Arista Networks Inc (ANET) 2019 Q2 法說會逐字稿

Welcome to the Second Quarter 2019 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 of the Arista website following this call.

I will now turn the call over to Mr. Chuck Elliott, Director of Business 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 second quarter ended June 30, 2019.

If you would 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 third quarter of the 2019 fiscal year, industry innovation, our market opportunity, the benefits of recent acquisitions and the impact of litigation, 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, Chuck.

Thank you, everyone, for joining us this afternoon for our second quarter 2019 earnings call.

Our profitability growth combination was once again demonstrated with a non-GAAP revenue of $608.3 million, while non-GAAP earnings per share grew to a record $2.44.

Services contributed 15.6% of revenue.

We delivered non-GAAP gross margins of 64.7%, influenced by our solid performance from our enterprise vertical.

We registered record number of new customers in calendar Q2 and continued to drive new customer logo expansion at the rate of 1 to 2 per day throughout the quarter.

In terms of verticals, the Cloud Titans segment remained our largest vertical.

The modern enterprise high-tech segment is now consistently becoming our second largest, with financials in third place, Tier 2 specialty cloud providers and service provider coming in at fourth and fifth place.

In terms of geography mix, in Q2 2019, the international contribution was 27%, while the Americas were at 73%.

We had a banner quarter for new products in Q2 2019.

We launched 2 400-gigabit product families during Q2, with the Arista R3 series for modular 7500 and 7280 models, as well as a brand-new Arista 7800 chassis family for 400-gig switching and routing based on the Broadcom Jericho2 silicon.

This is enabling our flagship EOS and uncompromised multi-terabyte capacity and availability.

We have now launched 10 400-gig platforms, and Arista has more 400-gig products than any other peer.

We have begun active product qualification with more meaningful 400-gig revenue really expected next year in 2020.

Given the recent industry news, I wanted to take this opportunity to comment on 400-gig optics.

So to join me on this is Andy Bechtolscheim, our Chief Development Officer and Chairman, who will speak more about this.

Andy?

Thanks, Jayshree.

The first observation on optics is that in the cloud, pluggable optics have led to a disaggregated business model between the switch and the optics vendors, with virtually all optics in the cloud being purchased directly from optics vendors.

Cloud providers typically qualify at least 3 optics vendors to ensure lowest cost and diversity of supply, and we don't see that changing with 400-gig.

In the case of 400-gig CR, which is the long distance optics that won't ship in volume until mid-2020, we're aware of 1 dozen optics module vendors that plan to offer compatible 400-gig CR modules, competing on the basis of price, quality and volume availability.

We do believe that 400-gig CR will be a very competitive market with competition that will drive unprecedented price performance improvements for 400-gig coherent optics.

We work closely with our largest customers to qualify all 400-gig optics, including 400-gig CR that are relevant to them, with the objective to deliver the most cost-effective 400-gig optics solutions to the market.

Thank you, Andy.

So speaking of new products, we also introduced Arista's first entry into the Cognitive Campus edge with our 720XP power-over ethernet switches and our new WiFi 6 offerings, all of which are supporting CloudVision for the campus, slow base telemetry and security segmentation services.

With this, Arista establishes an exciting and formal complete cognitive portfolio addressing the transitional changes in the campus security and IoT era.

We are in early field trials now and we expect more results in second half 2019.

So speaking of second half 2019, as you all know, we experienced some turbulence in Q2 2019 with the pause of a specific Cloud Titan set of orders.

They have now resumed spending, and we expect stabilization in second half 2019 for the overall Cloud Titan spend.

Certainly, second half will be an improvement over the first half, but we do not expect the cloud momentum to be a repeat of second half 2018.

Naturally, these trends are consistent with the annual cloud CapEx forecast reported in recent weeks.

Our enterprise segment is healthy, with growing interest in our campus and multi-cloud migrations.

On June 6, 2019, we celebrated our 5-year IPO anniversary at the New York Stock Exchange with both our premier customers and analysts.

Our deep collaboration with Microsoft was evidenced with CEO Satya Nadella joining me at our special event as our chief guest.

Together the 2 companies share a synergistic vision in cloud area networking.

We announced Microsoft Azure cloud integration, including vWAN and IoT Central.

As our customers migrate to a cloud-led strategy, bringing holistic clients to any cloud experience, we are seeing a compelling conviction in Arista as their strategic partner.

I'm proud to share that for the fifth consecutive year, we have also attained a status as the leader in Gartner's Magic Quadrant for data center networking, with our strongest showing yet in both vision and ability to execute.

And with that, I'd like to turn it over to Ita for more financial specifics.

Thanks, Jayshree, and good afternoon.

This analysis of our Q2 results and our guidance for Q3 '19 is based on non-GAAP and excludes all noncash stock-based compensation impacts, certain acquisition-related charges and other nonrecurring items.

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

Total revenues in Q2 were $608.3 million, up 17% year-over-year and above the midpoint of our guidance of $600 million to $610 million.

Service revenues remained strong, representing approximately 15.6% of revenue, up from 15.1% last quarter, reflecting a healthy level of renewal activity.

International revenues for the quarter came in at $162 million or 27% of total revenue, up from 26% in the prior period.

Overall gross margin in Q2 was 64.7%, above the midpoint of our guidance of 64% to 65% and up from 64.5% last quarter.

Gross margin in the period benefited from a lower cloud contribution combined with healthy enterprise and services performance.

Operating expenses for the quarter were $158.7 million or 26.1% of revenue, down slightly from last quarter at $160.7 million.

R&D spending came in at $101.7 million or 16.7% of revenue, down from $106.5 million last quarter.

This reflected lower levels of new product-related NRE and prototype spending in the period.

Sales and marketing expense was $45.1 million or 7.4% of revenue, up from last quarter with increased head count somewhat offset by some reductions in other sales costs.

Our G&A costs were $11.9 million or 2% of revenue, up slightly from last quarter.

Our operating income for the quarter was $235.1 million or 38.7% of revenue.

Other income and expense for the quarter was a favorable $13.8 million.

And our effective tax rate was lower at approximately 20%.

This resulted in net income for the quarter of $198.6 million or 32.7% of revenue.

Our diluted share number for the quarter was 81.3 million shares, resulting in a diluted earnings per share number for the quarter of $2.44, up 26.4% from the prior year.

Now turning to the balance sheet.

Cash, cash equivalents and investments ended the quarter at approximately $2.3 billion.

We repurchased $100 million of our common stock during the quarter at a weighted average price of $246 per share.

As a reminder, our Board of Directors have authorized a 3-year $1 billion stock repurchase program commencing in Q2 '19.

The program allows us to repurchase shares of our common stock opportunistically and will be funded from operating cash flows.

We generated $196 million of cash from operations in the second quarter, reflecting strong net income performance, offset by increased working capital requirements and a reduction in deferred revenue.

DSOs came in at 51 days, up from 41 days in Q1, reflecting the timing of billings in the period.

Inventory turns were 2.4x, down slightly from 2.5x last quarter.

Inventory decreased at $314.2 million in the quarter, down from $347.2 million in the prior period.

Our total deferred revenue balance was $502.2 million, down from $536.5 million in Q1.

Our product deferred revenue balance decreased by approximately $38 million in the quarter, reflecting customer acceptance of new features.

Accounts payable days were 37 days, down from 38 days in Q1, reflecting the timing of inventory receipts and payments.

Capital expenditures for the quarter were $3.4 million.

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

As expected, we experienced some softness and demand from our cloud customers in the second quarter.

While early indications are for improved demand from these customers in the September period, we believe that second half growth in this business will remain somewhat muted as compared to prior years.

We expect our enterprise and financial verticals to continue to perform well, offset by some declines in the service provider business.

On the gross margin front, we would reiterate our overall gross margin outlook of 63% to 65%, with customer mix being the key driver.

We will continue to manage investments in the business carefully, prioritizing growth and sales head count and resources as we look to expand our market coverage.

With this as a backdrop, our guidance for the third 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 $647 million to $657 million; gross margin of approximately 63% to 65%; operating margin of approximately 36%.

Our effective tax rate is expected to be approximately 20.5% with diluted shares of approximately $81.9 million.

I will now turn the call back to Chuck.

Chuck?

Thanks, 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 Jason Ader with William Blair.

Jayshree, on the campus side, can you provide any metrics, customer wins, anything that is worth the investment community knowing about in terms of tracking your progress there?

Thanks, Jason.

As you know, we introduced the products in June, at the Analyst Day on June 6, and we said most of them would be available in Q3.

So it's a little early to be giving customer wins.

But I can say with confidence that we are in very many early field trials with customers, to the tune of tens of them.

And Anshul and the team have been having very, very good interactions, and I fully expect that we'll have more results in Q3, Q4 and certainly much of next year.

And how do you respond to somebody that says okay, you guys have done well on the enterprise side with data center switching, but the campus side is a much more complex sale from the standpoint of, typically channels are -- more channels are involved, there's a lot -- there's wireless, there's potentially security, there's just a lot more going on with a campus, major campus deployment.

How do you help people get comfortable that you guys will be able to replicate some of the success you've had on the data center side?

Yes.

So step back for a moment and ask when did we get our success in the data center with enterprise?

It was 5 years after we started shipping products.

We didn't even report much on the data center.

We mostly focused on the [Nisha] financials and the cloud in the early years, right?

So I'm pretty sure we won't take 5 years to enter the campus market, but I'm here to say that the traction with enterprise will really come in 3 categories: the early adopters who already love our EOS, and therefore, that's going to be the fastest place of attraction.

The CloudVision, don't underestimate that, where they're looking for that single point of management and single pane of glass.

And then to your point, the third one will be new channels, new partners, new systems integrators.

So if you look at those 3 segments, we can start playing in 2 out of the 3 already.

So my response would be the campus technology in many ways is no different than the data center, it's very similar in layer 2, layer 3 protocols.

And customers who've appreciated us for the last 5 years at EOS and CloudVision are the first points of success for us.

The second point of success, which is new customers and new logos, will take longer.

But campus for us is a multiyear journey.

And as I said many times before, I'd be very happy if we did -- if our first deal was $100 million because I think it sows the right seeds for $0.5 billion and $1 billion in the future.

But none of us should think this is a overnight 1-quarter journey.

This is a 3-to-5-year journey.

Your next question comes from James Fish with Piper Jaffray.

Congrats on the quarter.

Just one for me is, Ita, how should we think about the impact of the 25% tariff on the gross margin guide?

Obviously, you're reiterating your 63% to 65%.

But if we were to get a trade deal tomorrow, I guess, how much would gross margins be positively impacted?

And or have you done enough to kind of offset it from a supply chain perspective?

Yes.

I think we've been working on improving the supply chain and addressing some of the issues with the supply chain.

And at the same time, obviously, we've had an adder to customers, which we've also been managing, and as the tariff rates have changed we've been fortunate enough, we've made enough progress that we've been able to kind of hold our margin to the increase of that customer adder.

And I think with the new news that we heard just before the call I think that's still the case, right?

We believe that we have done enough from a supply chain perspective that we should have minimal impact.

But I wouldn't think that there is a big swing in gross margin one way or the other, if either it went away completely tomorrow or we continue to see some changes in it.

I think we've done enough work where it's kind of mostly neutral from a gross margin P&L perspective.

James, just to add to that.

With the 63% to 65% range, we think the tariff can have impact on the gross margin, but it will be within that range of 63% to 65%.

And as Ita said, the team, the manufacturing team has done a tremendous amount of work.

We're not immune to the tariff.

We absolutely are affected by it.

And -- but I think the effect will be minimal.

Your next question comes from Rod Hall with Goldman Sachs.

I guess I'd just ask you, you had said that the large Cloud Titan orders had dropped to almost zero in the middle of March, and you didn't really have good visibility on when those might return, it seems like they have resumed spending per your comments.

When would you expect spending there to be back to normal?

Or is there some new level of normal?

Can you just give us some kind of an idea on what sort of visibility you have and how you see that progressing over the next few quarters?

I think, first of all, Rod, the new level of norm has changed.

We shouldn't use 2017 and '18 as our frame of reference, right?

So first half was a real adjustment for us to the norm in 2017 and '18.

Having said that, I think we -- you've all seen the CapEx reports and depending on whose CapEx you're talking about, they've all gone from double-digit growth to single-digit and some of them are negative.

So you can expect that the new norm is no more double-digit growth and is going to hover in the low single digits.

Anshul, did you like to add some more to that?

Sure.

We reiterated, we mentioned this last time, but I want to reiterate, which is there was no design loss.

The Cloud Titan has unpaused so they're back to the normal spending that they do, and the allocations were unchanged.

Is this [the quarter one] that you guys had called out?

Sorry, can you repeat the question?

So you guys had just, to clarify, Jayshree, you guys had said there was inventory and they were using inventory last quarter.

I just wondered where we are in the process of them utilizing that inventory.

Is it all done or they're still utilizing out of inventory as well?

I don't think we made an exclusive comment that they had extraordinary inventory levels.

It's typical to have some.

The real reason for our Q2 turbulence was a very conscious decision on the part of specific Cloud Titan to put orders on pause.

And those have resumed, and they resumed at levels that are improved over the first half, but nowhere close to the second half of 2018.

Inventory was not the reason.

Your next question comes from Ittai Kidron with Oppenheimer.

Congrats, ladies and Andy and Anshul, I guess.

Great quarter.

I guess I do want to drill down again on the cloud.

I just want to make sure I understand that you claim stabilization, I guess, I'm kind of wondering, has there been a change in the way they communicate with you?

Because it was a surprise last quarter, I guess, what makes you comfortable that they're not going to surprise you here?

And I know anything can happen, but has there been a change in the way you communicate with them such that it gives you confidence that there is a stabilization?

Ittai, that's a good question.

I'm not trying to imply that they couldn't make further changes on their business side, right?

What I'm trying to say, and again Anshul can clarify is, we're literally taking this one quarter at a time, and at this point, we see stabilization in Q3.

And anything can change in Q4, but if we had to predict, I think what -- the stabilization in Q3 could carry on to Q4, that's what they are saying.

Who knows what's going to happen in 2020, as your guess is as good as mine.

But maybe Anshul knows better.

You want to add something?

Well, I don't know anything about the future.

But the Q2 communication was sudden but was a very rare event for them.

And otherwise, our communication has been very normal, back and forth, that you expect between a customer and us.

And the engineering collaboration as well as planning for next gen designs.

So nothing is really extraordinary there.

Everything is very, very normal now.

Very good.

And Ita, just so you don't feel lonely here, a question on the OpEx.

At the midpoint of your guide, there's a quite a significant increase quarter-over-quarter in expenses.

Help me -- and I know you probably want to be some conservatism built in there, but nonetheless is there an unusual level of prototyping that's happening in the quarter?

Help me get my hands around how would I explain about a $20 million to $24 million quarter-over-quarter increase in OpEx, which is something that you've never really done?

I think that there's definitely some reserving the right to make some investments if we want to include it in there.

And I think the rest of it is, we did push hard on R&D.

We had talked about that on the last call, that we would prioritize sales and marketing and maybe push a little bit harder in R&D just given the quarter that we were heading into.

And obviously the intention is to kind of not to continue to do that.

So we will -- you will see some increase in R&D as we move through this next quarter.

Our R&D cycle, Ittai, is with new products, right, so we can't always time it.

And we've got so much new product coming out of our ears, it's one of the things we're very proud of and that has a natural impact on prototype expense.

Your next question comes from Alex Kurtz with KeyBanc Capital Markets.

Just a clarification in the question.

So Jayshree, on your comments about the large Cloud Titan and the resumption in the second half.

Do you see them changing how they use their networks and like the capacity and the utilization and "how hot they run their networks" because that one Cloud Titan obviously is chasing another Cloud Titan for business, and I think the underlying investor assumption is they will continue to invest to compete?

And so I'm just trying to understand, do you think, given all the understanding of their network that you have, are they making a pivot in how they run their network?

Again, Alex, I'll comment and Anshul is closer to it, so he can give more detail.

We have not seen any appreciable changes on [garshan] going to optimize for the last megabit of bandwidth or anything like that.

There is a -- there has been a general increase in spend due to the 100-gigabit common denominator across all layers of the leaf/spine network, including the data center interconnect.

That will vary on 400-gig.

Some cloud titans may stay on 100-gig longer, some may go to 400-gig faster, some may actually pick 200-gig.

So we do see the personality performance changes, but we don't see any major bandwidth planning down to the megabit at all.

One thing I'll add and Anshul can comment to that is, one thing we also see is, I've always said we're in the early innings, but that counts on the fact that the cloud titan is going to continue to invest in new regions and new locations for data centers.

More than your performance standard I expect we'll see more planning around where they put their data centers, and some of them may not open new data centers and some may rely on a more incremental strategy.

Alex, most of the commentary in the industry right now about optimizations, I believe, is tied more to compiler, compute virtualization optimizations.

As you know, the networking spend on switches and routers is in the range of 6% to 7% of their CapEx.

So they're not going to try and squeeze that and create a bottleneck which impacts the remaining 93%.

So the network has to run [at 03] and no one is trying to try and optimize beyond what we have managed to do by providing a very competitive offering.

And Ita, on the maintenance support number, it looked like it missed the consensus estimates by a pretty wide margin.

You obviously did well on product.

Was there something that, some mix on the balance sheet as far as deferred or that we should understand as far as why there was a disconnect there?

Anything that we should be aware of contextually around the maintenance and support execution in the quarter?

No.

I mean, I think if you look at the percentage of revenue, it was pretty consistent quarter-over-quarter.

You're talking about the services revenue on the income statement?

Yes.

Yes.

I mean, I don't think there was anything unusual there.

I mean, we don't guide it specifically, right?

It tends to be high in Q1 or Q4, but this was very normal.

Yes.

This is very normal, Alex.

As a percentage of revenue, it wasn't that different.

So I'll go back and look at the consensus numbers, but I think it's not a number we guide and I don't think the year -- the quarter-over-quarter trends looked particularly different to what we would have expected.

Your next question comes from Aaron Rakers with Wells Fargo.

Congratulations again on the quarter.

My question is actually on the enterprise, the traditional enterprise market.

There's clearly been some recent signals of a lot of choppiness.

One of the enterprise system companies preannounced tonight, we saw Intel's choppy results, et cetera.

I'm curious of how you -- what you've been seeing in that market and what gives you confidence that, that market will continue to grow at what sounds to be a very healthy pace through the back half of the calendar year?

I think the enterprise is a small -- we are new to this market, right.

And we are a recent entrant.

So we're not operating off the large base where we're a market leader or anything.

We're the newcomer.

So because we have a large TAM and because we have highly differentiated products, and I think also because there's an awful lot of enterprise fatigue with existing dominance of one vendor and lack of quality and all of that, we are seeing a unique situation despite the macro.

Not to say, if there is a really bad macro, we wouldn't see it.

But I think despite the macro, we're enjoying a little oasis in the desert, if you will.

Your next question comes from Jeff Kvaal with Nomura Instinet.

Question and a clarification I guess for me, please.

On the question, I'm wondering if you all have applied the same methodology to coming up with your guidance as you have in prior quarters.

It's sometimes tempting to adopt a more conservative assumption on close rates or what have you -- after a guide down?

And then secondly, the clarification is, when you say back to sustainable levels in 3Q, Jayshree, were you meaning over the course of the third quarter we'll be back at sustainable levels, or we're at full run rate August 1, game on?

So both your questions are intriguing.

I'm still kind of processing them.

So the first one is, did we guide like we normally do and were we being conservative?

I'm just translating your question.

Is that what you asked?

Yes.

Yes.

Okay.

So I think this word conservative and the label we have is a little bit of a misnomer.

I think we're guiding as best as we can and there's very little sandbagging going on.

We've -- particularly -- of course, we have customer concentration but particularly with the enterprise, there are so many more customers and so much more to forecast, we do our best in analyzing that forecast and this is our best effort.

I wouldn't think there is a lot of buffer in that.

That would be my caution to you, Jeff.

And in terms of the game on August or is it going to happen through the quarter, it is very difficult to ever predict a quarter, especially in its weakest, slowest summer months in some parts of the world.

So I wouldn't say it's game on in August, I would say it's going to be a profit through -- we will need all 3 months of the quarter to execute on this one.

Your next question comes from Thejeswi Venkatesh from UBS.

I wonder if you can comment on what you expect your largest customer to be as a percentage of sales in 2019?

Jayshree, I think earlier in the year you had indicated it would go back to historical levels, which many of us interpreted as 16% of sales instead of the 27% in 2018.

But as we get closer to the end of the year, you must have better visibility so I'm curious to hear that.

Yes, no, that's a good question.

I think my prediction of teens -- mid- to high teens is still what I think is our best estimate, 27% was wonderful, but a rare event.

And then a follow-up on 400-gig.

I know you generally said, you would expect that in 2020, but early part, latter part, and then can you parse how you're thinking about 400-gig between cloud -- routing and cloud switching?

Okay.

Good question.

400-gig, as you can tell from Andy and my talk, we're ready with the products and no problem with that.

But we have been often slowed down by the optics.

And as Andy said, I think he's predicting some of the optics to be in 2020.

Do you want to comment on that, Andy?

Yes.

One of the most important use cases for 400-gig is actually the 400-gig CR data center interconnect, and those optics will not be in volume production until mid-2020.

But we do expect customers to qualify these optics way before that.

So we cannot predict the exact timing here, but it's going to be in 2020.

Yes.

And as for switching versus routing, because a large number of these use cases will be data center interconnect, it will be both.

It will be hard to parse one versus the other.

They'll almost always want an option for routing or start right with routing in the beginning.

Your next question comes from James Faucette with Morgan Stanley.

I want to follow up on that 400-gig question.

Just -- I understand in terms of the availability of optics.

But can you help us understand what stage of evaluation of your equipment customers are in or potential customers are in?

And can they fully evaluate and qualify the products without the availability of those -- commercial availability of those optics?

Sure.

James, the way to look at this is, many of the new products for 400-gig are based on Jericho2 or Tomahawk 3 or other silicon.

There are many form factors with 100-gig ports as well.

So customers are busy qualifying them as 100-gig switches and routers first, they use them in existing designs, but with more efficiency with these new products.

And then they'll wait for optics like the ZR optics or [DRRF] optics to show up in volume before they can really use them as 400-gig.

But the transition is already starting in qualification, but again, you can expect 100-gig first.

And I think from a material impact on revenue in the industry, I think it's second half 2020.

Just as a quick follow-up, Jayshree, you talked about your go-to-market on enterprise.

But I'm wondering how you're thinking about today your needs for sales and support around those new products?

Is that something that you feel like you'll need to ramp up personnel ahead of sales or can you continue to be really efficient bringing on head count to support those customers kind of after commitments are already made?

Actually, that's a very good question.

I think we will ramp up salespeople ahead of sales.

But we can ramp the systems engineers and some of the support engineers post-sales after we get the wins, so a little bit of both.

So we're not applying the same discipline and conservatism, James, that you saw us do in the data center.

We're definitely adding head count.

And if you look at our sales and marketing as a percentage of revenue, it has increased, maybe not appreciably, because we're still holding the bar pretty high and making sure that the caliber and quality is not compromised just because we went and hired a bunch of people.

The other big thing I think that's going to play a huge part in this is partners.

We've never been viewed as a partner-friendly company, but we're very friendly with partners at the moment, and I think the campus is a key piece of that strategy and the partners see us as a key piece of that.

So the 2 will go hand in hand.

Your next question comes from Paul Silverstein with Cowen & Company.

I have got a handful of questions, if I may but...

She was trying to go from 100-gigabit to 400-gigabit speeds.

There she's doing a good job of it.

Most of these questions are clarifications, but let me fire away.

First off, regionally what are you seeing with respect to the quality of in demand on a regional basis?

I appreciate that a good chunk of your revenue comes from cloud, so perhaps the regional concept doesn't quite apply, but to the extent you are -- do have regional exposure, it looks like your non-U.

S. and your U.S. were about the same growth rate, in the high teens, what are you seeing regionally?

And then I've got some follow-ups.

The growth rate is very good especially in enterprise customers across all regions.

We're seeing better growth rate in terms of new customer acquisition in the international region.

So the big bet in larger customers tend to be more U.S. driven and the new customer logos tend to be more international driven, but all regions are growing nicely.

Jayshree, and again, I appreciate that you guys are a share gainer, so you're less economically -- less macro sensitive, but to the extent you are sensitive like any other company to a degree to macro trends.

In terms of the quality of budget, the quality of spend, any thoughts on what you are seeing -- we've seen from a number of other companies, most recently NetApp today, in terms of weakness from a demand perspective, any thoughts on that?

Once again, I appreciate you're a share gainer so maybe you don't see it the same way, but any thoughts you can share with us?

My experience with macro issues is we certainly won't be immune, but the way we will see it is that a lot of the activity we are seeing may not result in fast decisions.

So probably prolonged decision-making is -- would be my biggest worry should a macro set in.

Because usually when a macro sets in, customers tend to get conservative and then they don't want to make new decisions.

Have you seen that elongation yet?

Is that just a concern at this point or you are already seeing it?

Yes, I'm addressing your theoretical concern.

We have not seen it yet.

All right.

And then on services revenue, does it go without saying that the growth in your services revenue is going to slow consistent with the moderation growth in your product revenue?

Yes.

I think there is a linearity associated with that.

Yes.

Yes.

I mean, you'll get some ops that will -- just from renewals and stuff, but overall it will transfer to product.

All right.

2 more clarifications.

Jayshree, on your comment, or your reiteration, I think from the June analyst event with a $100 million forecast for enterprise campus.

My sense is that, that's drifted out a little bit from a timing perspective.

When you talk about $100 million, is that a calendar '20 outlook or is that the departure point September?

I recognize we're not talking about a ton of time in terms of the difference but what's that $100 million?

What period of time.

No, what I said at the Analyst Day, which was only 6 weeks ago, still holds, which is 4 quarters starting from Q3.

So Q3 -- second half of 2019 and first half of 2020, unless something changes substantially.

But we're still bullish on that and optimistic that we have the activity to result in that number.

Got it.

Ita, going back to the OpEx question that was asked earlier.

I would have thought your answer would have been as simple as, enterprise costs more money.

One of the beautiful things about cloud not just the concentration spin, but it was a relatively inexpensive market to address.

Enterprise, you've got to bulk up your sales force, your channel, and that costs money.

Not that you're projecting a dramatic increase, but correct me if I'm wrong, you're projecting an increase from 7.5% to 10% on sales and marketing with operating margin going down to 35%.

Are those still the operative numbers going forward?

And is that what's going on in terms of the increase in spend?

Yes.

I mean, I think quarter-over-quarter, there are other things in there, like we talked about with R&D and stuff, just on the quarterly trend, I think over the longer term, when you think about the model, I think that's what we described at the Analyst Day and that's the right way to think about it.

But again, we're not going to get to 10% overnight, right?

I mean, we're growing and then we're adding incrementally as a percentage of revenue, but it's not -- it won't become 10% overnight.

Paul, can we do that in the callback?

Your next question comes from Alex Henderson with Needham.

I'm sure I'm going to ask the question Paul was going to ask.

I was hoping you could talk a little bit about the market share trajectory.

So there's 2 -- there are 3 variables that were always the kind of underpinnings of the company's story.

One was the cloud growth.

Obviously, that's slowed quite a bit.

The other one was the ability to gain considerable amount of share annually.

Your share is fairly low.

Can you talk about, excluding cloud, what your expectation for the overall market growth for switching to be, and whether you can continue to pick up a point to 2 or 3 on market share annually?

And a follow-up on that question is, is there any change in that as we go into the 400-gig?

I think you've been pretty clear that you think you will continue to grow based off your software advantages 400 gig, could you address those 2 together?

Sure, Alex.

It's always difficult to predict market share.

But I don't think anything has substantially changed on total available market.

Yes, the cloud spend is reduced so we may see some shifting of TAM between 1 quarter and another.

And I think, our position both in the cloud and in our rate of enterprise design wins is only getting stronger.

So from a market share gain, since we've gone from zero to the teens rather quickly, probably our rate of gain, rate of gain will be slower, maybe more like 1 to 2 rather than 2 to 3 on an annual basis.

But I believe we'll continue to be a share gainer both in overall high-performance switching and especially in 100-gigabit Ethernet switching.

Great.

If I could throw one more question in, the Luxtera and Acacia acquisitions over at Cisco, if Andy is still around, any thoughts on why he did that...

Yes, Andy is here.

Any thoughts on why he did that -- they did that and how that might affect you?

And does there -- is there any concerns that as optics need to get closer to the switch chip that they may be positioning to have an advantage as we get to the 53 terabit switch chips that require the chips to butt up against the optics, can you give us any thoughts on where we're -- where they're going with that?

Right.

So we obviously don't want to speculate on a competitor's motivations or actions here.

But what I will say is that the optics field is intensely competitive.

There is plenty of suppliers both with silicon photonics, technology, the 400 gig DSPs that you need for that CR, that we don't see the competitive environment in optics changing at all.

And the pluggable phone pack in particular has just taken over the market over the last 10, 20 years, and we don't see that changing for all kinds of reasons, in particular your question on [50 0.2] our plan is to deliver that product with conventional pluggable optics which have [ever in stood] and will be the time to market product.

We do understand that certain people excited about co-packaging, but there's so many problems with co-packaged optics I wouldn't even know where to start.

So I'll leave it at that.

Your next question comes from Jim Suva with Citi.

It was great to hear Andy on the call at the beginning as well as pretty recently.

But Arista does a lot of things very much on purpose, so having Andy on the call as well as addressing the 400-gig topic, can you help us just maybe understand a little bit better.

Are you trying to like clarify some misperceptions or show that Arista is likely to gain more share or -- there was definitely a tempo change when he was speaking at the beginning of this call versus previously with Andy, and it's appreciated.

I'm just trying to figure out why and the excitement behind it as far as competitive standpoint or share or figure out the change in tone.

Tim (sic) [Jim], if you look at the analyst forecast for 400-gig, like at the lower forecast in particular, they make certain projections on where 400 gig will actually be deployed.

And the vast amount of 400-gig forecast is in cloud, obviously, cloud -- large cloud and small cloud.

So given our strong footprint in that market, we are very, I shouldn't say optimistic, but confident that we will have a good share of that business going forward.

400-gig as a technology is actually almost overkill for traditional enterprise.

I don't think you're going to see much 400-gig option in the enterprise anytime soon.

So this is a very much a cloud story.

And it's really when the cloud customers, the large customers are starting to deploy it is when you see the big ramp, and we do expect that in 2020.

And we have Andy frequently as a guest speaker.

I don't think there is any deliberate intent to do it differently than any other quarterly call, except investors love hearing from Andy and we love Andy.

And I think sometimes the 400-gig gets overhyped.

And I think bringing a dose of realism that Arista, the market leader and high-performance switching and especially 100-gig has more products than anyone else in 400 gig and is ready for that transition, but it will take time, is a pragmatic message.

Your next question comes from Simon Leopold with Raymond James.

Maybe to follow up on the 400-gig theme.

Understanding that the cloud will be the primary adopter.

I'd like to hear your thoughts on maybe compare and contrast the 400-gig cycle versus the 100-gig cycle for you reflecting on time?

And I guess where I'm coming from is, you're now the incumbent with 100-gig, you're sort of the one everybody wants to be, so it's a different position.

You're not the underdog anymore.

So I want to get that perspective.

And just as a clarification within this context.

My impression is that right now the 400-gig switches are being deployed as just really, really good 100-gig, high-density 100-gig, so not necessarily awaiting the optics.

So I just want to make sure that understanding is correct.

I'm going to kick it off and I'd love Anshul and Andy's detail on it as well.

When I step back and look at 10-gig migrations, it took about 8 years to happen.

And why did it take so long?

There was a very long tail because 1-gig was good enough and the compute in storage wasn't fast enough or large enough to require any better IO and the cloud hadn't happened.

The advent of cloud really pushed 25-gig, 40-gig, 50-gig and especially 100-gig.

So the 100-gig cycle, instead of taking 8 years, only took 2 or 3 years, and this is why we became such a market leader so quickly.

It all happened between 2016 and 2019.

When I look at 400-gig, I think you have to sort of look at it as, split it between the 10-gig cycle and the 100-gig cycle.

It will likely take 3 to 4 years to happen.

It will start first in the cloud and then it will migrate over time to other high-tech enterprise and cloud specialty providers as well.

So hopefully that gives you a sense of why 10-gig took too long, 100-gig happened very fast and 400-gig may be somewhere in the middle.

If you could add to that, so the 100-gig is still ramping as you may know.

Exactly.

So we're expecting very significant growth into next year and maybe even into 2021 on 100-gig ports, no mistake about that.

The reason why a cloud company would deploy 400-gig is because it's more cost-effective than 100-gig on a per bid basis, right?

And as you know, our 400-gig products have typically doubled the cost performance for 400-gig and 100-gig.

However, the optics, they are not at that level, right, so because the optics are still too expensive and arguably not, just not available in volume, a cloud company, even if they wanted to, could not deploy 400-gig today in volume.

It's just not possible.

And keep in mind that 100-gig is deploying in the cloud at a rate of, call it 10 million ports a year.

And it takes a long time to get to those kind of volumes on the optics.

So this is why we've been saying all along that it's a 2020 story until you even get to meaningful revenue in 400-gig.

One other thing I would...

Go ahead, Anshul.

Sorry.

This is important for everyone to understand that when we came out with 100-gig products in 2016 with the 7500R, our competition had already announced their 100-gig products.

It wasn't as if we had some huge advantage and we were the only one with a product and so on.

The market was very competitive.

You had to win on your own merit, on software, on partnership with the customer, on solving [the other] problems on quality, on support and so on.

And the exact same thing will repeat here.

And we feel very good about our position, and you've seen the kind of collaboration we've done with companies like Microsoft and Facebook recently, and I believe that will continue.

And is competition or the competitiveness, the price pressure any different in this cycle?

Because your competitors are saying that they're going to take market share in 400 gig.

Should we think this is any different than the 100 gig?

We've always competed against tough competitors.

That will continue.

I don't believe there's any different in dynamics overall...

We're not seeing new competition.

So it's the same competitors being aggressive.

Your next question comes from Samik Chatterjee with JPMorgan.

I just wanted to ask at a high level, Jayshree, how are you are thinking about kind of given the sluggishness in the cloud spend that you are seeing, how you think about diversification in the customer mix?

And particularly if you have any views of strategically where you want the customer mix to be kind of 5 years from now?

Is there more of an effort to kind of steer the business towards a particular customer mix to mitigate some of the volatility around the cloud?

And just a quick follow-up for Ita.

Maybe you addressed this, there's some headline today on [e off] about incremental tariffs on products that were exempt earlier.

Can you just clarify if there are any products that you're shipping from China that were exempt earlier?

I think this is a good question.

Our sales and go-to-market strategy is really shaping to be one that was on focused verticals to a horizontal enterprise where we will be much broader from a coverage, from a geography and from addressing a broader enterprise perspective.

And I think this will provide important diversification.

Campus was an important piece of that diversification.

Two years ago we asked our customers, should we be in the campus and they said no, and then this year when we asked them, they said you are late, so that tells you the thirst and the hunger for Arista technology to go beyond the data center so addressing a broader TAM and going out of our normal comfort verticals into a horizontal go-to-market is an important piece of these.

And Ita, you were going to answer the tariff question?

Yes.

I think the way to think about it is yes, it will have impact but between kind of the improvements we're making on the supply chain and other stuff, I think from a financial perspective, it kind of balances out, so it's a pretty minimal impact from a financial perspective even though we're continuing to turn the supply chain to respond.

Your next question comes from Mitch Steves with RBC Capital Markets.

Just a quick question from me.

You guys have talked about the annual numbers in terms of like the Street estimate so I just want some clarity there.

Are you guys comfortable with where Street estimates are at this point?

Just to get an idea for kind of like what seasonality in the back half looks like?

No, I think you guys have been more aggressive than we have in our guidance.

We're going 1 quarter at a time, but you all started the year at 30% and we at Analyst Day, as you know, Ita guided to mid- to high teens depending on the cloud spend, right, so.

Yes, I think, Mitch, the motto is that we're taking this a quarter at a time from our perspective.

It's difficult for us to go beyond that at this stage.

We're just kind of running the business a quarter at a time.

Got it.

And just to clarify on the Q4, kind of what expectations are, trying to just understand the half on half commentary.

Is it going to be essentially a few points below seasonal trends or do you think that it's going to be more than that?

Yes.

I mean, I think our -- the comments were kind of clear that we think second half over second half, it's a very different environment, right?

If you think about where we were at the second half last year it was very strong demand, we were building deferred et cetera, it's significantly different when you look at where we are the second half of this year.

That's what we're trying to communicate, right?

To put a specific number on it, we're not ready to do that for Q4 yet.

But I think there has been a significant change just from the momentum and the growth in the cloud part of the business.

Your next question comes from Hendi Susanto with Gabelli Research.

Jayshree, your Cognitive Campus will have general availability in Q3 2019.

Do you have updates and would you be able to share goals, timing and milestone in terms of early trials, early adopters in terms of verticals, integration with module and building an ecosystem of channel partners?

Yes, no.

I think we'll have more updates and results in -- towards the latter half of the 2019.

In terms of activity, it's been very high.

We've integrated Mojo into the company, and that's been a year now.

We've integrated it into our CloudVision.

The combination of our 720 XP [POE] switch and Mojo is really redefining a new Cognitive Campus layer.

The X3 [splines] are being very well received.

So what you're seeing here is Arista is having to position the new architectural shift to the next-generation campus in terms of network design and have the products tested at the same time.

So the activity level is very, very high.

The results we'll definitely share more with you in Q3 and Q4.

Your last question comes from Brian Yun, Deutsche Bank.

I also had a question on the campus opportunity.

So I've been hearing more and more that new enterprise campus deals are often led by discussions around Wi-Fi solutions.

Is that what you're seeing as well?

And if so is your -- I know you have the cognitive Wi-Fi portfolio in Mojo, but is the wireless portfolio right now pretty robust or is that an area where you are thinking about or would need to expand?

It's a good question.

We -- in the smaller enterprise sites, we often see that the conversation is led with Wi-Fi because they want to start with a small configuration of campus and they don't need to think of all the protocols.

We feel we have a very complete portfolio particularly with the introduction of Wi-Fi 6. But in larger enterprises it's actually the other way around.

Often the Wi-Fi has to integrate with other partners like ClearPass, and we lead more with the X3 [splines].

So it depends on the nature of the enterprise customers, but we see a bit of both.

This concludes the Arista Q2 2019 earnings call.

Thank you for all the good questions and for the opportunity to highlight our financial results and corporate achievements for you.

I also want to mention that we have posted a presentation, which provides additional information on our fiscal results, which you can access on the Investors section of our website.

We look forward to continuing the conversation with you during the quarter.

Thank you for joining, ladies and gentleman.

This concludes today's call.

You may now disconnect.