Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Unknown (00:00):
Metcalf, the founder
of Ethernet once once said at a
(00:03):
talk he said, I don't know whatcomes after Ethernet, but I'm
pretty sure it's going to becalled Ethernet.
Antone Gonsalves (00:09):
Hi, and
welcome to Tech News this week,
I'm your host, Antonio andsolvus, editor at large for
techtarget editorial, Ciscorecently released earnings that
spooked investors. Whatconvinced them to push the sell
button on Cisco stock, whichfell 10% in a single day, was
the company's lower thanexpected revenue and earnings
(00:31):
for the rest of the fiscal year.Here to help us understand
what's happening with Cisco isBob Albert, Networking Industry
analysts at Tech targets at anEnterprise Strategy Group.
Unknown (00:44):
Welcome, Bob. Thanks,
Anton. Great to be here.
Antone Gonsalves (00:47):
Yes. Nice to
have you. All right. So Cisco
beat analysts estimates lastquarter, but it's forecast for
this quarter and the rest of theyear convinced analysts to lower
their price targets. Whathappened? Or what did? Why did
everything changed so rapidly?Yeah,
Unknown (01:05):
I mean, I guess the The
interesting part is, I don't
know that it's so rapidly, Ithink this is the, you know, the
event that we've been seeing,building up to this for quite a
while now, with all the supplychain shortages, and
organizations wanting toovercome that. And pre ordering.
And, you know, across both thedistributor level, and the end
(01:27):
user customers, I think whatwe're just seeing is a bit of a
natural correction. And it's notjust a Cisco phenomenon. If
you've seen if you've beenfollowing the earnings
announcements all fall, you'veseen a lot of the networking
companies do the same thing,like, hey, we beat this quarter.
But we're just letting you know,down the road, things are going
to slow down as both thecustomers and the distributors
(01:50):
get to consume the product thatthey already have. So it's just
I honestly, I look at it as aminor correction in the market.
And I think, you know, it's notanything impactful of any of the
vendors products or anythinglike that. It's just more of a
there's a supply and demandissue. And right now, there's
plenty of supply, which is goingto weaken the demand for new
(02:12):
orders over the next couple ofquarters. And that'll correct
itself. And then we'll be backback online, as
Antone Gonsalves (02:18):
we all know,
Cisco is is a bellwether in the
tech industry. You know, CEO,Chuck Robbins, you know, he
insisted that the earnings thatdisappointed analysts, it didn't
reflect a slowdown in enterpriseIT spending in general. Right.
(02:39):
Now, investors, he said it wasstrictly what what you're
describing, you know, customersdealing with the backlog? Yeah,
the spending? Okay, well, butobviously investors didn't
believe them. They had doubts.This is so so why why the
(03:00):
disconnect? Is are is is is it?Is it enterprise IT spending? Is
it going to slow over the nextquarters? Or isn't it? Well,
Unknown (03:11):
I think like you said,
for all those products that
people have been that weresuffering from the backlog,
right, and pre ordering andordering as much as they could
to make sure that they couldmeet their demands have now find
themselves in a little bit ofsurplus. I think that coincides
from also from a, you know, WiFi perspective, there were a lot
(03:32):
of organizations over the lastcouple of, or at least this year
anyways, who spent a lot onreimagining their campus
environment, and preparing forthe return to work. And I think
we're seeing that a lot of thosethat maybe is a mature
environment as well. But like Isaid, from an overall
standpoint, we don't seeanything slowing down to that
(03:53):
rate from a distributedenvironment, right. So at the
Enterprise Strategy Group, werefer to the distributed cloud.
So applications moving tomultiple public clouds moving to
the edge being on private datacenters, employees still very
much engaged in hybrid workactivities. And then right, and
so they're working from home,they're working remote, they're
(04:14):
working in the office. So theone thing that's tying all that
together is everything needs tobe connected. So is there going
to be a slowdown for thenetwork? I don't think so. I
think as organizations continueto adopt hybrid and even return
to office, it's really going tocontinue to drive perhaps some
of that upgrade. Because what'shappening is the legacy gear
(04:36):
that they have can't handle thecollaboration bandwidth, right?
That video and voice traffic gointo every desktop, right? Think
about pre pandemic, you had avideo room. It was time for a
meeting and everyone was in theoffice and then you went to the
video room to talk to eitheranother company or someone who
was remote. The reality istoday's every desktop is a video
(04:56):
room. Right? And so and the wayEveryone connects that desktop
is wireless now, right? No oneruns wired to the desktop in any
new building, right prior to thepandemic, everyone was talking
about reimagining the officespace and completely wireless.
And so to enable thatorganizations are going to have
to move to Wi Fi six, ie Wi Fiseven, right to get that
(05:16):
additional capacity to supportthat, that video,
Antone Gonsalves (05:21):
and they are
going to need that kind of an
upgrade, given the fact that thereturn to Office is not full
staff. I mean, we're looking at,
Unknown (05:28):
it's not but like I
said that because it's not full
staff is exactly why they haveto do it. Because that means
what any, anytime they're goingto collaborate with somebody,
half of the people might not bein the office. And so again,
it's not. So if everyone'shaving this in the office, you
don't have enough video rooms toaccomplish that. So from what
I've heard, like said that thetalk track is typically around,
(05:50):
how do we reimagine the campusenvironment to enable those
collaboration apps to be usedfrom anywhere in the building
and not just specific videorooms, and then how this plays
out for the networkingcompanies, that means new
wireless access points. Andbecause these new wireless
access points accommodate the2.5, the five and the six
(06:11):
gigahertz range, they requiremore power, which means they
have to replace the wiredinfrastructure that's supporting
it, right. So they can get P OEplus and things like that. So
they have sufficient power topower the access points. And
they're multiple antennas, etc.So, so that's where that's where
you start seeing that change.Now, the question is where we
are not transition. Right, andwhether that's going to continue
(06:33):
to drive more, or people havealready bought for that. But
there's like said there'sthere's a few other things that
are that are driving it, aswell. I don't know if you want
to go there now or if you wantto stay ahead
Antone Gonsalves (06:46):
what because
yeah, what's so my strategy
group's research showing
Unknown (06:51):
us? Well, it's there.
It's not only the research, but
also the I've been on a lot ofcalls with the networking
companies Lately, they've beenhaving analyst days and things
like that talking about theirplans going forward. And there's
two little two little lettersthat keep cropping up. And all
those that happens to be a anAI. That happened. And so with,
with all of the, with all of theinterest around AI and all of
(07:16):
the activity that's going on,there's there's something that
we refer to it's like networkingfor AI. So how do you enable
those back end infrastructuresfor these AI environments that
these hyper scalars are puttingtogether these large enterprises
are putting together? And if youlook at it, you know, based on
who you talk to, it could beanywhere in that range of an
(07:39):
additional six to $10 billiondollar Tam, for these network
providers. So here's somethingnew that's going to drive
growth,
Antone Gonsalves (07:46):
what tam
stands for, what a
Unknown (07:48):
target addressable
market. Right. So that's, I
mean, they're going to make sixor 10 million, that's just
that's what it is to be split upamongst all the all the vendors.
And you know, a lot of themwill, you know, in that lower
end, one tends to be more towardjust the back end networks, or
the high performance back endnetworks. These are the ones
that will compete against theInfiniBand stack, did invidious
(08:09):
put together for that space. Andthis is where things like I know
if you're familiar with theultra Ethernet consortium, I am.
So something recently,
Antone Gonsalves (08:19):
Cisco and
Arista are members of that
consortium. And their purpose isto make Ethernet instead of
InfiniBand. The number one netCorrect? Yeah,
Unknown (08:32):
HPE. And there's a few
other companies on it, you can
go and see the whole list ofthem. And yeah, the idea is,
look, do you really want tointroduce another technology.
And we've all known if youfollow networking long enough,
there's been a lot oftechnologies that have been
introduced, but Ethernet isstill going. And so remind Bob
Bob Metcalf, the founder ofEthernet. Once once said at a
(08:53):
talk, he said, I don't know whatcomes after Ethernet, but I'm
pretty sure it's going to becalled Ethernet. And so and
that's what you're starting tosee with this ultra Ethernet,
right is that hey, how do we,how do we have a tech? Why
don't? Why would we try and putsomething completely new in
where people don't have theresource, the skill sets, etc.
When we could just modify andleverage Ethernet to do the same
(09:14):
thing, right? Let's keep thecost down, leverage the same
boxes, the same methodologies,there's a little bit of
difference, right? Because theperformance is required. And so
there's some new technologiesand that's what the that's what
the the UEC, the ultra Ethernetconsortium is all about how do
we modify Ethernet? How do wemodify the rocky stuff to make
it simpler to make it nonblocking all this kind of good
(09:36):
stuff to enable organizations touse that? And so, you know, like
I said, with Arista, with Ciscoand HPE, and you can imagine
others will join as well.There's a lot of other companies
in there, that they're reallygoing to be focused on. How do
we how do we incorporate thisand build it out? So that's a
big opportunity for a lot of thenetwork companies to expand
their Tam and get back to agrowth area right through Do
(10:00):
something that doesn't existdidn't exist six months ago 12?
Well, it did, but not in a wayit is now with everything
blowing up and all the attentionbeing spent on it.
Antone Gonsalves (10:09):
Okay, so who's
going to who's going to be the
buyers of this ultra Ethernet?When and if it comes, I mean, we
have I mean, when it comes toAI, enterprises, these based on,
on my reporting that enterprisesare not this will try to figure
(10:30):
out the return on investment onAI, they're still trying to
figure out what's required. Andbecause there's so much more
than even beyond the technology,what's required technology wise,
but to get the education ofemployees, how do you use it?
How do you secure it, I mean,these are issues that they're
struggling with. And as aresult, if they're under
pressure to do something rightaway, if a CIO is under pressure
(10:52):
from the board, or the CEO to dosomething right away, they're
just turning to Microsoft toSalesforce, and use AI. All
right, so you're talking aboutsomething about Ultra Ethernet?
I mean, is that really, fornetworking companies to sell
more to service providers andcloud, but then cloud providers
(11:13):
hyper scalar, scalars, asopposed to the universe? So
Unknown (11:17):
I think this is going
to follow a path, right?
Initially, it will be thosehyper scalars, right, the
service providers, the, youknow, really large enterprise,
who will be wanting to buildtheir own training and
inference, malt, rightenvironments. That's who's going
to adopt it. But like mosttechnologies we see, and even
hyper scalar technologies maketheir way down to the
(11:39):
enterprise. So the initialmarket is yes, it's about
enabling these hyper scalarsthat are using 10s, of 1000s of
deep of GPUs, and like justmassive environments, huge power
consumption, that's who will bethe initial customers, you'll
have some of the really largeenterprises doing that as well.
And then you'll see over time,right, that you'll see other
(12:02):
enterprises wanting to havetheir own environments wanting
to build things out. But you'reright, if you're, if you're an
SMB, or a smaller enterprise,and you can leverage, you know,
Ignite was this week. So it'stop of mind, right? If you can
do co pilot for 365 andMicrosoft graft and use their
studio to integrate with your,your SharePoint, for data,
right? Because it's all aboutthe data for AI, right? How do I
(12:23):
get the data? And mostimportantly, for all the
enterprises like you're saying,from a security perspective,
it's about how do we make surethat none of our data gets put
into a public API environment?So by building out those
private, we commonly refer tothem as SLM is right, the set of
the large language models andthe small language models, just
based on their own data, it'sthat they own and control. So
(12:47):
you're right, it's going to be aprocess. But definitely the
larger organizations, the hyperscalars are absolutely gonna
lead the charge on this. They'reobviously they're already there.
Right? They're already buyingthis. And so the question is,
how much of it goes to? How muchof it goes to InfiniBand? And
how much of it goes to Ethernetinitially, and then once like
UEC, is out? How much startsshifting over to to Ethernet? I
(13:09):
mean, the, the hyperscalers havecertainly shown that they don't
want to use proprietarytechnology, right? Everything's
white box, or some slightlymodified white box type of
technology, right? They want tobe able to reduce cost, they
want everything to bestandardized. So there's a lot
of, there's a lot of pluses forwhy they would want to use
(13:30):
Ethernet. Over InfiniBand.
Antone Gonsalves (13:33):
Okay, so let's
circle back to Cisco here. Were
one of the because Cisco, Cisco,if Cisco is going to get back on
track in terms of revenue tosatisfy their investors, it
seems like Splunk is, is acritical piece of that they're
spending $28 billion for thiscompany. How will Splunk change
(13:59):
their business? And, you know,give them their next round of
revenue?
Unknown (14:06):
Yeah, I think that's
another it's a really
interesting one. Obviously, thishas been talked about for a
couple of years, right? Whetherit's people have been waiting
for this to happen. So itobviously it presents a huge
opportunity for them from theobservability space. As I
mentioned earlier, thesedistributed cloud environments,
with applications everywhere,with workers everywhere, it
(14:27):
becomes increasingly difficult,right? The complexity increases,
the ability to have the abilityto have visibility, essentially,
right across these highlydistributed environment becomes
really challenging. So they'vebeen doing things on the network
side, like with 1000 eyes tolook at what's going on in the
across the internet, right?Being able to tie that in with
(14:49):
their devices, being able to getall that telemetry. While Splunk
has been working on how do Icollect all the log data from
everywhere? How do I collectapplication data? So there's
obviously there's a little bitof old up there, right? Because
Cisco has AP D and things likethat. But the overall strategy
here is how do we take all thetools that we have to collect
all the data, we have to providethat end to end visibility
(15:11):
across that environment. So thatwe're able to very quickly and
easily detect when there's anissue or to be able to collect
enough information and correlateit so that we can then you know,
proactively optimize thetraffic. And so what Splunk is
doing is really, you know, thethe addition of Splunk is
enabling them to have that,basically that full stack of
(15:36):
observability solution thatcollects from the was it melt,
right, the metrics, events, thelogs, the traces from the, from
the applications and from thenetwork. And I don't know that
anyone's got a better networkvisibility story than Cisco
does. Between all the you know,all the acquisitions, they've
made, all the AI focusedacquisitions, they've made all
(15:58):
the observability acquisitionsthey've made to be able to feed
all that data up, and then beable to, you know, crunch it and
provide ultimately what everyonewants, right? It's not data for
the sake of data. It's data forthe sake of getting actionable
intelligence, right? Being ableto do something with that data
that drives a better processthat drives better quality that
(16:20):
drives a better experience isultimately what the goal is
right.
Antone Gonsalves (16:24):
Bob, thanks a
lot for joining me. It's always
a pleasure to hear your insightand Cisco and everything else
networking. And for ourlisteners, that wraps up today's
podcast,