Arista up just 1% from Q4. Worried about it?

Thanks Saul
After selling before the ER
I bought back in after reading your posts.
I may be out for the next few earnings calls as it seems the market likes to sell off Anet on earnings (using the example of this time and last)
Thanks
MC

Thanks Saul, After selling before the ER I bought back in after reading your posts.

Hi Musicali, I just took some of my own advice and added a little bit to my large Arista position at $242.60. We’ll see how it goes.
Saul

2 Likes

Good call Saul.

1 Like

Not that it matters (or it shouldn’t) but yes I have been selling off Arista over the last two to three weeks and buying “hordes” of SHOP. That may surprise people given how I have been critical, but that is just my way of trying to dig into issues. I go with my first inclination, and then get to the truth of things (as best we can) by examining the issues.

Regardless, Arista was by far still my largest holding up until 2 days ago. I sold over half of my remaining shares, and yes, bought more SHOP. Arista is still a substantial holding, but much smaller than before.

The reason for this is part intuition, but it does not really take much intuition when less complex and far more straight-forward analytical thought can handle it. One recalls last earnings call when Arista was utterly wiped out. What was it, more than 20%. No biggie though. I just bought more. And it was a good decision as most buy decisions with Arista are, and have been for years.

But we are in a tough market at this time, when underperforming companies who hit bottom have bounced on earnings, and over performing companies with great earnings get cut on earnings. In such an environment, there was no way that Arista was going to outperform enough for a bounce, and the risk that companies guidance would actually come true was pretty high, as well, I believe Arista management for the most part (until I heard the parsing of words as A.J. described around white boxes. It was so un-Arista like).

Arista is in a transition period. And the shares are not going to greatly accelerate upward again until they give us some more security as to their dominant long-term position in the network, that white boxes are not slowly taking over many of their use cases (not their whole business obviously, but enough to cut into their growth rate on the edges and perhaps growing from there), the cEOS is a successful response to white boxes, that enterprise sales start to take off (it is clear, and Arista kind of admitted this) that enterprise sales are growing slowly, but they are not going as deep into the enterprise as they would like to be, and that the disconnect between titans and Arista is not a negative thing, but a lumpy thing.

All in all, momentum will change when Arista gets these things worked upon, but in comparison to other investments right now, it is more problematic.

Thus why I substantially trimmed. I will consider it I want to add or not here.

Tinker

11 Likes

For those still worried about ANET’s price action today, I encourage you to listen to the audio CC. It just might give you courage to the sticking point to add shares.

Jim

2 Likes

Why would one assume that increased spending by cloud titans would necessarily be proportional to their expenditure on networking equipment? Might not the big increase be storage or special processors or whatever?

1 Like

<<<Why would one assume that increased spending by cloud titans would necessarily be proportional to their expenditure on networking equipment? Might not the big increase be storage or special processors or whatever?>>>

Because more storage equates to more need to move the dat, ergo more switches, more routing functions.

More data centers being built equate need for more data to move around, ergo more switches, more routing functions.

Arista made the point that they are not necessarily correlated with titan expenditures. But all in all, I think an extremely good case can be made they are highly correlated.

One issue is AI. Titans appear to be going all in on AI in a big way. At the earnings conference Arista stated that AI was very early for them, and it sounded like they were not getting much additional business from it at this time. So to that extent, there may be lack of correlation.

But in general, when new data centers are going in, capacity is rapidly being put in, you need switches and routers that go hand in hand with it all. Having growth not correlate is something to consider, but to have an utter inverse correlation take place when Titans increase capital expenditures (and a good chunk of this is buildings and land and security and stuff - which may happen before actually building out the networking aspects of the data center) by 49%, and Arista growth slows to 25%, that is more than something to consider, but something worth digging into.

Who knows what we might find.

Tinker

2 Likes

Because more storage equates to more need to move the dat, ergo more switches, more routing functions.

Some increase yes, but not necessarily proportional. That is the assumption I am questioning. I think it quite possible that data centers could increase overall spending by say 50% and only have a 25% increase in the network spending because some other area of spending dominated the increase. If the increase is not proportional, then there is no reason to see the lower forecast for Arista as a negative or requiring some special explanation like white boxes.

4 Likes

could the Cloud titans be building the slow part, land buildings. power, etc well ahead of filling them up with electronics which can be acquired more quickly?
How much is seasonal, was Q4 2017 a quirk, or will it be repeated in Q4 2018?

At this point AI seems to be pre chasm.

2 Likes

AJ wrote - You can read (or better yet listen) to how the CEO equivocated when asked about competition and white boxes.

I listened to the call and posted the relevant part of the cc (about competition and white box adoption) in my last post here…

http://discussion.fool.com/guys-i-am-no-technical-expert-so-mayb…

What may seem like “equivocation” to you seems like careful articulation to me. She distinguishes between direct competition and white box adoption and says three things.

  1. Competition is aggressive - as it always has been.
  2. White box adoption by customers is not increasing.
  3. She throws in, perhaps for extra emphasis, that she is not seeing “any difference in competitive behavior due to white box.”

The analyst then seems satisfied with her answer. To suggest she is dodging the question and hiding something nefarious seems objectively inaccurate to me. I hear nothing sketchy or evasive in her tone.

BD

20 Likes

<<<“any difference in competitive behavior due to white box.”>>>

This is what A.J. is talking about. Prior to this statement she carefully defined that white boxes were not “competition” and then went on to say difference in “competitive” behavior. This is really parsing her response as it sounded more like to me she was talking about competitors like Cisco, and white boxes were not making a difference in how Cisco and the like were behaving.

I do not think I am reading too much into it. She is the one who carefully defined the term, and she is the one that really quickly blew off the very proper follow up regarding white boxes. Instead of simply saying white boxes have not materially affected our business or something direct, she said what she said how she said it.

I do not recall her saying that “white box adoption by customers is not increasing.” What she said was that white boxes were not “competition” because that was really a customer thing.

May be worthwhile to go back and re-review that part of the conference. May do so tonight. But that is what both A.J. and I were concerned about, and it was not a fluke the manner in which she responded. She knows that white box adoption is the #1 fear of investors. She was able to equivocate and get rid of the question, by pretending to answer it, all while never answering the question. Quite a talent to pull that off. I am sure it was prepped ahead of time. It Ms. Ullal also did not feel comfortable answering in this manner (was my impression). I think she is normally a far more straight forward person.

This white box question was an opportunity! She could have blown it out of the ball park by saying, no, no increase in white box usage by customers (but this would be a lie, as Microsoft, as an example, is dead serious about its SONIC initiative and clearly is using more white boxes).

Or hit a double out of it by saying, “yes, there is some more white box usage in certain, but limited use cases. cEOS is being well received running on these white boxes. But no, we are not breaking out software only sales at this time.”

I mean, honestly. That is why A.J. and I are talking about. Simple question, chance to get a double or even a home run (depending on the circumstances) and she simply does not answer the question by clever word parsing and self-defining words.

Yeah, white boxes are competition! They are competition by any definition but hers. A white box performs the same function that a Arista switch performs. A use case taken by a white box is one that Arista does not make a sale into.

Yes, there are some basic functions that Arista does not go after and that white boxes do. There they are not competing. But Cisco sells products that Arista does not, but that does not mean they are not still competitors in the markets that they do vy for together.

No, I am sorry. I think A.J. and I are correct. She needs to clarify this, or let it hang and we can make up our own fantasies about what it means.

Tinker

8 Likes

Tinker,

If the business model is making money off of hardware with software integrated into it, they will absolutely have issues.

There is no way around this. None. They must have a business that runs a seemless software across the cloud (east west traffic) and the white
boxes (north - south) traffic. Anything less and they will be displaced.

These are hard words, I do not use them lightly. This conversation will cause me to did deeper. Much deeper.

Cheers
Qazulight

6 Likes

They must have a business that runs a seemless software across the cloud (east west traffic) and the white boxes (north - south) traffic."

But, isn’t this exactly the point? If the N-S traffic is inherently less demanding (which it is), then lowest cost is going to drive purchases there. I would think Arista has no real interest in trying to compete in that market. So, the point is that Arista provides a cost effective, highly performant solution for the E-W traffic and the software for the white boxes on the N-S traffic that produces an overall integrated solution. So, neither Arista or Cisco is losing sales to white boxes because that isn’t a market in which they are competing. But, Arista is competitive both because their boxes are desirable for the E-W traffic and because of the integrated solution.

9 Likes

The analyst then seems satisfied with her answer. To suggest she is dodging the question and hiding something nefarious seems objectively inaccurate to me. I hear nothing sketchy or evasive in her tone.

I agree, Broadway Dan. Yesterday afternoon I’d considered picking up more shares in the after hours market, but decided to wait and listen to the CC transcript. Ended up picking up more shares this morning at a better price I could have gotten them for AH yesterday.

3 Likes

If the following is accurate, then the CEO’s statement is not only accurate, it makes perfect sense. It also means that Arista will continue to excel.

But, isn’t this exactly the point? If the N-S traffic is inherently less demanding (which it is), then lowest cost is going to drive purchases there. I would think Arista has no real interest in trying to compete in that market. So, the point is that Arista provides a cost effective, highly performant solution for the E-W traffic and the software for the white boxes on the N-S traffic that produces an overall integrated solution. So, neither Arista or Cisco is losing sales to white boxes because that isn’t a market in which they are competing. But, Arista is competitive both because their boxes are desirable for the E-W traffic and because of the integrated solution.

Cheers
Qazulight

6 Likes

Qazulight,

I started a quick search on use cases for white boxes. And it is like the biorhythms of Arista. Everything is about 18 months in arrears and then goes quite thereafter. Most of the talk on white boxes is dated 2016, some 2015, with Arista CEO coming up with a blog in 2014.

According to her blog, which I had read before some time ago, white boxes are for lower value, specific tasks, instead of general tasks. She did not use the term, but analogous to an ASIC sort of thing. She specifically mentioned for use around the security functions of a network.

A White Box OS model can work for a single, specific application in which one can narrow the use case to the specific topology of this single application, and can then develop for the narrow set of features required by this application. It has been reported that Google originally implemented this concept in 2004 (using several hundred internal engineers) to build a product. Facebook’s proposed design is another approach to open application control. These are both validations for open use cases in cloud networks.
Outside of these specific use cases, the need to be able to support
broader applications and mobile workloads and workflows using a Universal Cloud Network architecture has driven the need for a broad and rich set of features and capabilities. This includes the massive scaling of the MAC address table, layer-2 redundancy, in-service software upgrades at both the spine and the leaf, VXLAN and overlay network termination, integration with cloud orchestrators for virtualization and analytics, and improved network-wide manageability and visibility. White Boxes have limited support for these advanced capabilities.

In this new world of software driven cloud networking, the granular separation of control, management and data plane is needed. Each EOS function runs in its own restartable protected address space, in much the same way that daemons run in Linux. In fact, because we have preserved the Linux APIs, a wide variety of Linux tools (such as tftp or fping) can run directly on Arista switches, in addition to the functionality provided by EOS, to bring the best of both worlds.
White Box approaches and Arista Cloud Networks are indeed different segments that can co-exist for different applications as shown in the diagram below.

https://blogs.arista.com/blog/the-impact-of-white-box-on-clo…

Here is a more recent description of their uses:

In the meantime, white box solutions will continue to find their way into the enterprise for specific use cases. For example, Dominic Wilde, ?vice president and general manager of the data center networking business for Hewlett-Packard Enterprise (HPE), noted that white boxes are a good alternative for out-of-band management applications, methods of steering traffic in specific directions, and inspecting packets more deeply in support of a security application.
But, Wilde noted: “White boxes are not a panacea for every problem. We need clarity of purpose.”
Ultimately, white boxes will find their way into the data center. It just may not be at the level of adoption many vendors currently expect.

https://www.sdxcentral.com/articles/analysis/questions-linge…

Then I went to Gartner, which is pretty much the definitive word in technology trends and was not completely disappointed. According to Gartner there is delineation between white boxes as described above, and brite boxes (as not described above). A brite box is a white box, except it runs a proprietary and more sophisticated operating system, as Gartner mentioned like Cisco (IOS) or Arista (EOS, or would be c(EOS)).

Gartner predicted that brite boxes may be 10% of the market in 2018, up from 4% a few years earlier (roughly mirroring the marketshare rise of Arista actually).

Of poor news for Arista in regard, is that for the most part white boxes are best had by only the largest organizations (which is where Arista operates - thus the marketshare in such environments will be greater than 10% by mathematics), and they can be 6-7x cheaper by port. Enterprises are less price sensitive (probably very well trained by Cisco in regard - so even if they had the expertise they are less likely to care about a white box).

In the end it appears that white boxes and brite boxes (where we can hope that cEOS may play a very prominent role, but it is still early days there as Arista does not even mention any numbers or anything around it) are growing marketshare faster than Arista is at present, and this will probably accelerate moving forward. One might postulate that these boxes will take more from Cisco than Arista, given if you want to use such devices for cost savings, Cisco is by far the most expensive cog in this networking segment.

Alas, another 2014 article. Seems Gartner and Arista wrote the definitive pieces on this technology trend largely before everyone else.

https://blogs.gartner.com/andrew-lerner/2014/11/19/britefutu…

It does seem quite credible that Arista is losing some business with the cloud titans that they once may have had, at the low end. That would just be on the fringes of their business anyways. Because the cloud titans have the expertise to run these cheaper boxes with more sophistication than anyone else. But unless they are running brite boxes with a proprietary OS (or the open source one that AT&T tried to build - but then discovered they did not have the brain power in house to do so, so that is why they outsourced it - shall see how that comes along for them because they have a goal of something like 60% of their data center to be free and open source software) does not appear that any of the usual core business is being touched.

I like to wrestle with these issues as they come up and usually end up going in circles and finding out, not so much to worry about after all. Which is probably the case here. But good exercise to figure it out anyways and to know what to look for.

Tinker

27 Likes

Awesome info! Thanks Timker!

My take after reading this thread is that Arista will move into the software side of the white box segment. To me the real moat is Arista’s ability to build a seamless system across the virtual server center (east west) and the white boxes/brite boxes (north south) traffic. Having an intimate understanding of AT&T’s ability to attact and keep top software engineers and the output of second tier software engineers I would say that Arista has a significant cultural advantage.

(or the open source one that AT&T tried to build - but then discovered they did not have the brain power in house to do so, so that is why they outsourced it - shall see how that comes along for them because they have a goal of something like 60% of their data center to be free and open source software) does not appear that any of the usual core business is being touched.

The traditional core business is going to cloud type server centers. For example the new LTE switch going into Hawaii is a Virtualized server center that is running an AI app.

While the 2020 initiative had a goal of shutting down the entire circuit switched network by 2020; I do not believe that AT&T will reach that goal.

Looking at the financial statements of AT&T I would say that AT&T is very sensitive to price. (Also the budget cuts that have come down are pretty significant) It is my opinion that the Quality of Earnings of AT&T is falling year over year. I know at least one other telecom company (Century Link?) is in similar dire straits.

Having studied VMware and SIP (session initiation protocol) it is my opinion that virtualization and the software defined network is the defining technology today and for the next 7 years. 5g doesn’t happen without it.

Cheers
Qazulight

12 Likes

Excuse me Najdorf, but where are the shrinking margins?

The CEO said Gross Margins would be down ~1-2%+ this Quarter, over 1% at the midpoint. It’s in the conference call. That’s one of the major reasons the stock dropped so precipitously.

You don’t have to listen to me just stating the facts, but you should probably listen to the CEO if you’re long the stock. The market is forward-looking, last Q margins became irrelevant as soon as she announced next Q would be worse.

I don’t think it’s that hard to understand greater competition in this sector will lead to lower margins. Other firms aren’t just gonna give up and let ANET win.

No excuses necessary,

Naj

12 Likes

I don’t think it’s that hard to understand greater competition in this sector will lead to lower margins. Other firms aren’t just gonna give up and let ANET win.

got no dog in the hunt (though I want to investigate!), but what they actually said was:

Aaron Christopher Rakers Wells Fargo Securities, LLC, Research Division - MD of IT Hardware & Networking Equipment and Senior Analyst Yes, I just want to go on the gross margin line and understand. I think, this quarter, looking at the guidance of 62% to 64%, that would compare to 63% to 65% guidance ranges over the next – or the last couple of quarters. It looks like your product gross margin came down a little bit. So just curious of kind of what’s changing there, if there’s anything within the mix or assumptions there that resulted a little bit of a lower gross margin assumption?
------------------------------------------------------------------------------------- Ita M. Brennan Arista Networks, Inc. - CFO & Senior VP Yes, I mean, I think the biggest driver still for our gross margin is kind of within that 63% to 65% range is going to be customer mix, right? And as we head into Q2, we believe we will have a heavier cloud, the large customer base mix in Q2, and that’s causing us to be a little more cautious around the gross margin guide and the gross margin outlook. So it’s really a customer mix question. I think the 63% to 65% for the year, we still feel good about.

so a mix issue, not indicated as competition
my knowledge to judge this answer is ZERO

16 Likes

Some increase yes, but not necessarily proportional.

You might also add “not synchronized.” Network requirements come as a step function, not as linearly correlated growth to compute capacity. The network will have over capacity built into it with each enhancement, it is never (OK, make it seldom) planned to just meet current demand levels. The compute capacity (read processing and storage) will grow with little current impact on the network up to a point. Once it approaches saturation, the network will be extended to address the next stage of anticipated growth.

I would always expect a disconnect between the growth of cloud titans and the need to increase network capacity. I think the initial mirrored growth in both was an anomaly born from the rapid growth from a base of near zero. The industry has passed that stage. Growth at the cloud titans bodes well for Arista, but when the purchase commitments come home is unpredictable. Hence, Arista forecasts with respect to that of which they have reasonable certainty, not hopeful expectations.

1 Like