Episode Transcript
Transcripts are displayed as originally observed. Some content, including advertisements may have changed.
Use Ctrl + F to search
0:00
This episode is brought to you by TIGUS,
0:02
where we're changing the game in investment research.
0:04
Step away from outdated, inefficient methods and into
0:06
the future with our platform, proudly hosting over
0:08
100,000 transcripts with over 25,000 transcripts added just
0:10
this year alone. What
0:15
sets TIGUS apart? It's not just
0:17
the sheer volume, it's the unmatched speed
0:19
at which our library expands, consistently outstripping
0:21
competitors. Our platform grows eight
0:23
times faster and adds twice as much monthly
0:25
content as our competitors, putting us at the
0:28
forefront of the industry. Our collection
0:30
is investor-led, ensuring unparalleled quality and giving
0:32
you access to questions and topics investors
0:34
care most about. Plus, with 75% of
0:36
private market transcripts available
0:39
exclusively on TIGUS, we offer insights
0:41
you can't find elsewhere. Forget
0:43
the traditional way of doing things. With TIGUS,
0:45
you have the most comprehensive, insightful, and rapidly
0:48
growing transcript library at your fingertips. See
0:50
the difference that a vast,
0:52
quality-driven transcript library makes. Unlock
0:54
your free trial at tigus.com/Patrick.
0:58
You may have heard me reference the idea of
1:00
maniacs on a mission and how much that idea
1:02
excites me. Well, David Senra is
1:04
my favorite maniac on one of my favorite
1:06
missions with his weekly crafting of the Founders
1:08
Podcast. Through studying the lives of
1:11
legends, he weaves together insights across history
1:13
to distill ideas that you can use
1:15
in your work. Founders
1:17
reveals tried and true tactics, battle-tested
1:19
by the world's icons and has
1:21
David's infectious energy to accompany them.
1:24
With well over 300 episodes, your heroes are
1:26
surely in the lineup and his
1:28
recent episode on Oprah is particularly great. Founders
1:31
is a movement that you don't want to miss.
1:33
It's part of the Colossus network and you can
1:35
find your way to David's great podcast in the
1:37
show notes. Hello
1:42
and welcome everyone. I'm Patrick O'Shaughnessy and this
1:45
is Invest Like the Best. This
1:47
show is an open-ended exploration of markets,
1:49
ideas, stories, and strategies that will help
1:51
you better invest both your time and
1:53
your money. Invest Like the Best
1:55
is part of the Colossus family of podcasts and you
1:57
can access all our podcasts including the free free podcast.
2:00
including edited transcripts, show notes, and
2:02
other resources to keep learning at
2:04
joincolossus.com. Patrick
2:07
O'Shaughnessy is the CEO of Positive Some.
2:10
All opinions expressed by Patrick and podcast
2:13
guests are solely their own opinions and
2:15
do not reflect the opinion of Positive
2:17
Some. This podcast is for
2:19
informational purposes only and should not be
2:21
relied upon as a basis for investment
2:23
decisions. Clients of Positive Some
2:26
may maintain positions in the securities discussed
2:28
in this podcast. To
2:30
learn more, visit psum.vc. My
2:36
guest today is Pat Grady, a longtime growth
2:38
investor at Sequoia and one of the firm's
2:40
senior leaders. Pat has been
2:43
a part of a long list of
2:45
legendary investments ranging from Snowflake, Zoom, ServiceNow,
2:47
Qualtrics, Okta, HubSpot, Notion, and OpenAI among
2:50
many others. There aren't many investors
2:52
who reference as well as Pat, both inside and
2:54
outside of his firm. We talk
2:56
about investing, building an investing firm and
2:58
building enduring companies. Please enjoy this great
3:01
conversation with Pat Grady. Pat,
3:04
I'm lucky in this conversation in that I
3:07
know a lot of your partners have interviewed
3:09
them, understand the business, and I'm fascinated by
3:11
Sequoia's history and the way that you invest,
3:13
but I've spent less time on the growth
3:15
side. And so I'm really excited to do
3:17
this with you today. I thought
3:19
a fun place to begin would be to talk
3:21
about the nature of
3:24
internal, healthy peer pressure at
3:27
Sequoia. How
3:29
it's created and
3:31
maintained in a way that is on that line
3:35
of healthy and overly intense. I
3:38
would love to hear as much detail as you
3:40
can manage about where this
3:42
internal pressure comes from and
3:44
how you tend to that flame. I'll
3:47
start with where it comes from. And
3:50
the simple answer is Don
3:52
Valentine. Don Valentine's the founder
3:54
of Sequoia Capital, one of the many brilliant
3:56
things that he did to build an organization
3:59
that would endure. build
10:00
values that have led to this high standard you
10:02
hold for yourself? Yeah, I
10:04
think this is on our website. But one of the
10:06
things I remember my dad always saying when we were
10:09
growing up was when your values are clear, decision making
10:11
is easy. And I think that was a
10:13
line he got from Walt Disney or something like that. But
10:17
I've always really believed in that. I
10:19
put a lot of thought into family
10:22
values and the team values and try to
10:24
make sure that those serve as the guidelines.
10:26
So my wife Sarah and I have, we
10:28
actually have family values and we have four of them.
10:31
And we had a fairly complicated form that
10:33
our kids didn't really understand. And we simplified
10:35
them. And they sound like motherhood
10:37
and apple pie. They're just work hard, be
10:39
kind, think for yourself, family first. That's it.
10:42
It's pretty straightforward. But with those four, you can
10:44
pretty much derive the right
10:47
answer to any given question and any given
10:49
situation. Or if the girls do something that
10:51
they shouldn't have done, we can usually explain
10:54
why they shouldn't have done it in the context of
10:56
those values. Or here at
10:58
Sequoia, the two things we care about most
11:00
as a partnership or performance and teamwork, but
11:03
then each team operates in
11:05
a slightly different context. And so our
11:07
team, the growth team has its own
11:09
values, which are value number
11:11
one, aggressive, but humble, value number two,
11:14
demanding and supportive, value number three, high
11:17
give is at zero volt, value
11:19
number four, strong under scrutiny. Those
11:22
four things also collectively we think can
11:25
get you to most of the behaviors that are
11:28
required to be effective over the long term at
11:30
Sequoia. Which of those do
11:32
you think is the hardest to put into practice on the Sequoia
11:34
growth side? Which of those values? I
11:36
think they're all hard. I mean, they wouldn't be good values
11:38
if they weren't hard. I mean, I'll
11:40
take demanding and supportive, which our partner, Rubby,
11:42
I think deserves credit for bringing that one
11:44
to us. There are
11:47
plenty of organizations that are 10 out of
11:49
10 demanding, but they may not be great
11:51
places to work. There are plenty of organizations
11:53
that are 10 out of 10 supportive, but
11:55
they may not be high performance. Most
11:58
people think there's an inherent trade off between. these
12:00
two things, we do not. We
12:02
think that being demanding of one another is being supportive,
12:05
and that the best thing we can do for each
12:07
other is to demand excellence, first of
12:09
ourselves and then of each other. And
12:11
so trying to be 10 out of 10 demanding and
12:13
10 out of 10 supportive, it's
12:15
hard because it's not necessarily human nature, but
12:18
we think it's part of the key to
12:20
our performance here. Do you
12:22
think that some of these things you're talking
12:24
about can be taught or are they just
12:26
things that are innate in a person? And
12:28
you were just hardwired this way, I read
12:30
that you had an inside sales job and
12:32
you wanted to win that competition every day.
12:34
And at a young age, you
12:36
just had this, what seems like from the
12:38
outside pre-wiring to be
12:40
ultra competitive and just
12:42
look at the world a certain way. Is
12:44
that your experience with founders and with people that
12:46
you work with that either have it or you
12:48
don't? And it's just unfortunate for the people
12:51
that don't have it. We believe that
12:53
our business is an apprenticeship business. I think in some
12:55
ways life is an apprenticeship business. I
12:58
won the cosmic lottery to
13:00
be born to wonderful, kind,
13:03
hardworking, loving parents. And
13:06
I think when you start there, you're off to
13:08
a pretty good start. And I think
13:11
the stuff that they taught me was sort
13:13
of reaffirmed throughout my life. First at Boston
13:15
College, which is a Jesuit
13:17
Catholic institution where the
13:19
Jesuit motto is men and women for others.
13:21
And then the BC motto is ever to
13:23
excel. And if you put those two
13:25
things together, you have this concept of whatever
13:28
you do, you should do to the very best of your
13:30
ability. But don't forget that we're all
13:32
in this world together. And so you shouldn't do it
13:34
just for yourself. You should do it for the community
13:36
and the world around you. When
13:38
I graduated from school and
13:40
started working at Summit Partners
13:42
in Boston, I was
13:44
lucky enough to work for a guy
13:47
named John Carroll, who I
13:49
didn't know much about before going to work for him.
13:52
And it turned out that he was both
13:56
probably the best investor at
13:58
Summit. and the
14:01
best human being you could possibly hope to work
14:03
for. And so in my
14:05
first few months at Summit, there was a situation
14:07
that came up where somebody
14:09
was trying to take
14:12
credit for a thing that I had done, and
14:14
I wasn't sure if I should fight for credit,
14:16
just let it go. And
14:19
so I went to John Carroll or JC, and
14:21
I said, hey, JC, I've got this situation, what
14:23
would you do? And
14:25
his answer was, in my
14:28
career, I've always taken the
14:30
highest possible moral road, and
14:33
I have never regretted it. And
14:36
that was very clarifying for me. And
14:38
so I let the situation go, and it turns out that
14:40
it was for the best. And it
14:43
reinforced what I heard at BC, reinforced what I heard
14:45
from my parents. And then I got
14:47
to Sequoia, and I remember
14:49
before joining Sequoia, people had told me, Doug
14:52
Leone is the greatest salesperson you will ever
14:54
meet. And so I was very
14:56
excited. I was like, okay, I can't wait to see
14:58
the magic of Doug Leone, the salesman. And
15:01
my basic job when I got here was
15:03
find companies that were interesting enough to bring
15:05
Doug to meetings. And
15:07
so Doug and I did hundreds of meetings in
15:09
my first couple of years here, maybe thousands. We
15:11
did a ton of meetings. And
15:14
after my first couple dozen meetings with Doug,
15:17
the thing that blew me away was when we went to
15:19
a meeting and anything related to
15:21
an investment came up, where maybe you're supposed
15:23
to be negotiating the investment. Doug
15:26
was so unbelievably transparent, it
15:28
blew me away. And
15:31
I remember afterwards, I said, well, Doug, how do
15:33
you, for example, a negotiation with
15:35
Doug might be, you're gonna wanna
15:37
pay X, we're gonna wanna pay one half X.
15:39
Why don't we just call it 0.75X and call
15:42
it a day? And
15:45
the whole thing would take 30 seconds. And
15:47
the other person would say, yeah, okay, that
15:49
seems right. He
15:52
wasn't making stuff up. He was actually telling them what
15:54
we wanted to pay. And they was actually guessing what
15:56
they wanted. And then it was just meeting in the
15:58
middle at a price that we didn't. the
18:00
market dynamics, that's all very doable. But
18:02
I think the difference between the
18:04
point of view that we're going to have on the investment and
18:07
the point of view that somebody else is going to have on
18:09
the investment is not going to be a function of
18:11
better understanding the cohorts or
18:14
listening just a little bit more carefully to
18:16
the customers. I think it's going
18:18
to be a function of actually understanding the people. And
18:22
when that model ends at the out year,
18:24
five years from now, has
18:26
that founder gotten 10 times better? And do
18:28
they still have gas in the tank? Or
18:30
is that founder exhausted and just
18:32
barely clinging to life? And
18:35
so the way that I like to do that is
18:37
to go for a long
18:39
walk with founders and try to understand
18:42
who they are and all the
18:44
ways that don't show up on LinkedIn. What was their
18:46
childhood like? What experience in their
18:48
childhood most contributed to who they are today?
18:50
What characteristic did they take from their mom?
18:52
What characteristic did they take from their dad?
18:55
If they have brothers and sisters, how do
18:57
they view themselves in relation to their brothers
18:59
and sisters? And what was the happiest
19:01
moment from their childhood? What was the biggest mistake that
19:03
they made in their childhood? All those
19:06
sort of questions that in and of themselves may not
19:08
tell you much, but when you go
19:10
deeper and deeper and deeper, you really start to
19:12
understand who they are and what they value. And
19:15
if you can understand that, you start to understand what
19:17
drives them. And then you can start to
19:19
map that onto the business they're trying to build and
19:22
figure out if they're actually likely to build
19:24
something for the long-term or
19:26
if this is more of a passing fancy and they
19:29
want to play the game of entrepreneur versus actually building
19:31
something that matters. When
19:33
you want to win an investment
19:35
and it's competitive, meaning there are other
19:37
very talented investors, very smart
19:39
investors who can do a lot for
19:41
the business. There's this plethora of amazing
19:43
people out there. When you're
19:45
up against that sort of situation, how do you
19:47
win? I'm assuming your win rate at
19:50
Sequoia in general and for growth specifically is very
19:52
high and that you're seeing
19:54
most of the opportunities out there because of
19:56
the size of your business and platform. So
19:58
how do you keep... that win rate really
20:00
high. What are the things you do to
20:03
win when it's competitive? I
20:06
think one of our core beliefs
20:08
is that anybody can beat us on any given day. So
20:11
back to your comment earlier about how do we
20:13
keep that pressure on ourselves, we truly believe that.
20:16
And we believe that in
20:19
part because our competitors are
20:21
no joke. They're very smart people who
20:23
know what they're doing, who work really
20:25
hard, who have killer instincts, who
20:28
have a nice way with founders at lots
20:30
of other firms. And any given one of them
20:32
could beat us on any given day. And so
20:35
when we're in a competitive situation, we
20:37
can't just waltz in and hope that the
20:39
Sequoia Business Card is going to give us
20:41
the advantage. It is hand-to-hand combat, and we
20:43
have to be on our absolute best if
20:45
we have any hopes of making that investment.
20:48
So what does it mean to be at your absolute best? I
20:51
think the biggest mistake people make is selling
20:53
by telling founders how awesome you are. Founders
20:56
don't care how awesome you are. They want
20:59
to know how awesome they have a chance to become. The
21:01
thing that we try to do is not
21:03
to sell the merits of Sequoia and
21:05
all the wonderful value added hands-on company
21:07
building stuff we can do for you.
21:10
Maybe we'll sprinkle in a little bit of that here and there.
21:13
But the thing we really want to do is understand who
21:15
are you, what do you want
21:17
to become, and what is it you want to
21:19
build. And if we can understand
21:21
those things, and we can feed that
21:23
back to you to show you that we understand those things,
21:25
and to show you that we are interested in those things,
21:27
and we want to be a part of those things, and
21:30
maybe some of the resources that we have here
21:32
could help you in achieving those things. But the
21:34
main thing is not our resources. The main thing
21:37
is your vision and your dream. And
21:39
if you believe that we
21:41
believe in you, and we are going
21:43
to be here to support you, and we
21:46
shouldn't be trying to win an investment if that's not
21:48
true, and so we should be able to articulate that
21:50
to you in a very authentic way, if you
21:53
believe that, it starts to become fairly
21:55
straightforward. And so we hope
21:57
to be in situations where the right people are going
21:59
to be able to the
26:01
vision of the future that this founder hoped to
26:03
build with the company and this founder hoped to
26:05
build for him or herself. And
26:07
that was the most potent elixir
26:09
you could possibly imagine. And maybe
26:12
not 100% of the time, but 95% of the time
26:14
after Doug unfrolled one of those, the
26:19
founder would be desperate to work with us. And
26:22
then all we had to do is decide whether or not we wanted to.
26:24
So that was a pretty magical experience. What's
26:26
your version of that? Now you're the Doug. What
26:29
do you do in those meetings? Obviously
26:31
everyone's style is different. How have you honed it and
26:33
shaped it in your own way? Well,
26:36
first off, I will never be the Doug. They're the only
26:38
one. It
26:40
is not possible for there to be another duck. One
26:42
other trick that I had when I was in
26:44
my 20s and early 30s and having witnessed
26:47
Doug do that so many times, when
26:50
I would go into these meetings and that was
26:52
2007 when I joined here was the early days
26:54
of the cloud transition. Most of
26:57
the founders that I was meeting with, they'd been
26:59
in the enterprise for 20 years. And
27:01
so almost every meeting I went
27:03
into the founder was meaningfully older than I
27:05
was. I looked at a kid who
27:08
happened to have a very nice business card, but
27:10
otherwise no experience that could be relevant to them.
27:12
And so I had to
27:14
figure out how to get credibility. And
27:18
the way that I would usually do that, mirroring
27:20
what I'd learned from Doug, was to start the
27:22
meeting by saying something like, hey, I
27:24
don't know much about your market, but
27:26
it seems like you have a chance to do blah, blah,
27:29
blah. And the
27:31
blah, blah, blah would be whatever thesis I
27:33
might have come up with based on looking
27:35
at their website or talking to some of
27:37
their customers or studying some of their competitors
27:39
or whatever the case might be. And
27:42
that was usually enough to
27:44
get the benefit of the doubt and to
27:46
get them thinking, okay, this guy kind of
27:48
understands who I am and what I'm trying
27:51
to build here. And I would say that
27:53
if you fast forward to today, that's still
27:55
the most powerful thing you can do is to make
27:57
a founder feel like they're seen. You
27:59
understand. them and try to validate their
28:01
ambition through the way that you
28:04
describe their business, I think that's still the most
28:06
powerful thing that we can do. And
28:08
then beyond that, that might
28:10
get you a nice in-person interaction with the
28:12
founder, but then ultimately we are in a
28:14
product-led growth business. Our product is
28:17
the service that we provide to our founders.
28:19
And when our founders tell other founders what
28:21
they think of us, that's ultimately
28:23
the thing that gets them over the line.
28:25
And so our objective function
28:27
is to maximize net multiple money returns for
28:30
our limited partners, not to maximize founder and
28:32
PS, but if we can
28:34
have our cake and eat it too, that's the best
28:36
of both worlds. And so founder and PS and net
28:38
multiple money returns are the two metrics that we probably
28:41
care most about. Thinking about
28:43
the emails that you would write reminds me that
28:45
a bunch of people have mentioned to me how
28:47
good you are at memo writing. And
28:49
I know memos are taken very seriously
28:51
at Sequoia, but I'd love to hear
28:54
for you specifically how you
28:56
define great business or investing writing,
28:59
what you are seeking to do in those
29:01
memos and what you respect most
29:03
when you see it in other similar
29:05
memos. This comes back to
29:07
that expression, listening happens at the year. I
29:10
think there are a lot of memos that
29:12
get written speeches that get given where the
29:14
objective is to impress upon other people
29:16
how smart you are. And I think when
29:19
you write a memo or you give a talk or
29:21
whatever, you should do so with the service mindset. My
29:23
objective is not to impress upon anybody how
29:26
smart I am. My objective is for them
29:28
to understand. And if I
29:30
actually want them to understand, I need to
29:32
make it as simple as humanly possible. And
29:34
if that means sacrificing a few details for
29:36
the sake of clarity, not in
29:38
a way that distorts the picture, but in a
29:40
way that clarifies the picture because people aren't going
29:43
down rabbit holes that don't represent first order issues,
29:45
that's probably worth it. And so
29:47
a great investment memo is three pages, not 12 pages.
29:51
And when you're done with those three pages, you should have
29:53
an accurate point of view on everything that
29:55
is good, everything that is bad,
29:58
and the so what will When we mash
30:00
that soup altogether, should we make this
30:02
investment or not and why? And
30:05
I think you can do that in three pages generally. Take
30:08
a bunch of the exhibits, the cohort charts and everything
30:10
else, throw those in the appendix. You
30:12
can express that in one sentence. You don't need to have everybody
30:14
dig through all the data on their own. Go ahead and throw
30:16
that in the appendix in case anybody wants to go back there
30:18
and play with it. But keep
30:21
the narrative upfront, keep it tight,
30:23
keep it crisp, and make sure
30:25
that the thing you're optimizing for
30:27
is clarity and understanding, not trying
30:29
to impress upon people how smart you are. Do
30:32
you remember reading a memo, not one
30:34
you wrote, but another one about a
30:36
company that got you the most physically
30:38
excited? A hundred percent, yes.
30:41
The first one that comes to mind is in
30:43
2009, a
30:46
few weeks before the market bottomed
30:48
out, we got into business with
30:50
Airbnb at the seed stage. And
30:53
that seemed like a crazy idea, but it was working
30:55
and we fell in love with Brian and Joe and
30:57
Nate. And anyway, so we were lucky enough to be
30:59
in business there and be at the seed stage. Fast
31:02
forward to 2012, the
31:05
company was clearly working, even
31:08
as existing shareholders and board members, we weren't
31:10
sure exactly what to make of it. And
31:13
there was a growth realm coming together. So
31:15
the first memo comes out and
31:18
it's a typical investment
31:21
memo. It talks about the market and it
31:23
talks about the numbers and it talks about
31:25
the team and it talks about the competitors
31:27
and all that good stuff. And
31:30
there was a model at the end of it that
31:32
did your typical simple linear
31:34
extrapolation with decelerating growth
31:36
rates and maybe some flat margins.
31:40
And it got you a two and a half X return. And
31:43
so you read that and you sort of yawn. Okay, two
31:45
and a half X, who cares? And
31:47
so we decided on that Monday not
31:50
to move forward, but we were interested. So we decided to do
31:52
some more work the following Friday.
31:54
A second memo comes out page
31:56
one of the memo. Then to
31:58
become a one. $100 billion
32:01
company and here is why. And
32:03
it laid out with perfect clarity, why
32:07
they had a chance, not just
32:09
to be a two and a half X multiple
32:11
of money, but to return the fund a couple
32:13
times over because of the
32:15
size of the market, because of the structural
32:18
superiority of the business model, because
32:20
of the creativity and passion
32:22
and clarity and mission orientation
32:25
of the founders. And
32:27
it takes courage to do that because most of
32:29
the time when you look at a business that
32:31
was probably grossly overvalued by
32:33
any traditional metric at even a $2
32:36
billion entry price and say
32:38
that it will someday be a $100 billion business, you
32:41
get laughed out of the room. But
32:43
it turns out those are the only investments that
32:46
actually matter. And if
32:48
you don't have the conviction that the company has
32:50
a chance to be something truly special, you
32:53
shouldn't be recommending that we invest. You
32:55
sent me these really fascinating criteria for making
32:57
an investment at the growth stage one
33:00
of which is, I can't remember the statistical term,
33:02
leptocritosis or something like that. Some
33:05
very fancy sounding, kritosis metric
33:08
that basically is like what
33:10
you just said. We're not making a growth investment to
33:12
earn a three X return. That may happen a lot.
33:14
We're only gonna make an investment if we
33:16
feel even at the growth stage that it
33:18
has this crazy asymmetric upside. Why that specific
33:20
thing is one of the small handful of
33:23
things that was on your list of criteria.
33:25
Well, first I wanna just highlight the choice
33:27
of words there. The choice of words is
33:30
important. We have leptocritic return profiles. That
33:32
comes from Rulof. So Rulof might be the
33:34
only licensed actuary who is active in the
33:36
venture capital business today. And so
33:38
his actuarial statistics background blessed us
33:41
with that word. But leptocritic
33:43
return profile to your point basically means.
33:45
Fat right tail. Better
33:47
than typical chance that the 10 X plus
33:50
return. Yeah. The reason we do that
33:52
is because our experience has been if
33:54
you quote unquote underwrite to a three X return,
33:56
you end up with a two X. If
33:59
you think something actually. has a chance to be
34:01
a 10x plus for return, maybe you
34:03
end up with a 3x or a 5x or a 7x or something shorter
34:06
there. But if you don't think
34:08
that that upside potential is there, it's
34:10
probably just not a good enough company, or
34:12
it's probably just not a big enough market.
34:15
And so we don't underwrite
34:17
to a 10x expecting every investment to produce
34:19
a 10x return, but if we can't see
34:22
a 10x return, it's probably just not good
34:24
enough. There's three really
34:26
fascinating business criteria, which again, you're evaluating
34:28
companies when you can sink your teeth
34:30
into them. There's customers, there's revenue, there's
34:33
sometimes profit, there's a team, there's lots
34:35
you can dig into. And
34:37
I'd love to just spend a minute on each
34:39
of these concepts, because a lot of the things
34:41
you've talked about, they're just very simple and elegant,
34:43
but I'm sure that there's just tremendous amounts of
34:45
nuance and depth underneath the hood. The first is
34:47
the term you use is that it's an emerging
34:49
market leader, which sounds like it could be a
34:51
double entendre, like meet a few different things. So
34:54
maybe describe that one. Well,
34:56
first, the process by which we came to this
34:58
fact, so it's many years of iteration, but there
35:00
was an offsite we did, I think it was
35:02
called the Inn at Pelican L down in SoCal,
35:05
where we locked ourselves in a room for
35:07
about two days and spent
35:09
the entire two days just debating
35:11
these words on a whiteboard to
35:13
come up with exactly the right criteria
35:16
to define our investments. So the emerging
35:18
market leader, it is a little bit of a
35:20
double entendre. The thing that people
35:22
misunderstand about this most frequently is, oh,
35:25
okay, so we need to invest
35:27
in a company that is the market leader today. No,
35:30
we need to invest in the company that we believe
35:33
will be the market leader tomorrow. They
35:35
could be two people with an idea today.
35:37
It can be objectively number seven today. Google
35:40
was not the first search engine. Flextronics
35:42
was not the first contract manufacturer. There
35:44
are lots of examples of companies that were not
35:47
the first and did not start in a market
35:49
leadership position, but because they had
35:51
a better architecture or a better business model
35:53
or a better team, they ended up being
35:55
the market leaders over time. And
35:57
one thing that is sort of objectively true particularly
36:00
in the world of technology is whoever ends up
36:02
number one in the market doesn't
36:04
just have their proportional share of the market
36:06
cap, they have a disproportionate share of the
36:08
market cap. And so investing in number
36:10
two or number three in a market maybe can make a
36:13
little bit of money, but you're not going to produce outsized
36:15
returns for your limited partners. So it's really important for us
36:17
to invest in the companies that we think are going to
36:19
be number one in the market. The
36:21
double entendre is that we think the
36:24
market is emerging and
36:26
we think that the company is emerging
36:29
to become the leader of that market.
36:31
So the market itself might not have much of
36:33
a TAM today, but we have a
36:36
reason to believe that it's going to have a TAM tomorrow.
36:38
And a good example of that, I remember
36:40
when Okta was going public in 2017 while they were on
36:42
the road show, Forrester
36:45
published a report that said the TAM for cloud
36:47
identity was 150 million. Well, at
36:50
that time had, I think a little more
36:52
than 150 million of revenue. And
36:54
so people look
36:56
at a static moment in time and
37:00
have a hard time extrapolating that to five or 10 years from now
37:02
when the company has matured and trying to have that point of view
37:04
about where the market is going
37:08
coupled with the point of view about what position the company
37:10
is going to occupy in that market.
37:13
That is what ultimately gets us to whatever that out your
37:15
revenue projection might be. It's not a function of the financial
37:17
model. It's a function of the market dynamics in the company's
37:21
position. The second respecting
37:23
the specificity of the word choices
37:26
here is maybe the most
37:28
interesting one, which is unique and compelling value
37:30
proposition, which on the surface sounds like, yeah,
37:32
sure, like sounds great. But I think behind
37:34
each of those words is something that is
37:36
incredibly important as you evaluate the businesses. So
37:38
maybe describe why those specific words. Yeah.
37:40
So the first one emerging market leader is
37:43
a comment on revenue scale. The
37:45
second one, unique and compelling value
37:47
proposition is a comment on margin
37:49
structure. And so if
37:51
you have a unique value
37:53
proposition, that should show up in
37:55
gross margin. It should show up in
37:57
gross margin because if your product is truly unique, you
38:00
should be a price-setter, not a price-taker. And
38:02
if you are a price-setter, you should be able
38:05
to set a price that's gonna provide you
38:07
with nice gross margins. And so
38:09
unique value prop gets you to a good
38:11
gross margin. Compelling value
38:13
prop is a comment
38:15
on operating margin. If
38:18
your product is truly so compelling, you
38:21
shouldn't have to bludgeon people to death with sales and marketing
38:23
to get them to try it and to get them to
38:25
pay you for it. So if
38:27
it is truly compelling, that should
38:29
show up in the efficiency of your go-to-market
38:31
organization. Or maybe there's a number like new
38:33
ARR divided by sales and marketing. Or
38:36
there's a number like LTV to CAC. Or
38:38
there's a number like payback period. There should
38:40
be some number, or maybe 99% of
38:43
your new customers come in organically. There
38:45
should be some number that basically demonstrates how
38:47
compelling the value prop is that
38:50
leads to low sales and marketing, which in turn
38:52
leads to a high operating margin. And so if
38:54
you have an emerging market leader, chances are you'll
38:56
have good revenue scale. If you
38:58
have a unique and compelling value proposition, chances
39:01
are you'll have a nice margin structure associated
39:03
with that revenue scale. And those
39:05
are the ingredients that should ultimately determine
39:07
the quote unquote out-year financial model as
39:09
opposed to the typical linear extrapolation that
39:12
you might otherwise see. I'm
39:14
realizing now that the third one, which is
39:16
listed as sustainable competitive advantage, is the perfect
39:18
third domino, which is, okay, you get to
39:20
revenue scale, you've got good margins, how do
39:22
you protect them? Exactly. And so
39:24
what are you thinking when you're trying to suss
39:27
out the end state potential future
39:29
moat or sustainable competitive advantage? How do
39:31
you do that? It seems really
39:34
hard to know ahead of time. This
39:36
rhymes with the conversation we're having on people. And
39:39
the reason, this is a hotly debated
39:41
term, sustainable competitive advantage versus saying moats,
39:44
because moats is probably the more common
39:46
vernacular. The reason it's sustainable competitive
39:48
advantage and not moats, a
39:50
moat implies something that has
39:52
been built and will protect you forever
39:54
after, whereas the sustainable
39:57
competitive advantage is a bit
39:59
more dynamic. It is an advantage
40:01
that you are building every single day. And
40:04
the number one sustainable competitive advantage that we
40:06
see out of companies, it's not
40:08
a network effect. It's not an ecosystem advantage.
40:10
It's not some piece of IP that's impossible
40:12
for other people to replicate. It
40:15
is the DNA of the team. And
40:17
the canonical example of this was in 1999, 2000, the
40:21
smart money would have bet on eBay, but
40:23
it turns out you should have bet on Amazon. People
40:26
thought the marketplace business model of eBay was
40:28
so elegant and defensible, and
40:30
selling books online was a commodity business. It turns
40:33
out that one of those companies had Jeff Bezos
40:35
and the other one didn't. And
40:37
it was the founder and the DNA and the
40:39
culture that was created that led to the compounding
40:41
advantages over time. I think a more
40:43
modern example of that would be DoorDash. There were
40:45
plenty of people who tried to get into this
40:47
delivery business. There was only one Tony Hsu. And
40:50
so our partner Alfred who sponsored
40:52
that investment, he met Tony
40:54
at the seed stage and
40:57
liked him, didn't quite have conviction, kept in touch
40:59
with him. And then I remember Alfred told a
41:01
story about he happened to go to dinner with
41:03
Tony. He was at a
41:05
dinner sitting next to Tony prior to the Series
41:08
A and spent the entire dinner
41:10
talking with him about how DoorDash worked and
41:13
the level of detail and nuance and
41:15
grasp of the business fundamentals that Tony
41:17
had blew Alfred away. And
41:20
it was coming out of that that Alfred came back and
41:22
said, we have to make this investment, not
41:24
because of the business model, not because of
41:26
the market, but just because of the founder.
41:29
This is the sort of person who's gonna
41:31
create compounding advantages forever. Yeah,
41:33
having interviewed Tony, his command of
41:35
that business, which is a very
41:38
complicated business is truly unbelievable. Just
41:40
a special human. I've
41:42
also loved the way you've
41:44
articulated how you do
41:46
interviews, reference checks, and just evaluate
41:49
a person. You mentioned some of your
41:51
favorite questions to like ask on the long walk,
41:53
ask about their family, ask about how they make
41:55
decisions, all these sorts of things. There was a
41:57
couple in there that I'd love to actually take
41:59
and turn on. you. And one of them, which
42:01
I liked a lot was if you had this
42:03
magic wand that you could change something about yourself,
42:05
what you would change. I'm curious what your answer
42:07
is to that specific question. I
42:10
was afraid you're going to do this and I really
42:12
should have prepared to answer my own questions because I
42:14
haven't. But I
42:17
have always admired, I
42:20
guess it would be overly simplistic to say
42:22
extroverts. People who are
42:24
charming can command a room, natural
42:27
networkers are part of Carl Eschenbach
42:29
is a good example. Carl
42:31
is now spending 99.9% of his time as CEO
42:34
of Workday, but he still helps us
42:36
out on stuff from time to time. And Carl is
42:39
here at Sequoia for about seven years. But when he
42:41
walks into a room, it's as
42:43
if there's this hushed voice that follows him
42:45
just whispering, and it just
42:48
the presence that he has is unbelievable. And
42:50
his wife's the same as what he has the
42:52
same way. Carl and Anna just they light
42:54
up a room. Actually, my wife
42:56
is like that she lights up a room as well. I don't
42:59
light up a room.
43:02
I'm probably hiding in the corner hoping that somebody I
43:04
already know comes over and talks to me so I'm
43:06
not forced to go network. And
43:09
that's something that I have worked on over
43:12
and over and over again and forced myself
43:14
into awkward situations over and over and over
43:16
again. And as much as I've done it,
43:19
it's still not comfortable. I'm still not good
43:21
at it. And so that's probably the magic
43:23
wand is to make myself a little more
43:25
extroverted, a little more charming, a little more
43:27
able to light up the room the way
43:29
that some other people can. Well,
43:31
we have to lead into our strengths. And you
43:33
said earlier that you had this brute force mentality.
43:35
And I love this line about keeping
43:38
going until nothing surprises you when you're investigating
43:40
a person doing reference checks, I guess maybe
43:42
even investigating a business, maybe say a little
43:44
bit more about what that actually takes
43:47
and means to keep surprises
43:49
you. I just think it's like a really nice heuristic. We
43:52
get the question a lot from founders. What
43:55
are you looking for? And
43:57
what they want is a simple answer. They want
43:59
to hear where we're
44:01
looking for your first 10 POCs to convert,
44:04
or we're looking for X million of revenue
44:06
growing Y percent year over year. That's the
44:08
sort of stuff that they want. The answer
44:10
that I give them is a very frustrating
44:13
answer, but it's the real answer, which is
44:15
the thing that we're looking for is not
44:17
perfection. The thing that we're looking for is
44:19
clarity. Whatever story
44:21
you tell needs to be internally
44:23
consistent. Whatever evidence is
44:25
available to support that story needs
44:28
to support the story. It can't be disconfirming
44:30
with important aspects of the story. I mean,
44:33
it would be snowflake and zoom. So
44:35
we were lucky enough to get into business with
44:37
Erica zoom when the company was close to a
44:39
hundred million of revenue. It was probably 85, 90
44:41
in that neighborhood. And at that
44:43
scale, zoom was already 80% plus gross margin.
44:45
And as much as they were trying to
44:47
hire more people and burn cash, they just
44:49
couldn't do it. The money from customers was
44:51
coming in too fast. And so
44:54
they were trying to burn cash, but they kept
44:56
generating cash every single quarter. And so
44:58
zoom had perfect margins across the
45:00
board, exponential growth, snowflake. We're
45:02
lucky to get into business with when they were
45:05
just shy of 50 million of ARR. So
45:07
not quite the same scale, but similar snowflake
45:10
would have been more of the workday snowflake
45:12
at that time had maybe 50% gross margins
45:15
and burning a whole ton of cash. And
45:17
when we went from our first investment in snowflake
45:19
to our second investment in snowflake, which
45:21
is only about six months later, what
45:23
triggered it was bad news. So
45:26
we do these semi annual portfolio reviews where our
45:28
portfolio companies send us a bunch of information. The
45:31
information comes in from snowflake. They're
45:33
behind the revenue plan. Gross margins are
45:36
worse than expected free cash flow is
45:38
worse than expected. And we
45:40
got that and said, Oh man, I'm in big trouble
45:42
here. Maybe we shouldn't make this investment. And
45:45
so I got some time with Brad Floring
45:47
who is the still the VP of FPNA
45:49
at snowflake and asked him to
45:51
walk me through what was going on in the numbers. And
45:54
it turns out it was all good news. It
45:56
turns out the reason it was all good
45:59
news, which is hard to appreciate. Why was
46:01
revenue behind? Well, revenue is behind because they're
46:03
landing much bigger deals than they were expecting
46:05
to land, and those much
46:07
bigger deals take more time to ramp up.
46:09
It's the ramping up of that stuff that determines the
46:11
revenue, not the landing of the deals. Why
46:14
is the gross margin behind? Well, the gross
46:16
margin is behind because they're getting pulled globally
46:18
faster than they expected, which means opening up
46:20
availability zones in regions that are going to
46:22
be underutilized around the world, which means that
46:25
COGS utilization is not as high as you
46:27
might expect it to be. The
46:29
gross margins are also down because they're starting
46:31
to get asked to do these full-scale teradata
46:33
replacements in the enterprise far earlier than they
46:36
had anticipated, which means snapping up on professional
46:38
services, which goes into COGS. And
46:40
then operating margins are down because it turns out
46:42
that the per rep productivity was much higher than
46:44
they were expecting, and so they were loading up
46:47
on the sales organization because the reps were just
46:49
producing way faster and at a way higher level
46:51
than they were expecting. And so all
46:53
of the reason that the numbers were
46:55
bad turned out to be good
46:57
reasons. In that situation, we ended
47:00
up having clarity on why the numbers were
47:02
the numbers and what was actually happening in
47:05
the business. And as a result, we went
47:07
from an initial $15 million investment to another
47:09
$200 million investment a couple
47:11
months later. Anyway, the moral
47:13
of the story is you just keep asking questions until
47:15
the picture that's in your head becomes clear, and it
47:18
doesn't have to be perfect. It does have to
47:20
be clear. If it's not clear, you're
47:22
not going to have a good understanding of what risks
47:24
you are taking and what return you can expect in
47:26
exchange. Do you have a specific
47:28
goal in your mind when
47:30
you're doing a reference check, that specific
47:33
unit of investigation? Yes.
47:35
If it's a reference check on a person, I
47:38
want to understand the vector that is that
47:40
human being. We're lucky
47:42
to be surrounded by people who are pretty
47:44
creative and good thinkers. And so this is
47:46
a framework that we learned from Elon Musk
47:49
about the output of an organization is the
47:51
vector sum of its individuals. And
47:53
the point is that a vector has both magnitude and direction.
47:55
So you want to hire people with high magnitude, but then
47:57
you have to make sure they're all pointing in the same
47:59
direction. And so the thing that I
48:02
want to get out of a reference check is the magnitude
48:04
and the direction. And the magnitude
48:06
is almost a top grading exercise. How
48:09
good has this person been at each step of their
48:11
life? Were they the best person in their high school?
48:13
Were they the best person in their college? Were they
48:15
the best person in job number one? Were they the
48:17
best person in job number two? And
48:19
that's a fairly blunt way to look at it. There
48:21
are plenty of dimensions beyond did you have the
48:24
highest GPA or did you have the best performance
48:26
reviews that indicate exceptional
48:28
performance. And so this is in part
48:30
where the direction component comes in. Maybe
48:32
they were not the best person in their high school
48:34
because the thing they cared most about in high school
48:37
was building businesses on the side. And maybe they were
48:39
incredibly successful in that endeavor. Or maybe they just fell
48:41
in love with coding and it turned out that they
48:43
were a phenom in the open source world while they
48:46
were failing in their history class. And
48:48
so understanding what they actually care about and
48:50
trying to figure out whether they've been exceptional
48:53
at the things they really care about, that
48:55
defines the direction of the vector and the magnitude of
48:57
the vector. And that's the thing I'm trying to suss
48:59
out. I know you studied physics
49:01
in undergrad. The vector thing reminded me. And
49:04
there's a lot of physics envy in investing. A
49:06
want for formulas and variables. We've even
49:09
done it a little bit today. Here's
49:11
our three things. Where
49:13
does the physics background help you and
49:15
where does physics in general should be
49:17
left behind or fall short in the
49:19
world of investing? My
49:21
Catholic guilt compels me to knowledge that I
49:24
was only a physics major for two years
49:26
of college. I love switching to economics and
49:28
finance with a concentration in math. And so
49:31
I just want to be clear, but I
49:33
appreciate it. One way that
49:35
I think about this, I'll use the analogy
49:38
of the two critical ingredients in
49:41
an investment are the people in the market.
49:44
The market determines how big the company can
49:46
get and the people determine how big the
49:48
company will get. And I
49:51
think similarly, when you approach things
49:54
with the view of physics, you're
49:56
sort of understanding the rules of the system.
50:00
the individual agents who are operating in
50:02
that system that will
50:04
ultimately determine the outcome. And
50:06
so I think the physics point of view
50:08
is very complimentary to a much
50:10
more human point of view. And
50:13
if you can get both and you understand
50:15
the system level dynamics, but then you also
50:17
understand the individual actors within that system, that's
50:19
where I think you end up with the
50:21
highest likelihood of making a good decision. Could
50:24
we apply all of that to the world
50:27
of AI today, both the market and the
50:29
sorts of people that you're beginning
50:31
to see thrive in a frontier,
50:33
in a Wild West feeling part
50:36
of the world and incredibly exciting,
50:38
but uncertain new technology. I would
50:40
love you to just frame up first how you think
50:42
about the market and your shorthand
50:44
for what the opportunity is here and how to
50:47
think about it. And then I'd love to talk
50:49
about some more specifics, but maybe just starting broad
50:51
strokes, what has you excited? What has
50:53
you pausing? What has you most interested in the
50:55
world of AI? As context
50:57
for this, my first 10 years or
51:00
so at Sequoia were basically
51:02
focused on the cloud transition and
51:04
it was 2007 to 2017,
51:06
which is pretty wonderful time to have that as a focus. Come
51:10
2017 ish, it
51:12
felt like the vast majority
51:15
of first-class market opportunities in the world
51:17
of cloud software had already been occupied.
51:20
If you went to the other major tectonic shift that
51:22
was happening at the time, which was mobile, the top
51:24
10 apps in the app store had
51:26
been pretty static for a number of years. And
51:29
so it felt like we were getting fairly late cycle
51:33
as far as these technology platform shifts go. A
51:35
bunch of us here started trying to think
51:37
about what the next major platform shift might
51:40
be. And at the time,
51:42
our shorthand was data. And
51:44
the reason our shorthand was data is
51:46
because we just observed that the best
51:49
application experiences we were seeing tended
51:51
to be fueled by a pretty healthy dose of machine
51:53
learning. And so it felt like
51:55
the companies that were making use of all the
51:57
data that was available were just creating better experiences.
52:00
and creating better businesses than companies that were
52:02
not. And so we had this loose
52:04
hypothesis that the next major platform shift was going to
52:06
be something related to data. That was
52:09
part of what informed the Snowflake investment. That was
52:11
certainly what informed the Confluent investment, the DBT investment,
52:13
some of the other things in the modern data
52:15
stack. But it also led us
52:17
toward natural language processing, natural language understanding,
52:20
which is almost a predecessor term for
52:22
what we now think of as LLMs.
52:25
It led us to Hug and Face, it led us
52:27
to OpenAI, it eventually led us to a bunch of
52:30
different application or developer companies around that whole theme. But
52:33
I mention that because the thinking that's gone
52:35
into the AI theme for us really
52:38
began in earnest many years ago when we
52:40
were seeing the maturation of the cloud and
52:42
mobile cycle and trying to figure out what
52:44
might be next. I'd
52:47
say the reason that we have conviction in the AI
52:49
theme and sort of what it is we've actually thought
52:51
about kind of has to do
52:53
with the precedent conditions coupled
52:55
with just what we're observing in the environment.
52:57
And when I talk about the precedent conditions,
53:01
the idea of a neural net has been around for
53:04
literally 70 or 80 years, but
53:06
it hasn't been possible given
53:09
computes, given bandwidth, given
53:11
data, given talent. It
53:14
hasn't been possible to put it into practice the way
53:16
it is today until very recently. And
53:19
the major accelerant, of course, was the
53:21
release of chat GPT, which
53:23
we think will end up being this generation's
53:25
Netscape moment. You know, if you go back
53:28
to 1996 when the browser first came out
53:30
of Netscape, that opened the eyes of
53:32
everybody to the power of the Internet. I
53:34
think similarly when chat GPT came out in the fall of
53:36
2022, it opened
53:38
everybody's eyes to what was going on with LLMs
53:41
or AI more broadly and gave
53:43
people a visceral sense for what could be done. And so
53:46
that was sort of a step function increase
53:49
in the activity in this area. Earlier
53:51
that summer, we had stable diffusion. If you
53:53
remember, summer of 2022, stable diffusion came out.
53:56
All of a sudden, people are creating these fantastical images
53:58
and sending them around on Twitter. That
54:00
took the AI market from researchers
54:02
to researchers plus machine learning engineers.
54:05
When chat GPT came out that fall, it
54:07
was another step function increase and the people
54:09
were paying attention and it went from ML
54:11
engineers to all engineers, product managers, founders, consumers,
54:14
boardrooms of Fortune 500 companies. And
54:18
all that energy that's been focused on this
54:20
has started to lead to some pretty interesting
54:22
applications. How big do you think
54:24
this would be if I froze
54:26
the current frontier model
54:29
capabilities? If I said we're never
54:31
going to get anything better than GPT 4.0 or
54:33
Claude 3.0 or 4.0, the
54:37
best ones that are out there today, do you
54:39
think that it's still exciting
54:41
or is most of the
54:44
excitement dependent on continued and
54:46
successful scaling of the
54:48
quality and the capability of these things? I
54:51
think that's a fantastic question and we think
54:53
about that a lot. And the short
54:55
answer is I think if
54:58
you froze capabilities today and the
55:02
only thing that you invested in was
55:04
optimization, making it cheaper, making it faster,
55:06
making it easier. If
55:08
you did that, you would revolutionize almost
55:11
every industry on earth. I
55:13
think the capabilities that exist today
55:16
are so unbelievably powerful and have
55:18
only just begun to be harnessed.
55:21
There is an interesting question. If
55:23
you were Sam Altman, what would you do? I
55:26
think what he's doing currently is probably
55:29
the right move, which is let's
55:31
continue to be at the very bleeding
55:33
edge. Let's continue to produce the very
55:35
best models. And because we
55:37
have an advantage in aggregating capital on talent,
55:40
let's press that advantage and use it to
55:42
stay on the absolute bleeding edge. There's
55:45
an alternative version of the world. And I'm not
55:47
recommending this, but I'm saying it's possible. There's
55:50
an alternative version of the world where you say,
55:52
okay, we think we're starting
55:54
to see diminishing returns to scale, which
55:57
means maybe we've squeezed about as much
55:59
juice this architecture as we can squeeze,
56:02
which means we're going to change our attention
56:05
to a few other things. Number one, we're
56:07
going to have a small team of true
56:09
geniuses trying to figure out new architectures. Number
56:12
two, we're going to move
56:14
some of the compute from training into
56:16
inference. And so instead of spending a
56:18
bunch of compute building the model, we're
56:20
going to spend a little bit more running the
56:22
model, which is what people talk about when they
56:24
talk about planning and reasoning, which
56:26
basically means that the model can do more
56:29
sophisticated things when you are asking it questions.
56:32
And then the third thing that you could do,
56:34
if you believed that returns to scale were starting
56:36
to diminish, the third thing that you could do
56:38
is just straight up optimization and try to make
56:40
it fast, try to make it cheap, try to
56:42
make it easy, and in doing
56:44
so, just run away with the developer ecosystem.
56:46
So I think you could do that, and
56:49
you would have a cash generative business overnight.
56:52
You would still have a dominant market position, but you
56:54
would be taking the risk that
56:56
we have not started to see diminishing returns to
56:58
scale, or that those returns have not diminished to
57:00
the point where it invalidates the investment. So it's
57:03
an interesting alternative version of the AI world to think about,
57:05
but it's not the one that we're living in. How
57:08
would you handicap the scale question about
57:10
whether or not we've hit
57:12
that scale wall, whether we can come
57:14
up with creative ways of gathering more
57:17
novel data, or just other means
57:19
of breaking through this? The bitter lesson seems to be
57:21
the most interesting written piece about this, that you just
57:23
need more data and more scale of data, and the
57:25
thing will keep getting better. But we've used all the
57:28
data we have on the internet, or in the written
57:30
word, or whatever. So what do you
57:32
think the odds are that Wall exists versus us
57:34
just finding a way, because we're humans, and this
57:36
is our way to push through it? I
57:39
think people will always find ways to push through
57:41
it. One of the data points that I find
57:43
interesting, and different people have framed this in different
57:45
ways, there's a guy named John Carmack, who
57:48
may be the world's greatest living engineer. Some people think
57:50
that he is. And he
57:53
has sequestered himself with a couple
57:55
of other geniuses in the middle
57:57
of nowhere, reading old
57:59
research. papers trying to figure out if
58:01
the better architecture already exists.
58:04
It just hasn't been assembled in just the
58:06
right way. And the reason he's doing that
58:08
is because state-of-the-art LLMs
58:11
are about four orders of magnitude less
58:13
efficient than your brain. If you
58:16
think about the basic input as energy
58:18
and the basic output as computation, your
58:21
brain is 10,000 times more efficient
58:24
than a state-of-the-art LLM. Now,
58:26
Andre Carpathi has actually had
58:29
the same observation, but he thinks that it's six
58:31
orders of magnitude. So one
58:33
way or the other, these models are
58:35
dramatically less efficient than the human brain.
58:37
And so the reason that's important is that
58:39
nature has shown us that a better
58:41
architecture exists. And I have
58:44
to imagine long before we
58:47
get to anything that is universally agreed
58:49
upon as AGI, we're going to end
58:51
up with just a dramatically better architecture
58:53
that's going to be far more energy
58:55
efficient. It's not going to come
58:57
out of just optimizing transformers. It's probably not
58:59
going to come by just putting planning and
59:02
reasoning on top. It's probably going to be
59:04
some different base architecture. And I'm sure at
59:06
some point somebody will figure out what that
59:08
is. This feels like the
59:10
thing you said earlier, where if you bet on
59:12
the system, you get the two and a half
59:14
X Airbnb. If you bet on human ingenuity and
59:16
the collective founder of humanity, you get the $100
59:18
billion Airbnb or something. Robbie
59:22
said, do you explain this to him using a
59:24
unit called math percido? What does that mean? Yeah,
59:29
yeah, we're talking about this. And I don't know if
59:31
I've done a good job of this today, but one
59:33
of the things that I tend to be known for
59:35
here is trying to make things as simple as humanly
59:37
possible. Maths percido,
59:40
if the input is energy, one
59:43
fairly efficient source of energy for
59:45
human beings is a bag of
59:47
Cheetos. The output is computation. Math
59:50
is a form of computation. And
59:52
so Ravi and I were saying that the metric for
59:54
the efficiency of an LLM should be maths per cheeto.
59:57
And then at the moment, humans can do
59:59
way more math. for Cheeto than your best
1:00:01
LOM. Maybe use Harvey as
1:00:03
an example of, okay, let's just zoom. All this stuff
1:00:05
is exciting and I hope we break through all these
1:00:08
walls and all the ways you described, and that would
1:00:10
be so cool. But in the version of the world
1:00:12
where we just have what we have today and you
1:00:14
just got to build a useful application and a business
1:00:16
on top of it, Harvey seems like
1:00:18
a great example to just double click on, describe
1:00:20
what it does and how it's using the model.
1:00:23
And I would love you to just explain
1:00:25
what you've seen so far, what lessons that
1:00:28
business and product has taught you just to
1:00:30
zoom in on a real tangible example. Yeah.
1:00:32
So I think Harvey's a good example because
1:00:35
I think it is the first
1:00:38
and best example of a new
1:00:40
wave of application companies that will
1:00:42
come out of this AI tectonic
1:00:45
shift. They got started at
1:00:47
exactly the right time. So they were the very
1:00:49
first company to get access to GPT-4 to start
1:00:51
building on top of it. And
1:00:53
the founders come from the perfect
1:00:56
background. Winston comes from the world
1:00:58
of law. Gabe comes from AI
1:01:00
research and said together, they
1:01:03
understand both the problem and the solution.
1:01:05
And so it's a great example of founder market fit.
1:01:07
It's a great example of the why now and being
1:01:09
in exactly the right place at the right time. And
1:01:12
what they've built over the last 18 months
1:01:15
or so is the very
1:01:17
best legal assistant.
1:01:21
So it is not an AI lawyer
1:01:23
at the moment, it is a legal
1:01:25
assistant that can basically
1:01:28
do the work of a first year associate
1:01:30
at a big law firm. And
1:01:33
partners at big law firms have actually
1:01:35
AB tested the Harvey assistant versus
1:01:37
the associate. And the Harvey
1:01:40
assistant is just as good and immediate.
1:01:42
And so the task that might have taken six hours
1:01:45
instead of takes six seconds. And
1:01:47
so it's a pretty darn powerful assistant. The
1:01:50
ambition for the company is to
1:01:52
eventually use that to
1:01:55
democratize the world of law. If
1:01:57
you think about the legal world today,
1:01:59
it's a real person product. It's
1:02:02
very expensive and so
1:02:04
whether you're a company or an individual,
1:02:07
either you have a lot of money to spend on it
1:02:09
or you're probably not going to get a very good legal
1:02:11
service or you're probably not going to get any legal service.
1:02:14
And it turns out that with AI and the fullness
1:02:16
of time, we can provide world-class
1:02:19
legal services and we can
1:02:21
do it at a tiny fraction of the price. The
1:02:24
idea for Harvey is never to replace the
1:02:26
human beings, it's to dramatically
1:02:28
expand the market to a whole bunch of
1:02:30
people who don't have access to legal services
1:02:32
today. And so for the very
1:02:34
high-end law firms and that sort of thing, we're going
1:02:37
to be an assistant. For the rest
1:02:39
of the world, we're going to be the service hopefully.
1:02:42
And if we can pull that off,
1:02:44
I think it has a chance to be an incredibly important
1:02:46
company. When you're evaluating one of these
1:02:48
products because there's just not that many of them that
1:02:50
are built and fleshed out the stuffs a year old,
1:02:53
what are the things you're looking for that
1:02:55
are distinct from the same things you might look
1:02:57
for in what I'll call a non-AI product? Or
1:02:59
is it just all the same stuff that it
1:03:01
just solves a problem efficiently and elegantly
1:03:03
and it's just the same that there's just something
1:03:05
different under the hood? Unique
1:03:08
and compelling value problems. It comes
1:03:10
back to that. Market by market, there
1:03:12
are different pros and cons to
1:03:15
the different technical approaches, but
1:03:17
at the end of the day, the
1:03:19
technical approach only matters to the extent
1:03:21
it does something unique and compelling for
1:03:24
the customer. And so
1:03:26
we try to spend less time
1:03:28
underwriting the architecture and
1:03:30
more time underwriting the customer
1:03:32
and just really understanding what problem this solves
1:03:34
to them, why it's unique and compelling, how durable it
1:03:37
is, how else they might solve that problem, where they
1:03:39
see it going in the fullness of time, all that
1:03:41
good stuff. I actually think a mistake that
1:03:43
a lot of investors make, there are
1:03:45
a lot of investors who are very technical and strongly
1:03:48
weight their personal
1:03:50
opinions of the architecture of the
1:03:52
product and that's a useful
1:03:54
input. You just have to weight it appropriately because
1:03:57
again it's only as good as its impact on
1:03:59
the customer. And so we try
1:04:01
to be more customer-oriented and let's tack out
1:04:03
more customer back. In
1:04:05
the world of venture, it seems like
1:04:07
there is this almost magic pixie dust
1:04:09
that certain firms have that founders
1:04:11
seek out. There's a handful. We could probably name
1:04:14
them on the call pretty easily together and everyone's
1:04:16
guess that these names would be the same. Squyze
1:04:18
is certainly one of them. When
1:04:20
it comes to the maintenance of that magic
1:04:22
pixie dust that a few firms seem to
1:04:24
have where the winners keep winning and winning
1:04:27
begets itself because that brand grows and the
1:04:29
reputation grows, the role of the platform, the
1:04:31
way you talk about Sequoia's platform seems to
1:04:33
play a key role in the
1:04:35
odds that that pixie dust will persist
1:04:37
into the future. Can you describe the
1:04:40
platform strategy to building an investment firm
1:04:42
like Sequoia in a way that
1:04:44
maybe others building investment firms might be able to borrow
1:04:46
some of those concepts that have been effective for you?
1:04:49
And first, we'll define what we mean by
1:04:51
platform. So when I joined Sequoia, we had
1:04:53
14 people on the
1:04:55
investment team and
1:04:57
two people who I would call front
1:04:59
office operators. We had one person
1:05:01
in talent, one person in marketing. So we had
1:05:03
14 and two. If you
1:05:05
fast forward to today, we have 27
1:05:07
people on the investment team and
1:05:11
probably about 65 people
1:05:14
who I'd say are front office operators,
1:05:16
meaning marketing, talent,
1:05:18
engineering, product, data science,
1:05:21
design, and a
1:05:23
handful of other things, customer partnerships. And
1:05:27
so that group of operators is
1:05:29
really what we mean by the platform. That's the
1:05:31
bulk of what we mean by the platform. There
1:05:34
are two key advantages we get out
1:05:36
of that group. One is they
1:05:39
dramatically amplify the efforts of the investment
1:05:41
team. So one
1:05:45
concrete example of that is the amount
1:05:47
of information that we have. We have
1:05:49
a homegrown CRM system powered by a
1:05:51
homegrown data science system. The
1:05:53
information that we have available in that system
1:05:55
for a company that we've never met is
1:05:58
more than the information. we would have
1:06:00
had on the same company 15 years ago at the
1:06:03
time of making a final investment decision. And
1:06:07
so that's a massive amplification
1:06:09
of our ability to source and pick and work
1:06:12
things through the funnel that leads to an investment.
1:06:14
So that's one concrete example. So one thing that
1:06:16
we get out of the platform is an amplification
1:06:18
of our efforts as investors. The
1:06:20
second thing that we get out of the
1:06:23
platform is advantages that have a chance to
1:06:25
compound over time. So historically,
1:06:27
the only compounding advantage that you get
1:06:29
in a venture capital business is
1:06:32
your brand and your
1:06:34
culture and your network.
1:06:37
But all those things are somewhat ephemeral. One
1:06:39
bad decision can tarnish your brand. The
1:06:41
platform team is building things that
1:06:43
can compound over time. And one tangible example of
1:06:45
that is we have a clever way that
1:06:49
one of our talent partners came up with
1:06:51
to collect signals on people. We
1:06:54
now have a couple hundred thousand
1:06:56
people in our database on
1:06:59
which we've collected these proprietary signals that
1:07:01
are not available anywhere else. And
1:07:03
so our ability to take a look
1:07:05
at a company and pretty quickly get a good
1:07:07
sense for the talents inside the building and how
1:07:09
well they've hired based on the signals that are
1:07:11
already in our system. That's an advantage
1:07:13
that's going to just keep on compounding. We have a
1:07:15
couple hundred thousand today. Over time,
1:07:17
theoretically, we could have just about everybody
1:07:19
in the technology world in that database.
1:07:22
One other point that's worth mentioning here. The reason
1:07:25
we decided to invest in our platform has
1:07:27
to do with what we saw happening outside
1:07:29
the building and a strategic choice that we
1:07:32
made. So what we saw happening
1:07:34
outside the building was the democratization of
1:07:36
the means of production. And
1:07:38
what I mean by that is any founder
1:07:40
anywhere can now go online
1:07:42
to educate themselves about the basics
1:07:44
of technology and building a business
1:07:47
and become an internet entrepreneur overnight.
1:07:49
And as a result, the volume,
1:07:52
variety, and velocity of startups
1:07:54
has increased dramatically. But if
1:07:56
we were still just 14 people or today 27
1:07:58
investors trended. do our jobs, we
1:08:01
wouldn't be able to cover the universe of opportunities. We
1:08:03
wouldn't be able to make our way through them efficiently.
1:08:06
And so the strategic choice that we made was we could
1:08:08
have taken the path of, okay, well, let's
1:08:10
not have 27 investors. Let's have 270 investors. If
1:08:14
we have a big team, we can cover everything. And
1:08:17
the reason we specifically decided not to do that
1:08:20
is because at the end of the day, there are
1:08:22
only two things that you need a
1:08:24
human being to do in the world of investing.
1:08:27
Everything else can be automated, but the two things that you have
1:08:29
to have a human being do, number one,
1:08:32
build the relationship with the founder. And
1:08:35
number two, make the decision.
1:08:39
Doesn't matter how many inputs you have, somebody
1:08:41
has to take those inputs and make the
1:08:43
decision. The declarative statement, we should invest because.
1:08:46
And so if the two things that we have to have
1:08:48
human beings do are a relationship with a founder and
1:08:50
make the decision, if we
1:08:52
disperse the knowledge and experience of the
1:08:54
partnership across a couple of hundred people,
1:08:57
any given one of them is not going to be all that special. If
1:09:01
we concentrate the knowledge and the experience
1:09:03
of the partnership on the
1:09:05
smallest possible number of people, we
1:09:08
have a chance for each one of those people to
1:09:10
grow into something really special. And
1:09:13
if we're hunting outlier founders, they
1:09:15
don't want to deal with people who are just okay. They
1:09:18
want to deal with people who are outliers themselves. And
1:09:21
if we can hire people who already have
1:09:23
outlier characteristics and then
1:09:25
supercharge them with concentrated experience
1:09:27
and knowledge, we have a chance
1:09:30
to produce the next Doug Leone, the
1:09:32
next Rulak Bota, the next Alfred Lin.
1:09:35
What Mike Moritz story most
1:09:38
stands out in your memory where the story taught
1:09:41
you something interesting. April, 2010.
1:09:45
You answer these crazy facets, very impressive. Well,
1:09:47
this line does stand out in my hands.
1:09:51
April of 2010, we had this
1:09:53
partner named Chris Olson who
1:09:55
found at that time a young man
1:09:57
named Sebastian over in South Africa.
1:10:00
Stockholm and Chris
1:10:02
built a relationship with Sebastian and
1:10:05
started to fall in love with a company named Klarna.
1:10:08
And Chris asked Michael Moritz to parachute in
1:10:10
to help him win this competitive investment. At
1:10:12
that time, it was already a big deal
1:10:14
in Europe and it was competitive.
1:10:17
And so Chris and Michael Moritz end
1:10:19
up securing the opportunity to invest in
1:10:21
Klarna. And then they bring me
1:10:24
along for a week to try to do all
1:10:26
the diligence and meet the team and polish up
1:10:28
the final investment recommendation. So I'm in Stockholm with
1:10:30
Michael Moritz and Chris Olsen for a week. We're
1:10:32
spending all day at the company and
1:10:35
Michael is not set to work. Chris
1:10:37
is leading the conversation. I'm chipping in from
1:10:39
time to time and Michael is just sitting
1:10:41
there silent, just listening, taking it all in.
1:10:44
And Chris and I are desperate to know what he's thinking,
1:10:47
particularly because asking him to spend a week in Stockholm is kind
1:10:49
of a big ask and we wanna make sure we're not wasting
1:10:51
his time. So we finally get to dinner
1:10:54
on night two or three. And
1:10:57
at that point, Chris and I thought that the
1:10:59
major issues in the investment were things like what's
1:11:01
gonna happen with interest rates because remember this is
1:11:03
a bank with a balance sheet in the wake
1:11:05
of the global financial crisis. We're very concerned about
1:11:07
what's gonna happen with interest rates. We were concerned
1:11:09
about whether or not they'd ever be able to
1:11:11
make it into Germany. At that time, they had
1:11:13
a pretty strong position in the Nordics and Germany
1:11:15
was the big market that they were trying to
1:11:17
enter. And we go to dinner with
1:11:20
Michael Moritz. Chris works up the nerve to say, okay, what
1:11:22
did you think? In Moritz
1:11:24
in typical Mauritsian fashion and
1:11:26
there's an exhale and a long pause and
1:11:30
he says, the question is
1:11:33
whether they can get to
1:11:36
a few hundred million of net
1:11:38
income and
1:11:40
the answer will come down
1:11:42
to the strength of the engineering team. Now,
1:11:46
Chris and I, I don't know if we were showing a
1:11:49
few hundred million of revenue in the model that we had
1:11:51
built and we
1:11:53
certainly had not asked that many questions about the strength
1:11:55
of the engineering team. And
1:11:57
it turned out that Michael...
1:12:00
was exactly correct. And if you look at
1:12:02
the company today, it's an absolute behemoth. And
1:12:06
the strength of the engineering team was
1:12:09
so critical because the value
1:12:11
prop for this product was very
1:12:13
strong for merchants and
1:12:16
very strong for consumers. So
1:12:18
it was a no-brainer, except it
1:12:20
was a pain in the butt to implement. And
1:12:23
so the key was gonna be, could
1:12:25
you deal with the complexity of all
1:12:27
the different e-commerce systems and all the
1:12:30
different payment mechanisms and all the different
1:12:32
preferences of the customers? Could
1:12:34
you deal with that complexity in an
1:12:37
elegant way that is product
1:12:39
driven, not brute force
1:12:41
driven, to reduce the friction
1:12:44
for people to deploy this product? And
1:12:47
if you could, you are gonna
1:12:49
become ubiquitous. And if you
1:12:51
couldn't, you weren't. And it was gonna come down to
1:12:53
the strength of the engineering team. That
1:12:55
was a lesson for me because Chris and I, we
1:12:58
had planned for this trip. We had our long list
1:13:00
of all the different questions we wanted to ask. We
1:13:02
were frantically screwing about trying to do all of our
1:13:04
work on the investment. And Michael
1:13:06
got it down to the very simplest possible thing,
1:13:08
which turned out to be exactly the right thing.
1:13:10
And so I guess the lesson you've got to
1:13:12
do was, you gotta zoom out
1:13:15
and make sure that you're operating at the
1:13:17
right level of ambition. That was just 300
1:13:19
million net income thing. And
1:13:21
that you're actually focused on the first order issue,
1:13:23
which is the strength of the engineering team thing.
1:13:26
When you think about the sensations in
1:13:28
both your body and your mind of
1:13:30
the feeling of being desperate
1:13:33
to win, how would you describe
1:13:35
what that feels like? It's
1:13:37
funny you ask this because I've been concerned that
1:13:40
as I've gotten older, I've lost some of the
1:13:42
edge or some of the killer instinct. And then
1:13:44
I was comforted by the camping trip that we
1:13:46
went on with our founders a couple of weeks
1:13:48
ago, where I felt like
1:13:50
glimmers of it were still there. And they
1:13:52
showed up in the silliest possible way, which
1:13:54
we did this set of activities, one
1:13:57
of which was axe throwing. And
1:13:59
you're on a, a clock and you had to
1:14:01
get as many bullseyes as possible before the
1:14:03
time expired. And I was
1:14:05
on a team with a couple of our founders and
1:14:08
we realized that one of them was better than the
1:14:10
rest of us at throwing the axe. And so we
1:14:12
ended up doing division of labor where my job was
1:14:15
to sprint and retrieve the axes that
1:14:17
had been thrown. And his
1:14:19
job was to keep throwing them. And
1:14:21
at one point he got a few bullseyes in
1:14:24
a row and he turned around and raised his
1:14:26
arms in victory. And there are still a minute
1:14:28
or two left on the clock. And like a
1:14:30
crazy person, I ran back to him yelling, no,
1:14:32
no, no, there's more time. Keep throwing, keep throwing,
1:14:34
keep throwing. So I guess the feeling
1:14:37
is you get a little bit carried away with
1:14:39
yourself. And actually, I think this is one of
1:14:41
the things that makes Doug Leone so special because
1:14:43
he lives his entire life this way. You go
1:14:45
into a mode where you are purely
1:14:47
driven by the objective function, whatever
1:14:50
the thing is that you are trying to achieve.
1:14:53
That is the only thing that you can think about and
1:14:56
nothing else enters your consciousness. And
1:14:58
in that case, I had lost track of
1:15:00
the social graces of yelling at somebody to
1:15:03
put their arms down and throw more axes
1:15:05
because the objective function was the only thing that I
1:15:07
could see. And then the analogy to Doug is one
1:15:09
of the things that I think makes Doug so special.
1:15:12
Anytime you ask him to
1:15:14
do anything, personal discomfort, personal
1:15:18
risk does not
1:15:20
enter his calculation at all. If
1:15:22
it is physically possible for
1:15:25
him to do the thing that
1:15:28
is required to achieve
1:15:30
the mission, he will do it. And
1:15:33
I think about that as it's
1:15:35
the ultimate humility, not caring
1:15:37
about himself, his ego,
1:15:39
his comfort at all. It's
1:15:42
the ultimate service mentality.
1:15:44
It's the ultimate mission
1:15:47
orientation, where the only thing
1:15:49
that you can see is the mission and what needs
1:15:51
to be done to achieve the mission. And everything
1:15:54
else just doesn't register. When
1:15:56
you go into that zone, the only thing that you can
1:15:59
see is the mission. else doesn't register. I
1:16:01
think that's where that killer instinct comes in. If
1:16:04
you ever do retire and you're at a
1:16:06
retirement party, what do you hope people say
1:16:08
about you? I think the
1:16:10
themes have been consistent. I mentioned with
1:16:12
Boston College, there's the Jez and Amato, Men and
1:16:14
Women for Others, and then the DC motto Everto
1:16:16
Excels. You have that concept
1:16:18
of teamwork and that concept of performance. And
1:16:20
at Sequoia, very explicitly, the two things we
1:16:23
care about most are teamwork and performance. So
1:16:26
I think the thing that I would hope
1:16:28
to hear is that I was a top
1:16:30
performer, but also a top teammate. One
1:16:33
of the ways that most manifests on
1:16:36
a day-to-day basis, the performance thing
1:16:38
I feel like we've covered, you're an animal and you
1:16:41
want to win and do whatever it takes. On
1:16:43
the teammate side, where does that
1:16:45
most commonly manifest? And what have you learned about
1:16:48
it 15 plus years into doing
1:16:50
this? Our partner, Andrew Reed,
1:16:52
had this good line the other day, which
1:16:54
was, sometimes you need less leadership and more
1:16:56
leadership. And what he
1:16:58
meant by that was, sometimes
1:17:01
people think that leadership means telling other
1:17:03
people what to do, but
1:17:06
sometimes leadership actually just means doing
1:17:09
the work so that people
1:17:11
can see how it is supposed to be done.
1:17:15
And because we're in an apprenticeship
1:17:17
business, I think a lot
1:17:19
of what we need to do
1:17:21
is to just do the work, just do
1:17:23
the basics of blocking and tackling, doing the
1:17:25
job, and that that's more helpful
1:17:28
to the other people on the team than
1:17:30
any amount of one-on-ones or mentorship
1:17:33
or structured feedback or whatever else.
1:17:36
And so I think what being a good teammate
1:17:38
means, when we construct a
1:17:40
team to go after an investment, there
1:17:43
are two roles. There's the sponsor and there's the
1:17:45
wing person. And the sponsor's
1:17:47
job is to secure the investment
1:17:49
and to make the case internally.
1:17:51
The wing person's job is to support the
1:17:54
sponsor. I don't have to be the sponsor,
1:17:56
I can be the wing person. There can
1:17:58
be an investment that somebody else is sponsoring. And
1:18:00
my role is just to support them. And that might
1:18:02
mean that I'm the one building the financial model and
1:18:05
writing the memo and calling the customers. And they're the
1:18:07
one doing the fun stuff of romancing the founder and
1:18:09
making the case in the partner meeting and that sort
1:18:11
of thing. And so I think
1:18:14
what it means to be a good teammate comes
1:18:16
back to that mission orientation that we were talking
1:18:18
about earlier. Whatever the mission is,
1:18:20
whatever the job is that needs to be done,
1:18:23
just do the job. It doesn't matter what your
1:18:25
specific role is in achieving the mission. It just
1:18:27
matters that we achieve the mission. If
1:18:29
you think about the landscape of this
1:18:32
style of investing, it's mature. There's lots
1:18:34
of firms. When you started, it was
1:18:36
much smaller, both in people, firms, assets,
1:18:39
investments, et cetera. Do
1:18:41
you still think there are open zones
1:18:43
of opportunity to try new
1:18:45
concepts and reinvent the game a little
1:18:47
bit from the investing side? I guess
1:18:49
ask differently. If I forced you to
1:18:51
go start a new firm with none
1:18:53
of the benefits of the existing firm,
1:18:55
just you, how you would
1:18:57
approach that challenge where the goal was to win
1:19:00
and be successful and back great companies. How would
1:19:02
you go to market as a
1:19:04
new investor in this more mature
1:19:06
environment? It's hard for me
1:19:08
to think of anything other than what my wife
1:19:11
Sarah is doing with her firm conviction. Tell me
1:19:13
about it. I actually think what
1:19:15
she's doing is exactly right. And I'll give you
1:19:17
the specific example, but I can also generalize from
1:19:19
there. So Sarah was
1:19:22
a partner at Greylock for about a decade
1:19:24
and then left just under two years ago
1:19:26
to start a new firm, which is called
1:19:28
Conviction Partners. And the
1:19:30
reason she started it was because she
1:19:33
saw this new crop of what she calls
1:19:35
Software 3.0, which is
1:19:37
basically AI driven companies starting to
1:19:39
emerge and wanted to
1:19:42
build a firm that could be built from
1:19:44
the ground up to service that new crop
1:19:46
of entrepreneurs. The thing that
1:19:48
I think is so effective
1:19:50
and so special about what she's doing is
1:19:52
that unlike a lot of people in the
1:19:54
venture capital world, she believes
1:19:57
that being small is a weapon and
1:19:59
that you don't. get advantages out
1:20:01
of scale in the venture capital business,
1:20:03
you get advantages out of quality. So
1:20:06
she kept her first funds much smaller than it
1:20:09
could have been. She's kept her
1:20:11
team much smaller than it could
1:20:13
have been. She's kept her portfolio much smaller
1:20:15
than it could have been. In
1:20:17
each step of the way, she's optimized for quality.
1:20:21
The benefit that you get from
1:20:23
optimizing for quality is that if you
1:20:25
achieve high quality, the growth comes to
1:20:28
you. You look for
1:20:30
growth, you're not likely to reverse engineer quality.
1:20:32
If you look for quality, you're going to have plenty
1:20:34
of choices about how much you want to grow. And
1:20:37
so she's assembled an
1:20:40
exceptional portfolio. She's
1:20:42
now starting to think about fund number two, and it's
1:20:44
going to be the easiest thing in the world to
1:20:46
raise it. She's going to keep
1:20:48
it smaller than it needs to be because
1:20:50
again, she's not optimizing for assets under management.
1:20:52
She's optimizing for quality. And
1:20:54
I think that's a great way to build a business. And
1:20:56
then the way I would generalize
1:20:58
that is not just the point on quality, but
1:21:00
also she is currently known
1:21:03
for one thing, early
1:21:05
stage AI companies, early
1:21:08
defined as series A or earlier,
1:21:10
and AI defined as AI. If
1:21:14
you are an early stage AI company, there's
1:21:16
a pretty decent chance that you're going to think of her.
1:21:20
If you are not an early stage AI company,
1:21:22
there's a pretty decent chance that you're not. I
1:21:25
was on a board with a guy named Jeff Richards from
1:21:27
GGV, and he referred to this as the chicken issue. At
1:21:30
some point, some genius at Chick-fil-A probably said, hey,
1:21:32
if we put burgers on the menu, we can
1:21:34
attract more customers. And somebody else said, yeah, but
1:21:36
the thing we're known for is chicken. If you
1:21:38
want a burger, you should go somewhere else. Most
1:21:41
companies have a chicken issue where they want to do the
1:21:43
chicken and the burger. And just focusing on
1:21:45
the chicken is important. Yeah, it's funny. It's turtles all
1:21:47
the way down. It's probably the same advice for a
1:21:49
new technology software company too. You got to do one
1:21:51
thing and do it really well to get going. And
1:21:54
that just seems to be a universally good advice for
1:21:56
sure. Is there anything else
1:21:58
about your whole world? that
1:22:00
you wish was meaningfully different than it
1:22:02
was like system settings or just ways
1:22:05
of doing things or just norms that
1:22:08
Armed with that magic wand you would change
1:22:10
drastically Yes, I probably
1:22:12
won't articulate this in the best possible way
1:22:16
But my partner rule off expresses
1:22:18
this in a pretty good way where he says
1:22:20
look venture capital is not an asset class what
1:22:23
he means by that is less
1:22:25
than 1% of the
1:22:28
companies that get started End
1:22:30
up accounting for 99%
1:22:33
of the market cap that's created and I don't know if
1:22:35
those are the exact right numbers But it's something like that
1:22:38
and so if you want to approach this as
1:22:40
an asset class and buy an index of all
1:22:43
the startups You're gonna
1:22:45
get drowned out with noise and the vast majority
1:22:47
of those investments are gonna be no good because
1:22:49
the vast majority of those companies Don't
1:22:51
need to exist. They're not solving
1:22:53
an important problem or they're not doing it in
1:22:55
a unique and compelling way And
1:22:58
so when you approach venture capital as an asset class
1:23:01
you end up with companies
1:23:03
that don't need to exist funds that are not
1:23:06
going to perform in people
1:23:09
who are attracted to Maybe
1:23:13
the fame of being a popular
1:23:15
founder maybe the perceived riches of
1:23:17
having a successful exit but
1:23:20
they're not attracted to this necessarily
1:23:22
for the right reasons and And
1:23:25
when I say right reasons everything is relative right
1:23:28
in terms of what I think of as right
1:23:30
it's not objectively, right? It's just my personal opinion
1:23:32
or my personal point of view. But
1:23:34
if you say, okay, well, what are the right reasons? I
1:23:37
think if you are a founder who
1:23:40
really cares about some
1:23:42
problem in the world that is not being
1:23:44
solved in just the right way and You
1:23:47
want to dedicate the next couple decades of
1:23:49
your life to solving that problem? That's
1:23:52
a great reason to go build a business if
1:23:55
you are an investor who believes
1:23:57
that Entrepreneurs
1:24:00
ownership more so than any other force
1:24:03
shapes the future of the
1:24:05
world that we get to live in and
1:24:08
you want to dedicate your life to
1:24:11
serving those entrepreneurs so
1:24:14
that they can realize the maximum
1:24:16
possible impact of their dream,
1:24:19
that's a pretty good reason to be an investor. If
1:24:22
you're an entrepreneur who just wants to get
1:24:25
invited to the fancy parties and conferences and
1:24:27
issue press releases about your latest funding round
1:24:29
and tell your friends you're a unicorn, that's
1:24:31
not a great reason to be a founder.
1:24:35
And if you're an investor who wants
1:24:37
to maximize assets under management so you
1:24:39
can milk the fee stream and spend
1:24:41
your time on Twitter pontificating about the
1:24:43
future direction of AI so that you
1:24:45
can show up in news reports, that's
1:24:48
not a great reason to be an investor. And
1:24:50
so if I had a magic wand and I could change one thing
1:24:52
about the industry, I would try
1:24:55
to slice off some
1:24:58
of the people who are participating without
1:25:00
the most pure motives and
1:25:03
increase the concentration of founders who
1:25:05
really care about their customers and
1:25:08
investors who really care about their founders. Amen.
1:25:11
What does it take to achieve legendary
1:25:13
potential? I love that term
1:25:15
you use all the time. That word is
1:25:17
really important to Sequoia Legendary. It
1:25:19
implies effort and scope of
1:25:22
ambition and all the things that we've talked about
1:25:24
and that you just talked about. I
1:25:26
would love you to sum it all up with what
1:25:28
you think it takes and what you've watched it
1:25:31
take. This isn't theoretical. You've seen it happen many
1:25:33
times. What does it take to achieve that sort
1:25:35
of potential? Why is that so motivating to you?
1:25:38
So we had Max Rhodes, who is the
1:25:41
founder and CEO of Fair at our
1:25:43
offsite maybe last year. And
1:25:47
he particularly in the early days of
1:25:49
Fair was just legendary for
1:25:52
his work ethic and he still is, but he
1:25:54
was very legendary once upon a time. And
1:25:57
so we asked him, what is it that keeps you
1:25:59
going? when all of your
1:26:01
friends are out having fun or when you're exhausted
1:26:03
and frustrated and just want to quit. What is
1:26:05
it that keeps you going? And
1:26:07
he said it was a voice
1:26:09
echoing in the back of his head, happened
1:26:12
to be the voice of our former partner
1:26:14
Michael Moritz, who he once asked
1:26:16
this question, what is it that separates the
1:26:18
truly legendary companies from all the rest? And
1:26:21
the voice was saying, relentless
1:26:24
application of force. And
1:26:29
I think that's it. Of course, the
1:26:31
question is, what is it that causes you to
1:26:34
relentlessly apply the force? And
1:26:37
that gets to the core of who you are and what you
1:26:39
care about and why you're building this company
1:26:41
to begin with. And to my
1:26:43
earlier comments, if your motivation is to release
1:26:46
press releases about how you're now a unicorn,
1:26:49
that's probably not a durable enough
1:26:51
motivation to really keep going
1:26:53
when things get tough. And
1:26:55
your motivation doesn't have to be obsession around
1:26:57
the customer problem. Maybe your motivation is you
1:26:59
just really love building things or
1:27:02
you really prioritize craft and you really want
1:27:04
to build just a beautiful, amazing product that
1:27:06
people are going to love. Or
1:27:09
maybe you got into it for the wrong reasons,
1:27:11
but now that you have a couple hundred employees
1:27:14
and you realize that people are really
1:27:16
counting on you, you feel a
1:27:18
sense of responsibility and you really want to
1:27:20
do right by them. And so there
1:27:22
are plenty of reasons that might cause you to
1:27:25
relentlessly apply that force. But I would say
1:27:27
the thing that probably separates the legendary from
1:27:29
the rest is in fact
1:27:31
the relentless application of force. I
1:27:34
think you might know my traditional closing question for
1:27:36
everybody, which I love and is a very appropriate
1:27:38
one, given a lot of the values you've talked
1:27:40
about and just your own past. What is the
1:27:42
kindest thing that anyone's ever done for you? I
1:27:45
had a hard time coming up with a good answer
1:27:47
to this question, knowing that you're going to be asking
1:27:49
it. And the best thing
1:27:51
that I could come up with was
1:27:53
it's going to sound very generic, but I'll make
1:27:56
it a little bit more specific. And the very generic
1:27:58
form is giving me a chance. The
1:28:00
thing that's a little more
1:28:02
specific is I had this
1:28:04
nice scholarship in college and
1:28:06
I remember the
1:28:09
wife of the director of the scholarship program who
1:28:11
oversaw all of us to make sure we weren't
1:28:13
losing our way and just Incredibly sweet lady. I
1:28:15
remember one time She made
1:28:17
the comment, you know You're
1:28:20
kind of rough around the edges, but you clean
1:28:22
up pretty nice and I think
1:28:24
that's a fair comment. I think Particularly
1:28:27
high school college earlier in my career
1:28:29
rough around the edges might have been
1:28:31
a generous statement I
1:28:33
know Doug refers to the early version of
1:28:35
himself as insufferable. Maybe I was somewhere close
1:28:37
to that I was certainly a bit
1:28:39
prickly and a bit full of myself And
1:28:42
so I guess the kindest thing that anybody's ever done
1:28:44
for me is to see through that and to see
1:28:47
Whatever goodness or whatever positive attributes
1:28:49
might have been hiding inside and
1:28:51
to help those things come out
1:28:53
and flourish over time I'd start
1:28:55
with my parents that were strict
1:28:57
but not hard on me as
1:28:59
a kid and gave me the room to figure
1:29:02
out who I was and then
1:29:04
in college the people who were
1:29:06
kind enough to Give
1:29:08
me a scholarship or spend time with me
1:29:11
my first job and I mentioned John Carroll It's
1:29:14
funny I heard after the fact that there was general agreement
1:29:16
at some of the partners that I should be hired But
1:29:18
no one person actually wanted me on their team And he
1:29:21
was like, okay, I'll take him And
1:29:24
then I think here with Doug as I mentioned, I
1:29:26
was the youngest person we'd ever hired. I was experiment
1:29:28
I was far from perfect and I almost got
1:29:30
fired multiple times after I got here But Doug
1:29:33
was the one who took a risk on hiring
1:29:35
me and Jim Getz was the one
1:29:37
who stood up for me when I was Not doing so
1:29:39
well. Why did you almost get fired?
1:29:41
What was the closest? It wasn't that
1:29:43
I did something egregious It wasn't
1:29:45
a thing that caused me to get fired the
1:29:47
context of me joining We
1:29:49
were just making growth investing a
1:29:51
first-class citizen and I was
1:29:54
hired from Summit Partners Which is really good
1:29:56
growth equity firm and I misunderstood
1:29:58
my job as teach
1:30:00
Sequoia how to invest like Summit. The
1:30:04
thing that I should have done was
1:30:06
understand who Sequoia is, and then extend
1:30:08
Sequoia into growth. When
1:30:10
I first got here, it was like oil and water, where
1:30:12
I was just trying to rinse and repeat with the stuff
1:30:14
that I've earned at Summit, which was not the right stuff
1:30:17
to do as part of Sequoia. I
1:30:19
wasn't learning fast enough. At
1:30:21
one point, apparently, five of six general partners said
1:30:23
that I should be let go, and Jim Guess
1:30:25
was the one who said, over my dead body.
1:30:28
The reason Jim threw his body across the tracks when
1:30:30
everybody else wanted to fire me was, it
1:30:33
was a little bit skills or attributes or
1:30:35
whatever, but the biggest thing was intentions. He
1:30:37
could see that I desperately wanted
1:30:40
to do the right thing. Fascinating.
1:30:43
Pat, this has been a total blast and pleasure. I've learned
1:30:45
a lot. Thank you so much for your time. Awesome.
1:30:48
Thank you. If
1:30:50
you enjoyed this episode, check out joincolossus.com.
1:30:52
There you'll find every episode of this
1:30:54
podcast complete with transcripts, show notes, and
1:30:56
resources to keep learning. You can also
1:30:59
sign up for our newsletter, Colossus Weekly,
1:31:01
where we condense episodes to the big
1:31:03
ideas, quotations, and more, as well as
1:31:05
share the best content we find on
1:31:07
the internet every week. Thanks
1:31:09
for watching. I'll see you next time.
1:31:11
Bye.
Podchaser is the ultimate destination for podcast data, search, and discovery. Learn More