Arista Networks Inc (ANET) 2017 Q3 法說會逐字稿

Welcome to the Third Quarter 2017 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. Charles Yager, Director of Investor Relations. 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; Ita Brennan, Arista's Chief Financial Officer; and Marc Taxay, Arista's Senior Vice President and General Counsel.

This afternoon, Arista Networks issued a press release announcing results for its fiscal third quarter 2017. If you'd like a copy of the release, you can access it online at the company's 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 2017 fiscal year, industry innovations, our market opportunity and the impact of litigation, which are subject to the risks and uncertainties that we discuss in detail in our documents 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 statements apply as of today and should not -- 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, Charles. Thank you, everyone, for joining us this afternoon for our third quarter of 2017 earnings call. I am pleased to report that we had another solid quarter. We achieved record revenue of $437.6 million, growing 50.8% sequentially year-over-year, coupled with strong profits with a non-GAAP earnings per share of $1.62 and record operating margin of 38.6%. Services contribution was approximately 13% of overall sales.

From a geographic perspective, our customers in the Americas made up 71% of our total revenue, while the international theaters contributed higher than normal at 29%. This can be attributed to the higher shipments and organic growth this quarter. We delivered non-GAAP gross margins of 64.4%, well-balanced across our verticals and product mix. Our top customers were represented by all 5 verticals in the following order: cloud titan, Tier 1 and 2 service providers, cloud specialty providers, high-tech enterprises, and financials. Our new customer acquisition continues to be brisk especially in the international theaters. A key highlight in the high-tech enterprise vertical was the highest number of customer acquisitions in one quarter, signaling the mainstream arrival of million-dollar customers in the enterprise vertical.

In terms of new introductions in Q3, in this September, Arista introduced a very strategic offering, moving from legacy places in the network to places in the cloud, called PICs. Our Any Cloud software platform is based on our virtual EOS router and CloudVision cloud tracer for visibility metrics. We are truly enabling hybrid cloud networking with leading partnership offerings, including Amazon AWS, Microsoft Azure Stack, Google GCP, Oracle Cloud and Equinix Exchange. Arista Any Cloud is, indeed, a profound approach as it extends the compliance and visibility of our EOS from the private premise to the public cloud workload.

This September, Arista showcased our state-of-the-art media and entertainment offerings at the International Broadcast Conference in Amsterdam, including live, multi-vendor video workflows with over 15 technology partners. Our (inaudible) programmability and visibility is well suited for media broadcasters and content distribution, guaranteeing real-time delivery. Arista is truly migrating analog video, SDI, to digital IP for video editing, rendering and virtual reality.

In terms of Q3 recognition, we received some key awards that we're really proud of. The Bay Area News Group published the Bay Area News Top Workplaces, and Arista has been named #5 out of 95 companies in the large company category with more than 500 employees. We also received accolades from Fortune magazine as #10 out of 100 of the top fastest-growing companies in America.

I would like to comment a little bit now on cloud spending for that seems to be a popular and keen topic these days. As you all know, Arista has experienced unprecedented growth in the first 3 quarters of the year with stronger-than-expected demand from our major cloud titan customers. Due to this unforeseen demand for initial 100 gigabit routing and switching deployment, we have been experiencing almost 50% growth in the past 3 quarters instead of the more normal 20% to 30% consensus growth being modeled by our analysts. Following the July ITC order, we introduced our 945 workaround features in mid-September. Consequently, this lengthened our customer testing cycle and qualification time, extending the completion for 3 of our cloud titan customers into Q4 2017. Therefore, we saw a slightly lower cloud titan business in Q3 than we expected. I am comfortable that this near-term impact of certifications do not affect our cloud business trajectory.

In summary, Q3 once again reflects Arista's proven track record of execution despite some challenges we faced. In Q2 2017, we steadily gained market share in high-performance data center switching both in ports and in revenue according to all the major market reports. Crehan Research has reported that Arista had 14.2% and 15% in dollars and ports, respectively, of market share. Combined with Arista's superior quality and architectural [5A] innovations, we are experiencing a broader inflection and acceptance into the marketplace.

And with that, I'll turn it over to Ita, our Chief Financial Officer, for the Q3 2017 financial specifics. Ita?

Thanks, Jayshree, and good afternoon. This analysis of our Q3 results and our guidance for Q4 '17 is based on non-GAAP and excludes all noncash stock-based compensation expenses and legal costs associated with the ongoing lawsuits. A full reconciliation of our selected GAAP to non-GAAP results is provided in our earnings release.

Total revenues in Q3 were $437.6 million, up 15.8% year-over-year and well above our guidance of $405 million to $420 million. Service revenues represented approximately 13% of revenue, consistent with last quarter. International revenues came in at $128.3 million or 29% of total revenue, up from 25% in the prior quarter. While some of the shift in geographical mix represented the timing of U.S. shipment, we also experienced healthy growth in our in-region international businesses in the period.

Overall gross margin in Q3 was 64.4%, consistent with last quarter and favorable to the midpoint of our guidance of 61% to 64%. Higher-than-anticipated revenue levels, combined with the more favorable customer mix, contributed to this outperformance.

Operating expenses for the quarter were $112.9 million, down slightly from last quarter. R&D spending came in at $68.6 million or 15.7% of revenue, down from $70.9 million in the prior period. This reflects reduced prototype and NRE spending offset by continued headcount growth. Sales and marketing expense was $35.5 million or 8.1% of revenue, up from $34.6 million last quarter, with increased headcount and demo-related cost.

Our operating income for the quarter was $168.8 million or 38.6% of revenue. Other income and expense for the quarter was a favorable $1.4 million. And our effective tax rate was 24.7%, resulting in net income for the quarter of $128.2 million or 29.3%.

Our diluted share number for the quarter was 79.3 million shares, resulting in a diluted earnings per share number of $1.62, up 95% from the prior year.

Legal expenses associated with the ongoing lawsuits came in at $7.9 million for the quarter. As a reminder, these legal costs are -- and related expenses are excluded from our non-GAAP results discussed above.

Now turning to our balance sheet. Cash, cash equivalents and investments ended the quarter at approximately $1.3 billion. We generated $205.9 million of cash from operations in the September quarter. This reflects strong net income performance combined with a decrease in supply chain related working capital.

DSOs came in at 45 days, down from 61 days in Q2, reflecting the timing of billings in the quarter. Inventory turns were 1.7x, up from 1.6x in Q2. Inventory increased to -- decreased to $333.2 million in the quarter, down from $363.8 million in the prior period. This reflects reductions at both the raw materials and finished goods level as we continue to optimize our supply chain. In addition, consistent with last quarter, we maintained a further $32 million as inventory deposits recorded in other assets at the end of the quarter.

Our deferred revenue balance was $565.1 million, up from $554.5 million in Q2. This balance continues to be made up of short- and long-term service contracts and product-related deferrals associated with acceptance terms and future deliverables. Our products deferred revenue declined slightly in the quarter, contributing to our revenue outperformance. Accounts payable days were 19 days, down from 51 days in Q2, reflecting the timing of inventory receipts and payments. Capital expenditures for the quarter were $2.6 million.

Now turning to our outlook for the fourth quarter. We expect revenues to range from $450 million to $464 million, representing 40% plus year-over-year growth at the upper end of our range. We continue to work closely with customers on the qualification of our 945-related product redesign, and we expect this to have some impact in our business in the quarter. Based on our current outlook, depending on the timing of shipment and product acceptance, we expect our products deferred revenue balance to decline in the period.

Operationally, we now have access to international sourcing for our redesigned products following the recent Federal Circuit ruling. As part of our overall supply chain strategy, we will continue to leverage our U.S. manufacturing and some U.S. sourcing capabilities as we move forward.

We experienced significant operating leverage in Q3, with operating margins of 38.6%, well above our historical run rate. These margin levels reflect exceptional top line growth in the last number of quarters, with revenue growth outpacing spending. While we will remain cautious in relation to our spending ramp, we expect that over time, our operating expense investments will gravitate towards our long-term model of 20% R&D, 10% sales and marketing, and 3% G&A . In particular, with reference to the fourth quarter, we expect some significant R&D related investments that will result in higher spending in this area.

With this in the backdrop, our guidance for the fourth quarter, which is based on non-GAAP results and excludes any noncash stock-based compensation expenses and any legal costs associated with our ongoing lawsuits, is as follows: revenues of approximately $450 million to $464 million; gross margin of approximately 63% to 65%; and operating margin of approximately 30% to 32%. Our effective tax rate is expected to be approximately 27% with diluted shares of approximately 80.1 million. Please note that based on our current outlook, we expect costs associated with the ongoing lawsuits to be approximately $12 million for the quarter. As a reminder, these lawsuit-related costs are excluded from the third quarter non-GAAP outlook we provided above.

I will now turn the call back to Charles. Charles?

Thank you, Ita. We are now going to move to the Q&A portion of the Arista earnings call. (Operator Instructions)

(Operator Instructions) Your first question comes from the line of Stanley Kovler of Citi Research.

I just wanted to ask you specifically about cloud, as you anticipated. And the question is really -- some of your competitors were talking about challenges more in the routing area, and so there are some architectural shifts that are happening in the space. I know your customers, you said, are going through a period of digestion for your products. What's the opportunity for the regional routing piece of the business into those accounts? Is there an opportunity for additional growth or for growth to accelerate particularly when it comes to FlexRoute licenses and things like that?

I'll take the question, Stanley. Thank you. So absolutely, I think Arista has been experiencing a changing architecture and 100 Gb transition since we announced these products in 2016. So this is not new for us. And really, we pioneered this change with the FlexRoute license, where increasingly, what we're seeing as a trend across our customer base is more and more of them want to replace routers with routing. And frankly, legacy vendors are on the wrong side of this trend. So as they may be experiencing a contraction in their routers, we are experiencing an inflection in routing. And our FlexRoute licenses continue to increase. You may remember I had challenged the team to do over 100 customers with FlexRoute in the very first year, and we achieved that last summer and now we've exceeded 150 customers.

And if I could just follow up more on the deferred revenue side. Can you help us understand the flexibility you have in revenue recognition and walk us through once again the varying pieces of deferred revenue? And how we should think about that beyond the 2 quarters?

Yes. I mean, again, I think the makeup of deferred revenue is really 2 buckets, right. One is services, which is kind of short-term and long-term services, and that gets recognized as those services are delivered over time. And then we have the product deferred revenue, which is really around acceptance clauses with customers and the delivery of new features, all right. And I think the timing of that depends on when you ship and it depends on when the customers accept. In Q3, we saw a slight reduction in deferred. We would have -- still have exceeded the top end of our range without that. So I think it was pretty minimal. As we look into Q4 right now, we'll see what we come out. We expect that it will come down based on what we can see now, and then we'll see how it plays out as we go through the quarter.

Your next question comes from the line of James Faucette of Morgan Stanley.

I wanted to ask you, Ita, on the profitability. I know that you've had a tremendous couple of last quarters especially relative to your expectations for revenue, and you've also outlined kind of that you planned to get back to your operating model. But what about the gross margin line? And that also continue to do quite well and as a result, you're pulling through. Again, it looks like for the December quarter, total operating margins, that look to be well above your long-term target. Does it make sense for us to start thinking about your long-term target being closer to the high 20s or 30%? Or how should we think about that?

Yes, I mean, I think what we're seeing certainly when the gross margins are operating in the 63% to 65% is that much as we've thought kind of putting this out there as the target, it's becoming a 30% to 32% operating margin business. And I think we're comfortable with that certainly for the near term. Again, the investment pieces, I think we're clear on. It will take us time to get there just given the rate that we've grown at. But the gross margin, we'll see what it does, but I think those are the drivers, right.

Your next question comes from the line of Vijay Bhagavath of Deutsche Bank..

My question, Jayshree, is on -- 100 Gb port sizes are declining, but Arista is beating numbers, in fact, crushing numbers. So I'd like to get your bigger picture view, Jayshree, on what's driving the growth in 100 Gb ports. Is it workload growth? Is it build-outs of new data center capacity? Is it legacy port refresh to 100 Gb? And how do you prioritize these demand drivers in terms of port growth?

Vijay, that's a good question. I think the 100 Gb transition has surpassed even our expectations this year. And as you might know, most market studies show us now at #1 share with about 40% market share. I think the main reason we're seeing that is because the 100 Gb has become the Universal Spine not only in the data center, but it's sort of stretching, I should say, to incorporate a lot of the switching and routing for regional and data center interconnect. So what we're seeing is this trend that -- the architectural trend that we have been espousing for a couple of years now, where the LAN, the WAN, the Layer 2 and the Layer 3 are all coming together to be a highly elastic Universal Spine. And this year has been definitely an inflection year and a starting year. But I must -- underlying that, I see many years of robust growth of 100 Gb ahead, and we're just beginning this year. And with the entry of 400 Gb, I don't see that the 100 Gb will -- anyway, will change. So I think what you're seeing is a significant investment with our top customer base across all 5 verticals. And while we may be a little more aggressively penetrated in the cloud titans, it's really applying to all the verticals.

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

Appreciate it. Jayshree, if I could take you back to the Web 2.0 architectural sort of question, the specific question being, are any of the cloud titans, those top 4, 5, using FlexRoute at present? Or have any of them made awards to you to use FlexRoute going forward for the peering DCI use case? And a related question, I've seen the numbers. They speak for themselves, but are you seeing any inventory buildup? And how much visibility do you all have as to the quarterly rhythm of demand from -- especially that cloud titan group? I appreciate it.

Sure. Because I have a short-term memory, I'll take the first question and then I'll go to your second, so visibility with our cloud titans. We continue to have an extremely intimate relationship with them. We have not seen an appreciable change in the visibility or the competitive landscape. As we have repeatedly said, because they're in a high-velocity, high-growth mode, the visibility we get is typically good for 1 to 2 quarters. And even then, it can change. It was bigger than we saw in the first 3 quarters of this year, but it's a very strong communication. There's very strong intimacy. This is an area where not only Anshul Sadana, our Chief Customer Officer, participates, but our entire engineering team is very involved in the road map, in the demand, along with our systems engineers and sales teams. So I would say we've got good visibility for 1 or 2 quarters, and we've got medium to low visibility thereafter. And probably, that's just generally true. It's the nature of the business. In terms of FlexRoute license and routing in general, absolutely, some of our early adopters have been the cloud titans right from the beginning in 2016, and they continue to be major deployers.

Jayshree, those folks who are adopting FlexRoute, they were previously using an MX router or the ASR 9000? Any sense you can give us for what you displaced then?

That's correct. We were -- just to put this in context, until 2016 or 2017, we were only in the leaf and spine layer. With the arrival of the FlexRoute license and the ability to scale and improve and just basically provide more routing features, we entered additional layers of the spine for routing and data center interconnect, where traditionally, they were core routers either from Juniper or Cisco.

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

I guess, I have -- I guess it's multiple questions, but they all tie into one thing, which is your R&D investment. You've talked about some delays in certifying the new hardware for the cloud guys, and it's understandable since it wasn't just a simple software workaround. It was a little bit more complex than that. But that delay plus the fact that R&D expenses are now sequentially down for 2 quarters in a row, I guess this comes together into this question of, are you underinvesting? Are you at a point in time where if you don't ramp very quickly, it will hurt you soon somewhere -- sooner rather than later? And again, tying it to -- Ita, to your guidance, your guidance implies a massive expansion of OpEx on a quarter-over-quarter basis. And even last year in the fourth quarter, that has not happened. So unless you have an -- a unique ability to absorb hundreds of employees in an extremely short time frame, which I doubt you can because you are a very picky about who you pick, help me tie this together. How can you get to those expense levels? And on the flip side, Jayshree, how can you keep investing and keep up with everything that's coming up? And the legal certainly throws another wrinkle into this. Some resources have to be dedicated to that. When does this start hurting your business, the fact that you don't have the headcount available?

Okay. Well, we'll try to keep our answer shorter than your question, Ittai. How about that? I'm teasing you. Well, let me see. So I think our engineers know that they have unlimited hiring capabilities, but they have a very high bar. So we continue to hire very, very well in engineering headcount. But frankly, many of them don't make our bar. So if we are limiting hiring, it's not because we don't want to hire but because we are limited by the talent pool in a geography. So one of the things we are doing is expanding geographies beyond our headquarters here in Santa Clara, but I just want to high five and give kudos to our engineering team for dextrously developing new products, doing very good engineering, recruiting and still keeping the bar high. I'm very proud of that. Having said that, what seems to decline and go up and down are the other variable spend. We don't control, sometimes, the chip and when they come out and the associated prototype and the NRE spending to go with that. So I think in general, we're doing better in headcount on the engineering investment side, but we could do even better and we expect to in spending more money in Q4 in the non-headcount department. In terms of the legal, there is no question in mind that we dedicate a lot of engineering resources to legal. However, over the last 3 years, which is our third year of legal, we've been doing workaround for many, many years. And I would say more time was spent this quarter not on -- necessarily on just developing the legal workarounds. Clearly, we had to scramble to do more of that and get it out in September, but also in qualifying it. That's where our systems engineers and our sales and marketing resources were more tested. So overall, I'm comfortable and confident, Ittai, that we can continue the headcount ramp and that we can grow into new markets organically. And there is always a possibility we do things inorganically as well. Ita, do you want to add to that?

Yes. I mean, Ittai, I think the thing to remember is that the R&D headcount and that part of it has been growing consistently, right. And where you see the lumpiness is really all around prototype and NRE-type expenses, right. So I think in Q4, as we sit here today, we believe we'll spend some meaningful money on some of those kind of onetime items in the fourth quarter. And we'll continue to ramp headcount, right. I think over time, our goal is to get back to that 20% of revenue.

All right, good. Just a follow-up, Jayshree, as you look into your future road map for '18, has it been delayed because of the legal workaround work or pushed out?

In some cases -- yes, in some cases, we have had to push out some of our priorities for customers to prioritize legal workarounds that were more time-based, yes.

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

So I'll keep it short here. When we think about these 3 trials that got delayed on the 945, where these projects that were identified and were going to be revenue this quarter and now they'll just kind of move into Q4 once the 945's work has been certified, is that -- is there real line of sight on this business?

Look, as I said, we have pretty good line of sight with our cloud titans. And absolutely, there were projects that were identified. Let me reiterate and be really clear that we did not lose to any competitor. It just has -- it's taking a long time. It can take anywhere from 2 weeks to as much as 10 weeks to do these certifications. So lengthening the call cycle and getting them in workarounds in mid-September didn't give us much of a chance to certify them in Q3. So we expect most of them will move to Q4 or, worst case, Q1.

Have you seen any shift in top-of-rack versus aggregation mix within the cloud titans, say, over the last couple of quarters and what they're communicating to you about their plans going into 2018?

From everything I can see, the mix has stayed the same. Our -- obviously, our aggregation has moved more to 100 Gb. And our top-of-rack has moved to 10, 25 and 50 Gb in the least. So -- but no, we haven't seen any big change in mix.

Okay. And I don't know -- is Marc in the room? I just had a question about...

We need to limit the questions.

Can we go to the next question, Alex, and we'll get to you later?

Yes, we'll ask him to call back, no worries.

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

I wanted to see if we could follow up on this hot topic of what's going on in the cloud. So I understand your explanation. I was hoping we could get a little bit of quantification around why the cloud -- or the degree that the cloud titans were softer than you expected. And you cited the evaluation which, while it makes sense to me -- I guess I'm trying to understand the bigger industry picture and hoping maybe you could help us understand what's going on with other companies. So certainly, your direct competitors, where you're taking routing business, that makes sense why they would see softer business. But what about all the other companies in the supply chain and to web scale citing slower demand? Can you help me put that in perspective?

Well, it's difficult enough to talk about my company, let alone help you with other companies. I'll try, Simon. I think if I step back and look at the big picture, the reason Arista is less affected by what other companies are -- is -- frankly speaking, we're on the right side of these trends. We're developing 100 Gb. We're developing routing. We're developing the right optics and DWDM capability, and these are all the modern architectures the cloud wants to move to. So I guess we don't have the overhead of legacies, the best way I can say this. We're also underpenetrated although the cloud has been our most important vertical for some time now. The 100 Gb movement is -- it's very early stages. I'm guessing even with the titans who were some of the earliest adopters, it's in the -- still in the 10% to 20%. So while the others may have had more absorption -- and I can't really speak to or may not really have seen this architectural shift coming, I think Arista has been thoughtful that -- and built the right products. And these are all new products that we pretty much put out in the last 6 quarters. And I think that's -- we're benefiting from the trend. I think the second thing is we're not relying on one titan. It's really important to understand that when we say cloud, increasingly, it's not a vertical. It's a horizontal. Cloud principles are being deployed across all our 5 verticals. So our top 20 customers, 30 customers are all deploying cloud architecture. Now the scale varies greatly with the titans and is different, obviously, for the other customers. And so therefore, even if we experience like we did, a delay in certification and a lower contribution from the titans, the titans are still #1 in aggregate because of the number of customers there.

And can you help us with the quantification aspect, whether percent of revenue or how different it was from what you expected? Just any quantification element.

I said it was slightly lower from prior quarter. So hopefully, that quantifies it, yes, and then...

I was looking for a number.

Yes.

I don't know if you can put a number on that. I mean, I think at the end of the day, what we're saying is that we're working through the qualifications. We've seen some effect from that. But the business with these customers and the activity levels of these customers, we haven't seen any change from that.

(Operator Instructions)

Your next question comes from the line of Erik Suppiger of JMP Securities.

On the routing products, can you talk about what kind of discounting you've been offering and how the gross margins on that compare to your traditional switching products?

Yes.

Do you want to take it, Jayshree?

No. Sure, go ahead, Ita.

I think we've always said, Erik, that the routing is accretive, right, and we're definitely -- we're creating significant value there versus what's available prior to this shift. And I think we see that, that is accretive to the gross margin.

Do you -- we've heard that you're not discounting nearly as much as on the switching products. Can you talk about what kind of discipline you've had from a discounting perspective?

I think our discipline applies to both routing and switching in general. The customers see the value of our quality of software, the programmability, the underlying architecture of being network-wide, database-oriented, the analytics, the automation, the single binary image. So both of that applies to both switching and routing. Routing, in general, tends to be fewer boxes and fewer ports. So in terms -- but higher margin. And switching is connecting to the servers and storage. So it tends to have more pricing pressure.

Your next question comes from the line of Alex Henderson of Needham.

I was just hoping you could your clarify your comments. So on the one hand, you're saying that the qualifications for the cloud guys has taken a little longer and it's pushing business out of 3Q and then into 4Q. And then on the other hand, you're saying that that's not an impact on the numbers. And I'm not sure I understand what you mean by those 2 statements. It seemed to be contradictory.

We're saying that we didn't do as well in the cloud titan in Q3 as we did in Q2. That's what we're really saying. But that doesn't mean the other types of customers didn't contribute to our number. So all 5 verticals did well, and it was very balanced. Cloud titan is still our #1 vertical. So I don't think they're mutually contradictory statements. They're actually connected, right.

So if the cloud titans...

I'm just saying -- yes, I'm just saying (inaudible)

If the cloud titans are coming back in on the fourth quarter and the other guys don't change trajectory, why wouldn't it accelerate in the fourth quarter then?

I mean, we're growing at, like, 50% year-over-year on a quarter-over-quarter basis. I don't know that it accelerates off of that. I think Q3, we had some impact, and Jayshree said it was a slight impact from the qualifications and so on. And we will see what happens in Q4, right. I mean, we have work to do there, and we're working through that with them.

Your next question comes from the line of Mark Moskowitz of Barclays.

I have a 7-part question. I'm just kidding. One question from me. Kind of building off the last question actually. So given your commentary around certification, I'm just kind of curious in terms of, can you talk a little more about the other customers who are doing well, shouldering some of maybe the certification overhang from the cloud titans? Because I'm trying to get a sense as you move into 2018 and beyond, are you going to start standing firmer and taller on other verticals beyond just the cloud if everything kind of works in unison? Is that routing? Is that enterprise? Like what verticals are helping offset some of this cloud certification overhang?

Right. I think if I look into 2018, some of the fastest growing verticals, besides obviously a clear contribution from cloud titans that we expect, is service providers, high-tech enterprise and cloud specialty providers. All of these 3 are smaller today for us than cloud titans but have ample opportunity for growth. In the case of the service providers, they are going through a re-architecture. They are operationally very dependent on their legacy architecture. So it takes time to move. But we feel optimistic that there are some changes happening there and have already happened there by virtue of service providers now being our second largest vertical for 2 quarters in a row. The cloud specialty providers, they look a lot like the titans except their sizes are different. And probably my most exciting message to you would be, we feel like the enterprises have a lot of legacy fatigue, and they are starting to show some real momentum. And I expect that the strength we showed here with both international and with enterprise customers, this was our largest number of million-dollar enterprise customer quarter. I hope it's not a one-quarter trend and it continues and they will have some contribution in 2018. Mark, is that okay? Did I answer all 7 parts of your question?

Yes, thank you -- just one question, thank you.

Your next question comes from the line of Tejas Venkatesh from UBS.

I'm on for Steve Milunovich. I wanted to better understand the lengthened call cycles you're seeing in the cloud. Is that because your workaround includes using a different silicon vendor?

Not particularly. So just let me kind of step back and give you some description of the quarter because it really was an interesting quarter. We -- in the beginning of the quarter, in early July, we were shipping normally. And then on July 5, we got the cease-and-desist order. Right, Marc, around then?

Yes.

Okay. So we had a busy season because as soon as we got the cease-and-desist order, you might remember that these are the same 2 patents that were invalidated by the PTAB that now the International Trade Commission, ITC, issued an order on. So our engineering team had to scramble to get out a workaround on basically invalidated patents, that dichotomy going on. So we introduced a new EOS release, 4.19, around mid-September. And so now we are actively engaged. We only had 2 or 3 weeks in the quarter to certify, and we did do a lot of certifications. But these qualifications times can be related to software only. Or in some cases, in the case of the ACL workaround, it can be a combination of software and hardware. And both are features that are not transparent to the customer. In other words, the customer has to put it in, and it really depends on the use cases and scale and complexity. So if it's a simple network, they may not need to do much and 2 weeks may be fine. If it's more complex and they have a lot of ingress and egress [echoes], then you have to go through more qualification. So that's why you have that wide range because we're committed to non-infringing workarounds on invalidated patents, I might add. However, that engagement with the customer can be short or long.

And one final question. Could you comment on your software-only revenue and how meaningful it's becoming? You mentioned over 150 FlexRoute software licenses and you also have CloudVision.

Yes -- no, I -- we not only increased -- thank you for letting me speak about that. This is a proud moment because I think it's an indication of how the networking business is truly migrating to software-driven networking. We not only exceeded 150 routing customers, but we have now also crossed 200 CloudVision customers. And we have several trials going on. So my anticipation is while it is not yet reportable -- so it's not material revenue and, again, because it has -- it's a software subscription over multiple years, I expect in the next 2 years that this business can become approximately a 5% of our revenue. I guess you'll hold me to that. But check in, in 2 years.

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

I guess I wanted to ask about some of the latest developments on the routing side. Obviously, there's new silicon in the marketplace. I know you guys are continuingly kind of working on feature development. The question is, as you continue to drive things on the silicon side and feature side, talk about those features and capabilities and how they expand your TAM on routing.

Okay. Well, I think there are 3 aspects to how we've gone about our routing endeavor from a development point of view. One, as you rightly point out, is just using, pardon my French, some kickass merchant silicon, right. And what I mean by that is everybody gets access to the same silicon, but somehow, we've been able to get more routes, more data packs, more convergence, more scale, more services out of it. And this is a true engineering feat that we apply not just to silicon, but this is, indeed, one of our closest partnerships with Broadcom especially on the Jericho family and chipset to -- and it's something we've been doing across 3 generations, PETRA, Arad and Jericho, to really maximize the feed, speed scale and performance. And I cannot underscore that enough. The second thing is we augment it and we complement that with our own coprocessor silicon drivers, et cetera, and a set of capability that the silicon may not have to enhance it. And again, this is very important because in routing, there's routing features and then there is use cases that we go into. And so we've been going into some very specific use cases in the content, media, cloud, routing aggregation, high-performance core, even some 5G wireless. So that -- those become the way we really approach it so that we're not just building again yet another router 20 years later, but we're really zooming in on the importance of the features we have, including the third area, which is our EOS programmability and routing protocols. We've got just about every 4-letter acronym covered there, including a BGP stack that is much more robust today as we've been working on it for several years; control planes with VXLAN, EVPN; and segment routing and also MPLS features with some traffic engineering. So it's really been a 3-pronged approach to attack and provide simplicity and elegance in our architecture and yet state-of-the-art performance and protocol as well.

And then the TAM question. I guess I'm just curious about how much you think you're driving the TAM up.

I think we're contracting the router TAM and increasing the routing TAM, but it's always been difficult to call it out as a separate TAM because it sits on top of our switches. So I think the platform is the same but the software TAM is jumping on it. So it's a shared platform with multiple features on it.

Your last question comes from the line of Hendi Susanto of Gabelli & Company.

You call out strong international sales in Q3. Where do you see strong international sales in terms of regions, verticals and use cases? Are there certain use cases that drove strong international sales in Q3?

Yes -- no. This is one of our prouder moments, I think, because we had a lot of international shipments in Q3 and we had a particularly strong Asia Pac and EMEA. Mark Foss, you've been driving some of this. You want to comment on it?

Yes, sure. As mentioned, the Americas business was about 71% of revenue. It broke out to where EMEA was 18% and the APJ was about 11%. And as Jayshree mentioned, we had very strong quarter for new customers acquired internationally, with over half of the customers -- the new customers we acquired were international. But the organic international customer revenue has been growing faster than the company average. This removes out any global customers which may be headquartered in the United States. But the organic international customer revenue is actually very strong for us.

Yes. And I want to also put in a plug for the leadership team there, [Naresh] in Asia Pac; Guenther and Ken Kiser who are running EMEA and the international regions. We've been investing in them now for the last year or 1.5 years with Mark Foss' leadership. And I think we're finally seeing the fruits of our labor here. While I don't always expect it to be 29% because the cloud titans might contribute to change that mix, I think that's a sign of things to come.

So use cases are broad-based?

Yes, use cases and country adoption is broad-based.

This concludes the Arista Q3 2017 earnings call. I also want to mention that we had 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, ladies and gentlemen. This concludes today's call conference call. You may now disconnect.