Does ANET really belong in the doghouse?

Seems that many on this board have relegated Arista to the doghouse. Like, yeah, it’s been a star but we have faster growing companies in which to invest, or perhaps, well, you know, Arista management is smart and so when they say growth will be slow, they’ve got to right, or when they say they don’t know that tells me it’s not a business in which I want to invest. Etc, etc.

Is this really deserved? Are the prospects for ANET really so diminished now?

Since many on this board sold in May, the stock has had a steady more than 10% rise - from $250ish to almost $280.

Some of the recent gains are due to a new product announcement: the 7170. This is their first unit based on the Barefoot Tofino chips, which are technically superior to the currently standard Broadcom (or even Cavium) chips in that they’re far more programmable without sacrificing throughput.

Martin Hull, Arista’s area vice president for cloud sayid: “The key aspect is that unlike a lot of high-end switches and routers today, the packet processor is programmable. It’s more like a CPU. It opens up a whole set of use cases.” (… )

This article:… sums up the advantages pretty succinctly:

The programmability of the 7170 enables specific functions to be moved back into the network and because only those features have been programmed in, there is little to no performance hit on the network and it allows the compute pool to do more computing with fewer resources, saving money on hardware, power, cooling, and all the other costs associated with running a server pool.

As the first article continues:
Additionally, because the platform is EOS-based, the same device can be used in multiple roles, each with its own profile, ensuring consistent management and provisioning… “When you’re looking around for a box to run NAT, the 7170 can do it at a much lower cost,” said Hull.

Hull said the real beauty of the platform isn’t so much its programmability as its re-programmability. “The customer can change it on the fly depending on what software you load” he said. “It’s programmable and re-programmable. We’re addressing multiple use cases with the same device.”

Even with the advantages that being first to leverage the Barefoot programmable chips provides, Arista’s solution still has its software are the main advantage. Others will of course eventually come out with units using those chips, but Arista’s software makes programming them easy and safe:

One of the difficulties of using a programmable switch is that the programming isn’t easy, and messing around with the network pipeline could have some disastrous consequences. This is one reason why many customers, even technically advanced ones, have stayed away from the Cavium-based white boxes. One thing Arista has over a white box is the company is releasing the product with pre-built templates, allowing customers to quickly configure the switch for their specific needs. … By providing the templates, Arista de-risks the deployment of something that customers may have shied away from because of the programming challenges.

Arista is providing two templates today (Layer 2&3 switching and cloud bare-metal config), with more (telemetry and security) coming in August.

And, here’s the real lock-in for Artisa: their same EOS software runs on any silicon and abstracts the functionality out while also providing silicon-based specific features. Customers that use Arista can deploy a wide variety of switches running Broadcom, Fulcrum, Cavium, and now Barefoot silicon. The operating system, EOS, will be the same on each switch…

This also means companies running Arista have future upgrade paths with much less disruption. They can replace old boxes with new boxes running the same software. There’s no more hardware vendor lock-in. Of course, if they want they can just purchase Arista hardware as well, and for a software company Arista appears to be doing a bang up job with hardware, too,.

I haven’t sold any of my ANET, and don’t plan to unless they stop leading the pack innovatively. This board is really good at finding the current hot, J-pop band type, growth stock at the hockey stick inflection point, but to me Arista looks like it has unique long-lasting qualities that I find more attractive. More Rolling Stones than Baha Men (… ), if you get my drift. ;^)


Tom Gardner posted about Arista a few weeks ago. I don’t think I’m allowed to paste the whole message here (you can read it on the SA board), but he sums up ANET as:

“Not all stocks rise. :slight_smile: In fact, something like 10% of all public companies drive 90% of the investment results.

My bet is that Arista is among that 10% over the next decade… it certainly has been solid performer since coming public!”


Smorgasbord1, thank you for putting together these details! I believe the same as you (ANET is #3 by size in my portfolio) and am not selling, though I had not put together quite so coherent argument on my reasoning.

Good post. I will have to finish reading it later. Personally, I reduced my ANET position by about 2/3 rds. I mostly agree w your sentiment. I don’t see any reason to give up on the stock. I think it is fair to say that I found some more compelling things to do w the funds for now. Another example, I just ditched GWRE for AGN. I just see more catalyst for AGN over next 24 months. We’ll see. But boy, did I ditch ALGN too early. There, I just thought it had gotten too far ahead of itself.



I sold my remaining ANET yesterday. A bird in the hand…

I think they will continue to grow, they offer a better cheaper product. But the Enterprise and even more so campus is a fight against entrenched interests, Cisco experts who see who see themselves as being devalued by Arista. I don’t quite understand the Campus bit because 6 months or so ago I can remember the CEO saying they are not going to enter that market, but then they did. Any idea what changed?

Part of the innovation theory is that companies will rush to new tech only if they have such badly broken internal processes that executives are worried about losing their jobs. I don’t think that point has been reached with Arista. Nor has the mass market “do it because everybody else is doing it” been reached. Yet.
Still I basically agree with you and ANET is on my watch list in case of some phony FUD or a general market collapse

Thanks to a cooperative market and tsunami like waves of innovation, there are lots of high growth candidates today. More than I can ever remember. And thanks to Saul’s board and NPI board deeper and better discussion of them than ever before.


I think they will probably be a good investment, but the hyper growth may be past.

They aren’t SaaS, so they have to repeat their hardware sales all over again every year plus more for growth (using Saul’s NVIDIA analogy).

They don’t partake in the SD-WAN hype/market, which is getting dominated by others such as Cisco/Viptela or VMware/Velocloud. Zscaler announced great numbers yesterday and called out SD-WAN roll-outs as one of the key catalysts for their business growth, as an example.

They are cloud-heavy, and going back to the “repeat their hardware sales” comment, they are at greater risk if cloud companies go another direction. Cisco may have an inferior solution, but they can also drive a tremendous amount of business to Azure, and there is likely quid pro quo relationships there. On a client call yesterday for 2 hours with Cisco team on datacenter, and majority of the time was on Intersight (cloud-based infrastructure mgmt). If anyone thinks Cisco hasn’t been paying attention to the SaaS trends and valuation upgrades by companies that go that route (PANW and NTNX as examples) you are nuts.

That ties to Enterprise clients. It is not simply a matter of “Arista is better” or “Arista is cheaper” that would suddenly cause enterprise clients to rip out the millions of dollars of Cisco investments. Not to mention if Arista sales can’t get above the Network IT teams and sell directly to the C/V level, they probably aren’t going to get a lot of traction to have IT veterans with years of Cisco training/certs suddenly agree to a move that makes their skillset obsolete. What is more likely to happen: A company commits to significant on-prem networking overhaul, moving from Cisco to Arista, or company decides to offload more applications/workloads to the cloud? Probably the latter, which is still an Arista sweet spot.

Clients can only handle so many projects at one time and so much change at one time. Cloud, security, PaaS, moving to HCI away from legacy SAN, leveraging GPU-enhanced VDI, etc etc…

Again, probably still a good investment, but they have been given a solid valuation based on their previous growth. When I looked earlier in the year, I believe they were about 1/3 of Juniper’s revenues yet twice their market cap, as an example. If their growth slows, the market may readjust what they feel is a fair valuation of a networking company.

Now…if they merged/bought Pure, or somehow merged with Nutanix, that would be really interesting. But that is day-dreaming.



I agree. Arista has a lot of staying power and scaling capabilities and that is very attractive.

I think most of us here are pursuing the hot growth of the moment and SaaS are a generational shift that can last a while but still it is the thing of the moment until the next thing comes along.

As for Arista some slowing and accelerating will occur but they are on a path for long term growth and I have not sold any but bought more last month. I don’t think one can really say when ‘hyper-growth’ starts and when it ends.



Hi there!

I know everyone is in love with the SaaS or subscription model - heck, I love it as much as anyone. I think it’s a great way for software companies to maximizing their customer value over the long term. And it also gives Foolish investors an edge, bc it’s not as straight forward to value as more traditional business models.

I have to say though, that it confuses me sometimes when it is implied that SaaS has some kind of business advantage because they don’t have to repeat their sales all over again every year.

I don’t think that that’s a fair distinction, because at the end of the day what drives repeat sales is the quality (and price) of the service provided. That is true for SaaS as for any other business. With SaaS it’s not like you make a pact with the devil and now you have to buy the service every year until eternity. SaaS businesses have to perform and provide a good service every year like everyone else. If not there will be some competitor to pick up that sale.

For me customer retention has therefore less to do with the SaaS-model; that’s just a way to charge customers (although I admit it can make retention potentially easier - especially in the short run). The customer retention advantage has maybe more to do with the way software works - it’s just very painful for an organization to change software once it is implemented. That’s why it’s stickier in general.

As regards to ANET, I haven’t sold any shares yet. I get the concern over slowing growth and weaker guidance. I will definitely watch the next quarterly reports very carefully. But I still like the business a lot. The market they are in is still strong I think. Competitive advantages are intact. Management is great. My biggest concern is over multiple compression as a result of slower growth rates. But I generally try not to react too much on valuation if it is still reasonable - which I think it is.

I want to close by telling you that I’m so thankful to all the experienced investors here (especilly Saul) that share their concerns and also are transparent with their decisions! This is so kind and immensely valuable to me! Thank you so much!

I also admire how quick you are able to change your mind on a stock and take action, irrespectively of the size of the position. I guess I’m just a bit slower… :slight_smile:

All the best,


Excellent post here and thank you.

While Arista is a software company, one can’t label it as SaaS. Take a look at deferred revenues of $181M which isn’t a large portion of their sales. I’m not sure of the sales cycle for Arista. I recall extended lead times in the 10 week range at one point during patent related issues. Point is, I don’t believe their sales cycle is very long, in fact, quite short. Not sure how long implementation takes…

Dreamer’s post was very insightful from a professional in the field who understands how buying decisions are made (I gather). I do want to challenge that for my own benefit, however. Let’s keep in mind this is from a layman’s view.

Before Arista, Cisco dominated the switcher/router market. Arista has captured large swaths of 100GB and encroached on the 10GB market to date representing somewhere just north of 15% of the total market. Unless Cisco is going to turn the corner all of the sudden and start eating Arista’s lunch, why do we presume growth is going to stall? Let’s ignore management’s guidance for one moment please.

If Arista or other competitors don’t start eating Arista’s lunch, are all of their customers going to stop upgrading their systems? Another way to ask that would be, how fast is the market growing? If it is at a standstill it is one thing. Quite the other if it is growing quickly.

Any insight is appreciated.



If Arista or other competitors don’t start eating Arista’s lunch, are all of their customers going to stop upgrading their systems?

Craig Barrett, who at the time was the CEO of Intel, was asked if systems would ever be upgraded to the point that they maxxed out. Essentially, would people have enough one day. He responded, “I never met a 15 year old who thought he had enough bandwidth.”

Just his humble opinion.

Explorer Supernaut
You can see all my holdings here:

1 Like

why do we presume growth is going to stall? I don’t think it is going to stall. Just be harder selling to Enterprise , so 25% growth that management suggests may be it. . It is a better mousetrap but enterprise customers will in many cases hang on to old mousetraps as long as they can. Switching is desirable but not a imperative for their company survival.

In any case I like to buy what is gong up , and ANET has gone nowhere since February. It is still adjusting the idea of 25% growth vs 35% or more growth. OTOH Arista has a huge potential TAM, and maybe one of the few Saul type stocks that you might lock away and hold for years


Thanks Jeb. I see I made a mistake in what I wrote. “If Cisco or other competitors…” is what it should have been.

At any rate, you seemed to understand what I meant. Appreciate the input.

I’d be interested to know, too, what others think. Arista beats forecasts, always. What is different today? Why do we expect growth to slow to the mid-20s?

I’m asking this more from a technical perspective. I understand one shouldn’t necessarily ignore what the company is saying in terms of guidance.


1 Like

In any case I like to buy what is gong up , and ANET has gone nowhere since February.


I hear you. But I’m not sure one can rely on this always. I might even say that valuations are stretched but we are not in imminent danger of a huge correction. Something with just a bit more value may hold up and appreciate more than super high flyers right now. I keep looking for options that aren’t overprice and rising like crazy.

Your timeframe of “since February” is pretty small. NTNX lingered for a time. PSTG did. AYX did as well. Nothing goes up always. Somewhat of a long term view is needed.



<<<our timeframe of “since February” is pretty small. NTNX lingered for a time. PSTG did. AYX did as well. Nothing goes up always. Somewhat of a long term view is needed.>>.

ANET does not belong in the dog house, rather, like me apparently (as my biking times are more than 2 minutes off what I use to do just two years ago - and I have been getting worse not better lately despite effort) Arista belongs in the masters league now instead of prime time. It is just a little slower than once it was.

Saul made a decision with Nvidia on the same basis. I disagree that Nvidia is not still in its prime, but you can see why one might think it is not.

100gb was the inflection point for Arista. This is when Arista takes over the market. And sure enough, growth increased up to 50%, and there was years of runway. Unfortunately this 100GB inflection point petered out rather more quickly than anyone anticipated. 25% growth. stated by management, and results matched statements.

Arista also faces a more difficult future. It came up in a new disruptive market, cloud titans. A business that heretofore did not exist and Arista provided the solution necessary to make it all work (Cisco and the other vendors had no solution that would enable such scale at anything approaching economic reality or manageability).

Arista’s new markets going forward are more difficult. (1) Cloud titans are becoming more self-sufficient and white label (we are assured this is not an issue, but it is, at least on the margins), (2) enterprise market already exists, and Cisco is entrenched and dominates and has now had multiple years to create more competitive product. There will be no surprise, and there will be no heretofore business that heretofore did not exist. Even AT&T has lead the creation of an open source software defined operating system for data center that theoretically does not need EOS. Thus will be a more difficult market, and (3) Campus switching business. Same, except the campus market is the least complex and usually slowest moving, and again dominated by Cisco.

Thus the past looks better than the future.

But fear not, there is a long-term potential upside. Arista’s multiple has come way down. It still produces tons of cash. Unlike this year where comparable were difficult, next year comparables will be much easier. With the lower multiples, if Arista can get its cloud titan business back up to 30-35% growth from here (very possible) and achieve traction in enterprise (campus market starts in 2020 - but enterprise market has taken off slower than Arista would like it to - i.e., entrenched Cisco that now has had time to produce more competitive product and pricing) one can see some room for big upside.

Even more so, if 400GBs comes out a little earlier than expected, and just takes off like 100 GB did. Then some even more huge upside potential.

I sold ANET for the same reason Saul did, I wanted a faster horse just like I want to be that faster horse again (will probably entail reduction in beer and wine drinking - perhaps better to just give up food to achieve same result - ahhh, decisions we jocks have to make) anyways, I think that is the high level view.

Certainly, by this time next year, with the lower multiple ANET has and the much easier comps, surprising to the upside may start again. In fact it is quite likely with this company. And the surprises only now need to be in the context of their lower multiple and lowered expectations.




I think you missed some of the deferred revenue for ANET. Last I looked it was over $450 M. They have deferred revenue in two places on their balance sheet, short and long term.

What an incredible explanation, Tinker.

There must be a reason management is guiding as such. There must be. But even with your amazing report, I still don’t see it. Somewhere someone just said one should be cautious of the thinking they can see everything. I can assure you I don’t believe I see all.

You point to white boxes and such, but we’ve always said the key to Arista is software. I believe their new router displays that (7170 I think).

It isn’t adding up for me. Arista is a rather large position in my portfolio. My consideration to date has been to trim it in half and that still seems reasonable. I haven’t acted upon that decision yet.

Thanks again.


1 Like

I think the most salient point of this discussion was Tinker’s point about the 100 gig inflection point. For those of you to young to remember the heady Cisco days as the “Internet” was being built out, Cisco the stock was a money printing machine.

Once the inflection point was passed Cisco the company continued to dominate the switch/router/security/VOIP market. Cisco the stock was a serial destroyer of wealth over the next decade.

Although the technology is very solid and the demand is still good, the moat is being drained. Over the course of discussions on ANET several have commented on Cisco’s entrenchment in the minds of enterprise decision makers. It is not a trivial thing.

IMHO neither ANET or CSCO will do particularly well, they will just beat each others brains out.


I like to buy what is gong up , and ANET has gone nowhere since February.

Four months without going up happens to every stock, if that’s the only reason, that’s no reason to sell unless you are strictly a momentum investor.


I am very sensitive to this phenomenon. I rode the CSCO and CSCO acquisition wave from a split adjusted pennies cost basis to $80…and back down to $12 (taking some but not nearly enough profits along the way). After all CSCO was the internet and the internet was just beginning it’s heyday right???
That is why when ANET guided lower 2 quarters in a row I just had to leave. When I see the ramp to 400gb beginning to hit I will revisit the name. Maybe it will ramp higher near term but that’s the chance I take.



lots of criticism for me saying ANET has gone nowhere. Yes I do use momentum as one (but not the only one) of my stock picking criteria. Especially when something fundamental (decreased earnings growth expectations) happens at the some time
See this chart, clearly something happened around Feb.

Since I am not Warren Buffett I have to buy stocks , not companies. I was not as quick to cut back as some but probably got about the same price.
What has changed is clear- Selling into Enterprise is harder than selling into cloud selling.
I was particularly disturbed by news at they were entering the campus market.

There is large universe of stocks out there. If I sold ANET and used the funds to buy something else that went up more than ANET, the ANET sale was a good decision. Going up is the most bullish thing stock can do. I am in this business to make money, not to be “right”

Should Arista earnings climb and the stock price shrink to support levels I might buy it again. In retrospect Feb highs were a blow off top. My WAG is that this bull will be over before ANET reaches them gain. But I do not base investment decisions on WAG.

Four months without going up happens to every stock, not in the midst of a big bull market. I don’t bet on horses that lag behind the others . Unless there is a clear phony FUD event. Which was not the case with ANET.