Arista Networks Inc (ANET) 2020 Q3 法說會逐字稿

Welcome to the Third Quarter 2020 Arista Networks Financial Results Earnings Conference Call.

(Operator Instructions) 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.

I will now turn the call over to Mr. Curtis McKee, Director of Corporate and Investor Development.

Sir, 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 30, 2020.

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 2020 fiscal year, longer-term financial outlooks for 2021 and beyond, our total addressable market and strategy for addressing these market opportunities, the potential impact of COVID-19 on our business, our product innovation, and the benefits of recent 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, Curtis.

Thank you, everyone, for joining us this afternoon for our third quarter 2020 earnings call.

To start with, we all hope that you and your families are safe during the global pandemic.

At Arista, we recognize our role and responsibility in supporting global communications and cloud infrastructure during these challenging times.

We are working closely with our employees, supply chain contract manufacturers, customers and Arista Foundation to assist in business continuity initiatives and people's lives.

Back to Q3 2020.

We delivered $605.4 million for the quarter, with a non-GAAP earnings per share of $2.42.

[New product] services and software and support renewals contributed approximately 21% of revenue.

Our non-GAAP gross margins were 64.6%, influenced by software and services mix and a higher Asia Pacific contribution.

We registered a record number of new customer logos this quarter and $1 million customers, a direct result of our momentum in the enterprise vertical and campus traction this quarter.

In Q3 2020, cloud titans was our largest vertical.

The enterprise is once again consistently our second largest performer, followed by Tier 2 cloud service providers and financials tied for third place and service providers in fourth place.

With respect to sector trends, cloud titans are now approximately 37%; enterprise and financials also are approximately 37%; and providers, which is our cloud service providers combination, is approximately at 26%.

What is clear is Arista's cloud principles now apply to all sectors, and we are diversifying well across customers and verticals.

In terms of geography mix, international contribution was 25%, with the Americas at 75%.

The European business recovered from Q2, and the Asia business has had a particularly strong quarter.

In the absence of a physical Analyst Day this year, we would like to shed some light on our strategy for addressing our $30-plus billion TAM.

The past decade can be best characterized by Arista's momentum in cloud networking.

But in the past 4 quarters, we have experienced a triad storm of cloud titan volatility, COVID pandemic and deferred revenue-related comps.

Despite this tough reset year, we believe Arista will emerge stronger, not only returning to double-digit growth in 2021, but also aiming for consistent growth in the years beyond.

We expect our multiyear growth cycle from 3 major product line contributors: one, our core cloud and data center products, our largest business, building upon our flagship Arista EOS, with the hallmarks of leaf spine topology across our 5 major verticals.

Two, our second market is the network adjacencies, with routing replacing routers and our recent entry into the Cognitive Campus workspaces.

The routing market consists of core spine, edge and peering use cases for Tier 1, 2, 3 service, cloud and enterprises.

The campus brings a cognitive unified edge for wired and WiFi endpoints as well as new IoT devices in a 2-tier leaf spine or spline network.

We do expect to disrupt the campus and router incumbency of the last few decades in the next few years.

Our third category is network software and services based on software subscription models, such as Arista A-Care, CloudVision, CloudEOS router for multi-cloud, big switch monitoring and our latest entry into advanced network detection and response with the acquisition of Awake Security.

Elaborating on our focus on subscription-based software, this product line is typically multiyear contracts.

Customers are driving mandates for network automation, monitoring, visibility across their data sets.

We believe the recent acquisition of Awake Security is a strategic contributor that transforms the silo aspect of security into a seamless, secure network.

The moment the cloud blurs the perimeter and the increased use of IoT and shadow IT means that our CIOs and CFOs need a network [ground of fruits].

Simply storing the raw data or alerts isn't comprehensible data.

A new AI-driven, data-driven, state-driven software stack like Arista provides is foundational.

In our opinion, the security industry is going through a metamorphosis from point security to proactive, predictive, secure networking.

CloudVision, combined with Arista's 2020 acquisitions of Big Switch Networks for observability and Awake Security for autonomous threat hunting are very natural software additions to this category.

The Cognitive Campus is a priority for Arista's adjacent product line.

We are witnessing a massive transition in the COVID era to work from home where the boundary between the office, the home, the teleworker transit and the user is blurring into elastic and flexible workspaces.

Arista's recent introduction of the Cognitive Campus is a state- and data-driven model coupled with a unified dashboard for wired and wireless edge for next-generation zero-touch campus deployments.

This combined with the Awake network detection and response security, feeds into our campus visibility flow tracker for both IoT and OT applications.

We do expect our campus portfolio to double by the end of 2021 as we invest in both engineering and go-to-market model.

A fitting example is today's announcement of our flagship 750 Series campus chassis, delivering a combination of industry-leading chassis, density, performance for high-definition video, fail over resilience for rolling upgrades, cognitive PoE and secure encryption.

I would like to invite Andy Bechtolsheim, our Founder and Chief Development Officer, to highlight our launch today.

Andy?

Yes.

We like markets that are ripe for real innovation, and the campus networking market is a prime example of that.

With the launch of our 750 series modular chassis, we are introducing a next-generation platform, which delivers more performance, more security, more visibility and more power capabilities than any other product in its class.

The 750 has 400 gigabits per second uplift throughput, which is 5x the performance of our nearest competitor.

This level of performance is key to support WiFi 6, where each access point has up to 5 gigabit throughput.

With a single 750 chassis, we can support a full complement of 384 WiFi 6 access points.

In addition, the 750 offers MACsec encryption on every port, which supports secure communication from the WiFi through the entire campus network.

The 750 also breaks new ground in density, being 3U more compact than our closest competitor.

Combined with the legendary reliability of Arista's EOS offering system, the 750 offers unmatched performance, security, availability, visibility and power handling.

Thanks, Andy.

That was great.

I'm excited.

I could use one at home.

Next-generation routing is another key adjacent market for us, whereby we're bringing the union of 2 box-level approaches, layer 2 switches and layer 3 routers, together for rich protocol support, resiliency, scale and programmability.

This has been a 4- to 5-year endeavor for us to bring disruptive economics via advanced merchant silicon and Arista's modern software stack EOS.

It's time to see the fruits of this labor.

Arista customers now see us as a compelling and cloud and carrier-grade routing alternative to expensive legacy routers.

We're extending beyond classical leaf spine intra DC use cases to multi terabit routing offering lower power and higher density interconnect, accelerating WAN and inter-DC routing use cases.

Our investments in the simplification of Arista's routing stack with standards-based EVP and BTP and segment routing are yielding early traction.

We have won a few important Tier 2, Tier 3 service provider customers for many of these use cases, including Spine interconnect, multi-access edge, layer 2, layer 3 VPNs and peering use cases.

Our core universal cloud network design for data center continues to expand with Arista as the pioneer.

We have been the market leader, driving to the #1 position in the 100 gigabit ethernet switching market and early leadership in 400-gig with flexible software partnerships for SONiC open [cloud network] and FBOSS with our cloud titan customers.

We have been pushing the envelope of storage, compute and AI clusters at cloud scale.

This quarter, we delivered our vision managed in a time series state-driven database as a network-wide service hosted in the cloud.

Clearly, we're enabling cloud principles across the entire enterprise, with 5 As of agility, availability, automation, analytics and an AI- and API-driven network.

Anshul Sadana, our Chief Operating Officer, will add some more color on our cloud customer traction.

Anshul?

Thank you, Jayshree.

Arista has been a leader in cloud networking based on our nimble execution, product quality and ability to co-develop with our customers.

Cloud companies are highly impressed with the work we've done over the last decade and are engaging us more than ever to discuss their architectures for the future.

The advent of 30-gig 30s and related silicon is driving a product transition.

We are winning new RFPs for next-gen products, both 100-gig and 400-gig.

We have expanded our footprint from data center leaf spine and DCI to now encompass WAN and edge use cases traditionally supported by legacy routers.

While there is a lot of talk about white boxes and 400-gig, our customers have maintained status quo.

We are not seeing a change in position from this status quo.

Customers who use white boxes are continuing to use white boxes, whereas customers who use Arista switches are continuing to use our products today and for their future designs.

If anything, some use cases currently covered by internally developed white boxes may transition to our feature-rich products in a few years.

Cloud companies operate at a massive scale, and they are continuing to grow.

This scale makes them lean more on us for technology and support.

Our cloud customers see immense value in working with us, and we have high customer satisfaction here.

Our portfolio is highly competitive, and we are being told that we are ahead of competition.

Hence, it is unlikely that we will see significant share shift.

Savvy cloud titans are not influenced by pretty marketing slides.

Our R3 Series products have now been qualified by all our major cloud customers for several 100-gig and 400-gig use cases.

We will keep marching on.

Back to you, Jayshree.

Thanks, Anshul, and we will be marching on.

Well said on the cloud titans success.

So in summary, Arista's vision now transcends LAN, WAN, cloud networking, with clear diversification across customers, products and verticals.

It's rooted in a more software data-driven network built across customer data sets and managed by CloudVision.

We have built a transformative architecture harnessing the new trends in IT computing, unified edge, archiving data across the network.

Our customers resonate with this vision, and I couldn't be more upbeat on our strategy, our innovation, our quality, our support, and our customer migration from legacy silo places in the network to cognitive clients to cloud networking, which we call places in the cloud.

We have now deployed a cumulative of 40 million ports over the past 12 years.

With that, I will turn it over to Ita, our Chief Financial Officer, for more financial specifics.

Ita?

Thanks, Jayshree, and good afternoon.

This announcement of our Q3 results and our guidance for Q4 2020 is based on non-GAAP and excludes all noncash stock-based compensation impacts, certain acquisition-related charges and other nonrecurring items.

Full reconciliation of our selected GAAP to non-GAAP results is provided in our earnings release.

Total revenues in Q3 were $605.4 million, down 7.5% year-over-year, well above the upper end of our guidance of $570 million to $590 million and up 12% from the prior quarter.

While we saw some improvements on the supply chain front, shipments remained somewhat constrained, with some expected continuation of extended lead times into Q4.

Service and software support revenues represented approximately 21% of total revenue, down slightly from 22% last quarter.

International revenues for the quarter came in at $152.7 million or 25.2% of total revenue, up from 19.4% in the second quarter.

While the shift in geographical mix on a quarter-over-quarter and year-over-year basis was largely due to the location of deployments by our cloud titan customers, we did see some incremental improvement in our in-region businesses also.

Overall gross margin in Q3 was 64.6%, above the midpoint of our guidance of approximately 63% to 65% and consistent with last quarter.

We continue to recognize incremental COVID-related costs in the period, including elevated freight and component costs.

Operating expenses for the quarter were $159.4 million or 26.3% of revenue, up from last quarter at $144.1 million.

We began to increase operating expense investments during the third quarter as our top line performance for the year continued to improve.

R&D spending came in at $106.1 million or 17.5% of revenue, up from $91.6 million last quarter.

This reflected increased employee costs and increased new product introductions related to spending in the period.

Sales and marketing expenses were $43.1 million or 7.1% of revenue, up from $41.9 million last quarter, with increased variable compensation and other personnel costs.

As a reminder, we continue to benefit from lower COVID-related travel and marketing expenses.

Our G&A costs came in at $10.2 million or 1.7% of revenue, down slightly from last quarter at approximately $10.6 million.

Our operating income for the quarter was $231.5 million or 38.2% of our revenue.

Other income and expense for the quarter was a favorable $13.2 million, and our effective tax rate was approximately 21.6%.

This resulted in net income for the quarter of $192 million or 31.7% of revenue.

Our diluted share number was 79.3 million shares, resulting in diluted earnings per share for the quarter of $2.42, down 10% from the prior year.

Please note that included in other income and expense for the quarter was a onetime gain on the sale of investments of $9.4 million.

In addition, given current low interest rate environment, all other things being equal, we would expect other income of approximately $3 million per quarter throughout the coming year.

The acquisition of Awake Security closed on October 7, and we are now focused on integration and purchase accounting.

The acquisition will not have a material impact on the financials for the fourth quarter.

Now turning to the balance sheet.

Cash, cash equivalents and investments ended the quarter at approximately $2.85 billion.

We repurchased $167.3 million of our common stock during the third quarter at an average price of $208 per share.

As a reminder, we have now repurchased $661 million or 3.2 million shares against our Board authorization to repurchase $1 billion worth of shares over 3 years commencing in Q2 '19.

In terms of capital allocation, please expect us to continue to execute opportunistically against the remaining authorization.

We generated $215.1 million of cash from operations in the third quarter, reflecting solid net income performance and a consistent level of overall working capital investment.

We expect to continue to strategically increase inventory levels through the end of the year as we improve lead times and attempt to buffer against any future COVID-related supply chain disruptions.

DSOs came in at 46 days, down from 65 days in Q2, reflecting linearity of billings in the period.

Inventory turns were 2x, down from 2.3 last quarter.

Inventory increased to $438 million in the quarter, up from $327 million in the prior period, as we continue to buffer certain components and products.

Our total deferred revenue balance was $562 million, down from $577 million in Q2.

As a reminder, our deferred revenue balance is now almost exclusively services-related.

The level of services of our revenue is directly linked to the timing and term of service renewals, which can vary on a quarter-by-quarter basis.

Accounts payable days were 70 days, up from 59 days in Q2, reflecting the timing of inventory receipts and payments.

Capital expenditures for the quarter were $2.5 million.

Now turning to our outlook for the fourth quarter and beyond.

While we remain cautious around the impact of COVID-19 on the economy and our business, we have seen some incremental improvement in underlying business trends.

Activity in our enterprise and provider sectors has remained healthy, with increased win rates across, for us, a relatively underpenetrated part of the market.

In addition, we have continued to solidly and consistently execute against the needs of our cloud titan customers.

We believe a combination of these trends, combined with favorable year-over-year comparatives, supports the current consensus growth rate of 13% to 14% for fiscal 2021.

On the gross margin front, we will continue to reiterate our overall gross margin outlook of 53% to 55%, with customer mix remaining the key driver.

Turning to spending and investment.

While we will continue to carefully manage spending, we are committed to making go-forward, results-based investments in the business.

This includes continued go-to-market expansion to support enterprise and campus growth and investments in R&D to support innovation across the business.

While it won't happen overnight, especially in an environment of resumed top line growth, we would take this opportunity to remind you of our long-term operating margin target of plus or minus 35%.

Finally, our outlook discussion above and our guidance for Q4 reflects our current understanding of COVID-19 and its impact to our business and supply chain.

This is, however, inherently uncertain, and we will need to continue to monitor and attempt to mitigate new challenges as the situation unfolds.

With all of this as a backdrop, our guidance for the fourth quarter, which is based on non-GAAP results and excludes any noncash stock-based compensation impacts and other nonrecurring items is as follows: revenues of approximately $615 million to $635 million, gross margin of approximately 63% to 65%, operating margin of approximately 37%.

Our effective tax rate is expected to be approximately 21.9%, diluted shares of approximately 79 million shares.

I will now turn the call back to Curtis.

Curtis?

Thank you, Ita.

We are now going to move to the Q&A portion of the Arista earnings call.

(Operator Instructions) Thank you for your understanding.

Operator, you may go ahead.

(Operator Instructions) Your first question comes from the line of Samik Chatterjee from JPMorgan.

I'll just keep it simple.

Jayshree, I think everyone on the call would recognize that you sound a lot more confident about returning to double-digit growth than you did 90 days ago.

So I'm wondering what's driving that.

Is it some of the 400-gig wins coming in as you expected?

Or is it the underlying customer spending starting to improve, that's driving that increased confidence relative to what we heard a quarter ago?

Yes.

No, thank you, Samik.

I think when you look at the foundation and fundamentals of Arista, they didn't change, right?

We've always had superior products and very strong customer traction.

And -- but I think we got, our customers and Arista got used to the uncertainty of COVID, and COVID became a new norm, and people have to start planning their spend.

So we saw a very balanced and customer traction across all our verticals and all our sectors in Q3.

Whereas I wouldn't say the same for Q2 and Q1, where we were still figuring things out.

So I think the combination of an unchanged strategy, a highly differentiated product, and we're just winning in every sector.

And there's no silver bullet, but humming on all 4 cylinders gives us a newfound confidence.

Your next question comes from the line of David Vogt from UBS.

Good quarter.

Great guidance.

Maybe just a big picture industry question, if you guys could entertain me.

Can you share your thoughts on the proposed Marvell - Inphi transaction that was announced last week, and maybe what it might mean for the industry holistically as we kind of move forward into 2021?

That would be great.

I'll try.

I mean, I came from the semiconductor world 2 decades ago.

But as you know, there's been a lot of semi consolidation: Nvidia-Arm, Maxim analog digital and now Marvell Inphi.

I think the way to look at this is semiconductor companies are all -- the large ones are trying to get larger, and the smaller ones are producing some real innovative technology but need scale.

We are very impressed with Inphi.

They've been an important partner for us, and they have both very best-of-breed technology in the 30s and optic side.

And it's something that Marvell lacks.

So hopefully, some of the strength they bring, especially to the cloud, will help Marvell.

Your next question comes from the line of Sami Badri from Crédit Suisse.

I just wanted to just check on a comment made by Anshul earlier in the transcript regarding white boxes.

And you made the comment that some customers may deploy white boxes today and then eventually swap that out for a feature-rich Arista product in the future.

Now is there maybe a rationale behind why they would want to deploy data centers filled with white box infrastructure or white box equipment and then eventually swap it out?

Can you just explain to us the transition that would take place in design or just in the thought process for why they would do that after they've gone down the white box route?

Sure.

Sami, the world has talked about white boxes by just from one dimension for many, many years, but remember, our customers do that analysis in both directions, not just one.

And as their scale grows and their needs grow, the network is getting more complicated.

And in certain situations when they reach a point where they need even more functionality than is easy to build internally and they want to stay competitive in the market and not miss on timing, they do start looking out as well.

But as I mentioned, these are future trends that take a couple of years at least to happen.

But these are certainly [included in] discussions and issues going on in our space today.

Your next question comes from the line of Alex Kurtz from KeyBanc Capital Markets.

Glad to hear everyone's doing well and safe at Arista.

And congrats on the quarter, Jayshree.

Just on your comments about software and maintenance becoming roughly 20% of the business.

Can this growth in your portfolio of software products, whether designed internally or acquired, to maybe accelerate that software renewal base to where maybe in a couple of years, we're looking at 30%, 40% of the business, if you can go back to these big cloud titans and kind of demonstrate the value of these bigger software portfolios that you didn't have a couple of years ago and really expand that base of software?

Thanks, Alex.

Good question.

I think very observant of you to note that when we say it's 20% of our business, it's not just services.

It's software renewals and subscriptions.

So it's already starting to have a contribution.

Can it be greater, greater than 20, 25?

Yes, it would be harder to be 30 and 40, especially with the cloud titans because cloud titans tend to think they can and they have the resources to build many of these tools themselves.

So I would say all the other 4 verticals are more likely to embrace our software subscription, while the cloud titans may take it in bits and pieces.

They would take it in components but not in entirety.

But I certainly think this segment where we can have subscription multiyear contracts is an important part of our triad stool of core networking, adjacent networking and network as a service capability.

Your next question comes from the line of Tal Liani from Bank of America.

I have a question on the trends in the quarter.

Product growth, so overall revenues are down 7.5%, but it's not even between products and services.

Product growth is down about 13.5% year-over-year and services are up 26%.

So I'm trying to understand both sides.

I'm trying to understand the strong growth in services versus the steeper decline in products?

Yes.

So first, remember, again, services includes many of our software subscriptions and multiyear contracts.

So it's not just services.

So -- but -- whereas product is very clearly perpetual product, right?

So look, I think the comps with deferred revenue never helped us from last year to this year.

And -- but I wouldn't read too much into the trend, except to say we're getting stickier with services and software, and we can only do better with product.

Yes.

I mean, I think, Tal, the services should continue to grow and continue to be a more meaningful piece of the business and layer the software on top of that.

And then product is recovering, but it is recovering.

You can see that in the Q4 guide and into our commentary for next year as well.

So I think they're just different drivers at this particular point in time, but they should both start to grow as we go forward.

Your next question comes from the line of Fahad Najam from MKM Partners.

A couple of things.

If I look at your working capital, it seems like your inventories have gone up significantly, while your accounts payable are also [awful].

But can you just help us understand what's happening, why the inventories are up so much?

Are you expecting a significant forthcoming demand ramp in the next couple of quarters?

Yes.

No, I think, Fahad, what we're really doing is we're buffering against some of the uncertainties around COVID.

And if you look at the Q when we file it, you'll see that a good portion of that increase is still raw materials and components.

There was some uptick in the finished goods right at the end of the quarter for particular products, but we still have more work to do around expanding that to other products.

But our goal is to have sufficient buffers that if we do get some more shocks from COVID that we're able to react and that we're in a better position to react than we were maybe on the first wave.

So it really is just a focus around making sure we've got more optionality and flexibility.

Your next question comes from the line of Paul Silverstein from Cowen and Company.

Jayshree, I think you kind of covered this during the call, but I'm going to ask you if you could -- in your outlook for next year, the double-digit growth in your endorsement of the 13% to 14% consensus number, to get there, I think I heard you say you expect your campus portfolio to double by the end of 2021.

Are you saying you expect campus revenue to double?

And can I ask you to go through what your expectations are for the cloud, for enterprise and financials and for your service providers to get to that double-digit growth?

Paul, thank you.

Yes.

So as you know, we said last quarter that we have achieved -- this is our first year at campus.

There are still young kids here, and we've achieved our first 100 million.

And if you may recall, we said it would take us 4 to 6 quarters to double.

I don't yet know if it will be 4 or 6, but we are feeling pretty good, particularly with our product announcement today and the campus traction we're getting, that we didn't feel last quarter.

So yes, the goal is to definitely double that $100 million to $200 million by the end of 2021 in revenue, and let's hope we can do better.

In terms of segments, I think it's too early to call a breakdown, but we're very comfortable with the annual consensus for 2021.

Ita, do you want to answer that?

Yes.

No, I think, Paul, we're not going to try to do it by vertical at this point, right?

We feel like there is -- between the campus stuff that you talked about between the growth and services, cloud has resumed growth, and then the rest of the business has also been performing well.

I think between all of those drivers, you have multiple different ways to get there, and we'll see exactly how it plays out.

And I think we're not going to try to [pick] which verticals we're going to do what at this point.

Can I ask you this, I trust that to get there, cloud has to be healthy by definition.

No, I think -- I hope you noticed our upbeat tone, especially Anshul's, we started the year saying it will be flat to down, and then we said it will be flattish.

And at this point, I think we're feeling like cloud can be a growth -- cloud titans can also grow next year.

Your next question comes from the line of Jeff Kvaal from Wolfe Research.

I was wondering if you 2 wouldn't mind giving us a little bit of some of the thought process behind the guidance.

I guess, one part is you had -- there's another partner in the supply chain at one of your customers that cast a pall over 2020.

And I'm wondering if that particular issue in semi was resolved.

And then the other question is, could you help us a little bit with the assumption of 400 and when that starts to layer in to the 2020 -- 2021 outlook.

Yes.

No, I think -- so the -- again, the rationale for the guidance, I think if you look at the various pieces, pretty much what we talked about with Paul a few minutes ago, right?

I mean there are multiple different pieces of the business, and we've seen positive trends across those sectors, right?

Not all of those reflected in revenue today.

But in the enterprise vertical being good, solid wins that will drive some revenue traction into 2021.

And then for cloud, I think we feel like we've got -- we're in the window of starting to understand their plans for next year and feeling more confident around those drivers as well.

And then obviously, the rest of the business continues to grow, looking at service software, et cetera, the more ratable piece of the business.

So I think those are the building blocks, and we like to put those building blocks together in multiple different ways and feel good about the guidance, and that's kind of where we came out for next year.

Yes.

Jeff, I think Ita said it best.

We expect cloud titan trends to improve in 2021, based on both the CapEx projections, which for us isn't a huge indicator because we're a small number.

But bulk of these deployments will be 25-gig, 100-gig and even some 400-gig.

I think the white box and the overall competitive landscape, much as everyone [sees] it, is unchanged.

And we're seeing a status quo and look forward to share gains in new roles in campus, data center, routing and cloud.

Your next question comes from the line of George Notter from Jefferies.

Very interested to hear the commentary about routing and all the different use cases you're pursuing in that space.

Maybe you can kind of talk about your definition of success in routing?

And also, could you give us a sense for where you are now in routing applications, percentage of sales or amount of revenue you're driving there?

And then maybe looking forward, what use cases are you in today?

And what use cases will you be in going forward?

Yes.

Sure, George.

It's difficult to break it down, as I've always told you, because routing and switching often go together.

But what we saw as a trend is we've always sort of aimed for the big elephants or the Tier 1 service providers, and those take time.

But we started winning, particularly this year, we turned the corner on winning a lot of Tier 2 and Tier 3 providers, not only in the U.S., but internationally.

And these use cases tended to be telco cloud.

Some of them were just very happy with their multi-vendor combination of vxlan, VPN, BGP staff from us.

Some of them are peering use cases.

So these are all classical routing use cases with better disruptive economics, programmability, resilience and routing features that they have come to know and love from others, but that they could get better from us.

So -- and then also, we're doing very, very well in the cloud customer base as well.

We have been for some time.

So the combination of all this to say that routing is not just with the service providers, it is now permeating all our 5 verticals.

Got it.

And is there one product delivery or feature delivery that has allowed you to turn the corner in the service provider space?

Or anything you can attribute that to?

If you know the service providers as well as you do, it's never one feature.

It's a long list of them.

So we've been working it for 4 to 5 years, and I think the culmination of it has led to more success.

But there's always one more feature to do, George, as you know.

Your next question comes from the line of Rod Hall from Goldman Sachs.

Yes.

Thanks for the question.

I wanted to just ask a kind of a housekeeping question and then the main one.

The housekeeping is, do you guys think deferred revenue will become an issue in '21 because of 400-gig rollout?

Do you think we'll see product deferred revenue building up again?

And then I also wanted to go back to enterprise.

It seems like a time when enterprises, if anything, would be slowed spending that you guys have really accelerated at least sequential growth there.

And there's a lot of absolute additional revenue in Q3.

So just curious, do you think you're pulling any revenue forward on enterprise?

Or what -- how does that look like it continues the next few quarters for you?

Ita, you want to answer it?

Yes.

I think on the deferred, it's always difficult to forecast when we'll have customer requirements, et cetera, for acceptance.

I mean, it tends to be largely for some very differentiated or different product set, particularly under the current accounting.

So we'll see.

It's hard to predict that there would be some of that or not at this point, right?

But it's definitely -- it's more difficult to get to a deferred bar.

It would have to be a very differentiated product going forward.

But Rod, I'll just say it feels good to be back after a tough few quarters.

The enterprises love our product, and they want to buy more.

And if anything, I would say, opposite of pull in, we're still slightly supply constrained -- constrained versus demand and are experiencing some shortages that will hopefully improve by the end of the year.

So no pull-in for sure.

Your next question comes from the line of Jim Suva from Citigroup.

And really heartfelt congratulations to you and your team for very strong results and outlook during COVID.

My question is regarding the outlook for 2021, when you mentioned consensus up 13% to 14%, you're comfortable with that.

Does that bake into the most recent CapEx outlooks that we were provided last week?

It's notable that like Facebook gave an outlook of up -- materially up [in percent].

And I know that you're wrapping up and getting ready for earnings and things, but I believe their CapEx is supposed to be 30% to 40%.

But would that be additive?

Yes, Jim, no, I mean, this is our current view as of today of what we think we feel comfortable with for 2021, and then we'll see.

There's always going to be puts and takes and the CapEx to networking correlation has its challenges, right?

So we'll see where it goes from there.

That's what we see as of today.

Jim, thank you for the question.

As you know, the CapEx includes building leases, capital.

So the networking component is very, very small.

Ita and I were talking earlier, as I was with Anshul, too, it's generally less than or around 5%.

So it's hard to make any specific decisions based on CapEx.

But as an overall trend, we are pleased that the CapEx is going up next year, but then we'll keep watching this quarter-by-quarter, because it tends to be lumpy and volatile.

And congratulations again to you and your team.

Thank you, Jim.

Thanks, Jim.

Your next question comes from the line of James Fish from Piper Sandler.

Congrats on the quarter, ladies.

Ita, you went over the negatives of the higher freight cost but also the positives of lower travel.

To your best understanding, what was the net impact of COVID-19 on OpEx this last quarter?

And then DSOs would imply a more front-end-loaded quarter, that either insinuates a slowdown in September or was it just lack of needing to push more business that we have a strong backlog heading into Q4?

Yes.

No.

I mean, the DSO is actually the biggest driver this quarter was just we exited with a high balance at the end of last quarter, right?

We were very back-end-loaded last quarter, and then you can collect against that, and that drives some good traction through the quarter.

So I think we were -- Jayshree mentioned, we still have some supply constraints we were still receiving in inventory, and you saw that in the accounts payable as well, right?

So we were still a little bit back-end-loaded through the quarter, but we had a good balance to collect against coming into the quarter.

I don't know that I'm going to try to size exactly what the COVID impact is.

There is a little bit in gross margin, and you've seen that in the product margins.

And there's travel, some marketing expenses, et cetera, probably a couple of million or so that comes out of the sales and marketing line because of it as well.

Your next question comes from the line of Jason Ader from William Blair.

On the campus side, just wondering what types of customers are actually buying this stuff and then the stuff working from home.

It's just -- it surprised me that you're winning a lot of new [volume] there.

So that's kind of the first part of the question.

Second part of the question is the 750 series product, is this a big hole in your portfolio?

And do you think that it could actually accelerate some of your traction in the campus market?

First of all, you'll be pleasantly surprised as I am that even though people are not coming back to smart buildings or their headquarters, it's really vertical-dependent, as you know.

Some of the logistics and health care and some of the business-critical ones do need to go back and do need connectivity and performance, in a far more distributed elastic fashion.

So I think campus is back, just not the way we thought it would be.

The emphasis on unified wire, wireless rather than specific smart buildings is the change we're seeing.

And it also gives them -- what we noticed is it's also giving them a chance to plan better because if everybody's in the building you can't plan, but now you can do a lot more proof-of-concept testing.

Yes, I'm very excited with the 750.

There's a whole billions of installed base of some of my old products called the Catalyst that are ripe and old in age.

And Anshul, you want to comment on the beauty of that?

Absolutely.

Jason, we're very excited with what they're doing, but a lot of this is driven by feedback from customers.

There's immense interest in the 750 series, and that's broadening our portfolio.

When you look at the Fortune 1000 type of enterprises, they absolutely need high density and high performance even in campus.

And yes, there are certain companies thinking that since the employees are not in the office, let's not upgrade or refresh, but there are also certain companies thinking that there aren't enough employees in the office, so let's go ahead and refresh.

And that's really what's driving a lot of the change in the enterprise as well.

But the feature set, consistent POS, CloudVision, automation, a lot of the visibility and the benefits to give to people is important to them.

If your videoconference has jiggered, you don't want to be debugging in the middle of the day, you need the glitch to resolve by itself and move on, and those are the capabilities we deliver with our products.

Which is the catalyst comparable from Cisco?

With the [75 750G]?

I think there's a couple of them.

There's the 4000 series and the 6500 series.

And the CAS 9400.

And their current -- the presently shipping one is the 9400, but the older ones are the 4K and the 6500.

Your next question comes from the line of Ryan Koontz from Rosenblatt Securities.

I wonder if we could circle back to the campus opportunity again and look at the competitive front there.

Obviously, you guys have great technology.

And how do you feel you're progressing in terms of building out your channels and displacing the big incumbent there in terms of kind of influence over these larger enterprise Global 2000 type customers?

Thanks, Ryan.

I think we have to understand that we are in our second year of campus.

And if you look at our competitors, they're in their 10th or 15th or 20th year.

They're very mature.

So definitely, we've got work to do on go-to-market models.

So our upfront first area of focus there is our existing customers.

When we have over 6,500 of them, some [percentage] of them love our EOS and CloudVision and want to use us.

The second is we are making more progress on the channel side.

We're not signing up everybody, but we certainly see a strong international focus on channels.

And I think that will be something we will continue to invest in.

And the third is, for the last 2, 3 years, we have been investing in the enterprise go-to-market led by Chris Schmidt and Ashwin Kohli under [Ashwin].

So we're starting to see the fruits of our labor and just enterprise traction, whether it's in the data center or campus.

And it's taken us the better part of 2 to 3 years to achieve that.

Your next question comes from the line of Ittai Kidron from Oppenheimer.

A couple for me.

Maybe just kind of to drill down some of the questions were asked.

First, on the cloud, clearly, you're more bullish here.

So it's great to hear.

Can you clarify whether this includes Facebook sort of refresh?

Is that finally back on track?

And you've got a piece of that?

And then piggyback on the question that Jason had on the campus in new 750, congrats on that.

Cisco just refreshed 9400.

I guess when I look at your solution right now online, it looks like you have a couple of cool features there, always on POE, but a supervisor, data plane separation.

But it was my impression that this whole chassis campus category is on a massive decline.

Customers moving into fixed architectures, not modular at all.

So help me understand how much really is the opportunity here?

And when you talk about your confidence in doubling the campus business through next year, is this going to be a material contributor to that?

Or this is really for thereafter?

Oh, boy, how many questions were those?

Three within one.

Lovely, but Ittai we will keep our answers shorter than your questions.

How about that?

Anshul, do you want to take the cloud one or --

Absolutely.

I wouldn't say that the Facebook server refresh is in our guidance model, but all of our [cloud agents] network plans for the next year are in our models.

Let's decouple that a little bit, and we have the data we have today from our customers for using that.

So we're confident that there is growth coming back next year on that side.

Right.

And on the 750, having -- both Anshul and I have been involved a lot in chassis in our prior lives.

I think if there's anything that will be under constraint, it's the stackables.

Customers are moving to more and more distributed 1 RU.

And one of the things that probably got a little unnoticed in our announcement today is our 2 RU 96 port, which obviates the need for any stackable.

And then for the really high-density distributed buildings, which have large employees, large headquarters, you do need a 750.

You need the investment protection of 100-gig uplinks.

You need the density and footprint and power.

You need the operational power management.

You need the security.

You need the always on failover time for rolling upgrades.

So -- and I think what's happened, at least in our minds, is the CHAP-ing has gone from being a physical cable plant discussion, to much more of an automation, visibility, security discussion.

You need all the properties that you had in the data center.

So that market is coming to us.

So we feel good that there's -- it's a new product.

It will take time to qualify, but it is a contributor to our number in 2021.

Your next question comes from the line of Meta Marshall from Morgan Stanley.

Maybe wanted to touch on just kind of what sort of traction you're seeing with existing customers from your Big Switch and Awake Security acquisitions and just whether that could be -- I understand it's not material to Q4, but a more material contributor to 2021.

Yes.

Thanks, Meta.

I definitely think it's a strategic contributor to 2021 in bookings.

As you know, the revenue may follow as multiyear contracts in software, network software and subscription models.

So how material will it be in 2021?

Probably small.

But both those are very strategic to influencing our customers' decisions on the data center and campus.

And in general, helping with their operational -- their architectural experience and their operational experience, it's a huge differentiator.

So observability, monitoring, in-line data analysis capability and then the autonomous threat hunting, it's so important now because the firewall or the perimeter for the firewall is just collapsed.

So the ability to do that is we just closed a Big -- the Awake acquisition in October.

So it's too early to tell.

But the one thing we can tell is everyone's interested in it.

Anshul, you want to add to that?

Jayshree, you're absolutely right, with IoT and sensors in campuses and work environments, the monitoring is going to be very, very different.

And there's an unmet need in this space today, and I think Awake [is the way] with the AI technology [on there] as well as Big Switch in the market side.

Your next question comes from the line of Aaron Rakers from Wells Fargo.

Yes.

I want to go back to one of Ittai's questions.

And I can appreciate that you're not factoring in kind of Facebook and server refresh cycle in the context of your guidance.

But if you go back in time, how do you think about the context of server cycles and how that pulls through your business just to give a historical backdrop?

Because as we look into '21, you've got, obviously, a big refresh for Intel.

You've got [A&B pushed into server CPU] cycle.

I'm just curious to how you think about that pull-through effect on Arista's business?

Aaron, I don't think it's changed much.

First, what happens is you've got to get the buildings and the power and the cooling in, then you've got to get the servers.

And there's typically a 1- to 2-quarter lag on the network.

And so clearly, the server cycle comes ahead of us, which is why we felt the pain we did last year.

Hopefully, we'll feel the gain of it next year.

Your next question comes from the line of Pierre Ferragu from New Street Research.

Jayshree, you mentioned in your opening remarks that outside of your core switching market, you had growth opportunities in routing, campus and software and services.

And I was wondering over like a 3, maybe to 5-year time horizon, how much of your growth do you think is going to come from these additional segments versus how much is still going to come from switching, your core switching market?

Well, Ita is telling me not to answer this question.

[I see that cue as today]

From a vision perspective, we think those 2 segments, the new markets where we're underpenetrated, we're in the early innings in both network adjacencies and software and services, will grow faster than our core market.

How about I leave it at that?

Is that a good enough answer?

That's good enough answer.

I'll have to live with it.

Your next question comes from the line of Tim Long from Barclays.

Can we just hit on the telco vertical?

It seems like it's still towards the bottom of the pecking order, at least in your results.

So what's going on there?

When do you think you'll get that going?

Is it just increasing the routing use cases and further penetrating on that regard or maybe the addition of more security features, which often do well in the telco networks?

Thank you, Tim.

Look, I think unlike the cloud, where they will adopt our routing features even if we're missing 1 or 2 and then operationalize it, you know very well, service providers want every bell and whistle.

And as I said in my opening remarks, it's taking us the better part of 4, 5 years.

We're making very good traction in Tiers 2 and 3. Tier 1 is still taking time.

But in this year, we have won some early design wins in Tier 1 as well.

But the numbers are small.

So I think in order to make them big, they need to spend more from us.

Your next question comes from the line of Amit Daryanani from Evercore.

Perfect.

I guess, maybe if I just go back to this double-digit sales growth expectation for calendar '21.

Just wanted to understand, is that an organic statement?

Is that a total revenue growth statement?

How do we think about that?

And then broadly, when I hear you talk about calendar '21, it sounds like the growth vectors are getting much more diversified than they historically have been.

And I'm wondering how does that play out into your operating expenses and perhaps the need to expand your infrastructure further in calendar '21?

So just -- Amit, just to take the question upfront, it is absolutely based on products as we have now organically.

We're not factoring in any future products or M&A in that discussion.

And maybe, Ita, you want to talk about the...

Yes, I mean, investment model.

Yes.

I think on the investment side, with the top line returning to growth, that gives us a lot of room to make those investments.

And maybe a little bit more than that, we'll see, right?

But we certainly have the growth on the top line to do that.

My script did remind everybody that our target model is 35% operating margin.

And we're reserving the right to do that, should we find the right investments to make.

And we'll be very metric-driven around that as we go forward.

But that's kind of the target model longer term.

In the meantime, we'll have the benefit of top line growth to help with the investments in the near term.

Your next question comes from the line of Simon Leopold from Raymond James.

I wanted to take your sense on a scenario.

If we assume an existing Arista customer wants to upgrade a data center to 200-gig, not 400 but 200-gig, what does that mean for Arista?

Is this a line card change or a swap of chassis?

And if it is a swap of chassis, you risk losing share in that kind of upgrade scenario for one of your customers?

So Simon, while we haven't talked about 200-gig broadly, there are certain customers looking at these types of technologies.

And since we're staying with the 4 lean architecture at [4550].

The good news is all of the products we mentioned, the R3 series, some in the 7060 series, some in the 7300 series, all of them support not just 400-gig, but also 200 gig.

So we are aware of the need in certain cases of 200-gig now very, very well poised to actually grow or benefit from that as well.

Your next question comes from the line of Woo Jin Ho from Bloomberg Intelligence.

Oh, great, thank you for squeezing me in.

So an intermediate-term technology question here.

So the [chip] question has been asked.

So NVIDIA and Marvell have been talking a lot about DPUs in the data center.

It sounds as if [in can is] starting to get a little bit more complex with the architectures at least.

Does DPUs present a core growth opportunity for you in the cloud that we may not have considered in the past?

This is a bit too early to say whether DPUs will come disrupt or is it a lot of marketing in terms of offload engines.

Remember, offload engines, especially on servers, have been around for many, many years.

And these might have special instruction sets for the new types of offloads the world is looking at.

But so far, we don't see any major impact to networking like us, maybe even a benefit because that drives the change in architecture and actually helps us compared from the incumbency.

Yes.

I think they generally have been co-processors.

They don't take away the need for a CPU.

Your final question comes from the line of Jon Lopez from Vertical Group.

I just had one clarification for Ita on the deferred side.

The deferred has gone marginally lower sequentially each of the last 2 quarters.

And I just want to make sure I understand.

Are there puts and takes in that?

Is that now more, say, dependent on larger renewals that happen periodically through the course of the year?

So if you can just talk a second about that?

And also, is there anything in that trending that we should think about as we think about services revenue in 2021?

Yes.

No, that's exactly what it is, right?

I mean, it's almost all services now, and it's really going to move around more by the term and the timing of just those renewals, right?

Whether I renew 1 year, or 3 years, it doesn't really make any difference in the business, but it will show up in that deferred revenue line item, right?

The revenue will still be pretty consistent, right, because you're recognizing it over time.

But the deferred line will move around, right?

So if we have a big renewal on a multiyear contract, that number is going to go up.

If we renew just 1 year and come back to do the next later, it won't increase or it might decline slightly because we're recognizing some of that revenue.

So it's going to move around a bit, but it's not a business driver, right?

It's just more how are we negotiating and closing out [the term].

This concludes the Arista Q3 2020 Earnings Call.

We have posted a presentation which provides additional information on our fiscal results, which you can access on the Investors section of our website.

Thank you for joining us today.

And everybody, please be safe.

Thank you for joining, ladies and gentlemen.

That concludes today's call.

You may now disconnect.