS. Your line is now open.
Thanks guys for taking the question. I just wanna go back to the margin, performance. If I go back to 2019, I think your Titan mix was relatively similar to where we are today, maybe a little bit less. Your product gross margins were almost 300 basis points higher. Can we kind of disaggregate that delta? How much of that is from expedited freight supply chain versus somewhat structurally lower margins with some of the new Titans and/or Microsoft and Meta? Then on OpEx, just a quick question. Growth has been 15%-20% sort of on a year-over-year basis. Is that kind of how we should think about the business as we progress over a couple of year cycle? Thanks.
Yeah. I think on the gross margin piece, I mean, there is more cloud mix in the business, I think, than we've seen.
Than we've ever seen.
Than we've seen.
This is the highest mix we've had.
We also have, remember, don't forget the expedite and fees and stuff we've talked about, this 200-300 basis points. I mean, that is still there. I think that will be there for a while. You know, certainly in our Q4 guide. I think it's a combination of those two, right? When you think, you know, back in that time period, you know, you could run the business for cost, right? We were optimizing cost in every way. Today, we're kind of hamstrung because of supply. It's inefficient right now, to be honest, right? Particularly with these decommits. We need to get out of that, and then we can kind of start to drive for some improvements on the gross margin side as well. I think on the OpEx side, you know, if you look back historically, we've probably been in the 22% growth type range on OpEx. You know, except in times of disruption, right? I mean, that's probably not a bad way to think about it.
Got it. I need to maybe just quick follow-up. It sounds like your structural margins on the Titans are relatively unchanged over the last, you know, call it two to three years outside of the expedite fee and supply chain. Is that a fair characterization on the product side?
I think it depends on the product mix to some degree, right? I mean, you've got, Yeah, you'd have to do more disaggregation to really get to what's happening there, right? The product mix is a part of that as well.
Okay. Thanks, guys.
Your next question comes from the line of Amit Daryanani with Evercore. Your line is now open.
Good. Thanks for taking my question. Congrats on a good quarter. You know, I guess the question to you on the Cloud Titan side, and on a broader level, a vertical. You know, we're seeing an increased focus on profitability from the cloud companies, and I'm just wondering, as they start to focus perhaps more of their resources on their own core competencies, could you see a lot of these cloud companies starting to shift from, you know, building their own solution on the networking spine side, which is starting to buy them more often? I'd love to understand, you know, A, from your existing customer base, are you seeing an expansion of engagement with them beyond what you've traditionally done?
B, more importantly, are you seeing newer customers come to you that want to start buying from you other than building it organically themselves?
Okay. I'll take it. Thank you, Amit. I'll take it in two parts. The Cloud Titans have always had a combination of build and buy. I don't think that's changing. They continue to look at whether it's SONiC or FBOSS working closely with us, but in some layers, they are absolutely sourcing their own products and technology. At the same time, I think the Cloud Titans have to run very mission-critical networks, and they recognize the value and the partnership at an engineering level to not build themselves, but really co-develop with us. These, I would say, are not build or buy. They're really co-developed partnerships where they get the best of both.
Despite all of their focus on cost, believe me, they're getting their best of both because they're getting great cost for us without making the direct investment themselves, and it really is a unique partnership. Anshul, you want to say some more to that?
Sure. Amit, in general, these cloud companies keep exploring every possible way to go faster. There's a misunderstanding in the entire, whether it's on the Wall Street side of things. Sometimes people have said, "Hey, people want to find something that's cheaper because they can save money." The reality is they have to look at their total overall business impact, and if they can go faster and get a competitive advantage, that's a huge benefit to them. Most of these companies do not build their own products or stacks just to build it cheaper. They build it because there's some secret sauce. There's some IP. There's some integration. That's what you're seeing with the co-development we do with Meta and Microsoft, whether it's on the FBOSS SONiC side or some of the hardware we develop as well.
The other Titans explore the same use cases or similar themes with us on an ongoing basis. These are not new discussions to us, and I think we'll continue to execute well. Clearly, the majority of the business continues to come from our top two customers. We are happy with that. The others are engaged, as we've talked to on previously. Those are long-running projects. By the time they get to a decision point and actually impact, you have to wait several years, and we'll get there. There's nothing that's changed in the industry. People are, in general, happy with the way status quo has been maintained with a little bit of, let's explore other options in working with companies like us where possible.
Thanks, Amit. Operator, we can take the next question.
Your next question comes from the line of Aaron Rakers with Wells Fargo. Your line is now open.
Yeah. Thanks for taking the question. Congratulations as well on the quarter. I wanted to ask a little bit more of a technical question. You know, as we think about 400 Gbps and you think about the traction you're seeing with your Cloud Titans as well as some of the specialty cloud guys, I'm just curious if you could level set us on where you think we're at in terms of the 400 Gbps cycle. And on that topic as well, you know, any updated thoughts on this idea of the AI fabric networks representing an incremental adjacent growth driver for your business at the Cloud Titans? Thank you.
Unquestionably, I think, while majority of the enterprise is still on 10 and 40, we are seeing more and more combinations of 100 and 400 Gbps, not only in the Cloud Titans, but also in the rest of the customer base. Just to level set on the numbers, you may recall that we had 70 or so customers in 2020. We doubled to 300 in 2021, and if I had to guess, I would say we'd double again in 2022 in terms of 400 Gbps customers. The important thing to remember here is they're not just 400. They really are a permutation and combination of multiple speeds, multiple use cases, and together they're an important contributor, but it's never an isolation 100.
Some of them are uplinks, some of them are native connections, some of them, as Anshul alluded, are DCI routing solutions. You know, our market share in this combination of 100, 400, and in some cases 200 is number one for good reason because they really have to work together with the right software and right management. The AI spine cluster is emerging as a new use case. It's still early days, and we've talked about it some. We'll talk about it more at the analyst day, but more use case emergence for, you know, applications that really push the speed and latency and performance and predictable bandwidth of 400 Gbps is going to add additional fuel to our 400 Gbps demand.
Yep. Thank you.
Thank you. Thanks, Aaron. Thank you so much. Best wishes.
Your next question comes from the line of Jim Suva with Citigroup. Your line is now open.
Thank you. Jayshree and Ita, I noticed in your press release that you made a comment to new market expansion. Could you elaborate a little bit, like, what is that or is that gonna be a focus for the event on Thursday? 'Cause I assume it's not just continual strength of where you're seeing or maybe it is, but if you can give any expansion on that, 'cause I know in the past or and currently you've seen a lot of expansion in the campus growth. I saw the new market expansion literally word for word, and I gotta think there was some purpose behind that. Thank you.
Maybe you read more into my quote than I meant. You know, at the end of the day, Arista is viewed as a data center company. If you look at the diversified portfolio, we're trying to say we're entering many more new markets like routing, like campus, and even subsets of the market, like securing our solutions better with encryption, with better visibility and observability. Indeed, we'll talk more about new market expansion in a couple of days. Undoubtedly, we're moving from becoming a pure data center company to many more markets that center the data in different locations.
Thank you so much.
Thank you, Jim.
Your next question comes from the line of Samik Chatterjee with JP Morgan. Your line is now open.
Yep. Thank you. Hi. Thanks for squeezing me in here. I guess it's a similar question to the last one, but, Jayshree, I just wanted to understand, as we look at your growth with cloud customers, how should we think about sort of that growth relative to sort of how much of that growth is coming from the additional use cases you're addressing beyond traditional switching? So for example, like DCI or routing, how much of that is probably contributing to the strong growth you're seeing with cloud customers? As we think about sort of the next couple of years, how should we think about given the visibility you have in terms of designs, how is that sort of momentum progressing from here on? Just a quick follow-up there.
Outside of Facebook and Meta, as we think about the engagement with other cloud companies, is that on some of the new use cases, or is that traditional switching and leaf and spine?
First, I think it's super important to understand that even though it's leaf and spines, Samik, these data centers, whether you're building them at a regional level or a mega scale level, you know, the leaf and spine is really going to expand and explode as the Cloud Titans expand their presence. That growth, that organic growth that we began our journey with them will continue, and we will depend on that growth. In addition to that, as Anshul said, we will have new use cases on the WAN, the regional, the DCI, and especially AI spine, particularly with the Cloud Titans. Those new use cases may also expand to other specialty cloud providers.
I would say we would be looking for sources of growth in both areas, expansion of the existing data centers into new locations and more scale, as well as new use cases.
Thank you.
Six of one and half a dozen of the other. Thanks, Samik.
Your next question comes from the line of George Notter with Jefferies. Your line is now open.
Hi, guys. Thanks very much. I guess I wanted to go back to the question of the Cloud Titan customers potentially building inventory. You know, I guess on an earlier question, Anshul, you were pretty dismissive of that idea. I guess I'm just wondering, you know, what additional detail you might kinda have that supports that idea. You know, if I look at your business, you've been growing, you know, product sales, you know, 55%, 67% year-over-year the last couple of quarters. It's pretty outstanding growth. You know, given the mix shift in the company, I would guess that the Cloud Titan customers have been growing 70% or 80% year-over-year, right? You know, I know you guys are building your own inventories.
Many companies are building their own inventory in the supply chain environment. I do think it is important question, and I guess I'm wondering exactly, you know, how you know customers aren't building inventories of your stuff. Thanks.
Yeah. George, that is a good question. I get asked that internally daily because we have to worry about what's going on on the other side, our customers buffering. I would say the best sign is the number of phone calls I get or Jayshree gets or others in the company get on when the hell are you shipping. These customers are still roughly hand to mouth. You know, there are a few places where they can still calibrate for the computer to come up and so on. In most cases, customers are desperate.
The deployment teams are waiting for the gear to show up, and they light it up as quickly as possible. You have to realize where we came from, right? Since the start of COVID, the world has been constrained, and that constraint over such a long period generally means people can't execute their project timelines, and they're still short. At some point, this will recover. We're nowhere close to that in today's time. I think it'll take a long time. Maybe as Jayshree was saying, if supply recovers by the end of 2023, I think that's when customers will feel comfortable, and they will eventually have an opportunity to buffer up, but not yet. Not there today.
Okay. Thank you.
Thanks, George.
Your next question comes from the line of Sami Badri with Credit Suisse. Your line is now open.
Hi. Thank you. I wanted to go to gross margins. I think, Ita, you've made two references on this call regarding just the trajectory of margins. I think you said that we're probably not going to see an improvement in gross margins for at least a year from now. Then the other comment you made regarding the 200-300 basis points, I think you said it's gonna be there for a while. Does that imply that w e should be modeling gross margins at least through maybe the first half of 2023, a little bit more in line to the last couple of quarters. That's the first question. Second question is, you know, if you were to give us an idea on how many ports or how many switches with 400 Gbps ports shipped in 2022, could you give us kind of like a percentage of total shipments and maybe what that would look like in 2023?
Yes. Just on the gross margin, I think given where we are today, I think it's not a bad idea to, you know, to model that gross margin as being kind of constrained until we get out from under the constraints, honestly. I think that's a very real kind of cost that we're carrying, and it's not clear yet exactly when that starts to improve. Yeah, I think it's not a bad idea to hold it there until we're able to kind of show real, demonstrable improvement and more predictability, I guess, on these decommitted parts.
Yeah. To answer your question, I don't have exact numbers on 400 Gbps , but if I had to guess, even though we're gaining customers, most of them are as uplinks, so the number of ports is still small. I would say our penetration is well under 10%, maybe even 5%. That's just a good guess.
Then thank you for answering gross margin and that piece. When we go into 2023, are you seeing almost like a sharp increase in that from sub-10% to obviously much higher than that, or not quite yet?
No, no. If we go into 2023, the way to think of this is, people are still migrating from 10 Gbps . Ten and 40- 100 will still be the largest migration. As a result of that migration, you will see more deployment of 400 Gbps . 400 Gbps will still be small. Maybe it'll go from 5% to 10%.
Perfect. All right. Thank you very much.
Thanks.
Thanks.
Your next question comes from the line of James Fish with Piper Sandler. Your line is now open.
Hey, ladies. Nice quarter and launches as well. While it's a metric, Ita, I know you don't really want to emphasize as much as, you know, others do in your space, you still have, you know, visibility here and confidence in that visibility. Maybe asking it a different way than others, as supply chain here starts to normalize, how are you guys thinking about how long that backlog actually takes to turn back to a more normal level? Is it four, six, kind of eight q uarters that's currently in your thought process for next year? Are you seeing any change in cancellation rates? Then just lastly, what's going on with that services deferred revenue as in aggregate, we were down about $92 million, and product was only down $63 million. Trying to understand why services is being impacted here.
Yes. Maybe let's start with the services piece. I mean, it's just all about the timing of the of the service agreements, right? If you think about what deferred revenue is, you know, if I sign a contract on day one, and it's a two-year, three-year contract, you're gonna have the maximum amount of deferred revenue you can have. And then when it comes towards the renewal of that contract, you're gonna have the least amount of deferred revenue that you could have from a services perspective, right? It's purely on timing on the services side, right? The lumpier thing is obviously on the product, but the services is truly just the mechanics of how you know, how the services contracts flow, right?
I guess on the, you know, cancellations, et cetera, I mean, we have not seen any change in the business. I think we're, you know, to Anshul's point, I mean, we're still seeing customers very intently focused on getting products at this point. I'm not gonna try to age out the backlog. That's kind of, I think, we're not quite there or ready to do that, just given some of the uncertainties on supply and so on.
I think as lead times improve, we expect customer visibility to shrink. Right now, we're working hard on that lead time improvement. We're a year away from that.
Got it. Thanks, ladies.
Thank you.
Thanks, James.
Operator, we have time for one more question.
Your last question today comes from the line of Tom Blakey with KeyBanc. Your line is now open.
Hey, thanks for squeezing me in, and Happy Halloween, everybody.
Thank you.
Looking at your, I'm curious about the cadence of the Cloud Titan revenue. It seems like a big uptick here in the second half, if I'm characterizing that correctly, from kind of the mid- to high-30s to get to that 45% for the year, you need to imply, you know, kind of over 50%. If I do that simple arithmetic, it implies that maybe the cumulative revenue of service providers, specialty service providers, financial and enterprise would be kind of flattening out in the second half. Just maybe comment on my characterization. Is my characterization correct? As a follow-up, are we relying a little bit more on one of those particular Cloud Titans more than the other? A comment about a correlation to actual CapEx spend would be helpful too. Thank you.
Yeah. I mean, I think, you know, as you look at the year, I mean, obviously we've been ramping revenues and ramping shipments, so we have been ramping kind of with cloud as well, right? Cloud has grown faster. It started off from a lower base, obviously, so it did, you know, it did need to recover, if you like. It was at 30%, which is pretty much the lowest we've seen it last year. It's definitely recovered from there. I think on the rest of the business, the demand is there. The demand has certainly exceeded our expectations, you know, when we came into the year. Given the shipment constraints, et cetera, we've been trying to stick to the FIFO, first in, first out, as much as we can.
You know, that has been more constrained. I think we would expect to see that kind of start to improve hopefully here as we head into next year.
Thank you.
Great. Thank you.
I don't know, Meta versus Microsoft maybe into the second half.
Yeah. That's. We'll have to wait till year-end for that, I think.
Okay.
Thank you.
Thanks, everybody.
This concludes the Arista Networks Third Quarter 2022 Earnings Call. We have posted a presentation which provides additional information on our results, which you can access on the investors section of our website. Thank you for joining us today, and thank you for your interest in Arista.
Thank you for joining, ladies and gentlemen. This concludes today's call. You may now disconnect.