Episode Transcript
Transcripts are displayed as originally observed. Some content, including advertisements may have changed.
Use Ctrl + F to search
0:00
Cloudcast Media presents from the Massive
0:02
Studios in Raleigh, North Carolina. This
0:04
is the Cloudcast with Aaron Delb
0:06
and Brian Gracely, bringing you the
0:08
best of cloud computing from around
0:10
the world. Good
0:13
morning. Good evening, Webinar. Welcome back to
0:15
the Cloudcast. We're coming to you live from our Massive
0:17
Cloudcast Studios here in Raleigh, North Carolina. And
0:19
it is Aaron for a quick programming
0:22
note for this week. We are starting
0:24
our conversion of the podcast from audio
0:26
only to include video as well. By
0:29
the time this is out, or maybe
0:31
shortly after, the main segment
0:33
should also be published on our YouTube
0:35
channel. Now, you'll see
0:37
us phase this in over the next month
0:40
or so as we convert all of our
0:42
process and workflows to our new tools. You
0:45
should also see and hear
0:47
an upgrade in the audio
0:49
quality of the interviews going forward as well.
0:51
We're on a mission in 2024 to update
0:53
your listening experience. And as always,
0:55
we'd love to hear your feedback. So
0:58
with that, let's move on to
1:00
Cloud News of the Month for March. And
1:06
we're back. And Aaron, we are trying something
1:08
new, man. I know. I
1:10
know. After all these years of kind of
1:12
having the same tools and the same workflows,
1:14
we're branching out. Yeah.
1:16
Now, we
1:18
are actually recording this on April 1st. This is not an
1:20
April Fool's joke. For the
1:23
first time, we're going to start doing this in video.
1:25
And we talked about this a little bit on
1:27
the year-end show about do we switch the
1:29
show over to video? And I think while
1:32
we got some feedback for a number of people that said, hey, I
1:35
listen to it as audio only, you guys don't need to
1:37
do it in video. I think what we found is a
1:39
couple of things. Number one is some of
1:42
the tools that people are using to record
1:44
podcasts, combination of audio and video are vastly
1:46
superior than everything that we've been using. So
1:48
we're excited about trying to improve the quality
1:51
of the show. But number two, we
1:53
realize that we don't have to pick just one or
1:56
the other. We can really kind of do a combination
1:58
of both. So we will continue. to
2:00
publish the show through all the regular
2:02
podcast channels as audio only and
2:04
then we're going to start using some of the
2:07
video just because it does
2:09
allow us to get into some new channels that allow
2:11
us to grow the audience and that's something we've been
2:13
trying to do for a while so. I'm
2:15
sorry about this new thing man you and i have
2:17
to make sure that we clean up and look decent
2:19
for video. I know i know that was
2:21
the one of the things to i was like oh gosh we
2:24
gotta figure that out now. Well
2:27
and i think the other thing we've been hearing as we've
2:29
been doing guest shows as well is that our
2:31
audience is constantly like oh we're only doing this
2:33
as audio because i think people have become so
2:35
used to you know with zoom calls and everything
2:37
they. They feel fine doing
2:39
stuff is video as well so we're
2:42
going to start with that so today's today's first show
2:44
happens to be our cloud news of the month so
2:46
we're going to do cloud news of the month looking
2:48
back at march of 2024 do you want to jump
2:50
in and get a start again. Yeah
2:54
i'll jump right in so we kind of you
2:56
know it's funny when we started this we
2:58
figured the format would kind of develop over
3:01
time a little bit and i think we've.
3:03
Developed into this rhythm that is kind of gone
3:06
into like hey let's do the old fashioned news.
3:09
Let's do an AI speed round and then let's
3:11
hit a couple big trends yeah yeah and i
3:13
didn't need it we didn't necessarily force them into
3:15
like X number of trends every month
3:17
right like i think last month was a pretty busy
3:19
month and so we did like five trends this month.
3:23
It's a little slower than say that happens three
3:25
trends this month yeah that happens that happens so
3:27
why don't you jump in you can at least
3:29
take a few of the we call good old
3:31
fashioned cloud news. yeah absolutely so good
3:33
old fashioned cloud news and i'm not necessarily going
3:35
to go into a mall in super depth because
3:38
there's actually well gosh probably upwards
3:40
of 10 of them right now but. we'll pick out
3:42
a couple of them we'll talk about them, but he
3:44
just a list of all here real quickly red has
3:46
changes their licensing in addition to
3:48
that Linux foundation creates a fork off of
3:50
that. SPF
3:53
from the ftx fame. sentence
3:56
to 25 years in prison. crypto
4:01
continues but not ftx. Bitcoin
4:06
is up and so is the years in
4:09
prison for sbx. Exactly.
4:13
Observe raises 150 million in
4:15
a series B for more observability and
4:17
actually instead of reading them I do
4:19
want to stop and talk about this one real quick. I
4:22
was probably of any of the old
4:24
fashioned cloud news ones this one really. I'll
4:27
say surprise me i will say shocked but surprise
4:30
me a little bit. There's
4:34
still money in observability
4:36
and there's still hundreds of
4:38
millions. Yeah. What was your
4:40
thoughts on that when you read
4:43
it i was like well i mean first off
4:45
jeremy bert is the ceo is a friend of
4:47
you were always excited when people forgot about that
4:49
yeah absolutely excited when folks are. I was excited
4:51
when folks get new rounds of funding yeah i
4:54
was surprised for two reasons one it
4:56
felt like funding just as a whole was sort
4:58
of down so you know good
5:00
good to see you know we're always happy
5:02
to see funding or at least money flowing out
5:04
of the season to create pockets but or at
5:06
least into their bank accounts but
5:09
yeah i mean it's one of
5:11
those things where i think monitoring observability
5:14
and maybe there'll be a future name
5:16
for observability. It's it's
5:18
something that's never going away and i
5:20
think it's it's a perpetual it's
5:23
a perpetual problem it feels like
5:25
it's never solved completely and you
5:27
know i think maybe there will there will always be
5:30
funding for people who are saying look we we think
5:32
we've got a better mousetrap. And
5:34
you know data dog has had an incredible run
5:36
for a while i think they're continuing to do
5:39
very very well but
5:41
yeah maybe there there is some i mean i
5:43
mean observes thing is really sort of built
5:46
on a very different. Data
5:49
store i mean so they build on top
5:51
of cloud top of snowflake so
5:53
you know there's there's some of that there's some there's some
5:55
technology there they're trying to do sort of an all in
5:57
one which is different and i think they're starting to build
5:59
in. some AI stuff. So I imagine
6:02
some of the VCs are looking at and they're
6:04
going, okay, maybe there is a different mousetrap that
6:06
can be built. Fair. Fair.
6:08
Well, and you mentioned stuff like Snowflake relates to
6:10
the next story. You want to go, Brian? Yeah,
6:13
they, they're, so
6:15
I don't know if we covered this last month, their CEO
6:17
left, he announced his
6:20
retirement, which, you know, we were joking on one
6:22
of the Slack channels that there should be, there
6:24
should be like a whole conference for people who
6:26
just decide like, Hey, I've had enough as being
6:28
a high level executive, I'm just taking the money
6:30
and running and I'm going to go hopefully just
6:32
lay out a beach, which is kind of our
6:35
ambition at some point in time. But yeah, Frank
6:37
Flutman left not too long ago, month,
6:39
month and a half ago, and the market has just
6:41
killed it. No, like, I mean, down
6:44
like 35, 40% or something like that,
6:46
even though they've announced, you know, their earnings
6:49
or numbers are pretty good and so forth.
6:51
But yeah, they basically, they got
6:53
sort of killed at the end of the pandemic,
6:55
because everybody moved from storing their data five or
6:57
six times to only going to store it two
7:00
or three times. So you know, their revenues were
7:02
down. And then yeah, the market
7:04
is essentially like, we're not sure who this
7:06
new CEO is, because the last guy was
7:08
was so good. So that one's, that was
7:10
gonna be interesting. You know, we don't, we
7:12
don't like to sort of give stock picking
7:15
advice or anything, but it doesn't feel like
7:17
fundamentally, Snowflake is a really different company. So
7:20
you know, who knows, maybe this is just a
7:22
dip in an opportunity for them to come back. Yep,
7:24
absolutely. Next couple that
7:27
are on here are all sort of Kubernetes
7:30
or CNCF related. So the
7:33
CNCF is not CNCF, Kubernetes is celebrating
7:35
10 years. So 10 years. Do you
7:37
remember the very first time we had
7:39
like Joseph Jax, and there
7:41
was another guy and they came on and they
7:44
were like, hey, we're doing an event called cube
7:46
con. And we're like, you're doing an entire event
7:48
around the scheduler? Are you like, are you crazy?
7:50
And now there's 12,000 people in Paris last week
7:54
were at were at cube con. So, you know,
7:56
they had some vision. Congratulations to them. But yeah,
7:58
10 years. of Kubernetes.
8:01
KubeCon Paris had 12,000 or more
8:03
than 12,000 people, so the biggest show ever. So
8:06
that space is thriving,
8:08
although we still wonder
8:10
what exactly is driving it. But it does
8:12
feel like it's sort of in the sort
8:15
of stabilization stage and mainstream stage.
8:18
So probably a lot more focused
8:20
on customers, maybe less on some of the new
8:23
incubating technologies. Well, and I'll add
8:25
this to maybe a quick
8:28
opinion, maybe borderline hot take on all
8:30
of this is, and it
8:32
relates to another article that's in Cloud News, which is why
8:35
Kubernetes needs an LTS, which
8:37
I thought was a super fascinating article. But
8:41
why is KubeCon so popular? Like
8:43
it really does feel like this
8:47
is putting a bad label on it, but
8:49
it's become the premier infrastructure plumbing
8:51
conference that VMware used to be
8:54
in. And all
8:56
those people that went to shows needed somewhere to
8:58
go, and so now they all go to KubeCon.
9:01
The last KubeCon, I didn't go to KubeCon EU, but the last
9:03
one I went to, it's like you go
9:05
there and you see the same folks in the
9:07
same booths, they're just working for different companies. A
9:10
lot of folks have just migrated from
9:13
one area to another. Yeah, it'll be
9:15
interesting to watch. I did a Sunday
9:17
perspective this past week on Broadcom and VMware,
9:19
and it seemed to get a lot of
9:21
traction from folks. It will be interesting to
9:23
see if VMware continues
9:25
to have their event of Broadcom's like, nope,
9:27
that's just another expense that we can
9:29
get rid of. We sort of have captive customers, we
9:32
can send a press release out with the new
9:34
thing. So it'll be
9:36
interesting. I will say for anybody who
9:39
is a Kubernetes customers, we don't tend to do
9:41
a whole lot of like plugging our day job.
9:44
OpenShift, if you are interested in an LTS has
9:46
had Kubernetes LTS for four or five
9:48
years, even though the community can't figure out
9:50
how to do it. Red Hat has been
9:52
doing it for four or five years and
9:54
they'll do like three or three plus years
9:56
LTS. So if you need that, just it's
9:58
out there. Don't you don't. may not have to
10:00
wait for the community to get it. Yeah, and I'll
10:02
just add this too. Again, it's a really good
10:04
article to go read. It's really fascinating and all
10:06
the reasons why and the release schedules and it
10:08
goes into a good bit of detail on all
10:10
of it. Complicated. I mean, it's funny. Yeah, and
10:13
here's the thing though. I've talked to a bunch of
10:15
customers recently about it again with my day job and
10:18
almost everyone universally they're like, no, we just,
10:21
anytime we have to migrate, we just build
10:23
entire new. We never migrate or
10:25
upgrade anything. They just build new. Yeah.
10:27
Move over and I'm like, whoa. Yeah, from
10:30
being around Kubernetes for a while, there is
10:32
two sets of kinds of
10:34
people. There's the people that just blow up the cluster and
10:36
move it over to a new one. And for them, they
10:38
don't need LTS. But there are
10:40
a lot of services that
10:43
run your day to day life that are on clusters
10:45
that are still, I mean,
10:47
they are many, many, many years old. That
10:51
people upgrade the application but they don't necessarily
10:53
upgrade the Kubernetes or they leave it for
10:55
a long time. So yeah, it
10:58
is the craziest thing. We've talked about it forever. VMware
11:01
used to put out a release once, a main release
11:03
every once every year and then a minor every six
11:05
months. So it was like, maybe
11:07
every 18 months you would talk your touch,
11:09
your virtualization infrastructure and Kubernetes came
11:11
along and was like, nope, every three months you
11:13
get a new release. Boom. Boom.
11:16
Boom. Yeah. Well, I
11:18
think as far as good old fashioned
11:20
news, I think we covered a bunch of stuff. I mean, basically it's
11:22
like, there's lots of projects that are
11:25
still going through licensing changes. You'll
11:27
see a few of them in there. HashiCorp
11:30
went through it, Redis is going through it. I
11:33
saw MariaDB is potentially getting acquired by a private
11:36
equity company. So there's a lot going on in
11:38
that space. And then the CNCF sort of has
11:40
their, one of their pillar things for the year.
11:42
So it's good to kind of see all the
11:44
videos from CNCF were already posted.
11:46
So congratulations to that team for getting all of those
11:48
out so quickly. Yeah, absolutely.
11:51
All right, man, this is sort of your
11:53
baby. The innovation speed round. What's
11:56
new, you know, shoot. By the time people watch
11:58
this, there'll probably be 25 other. new things, but
12:00
what's new in the AI space? Well, it's so funny
12:02
too, you mentioned that because one of the articles and
12:04
actually I put it in the Microsoft section, I
12:07
literally popped it in before we hit record
12:10
because this stuff is pretty amazing
12:12
how much there is, and this is
12:14
just highlights, right? And this is where too, I would say,
12:17
tell everyone if you ever get really
12:19
bored, go to the weekly Cloud News
12:21
of the Week doc, which is again,
12:23
I don't know, it's up to 70,
12:26
80 pages now where we just keep adding links
12:28
that we find interesting week after week, but most
12:31
of them are AI, that like we've already admitted
12:33
on the Cloud News of the Month, and most of
12:35
them are AI related, but here's kind of the quick
12:37
rundown and quick speed around. There's
12:39
a really good article, and it's a really long article
12:41
by the way, about
12:43
the engineers that created the transformer paper,
12:46
the one that kind of changed all
12:48
of this, and really, really
12:50
good read, I would definitely recommend everyone go
12:52
check that one out. Amazon, Amazon seems to
12:54
have picked, and we've talked about this before,
12:56
picked who they wanna be investing in,
12:59
they've invested in, Anthropic is
13:01
now up to almost $3 billion.
13:05
Saudi Arabia plans a $40 billion push to
13:11
sovereign investment into AI. There's
13:15
also, this was interesting,
13:17
good friends of the show, Acorn Labs,
13:21
we had them on not too long ago. And
13:25
by the way, there is one or two folks out
13:28
there too, that still have listened to every show, or
13:30
plan to listen to every show, it's getting harder and
13:32
harder with over 800 shows now. Hey,
13:34
you don't need to listen to that one, because they pivoted.
13:37
They basically came on and said, we're
13:39
gonna reinvent PaaS, and
13:42
then like two months later, they're like, nope. We're
13:44
not doing that. Yeah, so they released something
13:46
called GPT Script. So I
13:48
imagine applying the existing technology, but more into
13:50
an AI use case than a PaaS use
13:52
case. And then
13:54
as things always get smaller and easier,
13:57
GPT too can now be squeezed into
13:59
an ex... Excel spreadsheet, which I
14:01
thought was pretty crazy. And
14:04
then CNCF publishes, and this was hot off the
14:07
press, CNCF publishes a cloud
14:09
native AI white paper. Yeah, yeah, yeah,
14:11
yeah. The buzz around KubeCon was everything
14:13
was sort of AI, platform engineering or
14:15
something else. I think we're gonna talk
14:17
about that in just a bit. I
14:19
was sort of surprised that a few of the numbers, and
14:22
this is the crazy thing, is like we're talking about numbers
14:24
in the billions in terms of investment,
14:26
but I was sort of surprised the
14:29
Amazon investment in anthropic
14:31
wasn't necessarily net new. I mean, they had
14:34
pledged, like previously, they were gonna do 4
14:36
billion. This was just sort of
14:38
like truing up the previous four, to
14:40
get them to 4 billion. They
14:42
didn't buy them. They're just sort of making yet another
14:45
sort of tranche of an investment. I'm
14:48
a little bit surprised. I don't
14:50
know if I'm surprised or if it's something that we should keep an eye
14:52
on, but like it seemed fairly low
14:54
to me, given the stakes
14:56
that are between, even
14:59
between Microsoft and Google and AWS, it
15:01
felt fairly low. And it is
15:03
one of those things where you sort of wonder, does
15:06
Amazon know something more than we do? Because they
15:08
did come out with, they did come out maybe
15:10
like a week or so ago, and
15:13
Adam Cielicki sort of said, hey, we've got
15:15
X thousands of people using bedrock, or
15:17
are they zero behind
15:20
the level of investment they've
15:23
got to make? I think it's an interesting thing
15:26
to look at. The other thing that was interesting,
15:28
I thought the Saudi Arabia number, Saudi
15:30
Arabia has this gigantic wealth fund, sovereign wealth
15:32
fund. I thought $40 billion
15:34
seemed a little bit low, because I think they
15:36
are trying to, from
15:39
what I've heard, they're trying to become like
15:41
a third place where chips are
15:43
being built. Obviously, they're trying
15:46
to reestablish something. They know that
15:49
what's going on in APAC has concerns with
15:51
what's going on in China. The US has
15:53
to figure out how to build up manufacturing
15:55
again. And they have
15:57
so much money, and $40 billion actually seems
15:59
to be seems sort of low, especially when
16:02
you consider that the
16:04
folks that run, what's the Japanese
16:06
SoftBank is talking about? Like $100
16:08
billion. And
16:11
we talked earlier, like Sam Altman's
16:13
thrown around some $7 trillion number,
16:15
which, or $7 trillion number, which
16:17
seems obscene. But $40
16:19
billion seems sort of low for, if they're
16:22
really trying to transform from being
16:25
completely oil-based or petroleum-based
16:27
to something completely different. But
16:29
maybe just me. No, makes
16:32
sense. Absolutely. All right. Let's
16:34
get into, we got three trends this month. Which one
16:36
do you want to jump on first? Let's do CubeGone.
16:39
I mean, we already talked about it a little bit. Let's kind
16:41
of do, and neither of us were able to make it, first
16:43
of all. But I see
16:45
some kind of outside reflections. I'm sure
16:47
you have some impressions or everything. I
16:51
will simply say, again, we talked about more
16:53
than 12,000 employees, or attendees,
16:55
sorry. I feel a
16:57
little bit like AI-washing. What's
17:01
your thoughts on that? Yeah. I mean,
17:03
people came back to it and they said, hey, there was
17:06
a lot of AI talk. I
17:08
think the reality is, and people
17:11
can get cynical about this or not, like this
17:14
year, every
17:16
keynote is going to be, at least the day
17:18
one keynote, is going to be dominated by AI.
17:21
Because if you don't come out and
17:24
make a statement about where you are, right?
17:26
So this is the yin and yang of
17:28
timing of big events and so forth. You
17:31
could say, well, last year, AWS did
17:33
a lot at reInvent. Well, that was
17:36
in November, right? If you
17:38
had missed that cycle, remember, it was
17:40
sort of like the March-Aprilish timeframe was
17:42
when chat GPT sort of blew up
17:44
for the world and then something going
17:47
on. So they
17:49
missed the AI cycle last time
17:51
with CNCF. And there wasn't necessarily
17:53
a lot to talk about. So
17:56
You really had to make some sort of statement. You Had
17:58
to give people some sense of the AI cycle. Because
18:00
to a certain extent you know that
18:02
the Cncs and the Cnc of community
18:04
is competing for people's attention, and if
18:06
it they're perceived to be like we're
18:08
just the infrastructure people, you just the
18:10
cooper. Now these people, or whatever you
18:12
you do run the risk at the
18:15
border to be like, oh, they've. Immediately
18:17
overnight because relevance and he i think the
18:19
reality is like tons of the say I
18:21
stuff does run topic of burnett ease their
18:24
see other schedulers in some advancements have been
18:26
going on as space is probably good to
18:28
have folks like a video showing up and
18:30
infusing some new ideas and capital. Other stuff
18:33
so I don't worry about it. Been washing
18:35
so much as as being like this year
18:37
you're going to hear about it. You can
18:39
put whatever filter you want on just whether
18:41
you think it's Bs or not be as
18:44
but like you're not going to get around
18:46
that right? So. You have to sort
18:48
expect that. Come in and it'll be interesting
18:50
to see where you know how the Cncs.
18:53
Adopts, To this because I don't. I
18:55
don't know of any like net new projects that
18:57
are probably drop into the Cnc or for a
18:59
i will be a single watch. The
19:01
green agree. About in
19:03
I didn't have much to do with it. I
19:05
had a huge work work deadline for for Art
19:08
with you know, our user conference coming up in
19:10
May right? and it was like it was like
19:12
all the presentations were do and I was track
19:14
lead and I was like oh man I just
19:17
don't have time to keep up with it
19:19
this year. Yeah there's a there's a really good
19:21
write up. we've got a link in the show
19:23
notes. ah I'm you have somebody basically went through
19:25
Daniel Bryan's who's been around the Cncs for very
19:28
long time or in a lot of different areas.
19:30
infrastructure developer stuff it's of as he wrote a
19:32
bay. A really really good Ah,
19:34
I'm. Kind. Of a summary article
19:36
is probably four or five pages long. Been a
19:39
kind of walk through yes the I stuff that
19:41
sort of everything else and sir where the highlights
19:43
were and were areas are spending and not expand.
19:45
It's definitely worth a read. A few are you
19:47
trying to get get a good sense What happened
19:49
at the at you com Paris. Nice.
19:52
What's. next you want your soft smile minutes
19:54
we sat spent a little that before
19:56
but i think it the trend continues
19:58
microsoft is real really branching out from
20:00
just OpenAI as a partner. Three
20:03
different proof points on that this last month.
20:07
Well they did an
20:09
interesting thing. Yeah, they
20:11
acquired InflectionAI, which
20:15
really brings in a team that's gonna run
20:17
almost consumer level AI, right? Right. And
20:21
then we can break these down a little bit more. But then I
20:23
thought this was super interesting. They hired a new
20:25
head of AI. So basically you almost think of
20:27
it as when
20:31
the OpenAI and Sam Altman stuff
20:33
kind of all went to crap and Sachi
20:36
was potentially gonna bring over Sam as head
20:38
of AI. Well that, they
20:40
decided not to keep that seat
20:42
warm for him. They actually went
20:44
and hired somebody. Well
20:47
the nuance of this one is, so, and we
20:50
talked about this last month as well, right? So last month
20:52
they made the investment in Mistral. And so that
20:54
was sort of the first time they were like,
20:56
hey, we need to hedge our bets. So
21:01
InflectionAI and hiring the new head,
21:03
those things are connected. So InflectionAI
21:05
was this sort of
21:07
consumer chatbot sort of company. And
21:10
we'll put some links in the show notes. There's some
21:13
really interesting behind the scenes of like who was their
21:15
investor. And they had bought like 24,000 GPUs for
21:19
this consumer service without really ever
21:21
scaling. Like it was a weird investment, I
21:23
think, and did he as an investor as
21:25
well. But they basically
21:27
hired the guy who was running it,
21:29
who was an, had been a big deal over
21:31
at Google Brain. And so
21:34
he's now in charge of consumer
21:36
AI for Microsoft. Now what that
21:38
means, whether it's like Xbox or
21:41
tied into GitHub or whatever, it
21:43
does sort of show them realizing
21:46
like, it's not
21:48
just technology. Like you almost have to like
21:50
buy the teams of people. And there's such
21:52
a small number of people. They're all either
21:54
ex-Facebook or Google people. And
21:56
so he's now, yeah, he's now the head of AI
21:58
for consumer. Now it'll be interesting. you see at some point
22:00
like does that sort of merge
22:02
together with the stuff happening in Azure and
22:04
is Azure considered consumer or enterprise or whatever
22:06
but they you know they
22:08
continue to put their money where
22:11
their mouth is now you know whether
22:13
or not they've over invested as we
22:15
talked about you know Amazon's level versus their level
22:17
will be interesting but yeah yeah it's
22:20
this is this
22:22
is for you know this is like a big
22:24
game of what's the what's the what's the board
22:26
game that's all about war where you just trying
22:28
to take people's land risk is it risk of
22:30
that one yet risk going on right now yeah
22:32
well in and along the lines that you mentioned
22:35
about building the last article
22:37
in this tune this trend is Microsoft and this just
22:39
happened or I just put this in right before we
22:41
hit record Microsoft might be
22:43
building a 100 billion dollar supercomputer
22:47
for AI I think
22:49
it was it Stargate was that the nickname of it
22:52
I apologize I didn't click into the link let
22:54
me do it real quick yeah Stargate um
22:57
so I mean hey who
23:00
knows you know now we were measuring supercomputers
23:02
in hundreds of billions which
23:05
it's a little crazy to me well that
23:07
and and that one that that concept is
23:09
is again it seems like
23:11
it's another hedge because on one hand you've
23:14
got Nvidia which we'll talk about next you
23:17
know has been talking about you know the
23:19
gigantic Nvidia systems and if he keeps talking
23:21
about we're not a chip company we're a
23:23
systems company I mean it does
23:25
that does almost feel like Microsoft is going
23:28
we we need to be smarter about what
23:30
these larger systems look like because it's very
23:32
possible that either we have to run them
23:34
highly efficient or our customers are
23:37
going to ask us to build that for
23:39
them so you know hundred billion dollars seems
23:41
like a gigantic investment unless they're
23:44
you know specifically going after like something like search
23:47
but you know the idea of building
23:50
large large-scale systems to hedge against Nvidia
23:52
or to potentially be a supplier for
23:54
your customers doesn't seem you know like
23:56
way out in left field yeah
23:58
then they say this could launch as soon as
24:00
2028. And
24:03
what I picture for those of the science fiction fans
24:05
out there who watch the West World show on what
24:07
used to be HBO, I just pictured
24:10
all it was in one of the seasons, I
24:12
think it was maybe third season, fourth season. They
24:14
had this entire building and it just had
24:17
this big red glowing ball inside it. Like
24:21
that's why, and it was their artificial
24:23
intelligence that ran the world basically. I
24:26
feel like that's what they're trying to build here.
24:28
You would think you literally need a
24:31
nuclear power plant right next to this thing to generate
24:33
enough power for it, right? Which is an article we
24:35
had in the past. They're actually
24:37
exploring exactly that. So maybe it is
24:39
all the threads coming together. Lots
24:41
of big thinking happening at Microsoft these days. All right,
24:44
last trend. NVIDIA had
24:47
their GTC event, which it
24:50
was interesting because I have
24:52
never really followed this thing because for the
24:54
longest time, NVIDIA to me was always about
24:56
gaming. Up until the real this
24:59
last year or so, it was like, yeah, they do GPUs, but
25:01
the use case is always sort of gaming and stuff. And so
25:03
I didn't really pay attention to it. Did
25:05
you see what GTC looked like?
25:07
It looked like a Taylor Swift concert.
25:09
Oh, yeah. And again,
25:12
I had known about them
25:14
in the past, but again, it was gaming or
25:16
it was the VDI and EUC. Those
25:20
are the two experiences I had with
25:22
NVIDIA previous to this. But yeah, suddenly
25:24
this is like, again,
25:26
we're talking about trade shows and conference
25:28
seasons and all that. I mean, this
25:30
is suddenly one of the trade shows
25:32
to be at now. And
25:35
it's a pretty amazing event and
25:37
it almost reminds me of the old splashy
25:39
days, right? Of like when you
25:41
had lots and lots of production go into
25:44
some of these trade shows and I think
25:46
as budgets tightened up, some of it left,
25:48
but NVIDIA, they did it all. Yeah,
25:50
I tried to keep up with that. I've watched
25:52
a few of the things. I've watched some of
25:55
the keynotes and so forth. And it seems to
25:57
me there's two sort of schools. thought.
26:01
There's the school of thought. I mean, what I mean by that
26:03
is like people writing up summaries of it and so forth. There's
26:06
the ones that are like breathless over, wow,
26:08
this is really big, and this is amazing
26:10
and so forth. And they tend to be
26:12
focused on, I think what NVIDIA did is
26:14
they said, this is our moment to
26:17
sort of kind of reinvent themselves as being
26:19
like our use cases AI as opposed to
26:21
in the past where I think
26:24
they were always kind of trying to find a use
26:26
case, you know, or you know, find a mega use
26:28
case. And so, you
26:30
know, apparently what they did was, you could
26:33
look at the you can look at the announcements and
26:35
so forth, and they're interesting. But what
26:37
they tried to do was sort of be
26:39
like, the world of AI now resides with
26:42
NVIDIA, right? They wanted to put themselves at the center of
26:44
the earth. And so there was a lot of people that
26:46
were sort of, you know, their write
26:48
ups were like breathless. They're like, Oh, look, they're doing
26:50
cars, and they're doing robots, and they're doing healthcare, and
26:53
they're doing all these things. And,
26:55
and to a certain extent, like those were there,
26:58
but not as in depth as people kind of
27:00
made them out to be. And then
27:03
there was some really interesting write ups. Ben Thompson
27:05
had a really interesting one. I put a link
27:07
in the show notes, Ben Thompson's the guy who
27:09
runs a strategy, strategy, does
27:11
a lot of coverage of like chips and hardware and
27:13
stuff, as well as some other internet stuff. And
27:16
he had a really interesting take that, up
27:19
until now, so everyone thinks of NVIDIA as
27:21
being like this GPU company, which obviously they
27:23
are. And right now, they're sort of cornering
27:25
the market because they're short supply. But
27:28
he said, you know, their their moat is
27:30
really been this thing called CUDA, which is
27:32
their their software stack, basically. And
27:34
he said, up until now, there's never been
27:36
anybody who's ever looked at CUDA as a
27:38
as an external competitor and said, we should
27:40
try and build like an alternative to that,
27:42
or like an open version of that, because
27:44
it just, you know, it was kind of
27:46
niched to gaming or niche to certain, you
27:48
know, certain kinds of use cases, but not not
27:51
massively horizontal. And his his argument
27:53
was basically, NVIDIA
27:56
is is a big dog right now, and they're
27:58
kind of setting the world. on fire, but
28:02
the level of competition that they're going to have
28:04
around CUDA for people that are like, we cannot
28:06
let one company dictate what this looks like, is
28:09
going to be really interesting. And that was that
28:11
was kind of his take is like they've, they've
28:14
unlocked the market to now go after their biggest
28:16
moat. And it'll be interesting to see if whether
28:18
it's like a bunch of open source, you know,
28:20
projects or, you know, one or two companies or
28:22
even one company trying to build an alternative. Because
28:25
if you look at all their announcements, they're
28:27
all tied to we only run on with
28:29
this hardware. And if we've seen anything
28:31
about the internet over the last 10 or 15 years,
28:34
like that, that is like
28:36
the magnet that is the, you
28:38
know, the mosquito, the moth to the flame,
28:40
if you will, for competition. Yeah, yeah. Well,
28:42
and yeah, and part of it
28:44
too, was I mean, their their latest, the latest
28:46
GPU, the Blackwell GPU, $30,000, right?
28:49
And so like,
28:51
again, if you're an Nvidia investor or
28:54
an Nvidia fan, hey, this is amazing.
28:56
But if you're outside looking in, like
28:59
there's so many people now,
29:02
either trying to build competitive
29:06
GPUs or something
29:08
that is a chip that's not
29:10
a GPU, but is, you know,
29:12
specialized for AI or all these
29:14
other things, right, there is so
29:16
many competitive forces coming from so
29:18
many different directions. Yeah, I mean,
29:21
it is, it is one of the fascinating things
29:23
about our industry. And what makes it really interesting
29:25
is, anytime you put a large bag of money
29:27
out somewhere, I mean, just everybody, I
29:29
mean, if you know what the old thing, like
29:32
one person's profit is the next
29:34
person's business model. Yeah, I mean, that's everybody
29:36
is, everybody's looking at Jensen
29:39
Wong, they're jealous of his, you know, coat, you
29:41
know, rack full of
29:43
black leather jackets, and they are, they're
29:46
coming out, you know, and everything's everything's gonna be
29:48
timing, because the stuff they built has been built
29:50
up for a long, long time. Yeah,
29:52
they played a long game and they're reaping the
29:55
benefits of it. But now
29:57
it'll be interesting to see how quick
30:00
the bullseye moved to them and
30:03
how long they can keep it up. Well,
30:05
and I think the other thing is like
30:07
timing is always everybody's worst enemy in technology, right?
30:09
Like the longer you, I don't
30:11
know, you can't manipulate the market or shape
30:13
the market to your will or invent new
30:16
things. And I mean, they've got
30:18
a market who is sort of captured. Everybody
30:20
wants their GPUs, but like if you can't
30:22
get them for a couple of years, like
30:24
it does create a lot of
30:26
people going like, well, what alternatives available to
30:28
me? Because nothing isn't an
30:31
acceptable answer. And so like
30:33
you said, like I think we're gonna find maybe
30:35
not on purpose, but just out of necessity,
30:39
you know, what is, what do they call it? It's
30:43
how it's the innovator's dilemma, right? It's the,
30:45
you know, I'm gonna build from the bottom.
30:48
It'll be limited feature set, but it's cheaper.
30:50
It's simpler, right? It's the stuff that Clayton
30:52
Christian Gen has written about for years and
30:54
years. So it's gonna be fascinating stuff. So
30:56
yeah, not a super, super busy March. Like
30:59
you said, it felt a little more balanced of cloud
31:01
and AI and so forth. So that's a good
31:04
thing. But yeah, any last thoughts
31:06
for March now that we are into April
31:08
and it's starting to warm up everywhere? Well,
31:11
we gotta talk about basketball. I mean,
31:13
for all of our non-American fans or
31:15
non-sports fans, we have to apologize real
31:18
quick, but we do have to talk
31:20
about March Madness. Sorry about your team,
31:22
man. Yeah, well, yeah, exactly. But I
31:25
mean, it's a pretty crazy
31:27
tournament. Is it crazy?
31:29
Well, a lot of great basketball going on
31:31
this year. Yeah, and the local team, NC
31:33
State here in Raleigh, major the final four,
31:35
they are the Cinderella for this
31:38
year. So that's kind of cool, you
31:40
know, just for the local area. So yeah,
31:42
I think we are, we're getting into Q2. Q2
31:46
is gonna be, you know, we'll see
31:48
earnings here in a few weeks for
31:50
what Q1 look like. Q1
31:53
always seems to set what people's Q2,
31:55
three and four budgets look like, depending on how well
31:57
or not as well as it went. So. I
32:00
think while the first quarter is always a
32:02
little bit slow, I think we're going to see a little
32:05
bit of build up for April and May. Summer
32:07
will be summer as it always is and then
32:09
it'll probably be sort of fast and furious for
32:11
Q3 and Q4. But
32:14
this video I think is not so hard, man. I think we
32:16
can do this. I think we can do this. You think so?
32:19
We'll see. I think we'll give it a try.
32:21
We haven't published it yet, so this may be one we bury. You never know.
32:23
But the good news is lots of smart people
32:25
are making these tools easier for us. So yeah,
32:27
you'll start to see these on the
32:30
TikToks and the Instagrams and the YouTubes and all
32:32
the other places. So especially YouTube, we've always had
32:34
a really good following on YouTube. We just kind
32:36
of stop publishing stuff. So maybe it'll follow us
32:38
along. If you like video, hopefully you'll
32:40
like this stuff and we'll snip it up into some
32:42
things and go from there. So welcome to what? Cloudcast
32:46
2.0 in terms of our media profile. Still
32:49
doing audio, but now with more pixels. Yep,
32:54
exactly. All right, man. We'll
32:57
wrap it up, folks. Thanks for listening. Thanks for
32:59
watching. Thanks for telling a friend about the show. And
33:01
hopefully if you are still in the bracket, hopefully your
33:03
teams do well in the final four and we will
33:06
talk to you next month. Thank
33:09
you for listening to the Cloudcast. Please
33:11
visit thecloudcast.net to find
33:13
more shows, show notes, videos
33:15
and everything social media.
Podchaser is the ultimate destination for podcast data, search, and discovery. Learn More