Arista Competition?

I am making an early return after one heck of a party last night and great work out today, now cuddled with hounds on the command chair in the living room. I lost your post post Kingran, but you asked is there anyone who may take marketshare from Arista in 100gb and 400 gb switches.

The answer is sure, Cisco is trying and has been trying for years, Juniper is trying and has been trying for years. They will continue to try. Huaweii (I probably spelled that wrong) of course is trying as well. HPE is not, because HPE sells Arista products to the enterprise. There is not any new and upcoming start up that I am aware of.

Cisco still has more than 50% of total switch ports in the market. Arista’s marketshare over all is small, and depending on where you get the statistic is not above 14% total, or as low as 5.3% or so.

A former bear analyst from Morgan Stanley had to take it all back and now sees Arista moving up to 19% total marketshare in a year or two or so (I forget the exact time frame - but he had to take it all back and eat crow, and so he did).

That is the overall market. However, although Arista has made a ton of money with just that small marketshare with 10 gb and 40 gb, it is really 100 gb and 400 gb market that Arista becomes more dominant. I will get into that infra.

First, in regard to 100 gb, it is so popular because it has more bandwidth, a tm ugh less cost per gigabyte and only 1.2x as much power consumption. The economics of 100 gb is utterly compelling. The economics of 400 gb is tbd, but presently will be more expensive, so it will take longer to get the same sort of compelling economic benefits for 400 gb and thus 400 gb will start out next year in only special use cases.

100 gb, according to Arista will be growing for years and have a tail longer than 10 gb, that was a decade and still going.

In 2016, there were 1 million 100 gb ports in the market, in 2017 5 million, in 2018 it is expected that there will be 10-12 million ports. I do not have the estimates for 2019, but it should continue growing, even as 400 gb makes its niche appearance. As you can see, the total ports for 2018 it at least double of the ports for 2017.

What makes this so compelling for Arista is that Arista’s market share is much higher in 100 gb than it was in 10 or 40 gb. Much higher actually. I have seen estimates as high as 50%, but at the last earnings call was said to be 25%, which I believe was short-hand for 27.5% marketshare. I believe his is the market leader. Total marketshare tbd as I look for better sources. That is much higher from the other marketshare numbers I mentioned Supra.

The growth numbers are self-explanatory, even though one would expect the price per port to decline each year, it won’t decline by nearly as much as the increase in volume. And it is expected that Arista’s marketshare will continue to increase, thus how Arista is able to grow far faster than any of its competitors and far faster than the market.

Arista is the beneficiary of the equivalent of Moore’s law in the switch market. What use to be non-growth market for many years is suddenly tossing out, just like CPUs did, faster and faster speeds for better economics, that are pushing upgrades to the networks.

Cisco, Juniper, et al., have all been trying for years to keep up with Arista, and Cisco identified Arista as a mortal threat years ago. Despite this, Cisco has had little success in defending marketshare against Arista. Both Cisco and Juniper have done rather awful lately actually. To the point that Cisco will no longer break out its switch revenues separately and instead lumps it in with a much larger category (this was Cisco’s largest product, and to do this is amazing and very telling).

There are several advantages that Arista has had, but the primary one is their EOS operating system that was built from the ground up, is an open standard so it can be used with any third party hardware, and is backward compatible.

Cisco, Juniper, and anyone else, will not have years to create their own OS from scratch, and instead have to patch together their OSes to make it work like Arista’s does. This create problems with backward compatibility, this makes it more difficult to work with third parties, and this makes it more difficult to manage the network.

Cisco, as for example, runs its OS on individual switches, and is of course difficult to work with anything it other Cisco products. Cisco is making efforts to be more Arista like, but imagine the innovator’s dilemma they have in regard.

Cisco is said to be “resurgent” by Morgan Stanley (same analyst who had to do a mea culpa regarding his long term bearish stance on Arista) in that Cisco has multiple teams working on 400 gb and will not be “Arista’d” again. This despite the fact that Cisco has been unable to stop the loss of marketshare in 100 gb. Will things change in 400 gb? Hey, I do not know.

Thus why Arista has done so well, and is doing even better, with 100 gb really being an inflection point where Arista’s competitive advantage grows, and why I have had such confidence in it. Things can always change obviously, and I do not have any more information in regard at this time.

Hope this helps.



Forgot to add that whereas Cisco software is focused on each switch, Arista’s is not focused on each switch but rather a centralized database that creates multiple advantages in management and cost of ownership.

This also does not take into account the inevitable switch towards Arista’s switches also being used to replace routines in many circumstances with tremendous economic benefits that appear to be so great that it is difficult to see how Cisco or Juniper, in the proper use cases where Arista routers can replace the Cisco or Juniper router, will be able to defend their marketshare without destroying their own businesses. I do not know the overall potential for switches to replace routers, but I assume that Arista is planning to continuall improve and increase the use cases and thus has an end goal of replacing as much of the router infrastructure as the technology will allow.

It has been said that Arista, at some point, can actually expand into managing optical. Arista at the last earnings call stated that they have no interest in entering the campus market (which consists of usually less demanding switch speeds) but they intend to disrupt the market without ever having to enter it. Thus, their long term plan that they are working on, piece by piece.

I am sure that Cisco, Juniper, et al are learning from Arista and that they will improve their offerings relative to what Arista offers, and we shall see if this creates a more competitive environment or not for Arista.



Thanks Tinker for that post. You need to work out and cuddle your hounds more often.


Yeah, thanks! Wish TMF would write an article kinda like this when they do recommendations.

Good stuff! Feel free to write additional posts on companies talked about here/bullish on :)!

Thanks again!


1 Like

Arista Competition?

Tinker, that is an absolutely wonderful analysis. I wish I could recommend it several times. Makes me glad I added to my already large position.


as an end goal of replacing as much of the router infrastructure as the technology will allow. that is what I gather from listening to the CEO who has also stated “we are not a router company” So fewer routers more switches ? Does that imply fewer but larger networks?

“Present networks are not designed to handle the new types of loads been put on them. Change in network usage is coming fast.” Not quote but an hypothesis .

if that is indeed true , the long history of disruptive innovation indicates that few incumbents will be able to handle this kind of change fast enough. Rather than all the technical details ,maybe that is most of what you need to know to bet on NVDA, PSTG and ANET. A bet where the odds are in your favor

I caught that Campus remark too but don’t understand how they will do it, snd how they can make money off it. Unless they will disrupt it by eliminating it in it’s present form… To the Cloud and beyond?


Thanks Saul, I appreciate it. Here is to such analysis becoming great prognostication for us all!

I use to put my buys and sells in real time, but found that people mostly forgot the great buys but never forget the rare, very rare, but material enough, wrong calls. So I have largely stopped doing it. But I can say the last 18 months, if you include also contribution I have made monthly has been the second best returns in my life (only better was QCOM to RMBS LEAPS, when I was a mere yelp of an investor). Been a very good stretch and I cannot expect things to continue as they have. But I can hold on for the longer term as I have no plans on living off of my portfolio any time soon. Which makes our perspectives different, although we both agree, there definitely may come a time to sell. But since I have become more passive of an investor as of late, my selling has been far less often (although I have tried on a few stocks and sold relatively quickly as they just did not feel right, thus also my hesitation to buy into new stocks as quickly these days).

This said, even with the recent drop in share price for ANET, if you go out 6 month returns, ANET more than doubles TTD, even after TTD’s recent 20% one day rise:…

Not real sure if that link, longest I have ever seen (although it may minimize upon posting) will work. But it is a chart of NVDA, ANET, and TTD.

Congrats for recent TTD holders! I am more in the camp with Denny on TTD. Not that is does not look quite fine, but I don’t like the slowing of growth (albeit a good argument is made that the slowing of growth in North America may be offset, overtime, and then exploded beyond (kind of like dark energy has done to the expansion of the universe over time (for those more physicist orientated) by TTD. It seems to me that due to all these walled gardens that Asia is where TTD will have to make its name. And I do not like the customer concentration. If Google wanted to, or Amazon wanted to (particularly with their entry into live broadcast television, and Amazon has started broadcasting NFL games live) Google or Amazon could offer a comparable product to TTD. TTD’s only advantage would be that it was a “neutral” player.

We all know what Amazon does to such players. Google, could just be a second source. I have just cowarded away from TTD. Seems it was the right call, absent selling ANET in the shorter term and taking the tax hit that would mostly have neutralized the advantages of selling and buying TTD over the last 3 months. Then again, #3 in such a large market, and perhaps dominance in Asia! That could be something to behold. But for now, I like what I got and have no interest in tax consequences.

I just found it interesting the relative time periods, stock gains, even with the recent big hit ANET took.

For ANET, the forward guidance that we hear at the next earnings conference will be like the report from the Department of Agriculture in the movie Trading Places on orange crop futures! Yes, for those who do not mind dating themselves.



Some comment gold from the guy who says he recruited John Chambers to Cisco.

1 Like

As a follow up to this thread, and the “resurgent” Cisco as was put out by an analyst who was clearly being a mouth piece to push Cisco propaganda (but of course many believe it, as teaching discriminating analysis of the press is not an often taught subject, and does require some subtlety and experience), this resurgent Cisco, disappointed to the point that a follow headline reads: “Cisco’s Slump Good for Arista, other Rivals”…

Those who follow the NPI board know that I have been making serious changes to my port lately. I use to be able to better shorter term time things, FUD events and the like, and actually did it so often started to get cocky sometimes. But that takes full-time and sometimes manic commitment and it controls your life. So more power to those who do so, but I have made a life choice to fall back to the “doing nothing again today,” philosophy until such time something compelling comes along that I need to do.

This was partially brought on by tax considerations, and basically there was nothing better to do. My port was SHOP (with thank you to Duma for introducing me), ANET, and NVDA (NVDA - I believe it was a Forbes article that I kept in mind, and then people panicked, the stock sold off, and I had to jump on it). As the year has turned however, things have changed. Not so much for NVDA, but I just was not going to do anything about it because of tax consequences. Then I finally finished my taxes at the end of April, and you know what, at least for this year, it did not hurt. I live well below my means, and there was plenty of money left in the tax kitty pool. So miracle! And what not. It freed me up to start better allocating my portfolio.

I had the best of the best (excluding biotech, which I made a decision to stay out of at this point in my life - another life style choice) but a new generation of companies was coming up, young eager, and it was like a football coach getting rid of some veterans who had another one or two good years left in them and bringing in the new blood. I am still in that process of settling it all down, and then I hope to be able to “do nothing” many times again (other than adding monthly) until something utterly compelling causes me to do something on the rare occasion.

In that vein of thought I reassessed my port, and independently it seems Saul and I came to similar conclusions. For one, Arista, I started selling Arista off from my SEP (tax free portion of my holdings) and eventually got around to selling it all off. One simply has to believe management, that 25% means 25%, and management has good vision into the next 2 quarters, and sure enough, they stuck with 25%. There are better places to be than in a 25% grower in the age of complete Big Data disruption.

But more than this it was forward looking. I went all in on Arista, 67% of my port at one point, when the “inflection” point of 100gb hit full stride. This was Arista’s time to stride, when its competitive advantages materially increased and Arista would become the leading marketshare holder of this rapidly growing product category. And so it was, and so it was beautiful to behold.

But wait, something happened! We should not be petering out of this cycle. The 100 gb transition has now become no different, really, than any other transition, and it is no longer extraordinary! That certainly did not fit with the investment thesis.

Further, Arista’s future now depends on moving to the enterprise and to the campus markets. This is totally different than the large and hyper scale cloud centers. Because these cloud centers were all new and disruptive and were defining themselves as they went, and were growing at enormous speed, and they had no time to sit and wait for competing bids, they needed solutions here and now. And no one had these solutions like Arista! Now that things have slowed down to more manageable levels, vendors can now take bids for projects, think of alternatives, and re-assert authority over their supply chain.

It is more difficult in the enterprise and in the campus. These are not new markets being remade. These are markets being upgraded and playing into the heart and strength of Cisco strongholds. There is no way in heck that Arista will have the same sort of dominance and growth rates in these new markets that they had when the world was young (hyperscale database tornado).

Thus, long way to say, why I sold out of Arista. A “resurgent” Cisco had nothing to do with it. It was time, and it was a compelling time, because there is an entire new generation of companies out there that are far more exciting, with much better forward looking prospects, and one cannot invest looking in the rear view mirror. And thus, once I finally got over this tax issue thing, able to move I was.