Arista Networks Inc (ANET) 2018 Q3 法說會逐字稿

Welcome to the Third Quarter 2018 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 Product and Investor Advocacy.

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 Andy Bechtolsheim, Arista's Chairman and Chief Development Officer.

This afternoon, Arista Networks issued a press release announcing the results for its fiscal third quarter 2018.

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 fourth quarter of the 2018 fiscal year, industry innovations, 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 would 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 to 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 2018 earnings call.

I am pleased to report that we had a record Q3, once again surpassing the consensus estimates.

We exceeded our guidance comfortably with a non-GAAP revenue of $563.3 million as we grew 28.7% year-over-year despite tough comparisons from Q3 2017.

Our non-GAAP earnings per share was $2.11, with services contribution at 13.8% of overall sales.

From a geographic perspective, our customers in the Americas contributed 72% of total revenue, while the rest of our international theaters performed quite well.

We delivered non-GAAP gross margins of 64.6%, exceeding our forecast due to product mix.

Our top 10 customers included all 5 verticals.

Cloud titans contributed extremely well in Q3 and ranked as our #1 vertical, followed by cloud specialized providers and enterprises tied at the #2 spot; and financials and service providers tied at third place.

Our new customer acquisition and million-dollar customers continues to be solid as well.

And the adoption of CloudVision and FlexRoute software exceeds our expectations.

We closed our first 2 acquisitions in Q3, both Mojo Networks for cognitive WiFi; and MetaMako, for low-latency, based in Sydney, Australia.

The acquisition of MetaMako plays a defining role in deepening Arista's heritage with next-generation low-latency platforms.

MetaMako's ultra-low latency focus based on unique FPGA designs delivers 5 to 50 nanoseconds with a predictable 70 picosecond time stamping accuracy.

Last month, Arista introduced the new 400-gigabit fixed switches, the 7060X4 Series.

This is based on the new Broadcom merchant silicon Tomahawk 3, with significant routing and buffering improvements.

It includes 12.8 terabits capacity in a single RU form factor with superior price performance, power efficiency, density and buffer memory, all supported with our proven, single image EOS.

EOS also brings differentiated traffic management, load balancing and resilience, and it eases the qualification of our customers in cloud-scale networks.

We expect the early customer trials to begin this quarter in Q4 and mainstream production in 2019.

I would like to take this opportunity to invite Andy, our Chairman and Chief Development Officer, to elaborate more on the details.

Andy?

Thanks, Jayshree.

400-gig Ethernet is the next step in the evolution of Ethernet, delivering 4 times better scalability and density and up to 2 times the price performance and power efficiency of our existing 100-gig Ethernet products.

One of the great things about 400-gig E is that it really showcases our ability to rapidly bring new switch silicon to market that is fully supported by our market-leading EOS network operating system.

This means that customers can deploy the latest Arista 400-gig Ethernet switches in their production networks with confidence.

Going forward, we expect rapid evolution of new merchant silicon for 400-gig.

Our ability to quickly release new switches based on the latest merchant silicon with fully supported EOS software is a key competitive advantage with short merchant switch silicon life cycles.

While we are very excited about the 400-gig growth opportunity, we do expect to see a lot of 400-gig qualification activity in the first half of 2019, with initial 400-gig production deployments in the second half.

The 400-gig ramp in 2019 is also constrained by the volume availability of 400-gig optics, which so far are only available in prototype quantities.

Please keep in mind that customers are not waiting for 400-gig to build out their networks.

We are still in the midst of a major network upgrade cycle to 100-gigabit Ethernet, which is expected to continue to ramp strongly next year, with industry analysts expecting shipments of more than 16 million 100-gig ports in 2019 compared to less than 1 million 400-gig ports.

So clearly, it will take some time for 400-gig to ramp up.

In summary, we at Arista are very excited about the benefits 400-gig offers to our customers, and we expect to take a leading role in the rollout of 400-gig Ethernet in 2019 and beyond.

Thanks, Andy.

I really appreciate your tenacity in driving optics and 400-gig, and I think you do that not only for Arista but the entire industry.

What is clear to us is that we are in the midst of a multiyear cycle for high-performance cloud networking for both 100 gig and emerging 400-gig gigabit Ethernet clients.

And so as I reflect upon our 2018 strategy, we are executing well across many fronts, including innovative platforms, the migration from securities being a silo to a holistic segmentation, our partnerships with VMware in micro-segmentation, and our multi-cloud zone segmentation support for Zscaler, Amazon AWS, Google GCP and Microsoft Azure.

We're also coping well with the 10% tariff effected by USTR on September 24, 2018, now affecting our networking products.

With judicious planning by our manufacturing teams, we are reducing our dependency on China-sourced components gradually and increasing our manufacturing capacity outside China next year.

Meanwhile, we have implemented a short-term tariff fee of 3.3% as we are absorbing some of the incurred costs with the expectation that we can mitigate them in the future.

It has been 10 years since Arista started shipping products.

And as I reflect over the next -- past decade, I am very proud of Arista's leadership, our board, our employees plus our team work and execution from startup phase in 2008 to the prestige of becoming an S&P 500 company this year.

I don't think any of us could have accurately predicted the pace and magnitude of Arista's results.

In Q3 2018, we exceeded a cumulative of 20 million cloud networking ports.

To give you a perspective on this exponential traction, it took us 5 years to attain our first 1 million ports or 5%, which means we shipped 95% in the next 5 years between 2013 and '18, which I think is quite a ramp indeed.

In particular, our cloud customers have transformed the face of networking forever by mandating Arista as the gold standard in technology, quality and support.

We continue to experience momentum not only in this vital sector, but the propagation of these cloud principles to next-generation data centers, LAN, WAN, campus enterprises and service providers with NFV pairing and routing attributes, turning legacy PINs into places in the cloud, or PICs, as we call it.

Now with that, I'd like to turn it over to Ita, our CFO, for greater details on Q3 2018.

Ita?

Thanks, Jayshree, and good afternoon.

This analysis of our Q3 results and our guidance for Q4 '18 is based on non-GAAP and excludes all noncash, stock-based compensation impacts, acquisition-related charges and certain lawsuit-related costs.

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

Total revenue in Q3 was $563.3 million, up 29% year-over-year and above our guidance of $540 million to $552 million.

We were pleased with overall demand in the quarter with ongoing strength across the business.

Service revenues for the quarter were approximately 13.8% of revenue, down from 14.4% last quarter, which had included an unusually high level of renewal activity.

International revenues for the quarter came in at $157 million or 28% of total revenue, up slightly from the prior period, reflecting strength in our in-region international businesses.

Our international base is still relatively small and will experience some volatility on a quarterly basis as the business develops.

Overall gross margin in Q3 was 64.6%, up from 64.5% last quarter and above the midpoint of our guidance of 63% to 65%.

This outperformance versus guidance primarily reflected a slightly higher revenue mix from our non-cloud customers.

Operating expenses for the quarter were $155.1 million, up from $143.9 million last quarter.

R&D spending came in at $105.6 million or 18.7% of revenue, up from $92.3 million in the prior period, reflecting incremental headcount and higher prototype and NRE spending in support of new products.

Sales and marketing expense was $41 million or 7.3% of revenue, up from $39.9 million last quarter due to increased headcount.

Our G&A costs were 1.7% of revenue and excluded some acquisition-related legal and accounting fees, as described below.

Our operating income for the quarter was $209 million or 37.1% of revenue.

Other income expense for the quarter was a favorable $8.6 million, and our effective tax rate was 21.3%.

This resulted in net income for the quarter of $171.3 million or 30.4% of revenue.

Our diluted share number for the quarter was 81 million shares, resulting in a diluted earnings per share number of $2.11, up 30% from the prior year.

We completed purchase accounting for the Mojo and MetaMako acquisitions in the period, with immaterial amounts of revenue and expense included in our non-GAAP results for the third quarter.

For those of you focused on our GAAP results, we recorded $3.4 million of acquisition-related expenses and $5.9 million of acquisition-related tax charges in the period, which we consider to be onetime in nature and which together with $1.6 million of amortization of acquired intangibles have been excluded from our non-GAAP results.

Now turning to the balance sheet.

Cash, cash equivalents and investments ended the quarter at approximately $1.7 billion, down from $1.9 billion last quarter.

As a reminder, although the $405 million charge related to the settlement of our lawsuit with Cisco was recorded as a non-GAAP expense in Q2 '18, the cash payment for this amount did not occur until the third quarter.

Excluding the Cisco payment, we generated $286 million of cash from operations for the period, reflecting strong net income performance and improved working capital metrics.

DSOs came in at 53 days, up from 46 days in Q2, reflecting the timing of billings and collections in the quarter.

Inventory turns were 3.2x, up from 2.7x in Q2.

Inventory decreased to $216.3 million in the quarter, down from $245.4 million in the prior period.

This reflects reductions primarily in raw materials buffers as we continued to optimize our supply chain.

In addition, we maintained a further $19.3 million of inventory deposits recorded in other assets compared to $25.3 million last quarter.

Our total deferred revenue balance was $529.9 million, up from $448.6 million in Q2.

Product deferred revenue increased by approximately $38 million in the quarter, largely related to customer certification of features reintroduced into the product following the expiration and invalidation of certain lawsuit-related patents.

Accounts payable days were 39 days, up from 26 days in Q2, reflecting the timing of inventory receipts and payments.

Capital expenditures for the quarter were $4.5 million.

And now turning to guidance.

As we look to the fourth quarter and beyond, we believe that we remain well positioned with our key cloud customers and continue to grow our presence across our other verticals.

The midpoint of our revenue guidance for the fourth quarter of $582 million to $594 million results in revenue growth for the full year 2018 of approximately 40%.

Turning to gross margin and the impact of the recent tariff announcements.

The operations team is working diligently to optimize our supply chain and mitigate the incremental costs for both Arista and our customers.

We expect these supply chain modifications to take effect throughout 2019 as we ramp new sources of supply.

In the interim, we've introduced a tariff adder, whereby we will pass a portion of these costs to our customers pending completion of the talks (inaudible).

We expect the impact on gross margins in the fourth quarter of 2018 to be somewhat muted as we ship backlogs and consume pre-tariff finished goods and component inventories.

Based on everything that we know now, we would reiterate our typical gross margin range of 63% to 65%, knowing that tariff impacts, et cetera, will limit our ability to outperform the 64% midpoint of this range.

With this as a backdrop, our guidance for the fourth quarter, which is based on non-GAAP and excludes any noncash stock-based compensation impact, amortization of acquisition-related intangibles and certain lawsuit-related costs is as follows: revenues of approximately $582 million to $594 million, gross margin of approximately 63% to 65%; operating margin of approximately 35%.

Our effective tax rate is expected to be approximately 21.5%, with diluted shares of approximately 81.3 million shares.

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.

Due to time constraints, I'd like to request that everyone please limit themselves to a single question.

(Operator Instructions) Your first question comes from Rod Hall with Goldman Sachs.

This is Balaji on for Rod Hall.

I had a question on the competitive landscape as we move into the 400G deployments.

And maybe if you could describe how you would characterize the changes.

Clearly, Cisco is keeping up with you guys at this point, or at least it looks like they are keeping up.

And they've also had a lot of engineering engagement.

So is there any difference there?

And maybe also just a commentary on the optic supply shortages that you said.

Is there any difference between OSFP and QSFP-DD?

I'll kick it off, Balaji and then I'll hand it to, of course, Andy, who is much more deeply entrenched in this.

We haven't seen any significant change in competitive landscape.

As the market leader in 100-gig Ethernet, I think everybody has declared their products and introductions just as they did in 100-gig.

Time will tell what the real capabilities of these products are, and how we match, but we're very confident of our outstanding capabilities and differentiators.

Particularly, and don't underestimate the importance of combining the right silicon with the right operating system and differentiated features.

Inconsistent drivers and discontinuity of OS can be extremely cumbersome for customers, so ease of qual is very important as well.

We also firmly believe that internally developed ASICs are not keeping up with merchant silicon and are often not competitive in the 400-gig market.

So Andy, do you want to add to that?

Yes.

The other thing is the demand for 400-gig clearly comes primarily from the cloud, where we have a strong footprint and there's virtually no demand from legacy enterprise for 400-gig.

Cloud customers, in particular, have little time for experimenting with new software platforms and greatly prefer to go with trusted solutions.

And what about the optics, Andy?

Oh, the optics are not yet in volume production, and it remains to be seen how quickly they ramp up here.

You realize there's a ramp-up cycle for those -- for the whole supply chain, the optics vendors have to order the parts, they have to make the models and so on.

They're waiting for purchase orders.

So our current belief is that the supply will only reach what you would consider volume in the second half of next year.

And we are very Switzerland and neutral.

Andy and the team are supporting both OSFP and DD-QSFP.

So both have advantages depending on where you're starting from, so Arista will support both.

Your next question comes from Erik Suppiger with JMP Securities.

Two points.

One, just on that last one.

Is your impression that your competitor for the 400-gig is using internal silicon -- internally developed silicon?

Or is it based on Broadcom as well?

And then secondly, on the tariff front, I just want to be clear, you're adding on a 3% charge for your U.S. customers.

Is that how you're approaching the tariff impact?

Okay.

So let me try and take on both questions, and I'll hand it to Ita for the clarification if needed.

On the competitive front, obviously, we are not the experts on implementation.

But to the best of our understanding, the announcement was only made yesterday.

Some of the models use internal silicon, and some of them use non-Broadcom merchant silicon.

So they are not really a direct comparison to Arista's introductions.

And on the tariff side, yes, we're applying a universal 3.3% adder worldwide because a lot of our components are affected worldwide as well as our PCBs.

So it's not just U.S.

Your next question comes from Srini Pajjuri with Macquarie Securities.

Jayshree, obviously, you've said before many times that the cloud CapEx, the correlation between your business and the overall CapEx is not very high, but there are definitely a lot of concerns in the investor community about CapEx slowing down over the next few quarters.

I just want to hear your thoughts about what you're seeing out there in the market.

What are your customers telling you about next year, if anything?

And given where we are in terms of how high the CapEx is, if it were to slow down, what do you think -- what kind of impact do you think you'll see in your business?

Sure.

It's a good question, Srini.

Even though it gets asked many times, it still remains one of the most popularly asked questions.

As you know, our cloud customers have been adopting Arista unabated for over 5 years.

It's not a 1-quarter phenomenon; it's a 25-quarter phenomenon.

We don't see any near-term signs of that changing nor the predicted concerns.

It is true that we don't track one-to-one with that cloud CapEx.

And remember now that despite the lumpiness of the cloud titans, and we have many of them, we are comfortable with the continued spend because there are multiple use cases, and we expect this to continue this quarter and early 2019 given our typical 2-quarter visibility.

Your next question comes from Ittai Kidron with Oppenheimer.

Congrats, ladies, and Andy on a good quarter.

I had 2 small ones.

First of all, you haven't talked about campus, that little product thing that went GA in the quarter.

So maybe you can help us talk about business activity there, volume, trial activity, pipeline, how does that look?

And then just clarification, Ita, on the tariff again.

I just want to make sure, if this tariff -- if you didn't put on the adder, would your revenue guidance be 3% lower than what it is?

I mean, I'm just trying to understand whether the guidance really captures an extra 3% just from a tariff perspective.

Yes.

I mean, I think in Q4, in particular, we have got backlog that was booked prior to the tariff, that I don't really think we can disaggregate it and start -- trying to carve out that 3%.

And as we go forward, obviously, the goal is to remediate the cost as much as possible and start to kind of normalize that for us and for customers over time.

So I wouldn't try to say it's somehow an adder to the top line.

It's more an offset.

Otherwise, we'd take a bigger hit in margin.

So we're absorbing some of the cost and passing on some is the way to look at it, Ittai.

So Ittai, to address your campus question, as we've often said, a lot of excitement on the architecture, a lot of excitement on the acquisition of Mojo.

Not material in revenue this year or the first half of 2019 has been our consistent statement and that is true.

Having said that, what we have, since our last update, we have introduced the X3s line, as we call it.

This is used both in campus use cases and in data center cases, comes in a modular and fixed form factor.

And that has been very well received and is in early trials in the campus.

Customers are excited about that.

Your next question comes from Sami Badri with Crédit Suisse.

Could you tell us what drove the strength in deferred revenue?

Just any color on the material step up?

Yes, I think the biggest driver relates to, as we came out of the lawsuit, we had actually invalidated some patents and we'd have some patents expire, and that gave us access to some technology that we'd redesigned out of the product previously.

And now we've obviously put that back into the product because we have access to that technology.

And for some customers there is a need to requalify that new design, that new product, and we're engaged in doing that currently.

Got it.

And then just a follow-up is regarding tariffs and your campus switching rollout.

Have you heard from customers opting into the Arista Networks switches, on mainly the campus switching side, simply because pricing is more favorable to adopt it, rather than some of the competing products that might be seeing a higher tariff rate?

Could we get any kind of color around that and customer behavior?

I mean, I think the general answer is no, the tariff is not a reason to choose or not choose Arista products.

And in particular, on the campus, majority of our business, almost all is data center, so there's no correlation to be made on data center products or campus products to the tariff.

Your next question comes from Jeff Kvaal with Nomura Instinet.

Yes.

A question and a clarification for me, please.

Andy, I'm wondering if you wouldn't mind comparing and contrasting what we ought to expect out of this 400-gig upgrade cycle versus 2017's 100-gig?

Is it as powerful?

Do you expect to gain as much share?

Any thoughts along those lines would be helpful.

And then, Ita, I guess, I wasn't quite clear.

It sounds as though the cloud titan mix was better in Jayshree's initial remarks.

But then, it sounds like the mix was away from cloud titan in your gross margin explanation.

So if you could clarify that for me, I'd be grateful.

Yes, let me just pick that one up first.

I mean, it was very slightly mixed towards the non-cloud.

It's not -- and if you saw the gross margin it was like 10 basis points.

So it wasn't a big shift one way or the other.

And then on your question on the 400-gig expectations.

I would like to refer you to the market analysts like Dell'Oro and others that have modeled and predicted this in detail.

One thing I can mention here is that the crossover in bandwidth shift between 400-gig and 100-gig is currently projected to be in 2021, so it will take some time for 400-gig to come up to the same level of bandwidth as even 100-gig.

100-gig, if you look at these reports, is ramping extremely strongly based on the fact that it's fully available, fully qualified, the optics have been full volume, et cetera.

So a large cloud company or a customer that wants to deploy a new data center really has no choice.

They will deploy this 100-gig today because that's the only way they can buy hundred thousands of optics and so on each quarter.

Whereas, on the 400-gig side, it will take some time to get to those kinds of volumes.

In addition, I would like to observe that for brownfield data centers, meaning if you upgrade an existing data center, it's easy to stay with the same speed throughout the data center, so most likely, those data centers will stay with 100-gig for some time.

For a greenfield data center, you have a choice.

You can start with 400-gig, but, again, only when those components are available in sufficient volume.

Your next question comes from James Faucette with Morgan Stanley.

I just wanted to ask one clarifying question, to build on Ittai's question for Ita is that, in the formulation of guidance for the December quarter, how much is contemplated to contribute from the acquisitions that closed in the third quarter?

And then, taking advantage of Andy being here on the call today, Andy, can you talk a little bit about the number of hyperscale customers, or what you consider to be hyperscale customers, that Arista has and how that's changing?

I guess, more importantly, how are the requirements changing for the newer customers in terms of what they're looking for from Arista?

And how is that the same or different from your traditional customers?

So James, I think on the acquisitions, again, I would remind you that Mojo is a SaaS model, so that's going to be a ratable rev rec model for us, so that's a relatively small contribution.

On the MetaMako side, that's already being rolled into our kind of financials vertical.

It's now becoming part of the offering there, and it's definitely been impactful with customers and opportunities to the customers, but we're not planning to really track that separately.

And I'll tell you that the combined impact on the quarter is small.

You've got the Mojo stuff is a SaaS ratable revenue amount, and then, we have kind of single digit coming out of the MetaMako side of the house.

So it's a small contribution at this point.

On the cloud customer question, we cannot disclose the name of our cloud customers.

But I think we have said repeatedly that the competitive environment in that market hasn't really changed.

Obviously, every cloud customer is extremely concerned about network reliability, resilience, uptime, et cetera.

And my belief is that our fundamental competitive advantage is our U.S. operating system, which delivers those qualities.

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

Hey guys, great quarter.

I just had 2 actually small ones.

So first, just on the cloud enterprise financials.

So of those 3, is it still the case that cloud is kind of the fastest-growing segment?

And then enterprise, I'm not looking for exact growth rates, just kind of a trajectory.

And then secondly, how much was the acquisition actually benefiting you guys?

Was it just a few million?

Just looking for a way to get a number around that.

Let me take the vertical question.

There's no doubt that the cloud segment, both the Tier 1 and Tier 2 cloud has been growing faster than any others if you put those 2 together.

However, the enterprise is the fastest because it's starting off a much smaller base, and we're accumulating customers and million-dollar accounts very rapidly there.

And there's a lot of interest on both the data center side, where we're succeeding in installing, and the campus side, which we hope to convert into success next year.

And your second question was?

Yes.

I think in Q3, the acquisitions contributed very little.

The MetaMako only came in kind of half through September.

Two weeks.

Yes.

And again, the Mojo acquisition was probably -- you've got probably about a whole month or a little bit more than that.

But again, it was a SaaS ratable model.

So they contributed very little to Q3.

Got you.

So I guess, just to clarify real quick to being less than a point for next quarter?

Is that roughly correct?

And again, like I said, we're not really going to start tracking this individually, particularly as we roll the products into the portfolio.

I think for Q4, what I said was it's single digit contribution -- low single digit contribution.

And perhaps it will help to mention -- if you look at both of these as tuck-in acquisitions, one is going to help our campus overall and one is already helping our financials since we are in high-frequency trading and low-latency applications already.

Your next question comes from Jason Ader with William Blair.

I wanted to ask about the federal vertical.

We had picked up that you guys are starting to gain some traction there.

So can you talk about where you are with federal?

What type of momentum that you're seeing?

And where could this business be in a couple of years?

Yes.

No, I think Arista is becoming more and more committed to the federal market not only in the U.S. but worldwide.

We have completed a lot of important certifications, so we see this as a big opportunity and one we fully intend to invest in.

Obviously, we will be responsible about reporting wins and losses.

Until something is public, we won't really comment on rumors or protests.

So can't say anything about specific wins and losses, but definitely an important segment for us worldwide.

We're doing well in many parts and many international theaters as well.

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

Congrats on a solid quarter here Jayshree and team.

Just a clarification about the cloud business.

I think historically, you've talked about Microsoft in kind of this 10% to 15% range.

I think your expectation is that it's above 10% this year.

Is that still how things are shaping up?

Yes.

Thanks, Alex.

Although the cloud titans is a composition of many customers, Microsoft has always been our #1 customer, and I believe it will continue to be our #1 in a very solid fashion in 2018 and will be well over 10%.

Okay, great.

Jayshree, could you take us through some of the early deals with the campus products?

I know it's early days, but just how are enterprise customers reaching you on these products?

And sort of deal size, scope of projects compared to what you've done in enterprise before?

Just kind of compare and contrast what you've seen so far.

Alex, it's probably a little early for the level of detail you are looking at, and I promise I'll answer that question next year better.

But when I've been personally involved in this, the pattern match I see is many of them have, like the data center, an architectural need to shift and change, where they've got the classic 3-tiered model, and they want to move to the leaf-spine or often a single-tier spline model and then have different device edge connectivity.

So they're looking to make that change.

And sometimes it's a brand-new building they're going to construct next year or a year after, or it's a brownfield.

The second pattern I'm seeing is that they are already very comfortable with Arista's spines or splines and they're using our EOS, and they're going, Oh, geez.

I don't need to build a separate campus box.

I can use the same spine or spline and enable campus features on that box, whether it's BGP routing or VXLAN or tunneling or security features.

So architecturally -- and the third thing we're seeing is CloudVision for the campus, is something they're very excited about.

We demonstrated some of that capability at the Gartner conference.

John McCool and Jeff Raymond and the team have done a fantastic job there.

And I think you will see these 3 being the anchors.

The design is changing, the cognitive management plane architecture, and our data center customers really want to expand their footprint with us into the campus.

Your next question comes from Aaron Rakers with Wells Fargo.

Congratulations on the quarter as well.

I wanted to ask maybe a longer-term strategic question around the MetaMako acquisition.

I'm just curious, as you kind of fold that into the product portfolio, and clearly a little bit differentiated in its usage of FPGAs, how do you see FPGAs fitting relative to merchant silicon?

And with that acquisition, is there a certain subset or addressable market that you can now address that previously you couldn't?

And what size would that be?

I think, Aaron, your question is very thought-provoking.

As you know, Arista's core being is to adopt and massively deploy merchant silicon on our extensible software.

But there are use cases that require deeper programmability.

One example, even before MetaMako, would be the P4 programmability we do on the Barefoot silicon.

And the FPGA definitely allows us to go capture more state and improve our latency, and really get to the heart of the application in many of these customers and certain verticals for electronic trading.

Andy, you may want to comment on this.

I know you've been deeply involved.

Yes.

Meaning, you can't beat the latency of an FPG for those types of applications, which include both Wall Street type applications and also very precise traffic monitoring and network visibility kind of applications.

So it may not be the biggest market but it's a very important one.

It's a key market for many of our customers.

And do you see this broadening across your product portfolio over time?

I think it would depend on how big the market gets.

But I think at this point, we'll keep it focused on the specific application-driven use cases.

Your next question comes from Samik Chatterjee with JPMorgan.

I just wanted to understand, in the 400-gig, you mentioned kind of the differentiation that you have related to some of your competitors like Cisco, et cetera.

How should I think about how this plays into the competitive dynamics with white box?

Particularly is the technology from those manufacturers keeping up?

Or should I think of 400-gig being an opportunity for you to gain share?

Samik, I think the white box is a bit of a tangential discussion on any speed.

There's really 2 types of players who deploy white box.

One is a captive deployment in cloud customers that are looking to build their own, and they're going to do the same thing whether it's 10-gig, 40-gig, 100-gig or 400-gig.

And the second is, maybe experimental HPC clusters, et cetera, where people may try this, we don't see 400-gig and white box really that connected.

In fact, that is one place I would tell you that the requirement for predictable performance and not compromising speeds, will require the best hardware and best software combination.

So that's not a combination that comes to mind as the first use case for 400.

Got it, got it.

And then just if I can follow-up on the --

I'm sorry, Samik.

We'll just restrict it to one question.

Your next question comes from Fahad Najam with Cowen and Company.

Can you remind us how your traction in the routing market is going and if you're hitting your target?

I think, if I recall, you indicated that you expect at least to hit $50 million in annual revenue from routing.

Are you still tracking to that target?

Or are you exceeding that?

Can you give any commentary on the routing adoption?

Sure.

I will give you some year-end numbers.

The way we track it, don't really do it by revenue, we do it by customers and FlexRoute licenses.

We've got 3 routing licenses that we track on, and we are doing well in the acceptance of that license, particularly in the cloud, service provider and enterprise markets.

It's going well, but I would like it to go even better.

Especially, in the service provider market, I think we can have more improved results.

So any update in terms of do you think you're taking share in the routing market right now as is?

Or is it still in cloud?

We don't -- we do believe we are displacing designs.

Taking share would mean we track that market.

We don't participate in the classic traditional router market.

But we're absolutely taking share in applying routing onto our switching platforms and increasing our switch market share in routing use cases.

Your next question comes from Simon Leopold with Raymond James.

I appreciate the commentary Andy offered on the 400-gig market and the timing being biased towards the second half of the year.

I'm wondering if you have your own perspective on how to size this particular market, partly because I'm confused.

When I look at the quoted data from Dell'Oro versus the IHS, they seem to be a very broad range, as high as 1.5 billion for 400-gig, and 2019 seems hard to believe.

So I was hoping to get your perspective on that.

And also some clarification on -- you talked about the starting price at $1,800.

That's a low price.

I just want to understand what's the customer getting for that?

Yes.

I believe the Dell'Oro numbers include optics connected to the port.

So you have to -- when we talk about $1,800, that's purely the switchboards excluding the optics, that's a very important difference.

The optics cost more than the switchboards typically, in some cases, significantly more.

So there's a gap there, right.

But going back to your question, I think the best thing you can do is read multiple market research reports.

Some of them talk more to the cloud people, in particular, or more on top of the cloud developments than perhaps others.

There's just a lot of momentum right now on 100-gigabit, which is also reflected in these reports.

And the reality is the 400-gig can only be deployed once the optics and all the systems are available in high volume, because a volume deployment for a large cloud customer is like 100,000 ports a quarter, right, and you can't buy 100,000 optics right now in a quarter.

So it takes a while for the supply chain to simply catch up with those kind of numbers.

And with 100-gig, all the optics and all the systems are available in high volume today and ramping.

So it takes time.

But both are growing, and they are incremental to each other.

I wouldn't say one is displacing the other one right now.

I think it's a really important point that Andy made.

If you go pattern match with how we did on 100-gig, there are some striking parallels.

We had some early trials in 2015, 2016, but it took 12 to 18 months for the market share lead we took and got because the whole ecosystem had to come into play.

Something similar happened with 25-gig as well.

Until the entire ecosystem comes to play, which takes 6 to 12 months, you don't see that ramp.

So I think all the market studies are pointing to that ramp in 2020 or 2021.

Your next question comes from James Fish with Piper Jaffray.

Congrats on the quarter, ladies and Andy.

Ita, this one is more for you.

It kind of looks like Q4 guide implies a low- to mid-single digit sequential increase for product revenue compared to typically a low double digit increase.

Is there a reason for the caution or conservatism?

Or is it more related to the tax impact there?

And specifically as well, is there any concern around hyperscaler spending as you look into Q4 and 2019?

I mean, I think that if you look at the guidance, we are largely at the upper end of the guidance with the 27% year-over-year growth rate.

I think that's in line with kind of what we had set as the expectations, and there's nothing -- I don't think there's anything unusual about the guidance in that sense.

I think that we're comfortable where we are from a business perspective, and that's pretty much in line actually with what we've laid out, kind of, right from the beginning, certainly for the middle of the year.

So I don't think there's anything unusual about the guidance.

It's not really trying to reflect any particular key driver.

And any other clarity around the hyperscaler spending for Q4 and 2019?

Yes.

I mean, I'll go back to what Jayshree said.

I think for what we have visibility to, I think we're comfortable with where we are.

We think we are well positioned.

And we haven't seen anything different in the business in kind of that time frame that we have visibility.

It's been business as usual.

Your next question comes from James Suva with Citi.

When we think about your campus deployment, I know you take a look -- it'll take a little bit of time to see how successful it is or not.

Can you at least update us about are you first targeting like your top 20 accounts or your certain regions?

Or are you going to all your sales force at once?

And then maybe for Ita, a question about the tariffs is, I would have thought that the ITC ban would have positioned Arista quite favorably for the sourcing and supply chain of how you do things.

Is that correct but still you just need the 3.3% tariff increase?

Yes.

So just to, James, to address your campus question.

The most natural conversation is with our customer base across all 5 verticals because they already know us, and love us, and are familiar with us, and can see use cases.

The next natural conversation is with the acquisition of Mojo, we're actually getting exposure to new customers.

And some of the interest in WiFi is separate from our customer base.

So that's also a second motion.

And from a sales and go-to-market, both Anshul and Manny are really focusing on campus as a mainstream effort.

There's no sideshow going on here.

And so we're building an entire sales expertise and especially SE expertise.

So where we are putting special emphasis is not all of our sales team understands WiFi and radio management, so we do have specialized SE expertise there.

But the rest of the sales is across the entire sales and marketing focus.

It's nothing unique to campus.

And to your other question, Jim, I think we've certainly diversified our supply base and our sourcing, probably more than maybe we would have if we hadn't come through some of the ITC stuff.

But there's still work to do.

I mean, we're still sourcing in China.

There are certain components that are still being sourced in China, et cetera.

So we do have work to do to mitigate some of those costs.

That's obviously top of mind for John McCool and his team to get that done as quickly as possible.

Your next question comes from Alex Henderson with Needham and Company.

I was hoping you could talk a little bit about the international portion of your business.

Obviously, you have very tough comps here and tough comps for the next couple of quarters.

But could you parse a little bit between the slowdown in that business between economic conditions versus the comps?

And just give us a little bit of color between Europe and APAC?

Yes.

You're absolutely right, Alex.

I think tough comps is the issue.

We did very well organically in our overall geography.

However, some of the volatility was defined by where the global customers and their spend resided.

So Asia-Pac was strong, EMEA was a little weak, but U.S. was strong across the board.

So it just turned out that way depending on where the cloud titans and the top 10 providers spend.

But the organic business is still intact and doing well.

Yes.

And I think you'll see this when we file the Q, but we do have some volatility back and forth between EMEA and APAC over the last couple of quarters, right.

Because again, the base is still relatively small.

So when you win a sizable deal or a couple of sizable deals in one or the other it's going to have an impact.

If I could just ask one clarification.

Jayshree, did you guys say low single digit millions or low single digit percentage in terms of the contribution from acquisitions?

It wasn't clear which were you referring to.

Millions.

Millions, millions, Alex.

It's still relatively small.

Single digit million.

Your next question comes from Hendi Susanto with Gabelli.

Thank you and great Q3 performance.

Jayshree, Arista defined cognitive campus as the next frontier.

I would like to understand more about your go-to-market strategy, how similar, how different it is with your core go-to-market and whether you will have some closest partners.

Yes.

No, that's a very good question, Hendi.

Obviously, the low-hanging easiest go-to-market is the one we already have, and we've now got a nice healthy base of over 5,000 customers.

We're going to leverage that.

However, that will be a necessary but not sufficient condition to participate in the campus.

We are expanding.

And one of Manny's initiatives is, in fact, to complement our direct customer focus with an elite channel strategy focus.

We're not going to pepper all the channels, but we've already had some channel capability and experience in our International theaters, but we will be adding more to that.

So the combination of our own sales and marketing investment in the campus and the channels will be a very important 1-2 step in 2019.

Your last question comes from Woo Jin Ho with Bloomberg Intelligence.

A couple, if I may.

So Jayshree, when you came out with the 100-gig switch, you guys innovated on the network architecture of the cloud.

Are there any considerations similar to that with 400-gig switching?

And then, Ita, in terms of the deferred revenue uptick, how much of that was the IT related that needs to be QA-ed?

And how should we think about the deferred revenue drawdown hitting the P&L over the next couple of quarters?

Yes.

I mean, I think again, the product growth is really related to the requalification.

As to when exactly that comes back, it's difficult to tell.

And again, that balance will move with new stuff versus old stuff, et cetera.

So I'm not trying to forecast that, if you like, as part of it.

But I will say we're not contemplating a significant downward move in that in our Q4 guidance.

But other than that, I think it's too early to try and forecast it beyond that.

And Woo Jin, just to wrap up the last question of the Q3.

If you look at the way we approached 100-gig, we approached it from a network design perspective, heavy software differentiation, bringing high availability, agility, automation, analytics into our 100-gig platforms, mix and match of both modular chassis and fixed form factor.

You can expect us to adopt a similar strategy for 400-gig over the next year.

This concludes the Arista Q3 2018 earnings call.

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

Thank you for joining, ladies and gentlemen.

This concludes today's call.

You may now disconnect.