Episode Transcript
Transcripts are displayed as originally observed. Some content, including advertisements may have changed.
Use Ctrl + F to search
0:00
Costco sells at a premium to
0:03
the profits that they're making is because
0:06
people know that. Even
0:09
in a recession, people are going to go buy
0:12
in Costco because they want to save money. Whereas
0:15
you can't say that to a business like Etsy.
0:20
When, when people are losing jobs,
0:23
they're not going to go buy handcrafted
0:26
items that are being shipped to their
0:28
homes. Costco
0:31
would probably see 5%
0:33
loss in revenues. If we get
0:35
a scenario of 2008, again,
0:38
it's, you'll probably see 50% loss in revenues,
0:41
right. It's not to say that it's a
0:43
bad business, but Costco
0:46
is a better business. I think
0:48
the same applies to the
0:51
cloud industry. Dribble
1:03
investments, thank you for joining us. I think,
1:05
you know, we've been wanting to do this episode for a little while.
1:08
I have somebody that. S you know, so-and-so expert
1:10
on AWS, and I think you're the best person.
1:12
So we have to begin with, if you could just quickly
1:14
describe what AWS is
1:17
and what makes it such a compelling business,
1:20
right. I get asked this questions all the
1:22
time by people who are not in the tech sector I
1:24
think Couple of weeks ago. One of my neighbor's
1:26
mother asked me what is the difference between Amazon
1:28
and Amazon ups? So it says, I thought everything
1:30
is the same. Right. But the simple
1:33
explanation is I think it goes back to
1:35
the history. Amazon started as.
1:37
Right dialer and
1:40
asked they started selling books
1:43
and they expanded their inventory
1:45
to CDs, music, other
1:48
items and the number of customers increased
1:50
the amount of data they housed within
1:52
their business. Exponential increase, right?
1:55
The customers increasing the products increased. Lot
1:57
of traffic means a lot of compute
2:00
and they. They
2:03
had to invent a lot
2:05
of technologies to scale to
2:07
Amazon, as a people with an Amazon call
2:10
it Amazon scale. And once
2:12
they achieve
2:14
the scale, someone came
2:17
to Bezos and be like, Hey, why
2:19
we have achieved this scale? We can either keep
2:21
it as a secret sauce, or we can actually
2:24
make this technology available for others, for other
2:26
people to utilize it. And
2:28
then we can. Make a business out of
2:30
it and base of like, that's a beautiful idea. And
2:33
I think in 2009 they found AWS
2:36
2008 where basically
2:39
they launched easy two, which is
2:41
the compute service that
2:43
AWS provides. And basically
2:46
at that point was basically a commodity type of
2:48
business. Right. Any, any.
2:52
Moment Bob shop to any
2:55
major retailer can post
2:57
their website on AWS silvers.
3:01
And they don't have to worry about
3:04
building their own data centers that that's how it'll
3:06
get started. And right now we just not
3:08
that anymore. Right now. It'll be much
3:11
more, it's not just a commodity company. You meet
3:13
in your computer servers kind of a deal. It is
3:16
a, it is a soft, it is an
3:18
enterprise that
3:21
has a distribution muscles too,
3:23
all over internet. That's how I would describe
3:25
it. I just wanted to ask a question
3:27
about that. So I built a, I built a few
3:29
things using AWS and it was incredible
3:32
to me how quickly it was, how easy
3:34
it was for me to come up with this idea,
3:37
write it in node, JavaScript,
3:40
and then launch it. I can't even
3:42
fathom of what the world looked like before that.
3:44
Do you have any insights? Right.
3:46
I wasn't really like I think back in 2010
3:49
or so when I started building
3:51
my own obligation, that's exactly my
3:53
experience. Right? So you
3:55
write code doctorals that hosts
3:58
of up silver, and then you write business
4:00
logic on top of it, and then you
4:03
actually have to And
4:05
every single competent of the whole stack,
4:07
right? You have to spin up, you have to go
4:09
find a separate entity
4:11
that hosts your DNS. You have to go
4:14
find a separate entity that holds your
4:16
URL assets, the CDN
4:18
cache, everyone, everything you
4:21
have to know of being a full stack developer
4:23
was such a pain.
4:25
Before it will be a C like, even if, even right
4:27
now, if you're a full-stack developer and you're not using
4:29
any of these integrated solution, it's not
4:31
easy at all. There's so many things that you
4:34
have to have in your mind. I think
4:36
one of the core monsters of Amazon is
4:38
making E-Trade. It's straight
4:40
away, the smallest of inconveniences
4:43
that the customer house, that's kind of how Amazon's
4:45
core mantra. Right. And
4:47
I think it really started chipping
4:49
in and automating the
4:51
smallest things that the developers
4:53
do. And then with
4:56
12 years of engineering, it has come to a point
4:58
where if you want to. Create
5:01
a software solution, not just to upside like a
5:03
software solution, you can pretty
5:05
much do it with a
5:08
two pizza team. Right? So I think the two pizza
5:10
rule that Bezos had kind of aligns
5:13
with AWS. That's why the
5:15
two pizza team actually works. If you think of it, not
5:17
many people talk about this. Anyone can say that,
5:20
oh, we are, we are go with lean
5:22
mechanisms and we have to be can
5:25
be more than. Nine
5:28
or six people. Right. But it
5:30
is not that easy if you
5:33
don't have
5:35
all of your base infrastructure obstructed
5:38
away by, you
5:40
know, your own DevOps being AWS,
5:42
it's not easy. Like I've worked in companies
5:45
where teams like about
5:48
25 to 30 people. Work
5:51
on a single segment of
5:53
the whole software stack. And
5:56
then this is one I'm
5:58
talking about one product, which is being worked
6:00
on by about
6:03
10 sub teams.
6:05
And each team has 20 people and each
6:08
team takes care of being digital component. And
6:12
you can't, those
6:14
types of those types of organizations are
6:17
slowly moving out too. AWS and
6:19
Azure, and it enables
6:21
them to just go with the, see the unit
6:24
of just 12 people
6:26
or 10 people, and they can build the whole
6:28
solution. And because you have the
6:31
DevOps by you. Okay.
6:34
No, you can ask her anything and he is there
6:36
to give you the solutions
6:38
as generic as possible. So it
6:41
has, I've started away a lot of complexities. That's
6:43
how I do it. This as one of my friends told
6:45
me, AWS is my personal
6:48
DevOps. Right. That's
6:50
how most people, yeah. So
6:52
I think, yeah, that's, that's when I first learned about that,
6:54
I think that's what sort of struck me. You have the explicit
6:56
cost of having to just hire a lot of people
6:59
to work on the, you know, each component
7:01
or you can go to AWS. That's like the explicit cost.
7:03
Obviously the implicit cost is all the things
7:06
that were not being built before, just because people could not
7:08
afford to build them. And so I think that is maybe the
7:10
greatest thing that AWS has sort
7:12
of given to the world. You
7:14
know, if you, if you have to discuss, and I think, I think
7:16
I'm glad you brought it up that it was sort of brought around in 2008,
7:19
as you look at the
7:21
AWS competitors. And I'm sure
7:24
there were a lot of differentiate differentiating factors that
7:26
AWS had back then, but what are the core
7:28
differentiating factors that AWS has now when
7:30
compared to Google or Microsoft,
7:32
et cetera? The
7:35
core differentiating factor is their products.
7:38
If you ask me just because
7:40
they are the first entrant they
7:43
have almost
7:48
anecdotally speaking five,
7:52
four to five different
7:55
products that achieve the same thing,
7:57
but different offs. Right.
8:00
Let's say, let's talk about compute here with
8:02
AWS. If you want to
8:04
run an application, being
8:07
computer go, I'm calling it compute.
8:09
You can either do it. And by
8:12
buying an easy two instance and hosting
8:14
application, there are,
8:17
you can spin up a
8:19
Lambda and Landa
8:21
can quickly go do its thing
8:23
and close it. Or you
8:25
can do it. And and Beanstalk
8:28
obstruction, which takes
8:30
care of your
8:32
easy to and pro obstructed
8:34
with somethings for you. Or you can use
8:36
a faggot service, like a plaster was
8:38
transit application in a contract way.
8:41
Right. I'm just, there are
8:44
at least three more in there. That
8:46
does the same thing, but each
8:48
one has a different trade-off and
8:51
this, this kind of, this
8:54
kind of empowers the customer
8:56
to say that to
8:59
basically choose have more robust
9:01
choices. Right? So any
9:03
decision, like any CIO and CTO,
9:05
when they make a decision, whether they want to go with AWS,
9:08
Azure or GCP almost
9:10
always, no one regulates choosing
9:12
AWS because the old adage
9:15
of last year, right. No one got fired because they chose
9:17
Oracle or IBM. Right. It's
9:20
the same thing, right? Just because of the fact that
9:22
they've been here for a long time and they've been actually
9:25
trading away or at these features
9:27
on and on and on again, they are have
9:29
a very robust set up products that they have.
9:32
I think that's the number one. Competitive advantage.
9:34
If you ask me the second is the community that I've
9:36
built around, that I could go stack
9:38
overflow. If you look up the tags
9:41
for AWS, Azure and GCP
9:43
you would find about actually that
9:45
you would find about 9,000 pages
9:48
worth of posts for AWS for
9:53
Azure and 2000 pages
9:55
per GC. And that's probably a good
9:57
proxy for the amount
9:59
of community and engagement you have. Right?
10:02
So in this commute, it
10:04
actually, it, it might not sound like
10:06
a lot for people who are not in the
10:08
industry, but it is a
10:10
big deal for for a software
10:13
engineer when he's trying to Try
10:15
out a prototype and he
10:17
knows that he can go ask a vague
10:19
questions or refer the questions that are already been
10:21
answered. And that's a huge deal. This
10:26
even just making
10:29
the decision, whether someone wants to use an AWS
10:31
technology with us, let's say some
10:33
of the solutions people tend to, okay,
10:36
let's do this because it has higher stuff. Right.
10:38
So that's number two. I would, I would call it
10:41
and the other competence
10:45
are easily replicable and
10:47
already replicated. For instance, if it takes support,
10:50
it used to be able to support was
10:52
kind of best in class.
10:55
They have dedicated teams that have,
10:57
that are rightly incentives. To
10:59
minimize the number of tickets that come to them,
11:01
that they're actually incentivized to reduce
11:04
costs for the customers like
11:06
AWS customers actually tell you,
11:08
you should do this, but it support. Sorry,
11:11
would actually tell you, you should do this
11:14
to reduce cost. You will not
11:16
hear that anywhere. Right.
11:18
Like any listeners will be like, oh, do
11:20
you want to use this extra feature? Right.
11:22
But sales people do that, but
11:25
it really supports incentives actually
11:27
said to reduce costs for you, solution
11:29
architects. It used to be that it'll
11:32
be as hard, one of the best supportings, but I think it was
11:34
a GCP or uptake game. I don't think
11:36
it's a competitor, a competitor advantage
11:38
anymore. I just want to add that. I
11:41
think. Azure
11:43
and GCP are going to be as
11:46
good of a businesses as AWS.
11:48
I just know it'll be very deeply because
11:51
I worked with hands-on. I
11:53
I just want to second what you said about community. So
11:55
like for the various hacker hackathon type projects
11:57
that I've built that involve AWS tutorials
12:00
and stack reflow answers are my number one determinant.
12:03
I do not want to spend a month. Of, of lead time
12:05
to build something I want to spend like three days.
12:07
And in those three days, I want to build a proof of concept,
12:10
at least. So that is by far
12:12
my biggest determinant for choosing what technology
12:14
I use. And AWS is just a, no
12:16
brainer from that front. So how
12:19
can Azure and Google
12:22
cloud possibly compete if AWS
12:24
has such a headstart, both in terms
12:26
of what their offerings are, but more importantly,
12:28
what their community. I
12:31
think they are already started chipping away.
12:33
They just, they just need to start follow
12:35
the playbook. Microsoft is so good at
12:37
replicating others products. I
12:40
mean, it's not a surprise, like if
12:42
Microsoft school business and it's basically replayed
12:44
replicating other people's innovation and they're very
12:46
good at it. And they've done that beautifully,
12:48
right. As there has been an
12:50
existence for 10 years. And
12:53
it's, it's not, no one has the dates to
12:55
move to cloud anymore because. Microsoft Microsoft's
12:57
person. So Microsoft actually made
13:00
it, made
13:02
a Mindshare from customer's perspective
13:04
that in moving to cloud is not a bad
13:06
thing. Like we can deliberate go because
13:09
Microsoft is there, right. Microsoft has
13:11
that kind of Mindshare. And
13:13
I think it has almost like 10 years from
13:15
now. I wouldn't be surprised if AWS
13:17
and Azure are how equal market.
13:20
And GCP right behind
13:22
that I would not be surprised at all. And
13:24
they just have to need to follow the
13:26
playbook that it will use how it's set up.
13:29
And I think both of them would, would
13:32
be just fine and B
13:35
any other and new entrance that they are
13:38
thinking to come into the school space that is
13:40
not going to happen in my mind. That's it,
13:42
the, the intubated scope. In
13:45
my opinion, there are a couple
13:47
of people that I am for the sport, right. For
13:49
OCI, Oracle OCI is
13:51
I am for that spot. Like Oracle is probably
13:54
has the cash flows to actually make it happen.
13:57
But I doubt it because
13:59
just because of the fact that or to have lost
14:01
its luster people kind
14:03
of view it as old school. And
14:06
I know that's, that's still to be decided,
14:08
but. The intermediate into heart hyper
14:10
scaling witnesses closed outside of China,
14:13
these three are going to be the only
14:15
layers. I, I did a
14:17
calculation sometime I go, but
14:20
I don't have that almost with me right now.
14:22
The, the CapEx, the
14:24
capital expenditures that Amazon
14:27
Azure and GCP have spent
14:30
thus far. Would
14:32
sink a country like Argentina,
14:35
but that's right.
14:38
That's the amount of
14:41
money they have to put into
14:43
this do bring to the scale
14:46
and only
14:49
cash gushing businesses
14:51
can afford to do that. No new
14:54
player. Can you and I cannot
14:56
go sit in a garage today and say,
14:58
that'd be going. Compete with
15:00
AWS and
15:02
no VC can support that kind of money.
15:04
Either only
15:07
businesses that are actually gushing
15:09
cash out on it. They're like, we
15:11
don't know what to do with this. And
15:14
we are going to buy with this cash towards this.
15:16
And that sort of happened with Amazon. Like with
15:18
Amazon, that's not the case. Amazon was, was
15:22
actually unprofitable for so
15:25
long. Because of the money though, that they have
15:27
to spend on AWS,
15:29
if not, I think retail personally,
15:31
I think retail would have started showing free
15:33
cash flows. In 2009, not
15:35
saying that's a good thing, but I
15:38
think all the cash that was coming from retail
15:40
was just straight up going to Catholics for AWS.
15:42
Right. And also came up, explore Amazon retail,
15:45
but for Google the S. The
15:48
amount of free cash flow that the generating
15:50
from Soche, it's just straight
15:52
up going to Cemex of AWS,
15:55
right? the
15:57
GCP and same with Microsoft,
15:59
but they've spent so much for the past 10
16:01
years and only
16:04
Azure and AWS
16:06
are making profits right now. Google is not making profits.
16:08
I think it will take another couple of years. So.
16:11
Spending that much money without taking
16:13
profits for 10 years. Not many
16:15
countries can do it. Let alone country companies.
16:18
So I think these are the only players in my
16:20
mind that going to be in this space. So
16:22
our Google. So what is, Google's
16:25
play Google and Microsoft play to gain
16:27
market share. Are they trying to migrate AWS
16:29
customers or are they going after the market
16:31
who hasn't even gone to the cloud? Yeah,
16:34
th this is white space. They are just going
16:36
with whites, less exploration. So with that,
16:38
it'll be a, no one is competing with each other right
16:41
now. They're just, no, I mean,
16:43
they are competing at some level, but. Looking
16:46
at schmoozing CEOs and CTOs,
16:48
and they're showing how much that they can save
16:50
and how much they can improve their productivity.
16:53
And they're just going after the old school
16:56
basically the non-tech part
16:58
of SMP 500. If you think about it, it's
17:03
not in a cloud right now
17:05
and they're just going out to the market, right. When
17:08
I say, like, imagine
17:10
American airlines, Southwest airlines,
17:12
Nordstrom, ExxonMobil
17:15
container store, you know
17:17
all these people. They're just going out to the whites because
17:19
I think the statistic is not even accurate
17:21
at this point. I think the statistics somewhere around
17:24
20% market penetration. Yeah. People
17:26
can't even judge because no
17:29
one knows how, how
17:31
big that spaces, but
17:33
I'm I'm estimating about,
17:35
you know, sound of a percent penetration rate
17:37
in another 10 years. And
17:39
most people would come to the realization that
17:42
having their
17:44
obligations on in cloud is much
17:46
more Beneficial for them than
17:49
doing their own their own private clouds. There are multiple
17:51
reasons for it. I can tell you, I can tell you a simple
17:53
example, like from the
17:56
customer benefit, right? It's
17:59
from customer standpoint, it's very straightforward
18:02
in their head. I
18:05
don't want to name names, but
18:07
there's a healthcare company. And it's
18:09
mandatory as a healthcare hospitals,
18:12
software solutions company that I worked for before.
18:15
And this company has been onboarding.
18:18
Customers, you know, left and right,
18:20
right. Like when I say customers, hospitals, hospital
18:22
solutions that relate to technical technological
18:25
hospitalizations, and it's
18:27
an old school company and they
18:29
have the old data centers obviously. And
18:32
the bottleneck is whenever
18:35
they have to add a new customer,
18:38
they need to build a new data
18:40
center to house the
18:42
extra day. Right. Like either
18:45
they have to add new CPU's because
18:47
the clusters we're not scaling like every
18:49
like, like administerial
18:51
level, the, the
18:53
jobs were failing because the
18:55
clusters couldn't sustain
18:58
that much amount of data. So they need to add
19:00
extra notes, but they don't have space
19:03
in the same data center to access the extra nodes.
19:05
So they have to do extensive. Their data centers
19:07
or build new data centers to add new customers.
19:10
So imagine the amount of bottleneck that
19:12
it provides for operations. Like if you can,
19:14
they could only, they have the capacity
19:16
to onboard 20 customers a year,
19:18
but they can only do 10, but
19:23
once they move to a
19:25
cloud provider like AWS or Azure,
19:28
they don't, they're not, they don't have this bottleneck
19:30
anymore, even though they end up spending the
19:33
same amount over the long run. Right.
19:35
If you are monitoring the initial spending
19:38
and you do a net present value and whatnot
19:40
probably will come to about the same
19:42
money, but they're, but
19:44
where they win is the productivity gain. They
19:46
don't have to wait two
19:49
years to onboard 20
19:51
customers. They just have to do, they
19:53
can do it as fast as they can. And
19:56
in, in technological space,
19:58
that's a huge advantage, right? No one is going.
20:01
Profitability straight up, everyone is going for market
20:04
share. And because in technology
20:06
in it, that is the most important
20:09
part. Like it's, it, it lends itself to
20:11
be a winner-take-all kind
20:14
of space and it's a huge
20:16
benefit for customers, but that alone
20:19
is a no brainer for customers and customers are
20:21
going to, in my opinion, any
20:24
new software solution that existing
20:26
Non-tech S and P 500 companies are going to
20:28
bold. I think they are going to wisely choose
20:30
to just put it in cloud. I think most
20:33
of them already decided to go that way, if
20:36
you, if you could sort of ground this conversation
20:38
and numbers, cause I think you said we have
20:40
currently 20% market penetration
20:42
between all the cloud providers. Eventually,
20:45
maybe in five to 10 years, we have 75%
20:48
market penetration. What does that mean in terms
20:50
of dollar. Sure.
20:52
I'll go those numbers. So
20:55
we can do this calculation
20:57
either way. We can either go
20:59
from the
21:01
global GDP and backtrack. To
21:05
what would be on the cloud
21:07
town, or we can look at the current
21:09
businesses and I'll play kegger on
21:11
top of it and see where they are. We can do both. Let's
21:13
go from the global GDP route, right. Right
21:16
now in 2021
21:18
2020. The global GDP is
21:20
113 trillion and we apply
21:23
a 6% projected kegger to
21:25
disclosure, GDP assuming 2%
21:27
inflation, three per person productivity
21:29
gains. You will end up with $180
21:33
trillion worth of global
21:35
GDP. That's every S and
21:37
on the side and $80 trillion.
21:41
If you assume that. Any
21:43
big, any industry? APR
21:46
seem that 5% of the industry's
21:48
spend are going to come from it
21:51
maintenance and that very well is the case.
21:53
Like if you go look at statistics even
21:55
agriculture industrials are spending
21:57
5% of their
21:59
spend on it at this point.
22:02
And it's only going to grow on automation, right?
22:04
Like you can think of even the smallest things like
22:06
irrigation, automating the irrigation part. That's
22:08
an, it. Right. If
22:11
any, any sort of automation is going to compromise
22:13
the span. If you have him 5%
22:15
of that, that's going to be $9 trillion. Right?
22:18
So 20, $39 trillion of
22:21
global GDP is going to be on, it
22:23
spent to be precise
22:26
and of the $9 trillion.
22:29
So now we are talking about automation
22:31
and software and it companies and hardware companies,
22:33
right? This is the, this is the industry space
22:35
that we are right now. General heuristic, if
22:38
you take usually take any companies
22:40
the gender, like if you look at the expenses about
22:42
80% will be their employee
22:44
base. 20% probably is
22:47
their backend infrastructure.
22:49
That's a, it's a Pennsylvania
22:51
heuristic that I would like to date. It probably
22:53
is more if you asked him 20%
22:57
oh 9 trillion, that is
23:00
$1.8 trillion. And
23:02
then $1.8 trillion
23:05
is the, is the money
23:07
the world is spending on it infrastructure.
23:11
And now this time. Is
23:15
includes private clouds and
23:17
public clouds, hyperscalers, and, you
23:19
know people who decide to have their own clouds.
23:22
And you assume by 20, 30 scientific
23:24
person or the people on public clouds and 20, if
23:26
I choose to still choose to have
23:29
their own clouds but that penetration, you are looking
23:31
at 1.3, two. And if you assume
23:33
let's say that I'm wrong and there is one
23:36
more new enter the space and
23:38
the top five players only taking up
23:40
some 20% of the one, 1.3
23:42
trillion that I'm talking about. That's
23:44
$1 trillion. So the top
23:47
five players, including the two from China
23:49
are how in cumulation
23:52
are making $1 trillion in revenue.
23:55
Right? And if you look at the first
23:57
day, Right now it's
23:59
about 50% AWS right now
24:02
in 2021, 40%, AWS, 30%
24:05
as you're about 10,
24:08
10, 5, 3%
24:11
the remaining. Right. But obviously the
24:13
smaller players are going much faster
24:15
than AWS and Azure. And
24:19
if you are seeing. It reduces loses
24:21
market share and they are 30%
24:23
by 2030, that would bring
24:25
the revenues to $300 billion.
24:28
By 2030 and
24:31
$300 billion. If you
24:33
put a free cashflow margin on top of that,
24:36
that will be $60 billion. To give
24:38
some context, all
24:40
of Amazon makes $30
24:43
billion in free cash flow in 2020.
24:46
So it was a loan
24:48
would make twice that amount of free
24:50
cash flow by 2030. If my assumptions
24:52
are right. Yeah. If
24:54
you go back. So, so this is a
24:57
reverse engineering plan the global tab,
24:59
right? But if you see
25:01
how feasible it is from the current revenue
25:03
numbers on, on an annualized basis,
25:06
it'll be places making $55
25:08
billion in revenue in
25:11
the past 12 months. And
25:14
so 55 billion to 300
25:16
billion revenues. So
25:18
that comes to somewhere about. 21
25:22
person, or if
25:24
I'm not mistaken someone can correct me. But
25:27
it's one day, one person kegger as
25:29
right now is growing at about
25:32
to T one
25:35
person. I think last quarter
25:37
was 35%, but
25:39
they're growing at about 30 bucks of giggle right now.
25:41
So even if you assume the first four years
25:43
are going to be, you know, From
25:46
somewhere around 30% to failing
25:48
to 15%, they'll
25:51
easily make 21% and
25:53
that would get there. And I think
25:55
they will get there. Same with. As their material,
25:58
if you have saved 30%, I mean, if
26:00
you have simply let's assume that as it is 25%
26:02
they would be going a little more faster,
26:05
but I think they'll get there in my opinion.
26:08
Do you think though that the clouds of 2035
26:11
will resemble the clot that we have today? Or
26:13
could there be something which completely changed
26:15
the orders of magnitude? There's
26:17
something that I think everyone
26:19
needs to keep an eye on by 2030.
26:23
I don't think anyone them,
26:26
anything fundamentally would change
26:28
given the incentives of people who are
26:30
involved? No one wants
26:32
the underlying infrastructures infrastructure
26:35
to change. If they are already spent so much money,
26:37
right? Every single player wants the
26:40
same infrastructure. They, people change
26:42
adopting just one lead. They see clear incentives.
26:45
With the newer chain, right? That's why people
26:47
who are have their obligations in
26:49
their private data centers are moving to AWS because
26:51
they're seeing a clear incentive to
26:54
move right now. There's
26:56
nothing in the horizon, in my opinion. That
26:58
can be an upstart, but. There will
27:00
be something for sure in
27:02
the next 15 years or so. I
27:04
don't know what that's going to be. Anyone who is
27:07
enlisting technology has to keep an eye on these
27:09
things for sure. One advantage
27:11
that you have is that
27:13
new thing might very well be coming from
27:15
these three players I
27:17
was just saying that on your note about hyper scaling
27:20
and Amazon growing at 30%, I'm sure as
27:22
you're, they're all growing at a similar rate, what
27:24
is the biggest constraint to their growth?
27:26
I was just reading an article recently that Amazon
27:29
is having trouble finding the requisite
27:31
talent to maintain this growth. Do you
27:33
think that is the biggest constraint or do you think there's some
27:35
other big constraint? I think Tyler. Is
27:40
certainly a problem. The, the first
27:42
five years of growth, I think they were able to achieve with
27:45
much less head count.
27:47
But right now the industry they're employing
27:50
more than 150,000
27:54
people. The top three players alone, more than
27:56
150,000 engineers are not
27:59
engineers, kind of 2000 people, right? It does
28:01
include sales. This includes several
28:03
people who are in the data centers. I think
28:05
they are figured out a way to
28:09
attract new talent. I
28:11
think the trouble. Is
28:14
going to be keeping the quality
28:16
as they grow into the
28:18
scale. One of the biggest risks,
28:20
in my opinion for hyperscalers is
28:23
losing trust. The biggest stress that's
28:25
what happened to take Oracle,
28:27
right? Customer started losing trust in
28:29
Oracle. Once they realized that Oracle is
28:31
not scaling, it's not performing
28:34
to the level that they thought that Oracle
28:36
products do. Almost Oracle became
28:38
synonymous with, you know, sluggishness, slow.
28:41
You are. Right. So, but,
28:45
so that's one of the biggest things. So they have to keep
28:47
on innovating and
28:49
pushing the boundaries to sustain this growth
28:51
or else something else is going to be
28:53
an upstart. That's number one. And
28:56
security is second biggest risk.
28:59
If customer starts seeing
29:01
hearing news from everywhere that,
29:04
oh, there's a data breach here. And apparently
29:07
it's from Azure or GCP. They
29:09
start having this mind share of unreliability
29:12
and it will be very hard
29:14
to sell to new customers. That's going
29:16
to be the second biggest hurdle.
29:18
But if they are able to
29:21
maintain these two basically when I say
29:23
these two, I mean reinventing themselves,
29:25
number one basically the day one culture that basis
29:28
talks about and the second one being wonder
29:30
bus and attention to security. I think
29:32
that'll be fine. Sure. You know, I,
29:35
the other thing about AWS that I wanted to touch on,
29:37
I know previously you've mentioned, so the recession
29:39
proofness of the business, can you
29:41
speak a little bit about that? Sure.
29:44
Any investments that
29:46
as an investor you make one
29:49
of the first things that to look at as downside
29:51
production, correct. One of the reasons
29:53
that Costco sells at
29:56
a premium to the
29:58
profits that they're making is because
30:01
people know that. Even
30:04
in a recession, people are going to go buy
30:07
in Costco because they want to save money. Whereas
30:10
you can't say that to a business like Etsy.
30:15
When, when people are losing jobs,
30:17
they're not going to go buy handcrafted
30:20
items that are being shipped to their
30:22
homes. Costco
30:26
would probably see 5%
30:28
loss in revenues. If we get
30:30
a scenario of 2008, again,
30:33
it's, you'll probably see 50% loss in revenues,
30:36
right. It's not to say that it's a
30:38
bad business, but Costco
30:40
is a better business. I think
30:43
the same applies to the
30:46
cloud industry. Yeah. If
30:49
companies are struggling to make revenues.
30:51
The first thing as
30:54
a CFO would look at the airline
30:56
items and say that, Hey, we are to
30:58
go lean. We should probably let
31:00
go of some people, or we
31:02
should let stop spending
31:04
on new ads.
31:06
We should stop spending on expansion.
31:09
Those are the places that they will cut their
31:11
money. The last
31:13
place that they would cut their money on
31:15
is to strip their website away. They
31:19
need their website up and running.
31:21
No matter what the businesses say,
31:24
you are a restaurant Thai
31:26
restaurant that that
31:30
is going through Solutionary period. You
31:33
will, you know, the go people
31:35
first, you will do anything. But the last
31:37
thing that you will do is Stopping for your
31:40
website whoever is hosting your upside, right?
31:44
And ultimately all of
31:46
this trickles down to being managed by hyperscalers
31:49
and ultimately it goes to themselves. They
31:51
are almost like utility.
31:54
In some sense you will not cut your water.
31:56
Supply. Water will be the last water
31:58
bill will be the last place that you would stop paying. Right?
32:01
So it's almost like a utility
32:03
in the, in some sense. And I think. Hyperscalers
32:06
would cruise through a neuroception every
32:08
period. And also
32:10
one other thing to note is
32:13
they know that they are in this power position
32:15
and they don't abuse it. They
32:18
usually give leeways to
32:20
the customer because they know
32:22
that the customers need their help.
32:24
They usually just like in old landlord
32:27
gives moratorium doing coordinate. COVID-19
32:30
hyperscalers say that. It's
32:32
okay. You don't have to pay us
32:34
for you know, so on silly if you're struggling,
32:37
but then, you know, you have to pay us when things are
32:39
getting better and they would, they all,
32:42
they always provide because scenarios
32:44
like that, because at
32:46
the end of the day, the customer's success
32:49
is hyperscaler success.
32:52
They want every single one of their customers
32:54
to succeed. Phenomenal. Right.
32:57
Do you, do you think Amazon does that? I know
32:59
you mentioned this earlier, which is they really want to
33:01
save the customer money. And even in the case of
33:03
a recession period, they don't take advantage of them. Do you
33:05
think that is because of anti-trust
33:07
fears, they're worried they're going to get broken up and so
33:09
they don't want to do you know, they don't do price gouging
33:12
of anything. No, I genuinely think
33:14
that's because it's in the best interest
33:16
of them. They don't, you
33:19
don't want your best customers dying. Right.
33:22
Ask tobacco industry right.
33:24
You want your best customers living and
33:28
they, I think they would, they should, if
33:31
any operational entity
33:33
would choose to do this because
33:35
it's only in their interest. Oh,
33:39
one thing that you've mentioned a few times
33:41
was the potential upstarts.
33:44
So. I wanted to get your
33:46
opinion on something like, say Heroku, which
33:48
just to give context, it's
33:50
kind of offers some similar things to
33:52
Amazon, but it's a much lower
33:55
learning curve, but it's also much, much
33:57
less powerful. So where do you think
34:00
things like Heroku fit into this whole
34:02
ecosystem? Hiroko
34:07
probably has been an existence
34:10
as long as it'll be. Correct
34:12
me if I'm wrong, right? Probably about
34:14
10 years now. I think they have chosen
34:16
their niche and
34:19
I think they are decided to play
34:21
that niche. I don't, I don't think they want
34:24
to play the hyper scaling
34:26
space. I think they want to stick to
34:28
these are the know
34:30
bare metal solutions that we are providing. And
34:34
we are leader in the space. And if you want
34:36
to host your Kassandra
34:40
instance and manage it with
34:42
your own DevOps team, come and use our
34:44
solution to host it. You don't have to
34:47
go build your data centers, but you can
34:49
have your own DevOps team to maintain your solution. Right.
34:53
And that's a niche. That's a niche
34:56
that there are some people who prefer
34:58
doing that. The communities that have
35:01
a specific use case to maintenance their own applications.
35:03
And they are very happy with using private,
35:06
private clouds or managed.
35:12
Commodity commodity type
35:14
public clouds like Heroku. I
35:16
think everyone is trying to find the
35:18
niche and be the biggest player there. I don't think
35:21
if Heroku had to compete, they would have
35:23
already competed in the space. I don't think they have,
35:25
they might. Since
35:28
it's a white space, no one is
35:30
competing with each other, but once
35:32
the industry gets a dementia rarity, I think will
35:34
be a lot interesting to see how
35:37
each player is deciding to
35:39
partner with each other. So
35:41
you mentioned like by 2035, we see
35:44
these big five players. Do you think it's possible
35:46
that another thing which could happen is we have the big
35:48
five and then we have a huge chunk of market share that's
35:50
taken by just like these small niche islands.
35:54
Right. I think so. I think private cloud,
35:57
private cloud is going to have a lot more
35:59
players just because of the fact that it's
36:01
a commodity industry, right. Like HPE
36:04
is going to however, it's share
36:07
Dallas going to have its share. I'll be embarrassed
36:09
going to have its share. Toshiba
36:12
build servers for enterprises with average
36:14
year. I think there's, there are, there are
36:16
lot more players at that private cloud
36:18
niche, and even Amazon has outpost,
36:21
AWS outpost that it
36:23
launched a year and a half, couple
36:25
of years ago basically rolls
36:27
their silvers in, into
36:30
your data centers. Then trans
36:32
erased application on top of it. So
36:34
I think it's an all quite in that space. It's
36:37
not in public cloud, in my opinion,
36:40
I think in one of your threads, I
36:42
think you put it really nicely, which is Amazon
36:44
has sort of has this distribution power, that trumps
36:47
product. And so if somebody is already
36:49
in the AWS ecosystem, they're more likely
36:51
to use a Redshift compared to snowflake,
36:54
even though Redshift isn't really as great
36:56
of a product as snowflake. And that ends up
36:58
being like a moat. So can you speak a little bit about that? Sure.
37:01
Yeah, the one allergy that I keep referring back
37:03
to is there is a reason Chrome
37:06
is the most used browser and Android
37:08
and safari is the most used browser
37:11
and iOS that does primarily
37:13
because distribution, but
37:15
it's not that close. Other
37:18
than Safaria Safaria is better than Crow. They're
37:20
good. But most people prefer
37:23
to use whatever that's coming default
37:25
them. It's not applicable
37:27
only to general public, but
37:29
to software engineer still. I'm a software engineer
37:32
and I'm building solutions and mobile solutions
37:34
using tools. Part of AWS and
37:37
new project comes into maintainable. And
37:40
let's say that it involves graphical
37:42
database. I need, you know, a database
37:45
that provides graphical relationships.
37:48
The first thing I'm going to be doing, this is
37:50
literally going to be my workflow. I
37:53
would go to Google. I would, I would search
37:55
alias graph database.
37:58
I'm not going to search best
38:01
graph database. There is a difference,
38:04
right? I'm going to search in this graph
38:06
database and I'm going to go click on the documentation
38:09
and read it. Probably I'll
38:11
do a sanity check. Neptune
38:15
databases, Neptune DB probably I'll
38:17
do a graph on a checkoff Neptune
38:19
versus whatever
38:21
that's out there. And probably read
38:23
one post that says that Neptune
38:26
is good enough. And I was like, that's good enough. And
38:30
there's a trust that's built in these ecosystem
38:32
and software engineers are like
38:34
like you said they want to get the job
38:36
done as fast as possible. But
38:39
good enough quality. The key is
38:41
good enough. Good enough. You're right. So,
38:43
and then people didn't do iterate
38:45
and improve their product. So the incentive.
38:50
I wouldn't say incentive, but the
38:53
human nature is to stick
38:55
with the dealer that they know. And
38:58
and because of that, most people are going to choose
39:00
whatever that's, that's included as part
39:02
of their toolkit. And
39:06
this is not to say that snowflakes
39:08
is a bad business snowflake. He
39:12
asked any person who use snowflake, they're
39:14
going to give you glowing reviews for
39:17
that. They're going to say it achieved a slow.
39:19
They're going to say it doesn't scale. Well snowflake
39:22
decouples compute and data
39:25
storage scales, beautiful pleats,
39:27
much faster. It provides a lot more
39:29
integration, much better than any other
39:31
solutions out there. It's all true. It's
39:33
probably it's the best thing ever.
39:36
Slicer. But the
39:38
native integration and
39:40
the distribution that
39:43
the NATO the tools
39:45
that come directly
39:48
through AWS are going to always
39:50
win in my opinion. And
39:53
it's one of the boon, or
39:55
I would say pain or software industry. It's
39:59
so easy to copy features, right?
40:03
It's there is no Payton pattern that
40:05
that is protecting snowflakes. Anything
40:07
else Snowflake's architecture? So
40:11
I, I, I bet that
40:13
Redshift team is already working on they already
40:15
released retrospect from that are decoupled
40:19
computer. Data storage.
40:21
And they probably would be
40:24
logging a year behind snowflake,
40:27
maybe forever, but they would keep on improving
40:29
and for the most dollar post it's good enough. Right.
40:32
And also the integration parts
40:35
integration is the key to all of it. Imagine
40:39
you are in WhatsApp in your phone
40:42
and you want to share
40:46
a link that you would right. In
40:48
Chrome to someone through WhatsApp. Imagine
40:51
not having the share
40:55
and send button
40:57
from Chrome that is integration,
41:00
direct integration, right? That
41:03
is a pain like a customer pain. And
41:05
those things happen all the time in
41:08
AWS. If the
41:10
third party tool. That is
41:12
being sold in market marketplace.
41:15
Most of those, they don't have one-click integration
41:17
with native AWS tools. Right.
41:20
Snowflake probably doesn't have one,
41:23
it probably does, but you know any like
41:25
a normal, a normal third party
41:28
solution, wouldn't have one click integration
41:30
with the kindnesses to
41:32
put data into it. But Richard
41:35
does. You
41:37
don't even like the whole no-code
41:39
movement. It's,
41:42
it's only possible because
41:44
of well defined
41:46
integrations. If
41:49
those integrations are not provided for
41:52
you by default, then there is
41:54
no no code. You have to code those
41:56
integrations yourself. So the
41:58
no code is basically. Strengthening
42:01
the business model that all these hyperscalers
42:03
are already having. So distribution trumps
42:05
quality, in my opinion. Great.
42:08
Great. You know, this has been amazing. I think I've
42:11
honestly went a lot about AWS. I it's one
42:13
of my sort of new favorite things to learn
42:15
about. So before we wrap up, I want
42:17
to touch a little bit on your background and
42:19
how you got into financial investing. I know
42:21
you come from an engineering background, you know, other
42:23
engineers that are listening out there who have
42:25
for their whole life really just coded or stuck
42:28
to hardware or whatnot. And now they're thinking,
42:30
well, what's, what's The technology itself.
42:32
And what's the business. How would you
42:34
sort of advise them to approach that
42:36
aspect of technology? I started
42:41
to learn about finance as
42:43
a hobby. I started researching
42:47
into what makes a business.
42:51
Durable hence more of my hands,
42:54
my Twitter handle and going
42:59
through these, going
43:02
through those research, the one thing again
43:05
and again, and all this was
43:09
a good business. Makes
43:11
all the. All
43:14
the stakeholders, not the shareholders, all
43:16
the stakeholders, happy
43:18
and content. Right. That's one of the key
43:21
things. And when I
43:23
learned this, the first thing that popped
43:26
in my head is wait, the industry that I
43:29
am working in, it does
43:31
that, right? Like if
43:33
you look at the customers are happy because
43:36
they get through, spend less.
43:40
Are much more productive. They, shareholders
43:42
are happy because this is one of those businesses
43:44
that are actually joining profits
43:46
left and right. And the
43:48
employees are happy because these DevOps
43:50
people the people who work in AWS,
43:53
they brought they, if not for that,
43:55
they probably would be ironing.
43:57
You know, all these repeats. Automations
43:59
that has to be done again and again in
44:01
a different company. Now they get
44:04
to use all that to create
44:06
solutions that actually empowers
44:08
the whole world. Right. It, it, it
44:10
kind of touches on
44:13
it, makes all the stakeholders happy. And
44:18
this is the framework that I
44:20
look any look at any business
44:22
like For people who offer them technological
44:25
background, they see a lot of new things
44:27
that are exciting, right? Like electric
44:29
cars, like you see Tesla is doing
44:33
phenomenal phenomenally with their
44:35
car manufacturing, with self-driving
44:38
and people are in love with the car, but
44:40
you have to ask yourself all these questions
44:43
too. How durable the company is,
44:45
how profitable the company is. And you have to
44:47
build these mental frameworks by basically,
44:50
you don't have to do anything new. Just listen
44:52
to you know, Warren buffet, Peter Lynch
44:55
Charlie monger Darcy and
44:57
all these people, and basically how these
44:59
mental frameworks in your head to judge
45:01
companies based on their qualities.
45:04
And once he judge companies
45:06
based on their qualities, you can pretty much come
45:08
to an understanding of, okay, would
45:11
this company, it boils down
45:13
to a couple of questions. Would this company
45:16
last 30 years,
45:19
what would it take this company to last 30
45:21
years? If
45:25
this company, if you say that this company
45:27
will last 30 years would there be other
45:29
companies coming in to take the profit share?
45:33
If that's the case, then
45:36
why would customers choose this company
45:38
or the other? Right. So it's a, once
45:41
you start asking the core question, you can dig
45:44
in again and again, and again, to
45:46
find why there's the durability
45:48
of this company, right. I think building
45:51
all these mental models in your
45:53
head is probably the number one starting
45:55
place. And then you can start
45:57
looking into the financial statements, just like
45:59
and look at their numbers and see how
46:02
much is their growth trajectory. How much
46:04
are they growing revenues, their profits,
46:07
and what market share they're having? What are the
46:09
economics. You can look at all those things at
46:12
the secondary level, but at departmental
46:14
level you should have a good understanding
46:16
of what. A
46:18
business, good business.
46:21
Sure. And I, you know, I I'll just add to that. I think
46:23
beyond, even beyond investing, it's good to
46:26
study these things as a person in tech,
46:28
just because you're probably leaving jobs every
46:30
few years. And as you determined. Sort
46:33
of what company you're heading into. It's
46:35
probably good to know, like the financial health
46:37
of this company and sort of the market they're in
46:39
and the sort of the bigger point of view,
46:41
the bigger picture of the, of the business
46:44
and the industry. So on that note dribble investments.
46:46
Great. Thank you so much for joining us. I will, we
46:48
will include sort of your Twitter handle and other
46:50
places that people can find your content in, in
46:52
the show notes. Otherwise, yeah. Thank you so much. This has
46:54
been really informational and education. Thanks
46:58
guys. I really appreciate you guys having me.
47:00
That's our episode for this week. Thank you so much for listening.
47:03
Make sure to subscribe to us and rate us on Apple
47:05
podcasts. We would really appreciate
47:07
the support. You can also follow me on
47:09
Twitter at F Z from Cupertino
47:12
and Busan. The ad next facade.
47:14
See you guys next week.
Podchaser is the ultimate destination for podcast data, search, and discovery. Learn More