Arista's Quarter-Speed-Bump, Detour Or Something Worse?

About: Arista Networks, Inc. (ANET)
by: Bert Hochfeld

Arista's latest earnings release contained a growth guidance reduction that lead to a noticeable share price pull-backi.

The forecast reduction was entirely a function of the cyclicality of Cloud Titan demand.

Arista continues to gain market share in its switching business.

The turn in the Cloud Titan capex cycle appears to be just on the horizon.

Arista is starting to see the fruits of its massive research and development investment and the growth of its enterprise business is accelerating and is likely to continue that trend.

Arista’s outlook-just how bad is it?

Just about everyone interested in the matter has seen Arista’s (ANET) quarter by this point and seen the share price implosion in the wake of the weak guidance for this company. It was a huge surprise and disappointment, and no one was more surprised than this writer. The quarter itself, as reported wasn’t terrible, although neither was it a blowout. The upside to EPS came from margins-particularly the improvement in gross margins. It is worth noting that despite the demand speed bump, gross margins, due primarily to the customer mix, have been rising and not falling. And DSO fell noticeably year on year; the quarter was not back-end loaded and the company did nothing unnatural to achieve the numbers that were reported. The company did see an increase in inventories based on the supply chain ramp for the production of new products.

Arista’s problems, such as they are, are not competitive or operational but relate to demand in one component of its business, Cloud Titans. The problem, of course, is that Cloud Titans are the company’s most significant business component. And Cloud Titans rarely have a smooth demand pattern-although the comments about one of the Titans halting orders seem a little bit more volatile than one might reasonably anticipate.

At this point, ANET shares have lost 20%+ of their value from the high that they made in mid-April. That compares to a total pullback of less than 1% for the IGV over that same time That is, I believe, a substantial and unwarranted reaction to the earnings release and the guidance based on the information currently available. ANET shares at this point are still up by about 25% YTD, but they are essentially flat over the past year which has resulted in a substantially improved relative valuation that is now noticeably below average for the company’s growth cohort and far above average when looking at the relationship of free cash flow to growth. We have, to be sure, recommended the name on several occasions-and continue to do so.

I am not likely to claim that I know more about the network switching business than the folks that run ANET. I have said before that the two of this company leaders, Jayshree Ullal and Andy Bechtolsheim, are amongst the crème de ’la crème of the networking world. That remains the case and will remain the case for the foreseeable future. And so far as it goes, I am inclined to believe that Anshul Sadana, Arista’s new COO, and long-serving networking veteran with lengthy careers at both Cisco (CSCO) and Arista, knows more about the day to day activity in the network switching world than anyone else currently in the space.

It is obvious that none of these executives has formed an opinion at this time about the specifics of demand from the Cloud Titans in the short term. And of course there is a reason for that: there is simply no data regarding cyclicality on which they might rely to form an informed judgement. It is foolish to make commitments to investors or other stakeholders regarding the cadence of Cloud Titan capex, because trying to forecast things that have really no confirmed historical pattern is just not something that makes sense. My feeling in composing this article, has to be If the executives at ANET are unwilling to forecast beyond Q2, it would be more than a bit presumptuous for me to try to provide a dispositive forecast-although I have suggested, quite specifically growth rate ranges that seem likely to me.

But while they haven’t quite formed an opinion regarding the length of the slow-down in Cloud Titan capex, management did talk about the digestion period coming to an end by the end of Q2. That comment by the company's CEO ought to be more considered by investors, than the reluctance of management to talk about specific forecast components beyond Q2.

Obviously, the extent to which capacity is consumed by customers of the ilk of the Cloud Titans is the governing factor in the level of their capex. Cloud Titans need data switches to support the growth in transactions-over time there is a close correlation between transactions and Cloud Titan capex and that relationship can’t change materially. So, if the consumption phase of the cycle is coming to an end, then Cloud Titan growth will, of necessity, re accelerate.

During the course of the ANET call, management essentially withdrew its forecast of 21% growth for all of 2019-simply because they stated that they lacked the visibility with regards to Cloud Titan demand to support that number-although, to be sure, even in February, Arista executives were not terribly specific about their growth expectations.. Analysts, in turn, have now reduced their consensus revenue growth estimate to about 19.5% growth for both this year and next year. At this point, according to management, it has some visibility of Cloud Titan demand for a quarter or two, and little visibility beyond that.. And just for the record, I think those commentators who want to second guess the management of this company regarding visibility and the consumption of data switches are barking up the wrong tree and are doing so with really unfortunate methodology. My belief, as I express below, is that the current growth consensus will prove to be quite inaccurate. I think growth for Arista will re accelerate in the 2nd half of this year, along with demand from the Cloud Titans, and then will reach 30% or more in 2020. An unorthodox view-perhaps, but I think it is better supported by the evidence than the view embodied in the current consensus.

Summarizing various surveys regarding Capex in Cloud world

Obviously I do not try to form an opinion about capex for Cloud Titans in a vacuum. One can look at surveys or guess, or do something else. I have think looking at surveys is likely to be more accurate in compiling a forecast than other alternatives. At this point, the investment issue for ANET is pretty much a function of just how soon there will be a recovery in Capex by the large cloud vendors. The important content that I have drawn from looking at the numbers, and the forecast and one can find a variety of different surveys, often with conclusions that are not entirely congruent is that the capex for the Cloud Titans will increase markedly in the second half of this year and will show very strong year over year growth approximating 25% next year. Given my own limitation of resources, when it comes to surveys I have to rely on 3rd parties such as Gartner and IDC, but also on various brokerages that have their own survey teams. Brokerage teams actually do try to specifically forecast capex for the 4 most significant Cloud Titans. I have no reason to suspect their conclusions.. One of the articles on SA suggested that they believed it unlikely to see a 2H 2019 capex acceleration. That is not the conclusion that was reached by the surveys.

Since the article in question didn’t rely on much more than the judgement of its author, I will be happy to stake my investment thesis on the surveys that I follow. In particular, I anticipate that there will be a modest growth ramp in 2019. According to the latest survey I have read, 1H-2019 growth will be about 4%. It is that modest growth rate that has done in the current growth outlook for Arista. Indeed, looking at the expected growth for Cloud capex at 4%, and Arista’s revised Q2 growth expectation of around 16%, shows that the company is both likely taking share in its market segment and is also enjoying sustained material growth outside of the Cloud Titan space. The Dell’Oro surveys suggest that Arista has built its market share to around 17%, while Cisco’s market share in the high-speed data center switching market has fallen to about 49%. Cisco is still providing a noticeable price umbrella, and because of that and because of Arista technology, it is probably a low-odds bet to suggest that Arista’s share will continue to rise.

As many readers may be aware, 400 Gbps switches have recently become available. The absorption of these switches is likely, at least according to the survey of the 650 Group, to follow much the same trajectory as the acceptance of 100 Gbps switches. I would imagine that some of the strong growth being projected by surveys for 2020 is a reflection of the acceptance of 400 Gbps switches.

As was the case with the acceptance of the 100 Gbps switches, I would expect that Arista will have a larger share of the newer technology than it has built in the slower switch market-primarily because its software makes it easier to integrate the new technology into networks that are using the older technology switches.

I have written in the past about the advantages that Arista has when it comes to cloud scale architecture. Things like stateful orientation, software reliability and scalable architecture are well reviewed and avoid the process failures inherent in the so-called “spaghetti” codes of past offerings. The use of merchant silicon is another significant advantage for Arista and is seen in the company’s high level of gross margins. I think it is important to re-state-the shortfall that Arista is guiding to has nothing to do with competitive positioning or sales execution-it is solely a function of Cloud capex.

The estimates I have most recently seen suggest that capex for cloud service providers will rise by 19% in 2H and then rise by 25% in 2020. I think that a ramp such as that, would take Arista growth rates far above current published First Call consensus expectations. The survey work I have seen is backed up by committed plans of the cloud titans to open new data centers and availability zones. The data centers are obviously going to be commissioned with operating hardware. Arista will supply am increasing share of the data switches in new data centers; the Cloud Titans and Arista have a strong client/supplier relationship that is clearly not coming un-done. It wouldn’t be terribly surprising to see Arista Cloud Titan revenue growth rise from 10% or so this year to as much as 40% next year-levels that are both below and above a 20%+ growth trend and include market share gains for Arista.

Arista has lots of other revenue sources

Arista has been a child of the cloud basically since it started. It was designed to provide product to the cloud and specifically to provide switches to companies such as Amazon (AMZN), Google (GOOG), Facebook (FB) and Microsoft (MSFT)-the 4 cloud titans. But it sells to other users as well and over time, and despite the torrid growth of the cloud, it will be likely that the other client groups will surpass the size of Cloud Titans as a revenue source for Arista.

Probably the most significant potential for Arista is in the Enterprise area, the company’s #2 revenue generator. There has been no growth slowdown in the enterprise area and given the size of the opportunity and the advantages offered by Arista, I have no reason to believe there will be any growth slowdown, either now or into the future. The company had a record number of new $1 million customers in Q1 as it did in Q4, and these are, for the most part, in the enterprise segment. New customers are being closed at a rate of 1/2/day and as Arista becomes established at these customers, it will ultimately generate substantial revenues; much of the evolution of enterprise customers involves Arista landing and then expanding throughout the breadth of enterprises.

I have written in past articles about the potential for Arista’s campus switch, a product designed for enterprise users. I think it is reasonable to expect that 2019 will be a year during which Arista is qualified and accepted at many enterprises and continues to take noticeable share from Cisco, Juniper (JNPR) and other vendors in the enterprise space.

Arista classifies financial services as a separate vertical apart from the enterprise. And financial services is probably not the strongest vertical at this point. That said, many of the newest Arista products, which will be further discussed by the company during its June analyst day, are likely to resonate with those customers. The company has a segment that it describes as Cloud Specialty and Hosting Providers. This is a relatively small segment by revenue at this point but is growing a bit faster than the cloud titan space this year. Finally, the company’s smallest vertical is the service provider market and this segment has seen lackluster performance and is not expected to improve any time soon.

Arista obviously needs to see rapid growth in its Cloud Titan revenue vertical to achieve 30% growth-and that will happen, I believe, but it can grow at mid-teens rates because of the success it is enjoying, primarily in the enterprise area.

Arista, like all IT vendors is a product company in that what it sells, for the most part are products. So, the advent of new products which management suggested to be robust, will have a noticeable impact on demand. The company is seeing its software offerings show significant growth.

Arista has most recently started to make acquisitions. The acquisitions it has made are technology based. One of these acquisitions, Metamako, is starting to gain traction with enterprise and financial services customers. There are other products that don’t get lots of attention such as CloudVision which is a software telemetry solution that is already starting to add to percentage growth rates. CloudVision bookings doubled last quarter.

Valuation and Margins-Another couple of legs in the evaluation stool.

As noted, I have used a 3 year-growth rate of 31% in evaluating Anet’s valuation. I imagine some observers/readers will quarrel with that kind of outlook. I do think, looking at the specifics of the product offerings and the different verticals, suggests that 31% is quite reasonable and is most likely to be surpassed in 2020. But the fact is, one needn’t quite believe that Arista has a 3-year, 31% growth rate in order to make an investment case for the shares.

ANET shares are currently valued at 6.95X EV/S based on a 12-month forward revenue estimate of $2.8 billion. That is really not an aggressive revenue estimate if there is any kind of recovery in Cloud Capex. But an EV/S ratio of 6.95 is actually the average for growth percentages in the mid-high teens.

But more than that, ANET is far more profitable than almost all other IT vendors. In fact, based on my own analysis, its free cash flow margins are only surpassed by those of Palo-Alto (PANW), and only rivaled by Adobe (ADBE), Veeva (VEEV) and perhaps (TEAM). And those companies all have much higher EV/S ratios based on their profitability compared to ANET’s valuation.

ANET is very profitable because its gross margins are noticeably higher than those of most other hardware vendors. Non-GAAP margins reached 64.5% last quarter and are essentially forecast to remain at that level. The increase in year-over-year gross margins is a function of the mix switch from Cloud Titan to enterprise. Not terribly surprisingly, Enterprise, with smaller average deals, has a higher gross margin than Cloud Titan.

Like a few other companies I follow-most notably Atlassian-ANET has been able to trade its investment in research and development, for a very low rate of spending on sales and marketing. Sales and marketing spend, which grew by, 21% on a GAAP basis last quarter, is only 8.6% of revenues. Cisco, by comparison, has a sales and marketing spend ratio of 19.2% on a GAAP basis. Arista spent about 21% of revenues on research and development last quarter, up a bit more than 17% year-on-year. Cisco, by comparison, spent 12.7% of revenues on research and development.

Essentially, it is this trade that has allowed Arista to continue to exceed Cisco‘s operating margin, a notable achievement given the relative size of the two companies (Cisco is 20X+ the size of Arista).

Arista’s free cash flow margin will closely track its non-GAAP operating margin. The company, somewhat remarkably, has a very low ratio of stock-based comp-far less than most other high-growth tech vendors. Last quarter, stock-based comp was 4% of revenue and 14% of cash flow. Cash flow contracted last quarter, mainly because of a noticeable increase in inventories. The rise in inventories was mainly a product of the company beginning to prepare for production of some of the new products-both announced and unannounced. Over the course of the year, the increase in inventories will likely reverse, and thus the free cash flow margin will continue to rise, albeit at modest levels.

It seems to me, that ANET offers investors a combination of growth and profitability that is not to be ignored. It is, to be sure, a hardware company, and it does have customer concentration. But I really would like to be selling solutions that enable the continued growth of Azure and Facebook and Amazon. It is a nice spot. The company’s diversification efforts are under-appreciated, at least by some commentators, and its very successful trade between development spend and sales and marketing is not reflected by its valuation.

The shares have been pretty much eviscerated by the company’s growth guidance forecast and it is that forecast that has de-risked the shares and presented an excellent entry point. I am a strong believer as I have been for some time, that Arista will continue to provide investors with positive alpha.

Disclosure: I am/we are long ANET. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.