Arista Networks Inc (ANET) 2014 Q3 法說會逐字稿

Welcome to the third-quarter 2014 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 Investor Relations. Sir, you may begin.

Thank you, Operator. Good afternoon, everyone, and thank you for joining us. With me on today's call are Jayshree Ullal, Arista Networks' President and Chief Executive Officer; Andy Bechtolsheim, Arista's Founder and Chief Development Officer; and Kelyn Brannon, Chief Financial Officer. This afternoon Arista Networks issued a press release announcing the results for its fiscal third-quarter ended September 30, 2014. 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 FY14 and industry innovation, which are subject to the risks and uncertainties that we discuss in detail in our documents filed with the SEC, specifically in our Form 10-Q and recent prospectus, 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. What that, I will turn the call over to Jayshree.

Thank you, Chuck. Thank you, everyone, for joining us this afternoon for our second earnings call as a public Company. I am pleased to report that we had a good third quarter. Strong customer demand drove results that exceeded the consensus estimates. From a geography perspective, our customers in the Americas generated 81% of our sales. We are investing to improve our global expansion. Revenue grew 53% year-over-year to a record $155.5 million driven by our flagship 7500E-Series Spine and X-Series Spline platform.

We delivered non-GAAP gross margins of 65.2% resulting in a non-GAAP earnings per share of $0.40 as we grew profitably in a competitive and dynamic industry. We now have 2,885 customers going from one new customer a day last year in 2013 to more like one-to-two-a-day at the present. We have now shipped a cumulative of 3 million cloud networking ports. Our first million ports took us approximately 48 months, or four years. Our second million took about 18 months, and our third million has arrived at about nine months, more than five times faster than the original million took us.

This quarter, we further expanded our product line and continued aggressively to drive open standards in conjunction with key partners. We announced seven new Leaf platforms ranging from our cost-effective 7010 to our 128-port 7050X all available now, and all based on one common US image. We delivered Arista's EOS 4.14 software release for enhanced network virtualization, virtual extensible LAN, VXLAN, and resiliency using equal cost multi-path, ECMP, and layer-3 functionality. We identified and quickly fixed the Shellshock Linux BASH vulnerability across all our platforms in just five days from first disclosure by September 29, 2014. Our single image software architecture across our products made it possible to do this quickly and efficiently.

We formally announced our joint efforts with VMWare for total, network-wide orchestration including key initiatives for EVO:RAIL convergence restructure at VMWorld in San Francisco. Our other three ecosystem announcements were: participation in A10 security alliance, the Orion alliance with nebula and SYNNEX for turnkey private cloud solutions, and riverbed-ready alliance for cloud application-aware management.

Arista's newly introduced 7280 leaf switch with 100-gigabit ethernet uplinks and large buffers continues to be well received by the market as an industry first. We are undergoing many pilots with this switch in high-end storage and streaming content applications.

Today's wildcard topic is Arista's 10th anniversary. As we celebrate our first decade, I would characterize Arista's journey in three phases. Our first phase as we began as Arastra Networks in Palo Alto in October, 2004 with innovative founders and funding. Andy, David, and Ken pioneered a new class of software that the industry calls SDN but is now adopted by Arista customers as Arista EOS, our extensible operating system.

In the next phase, 2009 and beyond, Arista has fulfilled its disruptive vision for our customers with software-driven, cloud-scale networking, both for cloud titan, hyper-scale web customers, as well as the financial. In our next phase, 2014 and beyond, Arista EOS is indeed a fitting example of our Company's evolution from products to platform as we move into mainstream acceptance, growth, and market share. Our strategy is based on a foundational R&D investment with merchant silicon diversity, distributed systems, stage-driven software, agile releases, and modern methods for tools, kernel, and open programmable functions. Now, I would like to turn it over to Andy Bechtolsheim, Arista's Founder, Chairman, and Chief Development Officer as our guest speaker to look back at these 10 memorable years. Andy?

Thank you, Jayshree. When we started Arista 10 years ago, we had three fundamental beliefs. First, we believed that great software was the most important element in building a differentiated Ethernet switching Company. Second, we believed that merchant-switch silicon would allow us to build superior switches compared to traditional ASIC designs. And, third, we believed that cloud data centers would continue to grow rapidly and would require network architectures with a scale and robustness that is fundamentally different from traditional legacy network designs. All of these three beliefs have proven to be true and have guided us well.

On the software side, we created the most programmable and resilient networking stack in the industry, our extensible operating system. On the silicon side, working with our merchant silicon suppliers has allowed us to deliver superior performance and pull away from proprietary [EC] switch designs, and our cloud customers are using our products to build cloud networks on a scale that is simply unprecedented. In summary, we have built the right architecture for the cloud and continue going forward with the same focus on software, best-of-breed silicon, and scale.

I now want to cover some of the highlights in our history. We spent roughly the first four years in software and hardware product development. Started shipping our first product, which was the [ultra-latency] 7120 4S in Q3 of 2008. Jayshree and myself joined the Company at that time, and this is when we officially launched the Company as well. Our initial market focus was on high performance computing and on financial applications involving high-frequency trading, which despite the financial crisis in 2008, took off rapidly and established Arista as the leading provider of switches for high-frequency trading.

However, our real ambition from the beginning was to build large-scale cloud networks. For this, we developed the Arista 7500 modular switch that was the first product that allowed customers to build cost-effective, large-scale, leaf spine architectures. The Arista 7500 received the best of Interop grand prize and the best of networking award in 2010, and then again in 2013 for the 7500E update, which supports up to 280 40 gig ports, making the Arista 7500 the only product in the history of Interop that has received this very prestigious award, not just once but twice.

We now have over 20 switching products in our spline X series, which includes the 7050X, 7250X, and the 7200X families that all have become extremely popular building blocks for building highly scalable, programmable, and reliable enterprise and cloud networks. On the software side, we have driven programmability into all levels of the network stack, which has allowed us to integrate with a wide variety of third-party management tools, including Splunk, Chef, and Puppet. We have innovated with features such as zero-touch provisioning, zero-touch replacement, LANz, DANz, and unprecedented network visibility.

A key milestone in the history of our Company has been the introduction of the VXLAN's [physucation] in 2012, which we co-authored with VMWare. VXLAN allowed the dynamic creation of essentially unlimited number of layer-2 networks over layer-3 networks eliminating the legacy silo problem. We have also partnered with [Worldcom], Google, Microsoft and others to define a standard for 25- and 50-gigabit ethernet, which we believe will become very important as soon as it's commercially available.

Which brings us to the present. We are very excited about our new products in development, and our innovation pipeline has never been stronger. Our open, software-driven approach to networking based on the programmability of our extensive offering system is resonating with customers and partners who are looking to build best-of-breed, scalable, and resilient networks. And, we are looking forward to the next 10 years on this journey. With this, back to Jayshree.

Thanks, Andy. So, while our peers are still trying to mimic Arista's approach, they fundamentally lack the architecture to achieve this. As Andy described, our journey has just begun. We are looking forward to serving our shareholders and sharing our inflection points and aspirations in the continued journey and quest for this. With that, I'd like to turn it over to Kelyn Brannon, our Chief Financial Officer at Arista, for more details on Q3.

Thank you, Jayshree. Good afternoon, everyone. I'll walk through our financial results for Q3 and our guidance for Q4. I'd like to note that except for revenue figures that our GAAP, all financial figures are non-GAAP unless stated otherwise. A reconciliation of selected GAAP to non-GAAP results is provided in our earnings release.

Q3 revenue grew 13% sequentially and 53% over the prior-year quarter to a record $155.5 million. Third-quarter comparisons year over year shows continued strong growth on top of a record Q3 2013 with continued significant spend by our four principle verticals. International revenues comprised 19% of total revenues in the third quarter, a decrease from 25% in the prior quarter of 2014 and an increase from 16% in Q3 2013. The geographic mix of revenue was 81% in the Americas, 14% EMEA, and 5% APAC. The sequential decrease reflects lower deployments by our US-based global customers into their international data centers. The year-over-year quarterly increase reflects our US-based [global] customers deploying to their international data centers as well as an expanding international customer base.

Gross margin for the third quarter was 65.2%, down from 67.9% in the prior quarter, and up from 64.1% the year-ago quarter. The sequential decline primarily reflects an increase in the number of large-volume customers as well as an increase in overhead costs. In comparison to the prior-year quarter, margin improvement resulted from reductions in required warranty and inventory-related reserves and scaling operations with our revenue growth.

Our performance in Q3 clearly reflects movement in gross margin in accordance with our long-term business plan. We expect our gross margin to average in the low- to mid-60% range in the long term. However, it can fluctuate up and down from quarter to quarter due to customer mix, product mix, and the seasonality of our business.

Turning to operating expenses. R&D spending for Q3 was $31.9 million increasing sequentially from $31.4 million in the prior quarter and an increase from $25.3 million in the corresponding period in 2013. The sequential and year-over-year increases reflect headcount editions offset by timing of protocol expenses. Sales and marketing spending for Q3 was $18.6 million relatively flat compared to the $18.8 million in the prior quarter and an increase from the $13.8 million in the corresponding period of 2013. The sequential quarter change reflects seasonally fewer marketing activities in the quarter offset by additional sales and sales engineer headcount and higher commissions in line with revenue growth. In comparison to the prior-year quarter, the increase reflects headcount additions and higher commissions as we expand our sales and marketing activities to support the growth of our business.

G&A spending for Q3 was $9 million, increasing sequentially from $6.2 million in the prior quarter and an increase from $4.8 million in the corresponding period of 2013. The sequential quarter increase reflects additional accruals for the corporate bonus program and increases in insurance and taxes. In comparison to the prior-year quarter, the increase was driven by costs associated with becoming a fully operational public Company, litigation expense, and a corporate bonus accrual.

Operating income for the third quarter of 2014 was $42 million, increasing sequentially from $37.4 million in the prior quarter and an increase from $21.2 million in the corresponding period of 2013. Our effective tax rate for Q3 was 30.5%. Net income for Q3 was approximately $28.1 million, or $0.40 for per diluted share, using 69.7 million shares compared with a net income of $14.4 million or fully diluted earnings of $0.23 per share in Q3 2013 assuming the full quarter conversion of our IPO shares and the conversion of our notes payable and preferred shares to common shares. Our results for Q3 were significantly impacted by continuing movement of OpEx spend into future quarters from the timing of headcount and prototype expenses as well as certain accounting and finance compliance activities.

Turning to the balance sheet. We have cash, cash equivalents, and investments of $408.6 million at the end of Q3. Cash flow from operations and free cash flow for the nine-month period ended September 30, 2014 was $98.7 million and $87.9 million, respectively. Capital expenditures in the nine months totaled $10.8 million and were primarily related to purchases of development, testing, and manufacturing equipment. The accounts receivable balance was $84.1 million, an increase from the prior quarter balance of $67.9 million. Day sales outstanding was 50 days in Q3, up from 45 days last quarter.

Current and non-current deferred revenue was $77.7 million, an increase of $16 million over the prior quarter and primarily resulted from new service agreements and renewals. Inventory was $62.6 million, down from $71.1 million in Q2 2014, yielding turns of three. The decrease primarily resulted from a reclassification of spare parts from inventory to other assets.

Let me now move to our guidance. For the fourth quarter of FY14, our revenue target is $160 million to $168 million. Gross margin is anticipated to be in the 63% to 65% range, and we anticipate operating margins in the range of 22% to 25%. We estimate DSO will be in the low 50s and inventory turns remaining around 3 per year. As a reminder, Arista will not comment on its financial guidance during the quarter unless it is done through an explicit public disclosure. With that, I'll turn the call over to the operator for Q&A.

(Operator Instructions)

Sanjiv Wadhwani, Stifel Nicolaus & Company.

Thanks so much. A couple of questions. I know in the prepared commentary, you talked about all of your four verticals doing nicely. I was wondering if you could highlight maybe one or two that did exceptionally well, and then, looking out over the next 12 months what your expectations are? And then, second question. Competitively, Jayshree, I was wondering if you update us on what's going on competitively given that your biggest competitor obviously has volumes that are picking up with their new products. Thanks

Thank you, Sanjiv. Regarding our four verticals, the pattern was very similar to last quarter. We are continuing to see a very nice balance across all four of them. Usually, we expect more lumpiness with our cloud titan. But, both calendar Q2 and Q3, we have certainly received important orders from our cloud titan, but that's been very nicely balance by two other important verticals, our financials and our service providers and tier 2 cloud providers. So, I would say to you that it was very smooth and balanced across our high-tech enterprise, financials, cloud Titans, and service providers.

In terms of competitively, no major changes. We continue to see aggressive pricing behavior and aggressive discounting whenever Arista is in a deal. We continue to coexist in many accounts. Many of our customers are demanding and necessitating better technical solutions. And, also wanting to deal with source vendors.

Got it. Thank you, great quarter.

Thank you, Sanjiv.

James Faucette, Morgan Stanley.

Great, thanks. I just wanted to follow up on the question on the customer base. Just wondering if that balance in business that you have seen the last couple of quarters as you look into the December quarter and the visibility you may have beyond. Is that balance persisting across different verticals? Or, should we expect a little bit more lumpiness, as you said, particularly from the cloud Titans in the next couple of quarters.

And then, I just wanted maybe a clarification from you, Kelyn. You talked a little bit about OpEx coming in as maybe being delayed or pushed out a little bit. Related especially around new hiring, et cetera. Should we take that to mean that things were pushed from the September quarter to the December quarter? Or, that they are being pushed from the December quarter out into 2015? Thank you very much.

Thank you, James. I'll take your first question, and Kelyn will take the second. As you might recall, we experienced lumpiness with the cloud Titans in Q1 where one of our customers, Microsoft especially, was a large piece of our customer concentration. We have seen more balance in Q2 and Q3. It would not surprise me in Q4 to once again experience some lumpiness.

That typically ends up being our strong quarter our largest quarter usually for the year. And also, typically a quarter where we experience large customer concentration purchases. It may not necessarily be one particular titan. It could be a different one, but we fully expect that. In fact, we are pleasantly surprised to see the smoothness and balance the last two quarters.

And then, James, to your question on OpEx. We have seen a pushout from Q3 into Q4, particularly around as we think about ramping up our SOX compliance work and both remediation work as we work through our SOX compliance. And, it will also push out into the first half of 2015. And then, as far as hiring is concerned, we continue to have a very high bar as we hire, and we are out bringing more folks in and all of our functions, primarily around sales and marketing and R&D. But, we're going to continue to keep our standard high and hire thoughtfully.

Brian Modoff, Deutsche Bank

Yes. Just a few questions. One on the -- did you have any 10% customers in the quarter? And, second is around pricing? What you're seeing on the pricing environment competitively? What kind of pricing pressures you are seeing on a per quarter basis? Is it within your expectations, or are you seeing any more aggression out of Cisco or maybe even Juniper? And then, for Andy. The 25- and 50-gig product share you're looking at bringing to market, is that targeted at the cloud Titans again? When should we expect to see those on the market? Thanks.

Okay. So, hello, Brian. I'll take the first question, and then, Andy, of course, will take the 25-, 50-gig. In terms of was there any customer concentration, we continue to only have one customer who contributes to that concentration, and while their percentage contribution continues to come down -- I think we shared with you last time -- it has come down into the teens and double digits. We did not see anything more in terms of contribution greater than 10%.

In terms of pricing pressure, the 10-gig market, as you well know, has been exploding in ports, and we, Arista, have been progressive and almost proactive in bringing down the cost of 10-gigabit port so that customers can more quickly move to 1-gig. So, we do see, not pricing pressure, but aggressive price points that Arista themselves are providing on 10-gig. We have also witnessed the -- a greater acceptance of 10-GBASE-T as an alternative to 1-gig.

But, I would not say we're getting any greater pricing pressure than we always get and we consistently get from our competition. The number one and only competitor we really see is the major market leader that you know in the industry, and everybody else we see occasionally and rarely.

On the 25- and 50- question. This is a feature or function that is supported by future next generation silicon, which is in development. We cannot comment on this call about the availability dates on such products. But, the key thing here and the reason we built this consortium is we have to expect that the silicon vendors can build the silicon to so when it arrives as a product in the market it's actually fully compatible. And, this is the effort that we are participating with to ensure the earliest availability of such products.

Okay, and then, targeted customers for that? Well, it is available to any customer, of course. We do believe the cloud -- the large cloud customers will be the early adopters. Okay. Great, thanks.

Kulbinder Garcha, Credit Suisse.

Thanks. I have a clarification. Jayshree, did you say that you did have a 10% customer? Or, Microsoft was an over 10% customer this quarter. I just want to clarify that before I have my main question.

Just to clarify, we only had one 10% customer, and that's Microsoft.

Got it. Okay. My question for you is just in terms of -- given the customer traction that you are seeing. Can you speak about the various forms of distribution for Arista's product portfolio that you are beginning to think about? Maybe more international diversification, that kind of thing?

And, my question for Andy is -- I think around the time of the IPO, you did mention that Arista given the technology advantage you have would look at adjacencies. Where is the thinking on the product roadmap? Any discussion there would be helpful as we look into next year? Thanks.

Okay, I am going to take the question for Andy. While we've talked about long-term adjacencies, we have been very clear, both in the roadshow and our public calls, that our maniacal focus is the data center. Especially for 10-gig, 40-gig, and 100-gig. This is such a tremendously large market opportunity, anywhere from $6 billion to $10 billion in the next few years. That Arista wants to be focused there and be the absolute best-of-breed there, and today we are the number two player. And, we look forward to improving that market share.

And then, on the distribution?

On the international, Kulbinder, there's clearly room for improvement here. This is an area that will take time. It's not going to happen overnight in the quarter, but it's an area we're making investments this year. And, we look to see the fruits of our labor in the next couple of years.

Thank you. Good quarter.

Thank you, Kulbinder.

Alex Henderson, Needham & Company.

Thanks. Just two data points. Could you give us the headcount, and what you think the tax rate is going to look like as we get closer to next year here? And then, operationally, I was wondering if you could give us an update on your thinking on two technical issues. One, what is going on with the customers' attitudes toward the BiDi adoption of 40-gig multi-mode versus the 100-gig single mode.

And, second, obviously ACI has now been out for almost a year. I would assume that you have been getting a lot of feedback on issues, whether there are challenges to actually using ACI and whether your customer base has finally gotten through the testing and evaluation process on that product?

Okay, Alex, you asked three questions there. I'll parse those, and Kelyn is going to answer the first one. You asked about the employee count. We typically don't talk about specific numbers. But, Kelyn?

We don't on a quarterly basis talk about specific numbers. But, the second part of your question was on tax. So, as you know, I indicated that we were at 30.5% effective tax rate. If I look out to Q4 on a non-GAAP ETR, I would expect between 29% and 31%, and on a GAAP, the ETR to be approximately 31% to 33%.

If I think about next year and I make an assumption that my international revenue mix stays very similar to what is in 2014, our tax rate will be very similar, 2015 to 2014. Having said that, if the R&D credit gets initiated in, let's say, 2015, on a GAAP and non-GAAP basis, that would have about 350 basis points. Or, a 3.5% impact on the ETR, but you would see that if the R&D comes in, of course, every high-tech company out there would benefit in a very similar way. But, does that give you what you need?

That's plenty. Thank you.

Okay, and I'm going to turn it to Andy for the BiDi question, and then I'll answer the ACI.

So, the Cisco BiDi optics is a completely proprietary-type design that goes over a duplex fiber for 40-gig, multi-modes. Maybe 80 meter ridge. Last quarter, we introduced a much superior optics, which we call the universal 40-gig optics, which has three significant advantages.

Number one, it has a longer reach over multi-modes, 150 meters or more. Second, it's standards-based. It uses the 40-gig LR4 standard, but the best part is you can actually use it with both multi-mode and single mode, which is why we call it universal. So, the customer can make a choice how they deploy this initially multi-mode going through with single mode. We do believe this is a much better solution, and there really is no other advantage to BiDi.

On the ACI question, I think answered this before. We just had a 10th anniversary customer event. Let me give you first the customer perspective, which is they are pretty underwhelmed, and many of them don't understand what exactly the problem is that one is trying to solve with it. Because what's really happening in the new data center market is, you cannot really tailor and customize policy per application. The applications are pretty universal whether it's workload for big data or hadoop or virtualization or VDI or high-performance media transactions. And, what customers really want is a universal workload architecture with really good work flow visibility and good workload orchestration.

So, having an open, non-vendor lock-in, nonproprietary and highly programmable underlay and infrastructure has been fundamental. This has been Arista's mission. This is what we provide in our products, and we have the right APIs to work with different types of overlays whether they are controllers that are single function devices or more broader network management devices.

So, quite frankly, we have not seen any enthusiastic reception of ACI. It's possible that we are not witnessing the typical traditional enterprise customer that may be interested in ACI. But, given it's been near a year, the traction and the competitive nature of it versus Arista has been minimal.

One must also understand there's a difference between announcing it and shipping it. I think while it has been a year since it has been announced, it probably hasn't been shipping for more than a quarter. So, it's possible it has also been too early.

The last part of that question was if they had finished evaluating it and are now making decisions?

I can't really answer that because I haven't seen examples of that.

Thank you very much.

Thank you.

Mark Sue, RBC Capital Markets.

Thank you. I'm trying to get to map out historically how networking companies have historically seen success in the US and then subsequent growth overseas. Perhaps, Jayshree, the sequence of opportunities and the replication of data center wins overseas, since they are several years behind us in the US? And, how you see the length of the runway recognizing you're making investments now? And, what -- when you win data center-type of deals over here, how that portends for Arista wins over there?

Thank you, Mark.

I think it's fair to characterize that one of the reasons our international business growth can improve better is because we are doing so well in the US. Our growth in the US is so good it's hiding and shadowing any contributions we make internationally. But, if you actually parse our international business like I've had a chance to do, we are being well received in the more developed countries. For example, if we look in Europe -- well received in United Kingdom, Germany, Nordics, Netherlands -- those are typically the places. In Asia, I would point to Australia, New Zealand, Korea, and Japan are three examples. So, you would expect us to invest here and do much better than we are doing even now in these countries next year and the year after.

Having said that, I think as you rightly pointed out, there is natural risk aversion to new technologies in some of these international countries. And, I think the success of the US will naturally help in followers internationally. I believe our timing and our investment is appropriate right now. It isn't really missed opportunity. We expect to see the fruits of our labor in the next one to three years.

Okay, that's helpful. Secondly, just on the lockup. We're less than one month away before the official expiration. Just the thinking behind the early lockup?

Okay. I'll explain it, and then I'm going to turn it over to Kelyn for further. There are two complaints we received on -- during our IPO offering and beyond. Which is we didn't have enough shares on the float, and there was tremendous volatility on our stock. We believe by releasing the lockup, we can address for our investors both of those concerns, and we took a longer term view on it.

We also believe that we are helping some of our long-term employees, many of who have been here much more than five years. Especially our rank-and-file employees. Please note that none of the senior officers or directors or executives and large shareholders will be selling. So, we wanted to make sure that there was enough time in the tax year for both investors to invest in Arista and for employees to have some liquidity.

Kelyn, you want to add to that?

No, I think -- when I think about the employee base, and we enter into Q4, the lockup would have expired on December 2, and the trading would have began on December 3. It really only provides eight trading days for our employee base and our shareholders that have been subject to this lockup to trade before the blackout period begins on December 15. So, by this slightly early release, we're allowing our employees, not non-employees, but our rank-and-file, to have the opportunity to have an additional 15 days of trading before we go into the blackout period.

Okay, understood. Thank you and good luck.

Thank you.

Ben Reitzes, Barclays.

Yes, good afternoon. Thanks. Jayshree, can you comment on your partnerships? Particularly VMware. How is it going so far, and when do you -- or do you expect it to be material? And, how that develops? And then, one follow-up is with EMC now taking in the VCE, is there a potential that you could do some business with them within their conversion infrastructure when perhaps it was previously all Cisco all the time? Thanks.

Thanks, Ben. I would characterize our VMware partnership as very strong, both on the engineering side, and now more recently on the go-to-market side. Most people think that it is only an NSX integration with the prior Nicira acquisition, but it actually goes well beyond that for a lot of the EFX and market leadership that VMware has on several virtualizations with vCenter and VCIR. So, we actually have multiple touch points with VMware, in NSX, ESX, log-in site, we see ops and we continue to drive for the standards and technology with them.

In terms of go-to-market, it's too early to call material revenue, but we are seeing a lot of customer activity between the two sales teams. Tremendous synergy, and it's also being very welcomed by our customers who are looking for building that is best-of-breed cloud STACK, and they don't want to do it with each silo individually themselves.

Which sort of leads to your second question which is converged systems. I think the days of converged systems coming from one vendor and then selling for mediocrity is changing to more best-of-breed storage -- scale-out storage, best-of-breed network virtualization, and best-of-breed programmable networking. So, we are optimistic about working with a number of storage vendors. EMC has a lot of new announcements coming that we work closely with as well as other scale-out storage partners based on flash and scale-out [SCN] storage software

Okay, thank you very much.

Thank you, Ben.

Ehud Gelblum, Citigroup.

Couple of questions. Kelyn, on a clarification first. Can you give us a sense at least of what the services breakout is from product. I know your revenue historically has pretty much all product. There are some services, but it is small compared to other companies. Just give us a sense as to what that is? Has it gone up, or has it gone down. Do you expect it -- where is that percentage-wise? And, give us a sense if you think it stays that way.

Then, on your large customer, Microsoft, that was, again, a 10% customer, the year progressing as you had thought with the lumpiness having coming down in Q1, and, you're hinting it may go up a little bit in Q4. As we look at next year, should we be looking for a dominant player like Microsoft had been in 2013? Or, do you think that balance continues out into next year.

And then, two more things, if I could. Kelyn, you had also -- appreciate the forecast on taxes and what we should be looking for on taxes. If I remember right, there was some procedure that you're working on, or thinking about working on with respect to creating new tax structures internationally. With all the noise about tax inversions and Irish tax code over the last quarter, did some of that change your thinking of terms of what you could end up doing of the long-term?

And then, finally Jayshree on -- again, going back to the IPO. There was some talk and discussion about whether any in a white-box-switch world you'd be willing to sell EOS as a standalone product to put on anyone else's hardware? And, any follow-up thoughts you may have on whether that is still a possibility and what it would take to get there. I appreciate it

I'm tired from listening to the question. Now I have to answer it.

I'll walk through the first three in a sprightly manner. So, services today continues to be less than 10%. Therefore, we don't break it out. Services, as you saw, continue to grow. You saw the increase in deferred revenue that I talked about in my remarks. We expect it to continue to grow. It's going to cross over the 10% threshold at some point, and we will break it out. We're very pleased with the new services attached that came through the quarter and our renewals.

Moving on to Microsoft. As we mentioned in the Q2 call, we expect them to still be a 10% customer. We don't disclose quarterly. We will annually. We expect it, though, to come down to be in the teens, and as far as how we think about Microsoft this year as how we were planning and how we discussed it during the IPO and in Q2. It's playing out as we had expected.

As far as looking forward to next year and thinking about could someone else be a 10%? Could Microsoft? As we look, one of the nice things that we are seeing is an ever-increasing customer base that we have more volume customers today than we did last quarter or the quarter before and we continue to grow the base. And, as we continue to grow the base, the nice impact, or effect that has on us is that we don't suffer from the same type of customer concentration.

Okay. To answer your question on white box. First of all, let me go back to some history on the Company since we are on our 10th anniversary. When Andy and Ken and David started the Company, the initial thinking they had was that they would build a software-only Company. So, EUS has been architected as a very highly resilient, highly available, highly programmable software. And, that's what the E in EUS stands for -- extensible.

Now what they found is that our customers were looking for us to provide the full integrated experience of software, silicon diversity with our merchant silicon. But, we've always had enough traction there that can work with different types of silicon and different types of hardware if we needed to and if our customers demand that.

We have basically brought forward customized components of that over time. For example, our virtual EUS, our VEUS image, is a VM image -- a virtual machine image -- that any customer can implement on their control plane. We've been shipping that, and we have been improving that. This is something you can look at as basically a control plane white box experience that people want to play with our software.

You will see us as we evolve EUS as a product, EUS as a platform. Also, more customizations, more APIs, more SDKs, and more ability for customers to build on top of it. And, this is something that I alluded to last quarter with Facebook on the wedge example. A number of customers will take our version of EUS and customize it for a variety of their applications to get better application control. But, I don't look at that as white box. I look at that as their custom box, but using EUS as a fundamental foundation.

And then, Hudi, I just realized there was one question that you asked about our tax structure and our thoughts around that. And, the recent announcement out of Ireland. Our tax structure has been well in place since 2008, and there has not been any changes to that.

In the announcement, they talked about Companies that do have this type of tax structure are grandfathered in until 2020. As we look at this, they did reemphasize again that the corporate tax rate of 12% in Ireland is not changing. And so, we continue to want to invest both in A&L, our Irish subsidiary and just internationally, and continue to grow our international business. And, there's not going to be any changes at this time in the short- to near-term as we think about our tax structure.

Okay, thanks. Very helpful. Appreciate it.

Giorgio Krakopol, SunTrust.

Excellent on the cloud guidance, the main question is example how the ramp of this accounts look like in the absence of Microsoft growth? Can you give us a sense in what inning we are with this group collectively? And if the ramp so far resembles around your experience with Microsoft? And then, more specifically is there one or more of these accounts further along with their deployments? And how material are they in the mix?

Thanks, Giorgio. So obviously Microsoft is a high contributor to our growth and we continue to expect Microsoft by the way to be a 10% customer next year in 2015, as well. But having said that clearly our growth and contribution, in fact equal or faster growth is coming from a number of major customers in all four verticals, including other cloud Titans, tier 2 cloud providers service providers and financials. While these are not as large in size as a lot of Microsoft, they are definitely growing equal or faster.

Great. And then if I got look back in the last two years, Arista grew sequentially in the March quarter despite weak seasonality from service providers, and overall January and February slowed deployment activity among your customers. When I look ahead to your March quarter of 2015, do you expect a repeat of your historical performance, or do you think that business normalizes your customer spending patterns?

Giorgio, I am not smart enough to tell you how we'll do next quarter, let alone next year. But, I can say that Q1 tends to be seasonally our slowest quarter. We tend to be flat Q4 to Q1, sometimes even down. Q2 and Q3 are about the same, and then, Q4 is our usually our strongest quarter. This is the pattern we have seen so far.

Great. Thank you.

Thank you.

Jess Lubert, Wells Fargo Securities.

Hi. Thanks for taking my question and congratulations on a nice quarter. Question is also on 25- and 50-gig and now that Juniper, Cisco, and others are developing solutions, do you believe any competitive advantage here that maybe enables you to gain share as these solutions come to market? And then, I was hoping to clarify on the lock-up if the release applied to the David Sheraton Trust, or if they were still restricted at the moment? Thanks.

Andy will take the first question on the 25-gig and then Kelyn on the lockup.

The 25-gig multi-vendor standard consortium came into being early this year among companies who are highly motivated to bring this type of solution to market as quickly as possible. So, it is available in next generation merchant silicon. Having said that, unless you were very insightful and planning ahead a long time for your ec upgrades, you are not going to see that in any non-merchant implementations anytime soon. So this will create somewhat of a split in the market between the merchants growth that has the capability to support 25- and 50-gig and legacy switches that will not have that capability.

Is it your thought that these other vendors are not planning to leverage the merchant silicon. It if look at some of the high end switches coming out from Cisco, Juniper, Brocade, they are also using some of the same merchants silicon. Is that any reason to believe they wouldn't be moving down the same path?

I didn't say that. What I said is that anybody using the merchant silicon would have access to these 25- and 50-gig nodes. However, legacy ASIC switches proprietary switch designs did not anticipate this -- will not see 25-gig any time soon.

On to your question on the lockup. Again, I want to reemphasize again this partial release does not apply to any executive officers, any directors, any employees that centered rule 10b51 trading plans, or non-employee stockholders. So this partial release is not available to the David Sheraton Trust.

Subu Subramanian, The Juda Group.

Thank you. I have two questions. First, Jayshree, you mentioned strength in the chassis switches, the 7500EX series line. I was trying to understand, is the mix between module platforms and text platforms changing for you towards the modular platforms side? And can you tell what price before trends those two segments? And then, more broadly, if you look at 2015, this year was a year of some transitions in the competitive products side, as well as 10-gig options accelerating. How are you looking at 2015 for the market itself, recognizing, of course, you are not going to revenues at this point?

Yes, thank you. You and I talked about the before the mix of modular [trying to leave] has remained relatively the same. What I would add to that is 7500E acceptance has been very good. What's very clear to us, Andy, Hugh, Ken and team have some done real-world and simulation testing, and this is the only platform in the market with 10-gig, 40-gig, 100-gig, adequate density, poor density capabilities, small footprint and the right size of buffers; almost thousand X buffering than anybody else. And people think large buffers makes latency. Actually, large buffers means better application latency for your customers. So, I think the acceptance of the 7500, particularly in the spline -- particular in 104000, and particularly in big data applications by our customers has been heart warming for us.

We also are seeing the spline in 7300, and they really are taking two different market segments in the chassis and spline world. In terms of price per port, no major changes quarter to quarter we have always seen changes year over year. We continue to aggressively offer price supports and ASPs that can be in the range of 10% to 20% over year, and you can see that in 10-gigabit ethernet price support trends. We see 10-gigabit Internet, especially 10GBASE-T as being options for top-of-rack server and leaf connectivity and the spines are certainly adopting more 40- and 100-gig.

In 2015, broadly in the industry, how do you think about that?

Yes, saying what I did, I meant that for 2015, we are not seeing major change in 10-gig being the leaf of choice and 40-gig and 100 being the uplink and spine of choice.

Right. Thank you very much

Thanks, Subu.

Paul Silverstein, Cowen and Company.

Thanks. Jayshree, if I am looking at the numbers correctly, and I apologize if you said this earlier, but it looks like your revenue in the Asia-Pacific region was down by about $6 million. If I look at the trend line, on would have expected it to be over $10 million in terms of the delta. It would seem what would have expected and what you have put up. Is that a function -- were there a couple of major customers that contributed a significant piece of the $14 million you did in June and they stepped back? Just volatility? Can you tell us what's going on there, and I do have a couple of others.

Okay, thanks Paul. APJ business actually did quite well, but as you know, we also attribute our global customer footprint to the geographies and that's where we saw a shift. So, some of our larger providers, our Titans will purchase for a data center located in Asia one time, or Latin America another time, or Europe a third time, or US. That's what affected the mix, and that's the change you saw between Q2 and Q3.

We haven't seen a significant decline or increase, for that matter, in APJ business. We're doing okay, but we could do a lot better. But that particular change was due to mix of specific customers (multiple speakers) data centers. We are more beholden to where the data centers are located in terms of purchase.

Got it. Thanks. And then, on Microsoft, If I understood your commentary correctly, it sounds like there is a change in your Outlook. I think previously you been forecasting or expecting Microsoft to be relatively flat. I though I hear you say earlier in the call that you now expected -- or that you expected to be up, albeit the rest of the customer base to be far stronger growth, not unexpectedly. But, if that in fact a change in what you're saying, or what you're expecting from Microsoft?

No change. I was simply highlighting the fact that they were very strong in Q1, and, obviously, as a percentage of our revenue the upcoming down every quarter from the 22% down -- 22%? In Q1 -- down, and we expect that to continue. However, Q4 for every customer, including Microsoft tends to be a stronger quarter

Two quick others. I know you don't typically break it out, but can you give us some insight on the revenue contribution from new customers versus existing customers. I think last year was really about 85/15 existing versus new. Can you update that?

We try to give you some annual guidance, but we don't do that when the quarter basis. But, I think we continue to derive a stronger amount of revenue from our top 50 and top 100 customers. Is it 85%? Probably not that high, because our new customer acquisition is also growing at the remarkable rate. But we don't do the breakdowns by quarter, Alex.

One last question, if I may. I have to ask you, but I know Doug Gurule and a couple of other senior, somewhat senior executives left recently. I trust, or would hope that as a function of the IPO, and folks who have been working hard wanting to cash out. Can you share with us any thoughts?

Absolutely, we lost a couple of good talented executives this quarter, who have been with us five years: Doug Gurule, formerly our VP of marketing, then more recently our VP assistant engineer working for [Andres Sedana] our senior VP, as well as David Watkins, who was running our EMEA region for Mark Smith.

And look, our attrition is extremely low; it's in single digits, and we fully expect that some individuals who had tremendous financial success and also wish to move on to other things -- in the case of Dave, he is going to be retiring. In the case of Doug, he is resting for now, but I'm sure he will explore life beyond Arista. So we wish both of them very well, and we would rather explore the things and invest in rest. I think it was extremely friendly and in the call parting in both cases, and a natural part of the evolution of the Company.

Appreciate it.

Thanks, Paul.

Michael Genovese, MKM Partners.

Hi, thanks. Do you have any appreciable amount of revenues many vertical outside of your main floor? Is there any quantification you could give us there? And, on a related note what about VAR expansion, as you move from a more direct to also a VAR channel? Can you give us any update on how many VARs you added in the quarter or any other metrics we could use to track that?

Thanks, Michael.

We did not have -- outside the four verticals, since we lump a lot of our customers into the high-tech enterprise, there is not one not really specifically identify [a whole of] this quarter. So, the answer to your question is, no there's no particular one to single out. In terms of VARs, this is an area that Mark Smith has been sowing the seeds and putting some groundwork. We typically will fill almost 100% through distribution of VARs and resellers. But, a lot of our selling efforts have been direct to date, because of the complexity our technology and the intimacy our customers require directly with us, quite honestly and frankly for the large purchases they do. We do believe in 2015 and beyond, we will be selectively working with value-added resellers worldwide and really picking two tiers of distribution: one is what we call the elite VARs, where we work closely with them and they would be an extension of our sales organization, and the other would be quarter distribution for fulfillment, like we do today. So stay tuned: perhaps in one of our wildcard topics we will actually cover this in more detail when we are ready. But this is definitely an area of strategy and expansion and opportunity for Arista.

Great. Thanks. And then finally for me, on the -- looking at the December lockup, could you tell us where the 10 B51 plans have been established for senior executives? Any detail you give on that? And how should we think about the share count as we move into 2015 in the first half of the second half of the year? How should we model share count?

So, if we think about the 10 B51 plans, I will tell you that on a number of our executives have entered into 10 B51 plans, and we, of course are supporting that and encouraging that. As I think about the share count for Q4, given the lockup volatility and stock price, thinking about the treasury stock method and how you calculated it, and thinking through exercisable options that get exercised and then are not a part of the fully diluted calculation. If I was making a range, I would say share count would be for Q4 between 70 million and 71 million. But there's -- in this particular quarter, because of the lockups expiring, there is more moving parts in it.

We think that the share count, obviously, continuing to move up. So, by the time we got to Q4 2015, I know you don't guide that far out, but do you want to give us any goalposts around share count the year out?

Not at this time, we will address that in our Q1 call.

Okay, thanks a lot.

Jeff Kvaal, Northland Securities.

Yes, thanks everyone very much. I was going to ask you about revenue growth trajectory a little bit as well. Obviously, being on north of 50% is a great number to be at. At the same time, I think we can see that the revenue growth trajectory has slowed quite a bit from 2013 and 2014. Could you perhaps talk to us a little bit about what sorts of things are underlying what you think the growth trajectory may look like for going forward, and maybe that Microsoft for example, being a sizable chunk of your revenue is a flatish and that is dragging things down. It maybe that you are expecting a big pickup in Europe. How you're feeling about how the trajectory has been, and what should we look for going forward? And then, I have follow-up as well? Thank you.

When I look at the revenue trajectory, Jeff, I think what you're really seeing here is the [below] large numbers. We obviously can't grow on the base of what's going to be a $500 million year the way we did when we were $250 million or $300 million. I think you're seeing a natural decline in percentages because of the large gross, but you're not seeing any fundamental changes in the fact that Arista is growing much faster than the market growth of our segment and Arista's customer acquisition and penetration into these customers continues to be very strong. So, I wouldn't read too much into any particular vertical or our growth rate itself, but barring any kind of recession or crisis or change in market conditions, I would say our growth is well above the average of our market segment.

Okay, thank you. That's helpful. And then Kelyn, my follow-up is for you. I think last quarter you moved a little bit of product-deferred revenue from deferred into the actual revenue line. Are there any similar movements underfoot in this quarter?

Not at all. As I said in my remarks deferred revenue is primarily made up of service -- service-deferred revenue, so no, there has not have been any movement of product-deferred revenue into revenue

Okay, thanks. Thank you very both very much.

Thank you, Jeff

Simon Leopold, Raymond James.

Thank you very much. A couple of hopefully relatively simple questions. One is, can you tell us the number of shares that would be available for this early lockup expiration that are held by the employees? That's the first one. The second one, and I'm sorry if there was an update that I missed, but just wondering if there's any status you can offer on the lawsuits with OptumSoft, even any insight what kind of legal expenses you are incurring? And then the last thing, I was hoping to get a better understanding what you did discuss earlier, was the lower operating expenses in the quarter. It sounds like you're having a little bit of trouble, in terms of hiring people the quality you'd like the hire? And, I guess what I'm trying to understand is, how does that change? How do you manage to find the people and hire to have the resources you need to grow the Company?

I'll take the first one, Jayshree, on the number of shares. So, if we look at the release and if you go to into the 8K that we file we have detailed in their the exact numbers. But, we will release 3.7 million shares of common stock, and 1.8 million of vested options on 11/11 for a total of $5.5 million. And again I want to remind everyone that this is a release only 50% of what that holding is for employees only.

I will answer the second two questions. We ask on OptumSoft: there was not much to discuss there, so there was no real changes in status. The trial is still in discovery mode, and the hearing date has been set for 2016, so there isn't a quarterly report really. No news is no news.

The legal expenses have been as expected?

Our expenses have been as expected, and they are included in our performer numbers, and not been pulled out. So this is really no news there.

Okay, and then, the last part. Thank you.

On the last part, on having difficulty in hiring, I would not say we are having difficulty hiring, but I would say were holding the quality of our hires and that bar very high. And also, some of our expenses are moving up based on our new provided baseline introductions. So it is that combination that's contributing to our expenses being lower the we are planning. So, in terms of hiring, some of the things we have done as a post-IPO company, is move away from stock options to more to more RSUs, to better adjust. We have also, for the first time, instituted a bonus plan for 2014 that will obviously continue into 2015.

We're looking at engineering to expand our geographic location beyond our headquarters in Santa Clara. Today we have locations in Canada and India, but we will expand beyond, that as well, and stay tuned. In terms of sales and marketing were aggressively hiring, but we're always favored experienced high-quality hires sheer coverage models. So we will continue to deepen our four articles as well as internationally and really do smart sales and marketing.

And, just a follow-up on that very briefly, your guidance the supply that operating expenses are growing significantly faster than sales at the midpoint. I want to verify that's your expectation for the December quarter?

We haven't executed as well to this plan, Simon, but I can't promise that will be our actuals.

As I said in my remarks, we have really kicked off the SOX (technical difficulties). There are other ancillary things that will be falling through G&A as we continue to be ready to be SOX compliant. We're continuing to make investments in our ERP systems, not only for SOX but also to prepare ourselves to be able to scale as we continue to grow our top line. So, you will see OpEx growing as we continue down the path of getting completely compliant and being a public Company.

Thank you

Ittai Kidron, Oppenheimer.

Thanks my questions have been asked. I do want to ask about the gross margins. They have fallen to where you expected them to fall longer term, that's correct. But can you talk a little bit about the disparity in gross margins between large deals and smaller deals? I am just a little bit concerned that implied in your gross margins that large deals, your gross margins are much lower, and as you start becoming bigger component of your customers and they start relying more and more on you, is there a risk that the gross margin range starts drifting down even lower going forward?

Thank you. First of all, gross margins we been rethinking would come down, and you didn't believe, and I guess we finally did show you we can come down in gross margins. So, there are two reasons our gross margins came down. One was the customer mix, and the other was overhead in operating COGS that Kelyn will cover. I will talk about the first one.

There's no question that our gross margin and our pricing to customers is tied to volume. And there is a difference between a customer who buys a few units and a customer who buys thousands, if not millions of ports. So, there's a delta. And however, the delta alone does that contribute to our gross margin. What happens this quarter was, a combination of both that customer mix, as well as our overhead. And both were equal contributors to our decline in gross margin.

Kelyn, you want to add to that?

That's absolutely correct. So again, it was the customer mix and the volume discounts which we are very pleased to see. We are expanding footprint. And then, secondly, we did have some increased manufacturing overhead that we absorbed into OCOGS.

And as I have always said, when we on the market growth phase were absolutely going to make some trade-offs for gross margins in favor of market share. We'll do that in a thoughtful and balanced and sensible fashion, but we will absolutely do that.

Very good. And second, following up on Simon's question regarding hiring. I understand you're being selected, and as you should. I guess the question is, with the stock being at $80 from a stock option standpoint, do you find yourself be somewhat limited in how generous you can be in bringing in the right talent? Is hiring suddenly becoming very expensive for you just because of where your stock is right now?

I think we hire a different profile of folks who are less about the stock options, and more about joining a Company that isn't risky. So, it's a different profile of still very capable people who see the upside and the growth and the future: the next 10 years to be very exciting and can be very rewarding as well, both for the career and financially. But, you bring up a good point that stock options at $80, we are less likely to grant stock options, and more likely to have base salary, bonuses, and RSUs.

Good luck.

Thank you.

John Lucia, JMP Securities.

Hello. Thanks for taking my questions. I have two questions: the first one is on international. You noted lower deployments internationally by US-based global customers cause weakness internationally. I just want to understand why you think that's happening? What's driving that? And also, what kind of customer concentration there was with those lower deployments internationally? Was it just a handful of global customers that didn't deploy, or was it a broader issue?

I'll take another stab at that answer. So, if you think about it, as our global -- our US-based global customers deploy, they are not going to necessarily consistently deploy every quarter in Singapore or Japan. They are going to deploy in a data center in one particular geographic region. And then, they are going to deploy in another geographic region. So we really haven't seen -- if you think about what we saw was some significant deployments that happened in APJ last quarter. But, if you look at our overall international mix, this quarter to previous quarters and look at that trend, we still run around that 80/20 split and have consistently been there. And again, our percentage, our international revenue as a percent of revenue gets overshadowed by the rapid growth of what's happening in the US in the top line there. But, there has not been absolutely no deterioration in our international business.

Okay, that makes sense. My second question, you noted the quarter was well-balanced from a vertical perspective, but can you speak what you're seeing the most growth from new customers? Last quarter, you indicated some good traction with the high-tech, in particular. So if you can update us there, that would be helpful.

We did see growth across all four verticals. If there was one I would want to point out, it would be the service providers and tier 2 cloud providers. We saw good growth there

Okay, thank you.

All right. This concludes the Arista Q2 2014 earnings call. I'd like to say thank you to everyone for joining us today.

Thank you for joining us, ladies and gentlemen. This does conclude today's call. You may disconnect.