Latest / The Learning Corner by Precursor / Episode #1: Welcome To The Learning Corner

Transcript

Episode #1: Welcome To The Learning Corner · The Learning Corner by Precursor

0:00
Hold on one second.
0:01
Hey, buddy.
0:01
I'm recording a podcast right now.
0:03
I'll come see you before you go to school though.
0:04
Okay?
0:05
I'm recording though.
0:05
We're gonna have to edit this out.
0:06
So
0:08
It's the week of September 23rd.
0:11
Welcome to the Learning Corner, a Precursor Ventures podcast.
0:18
Welcome to the Learning Corner by Precursor, where we talk about the articles and Goodreads
0:25
that caught our teens' attention over the past week.
0:28
I'm Mia Farnham.
0:29
I'm a principal here at Precursor Ventures and I'm joined by our VP and managing partner,
0:35
Charles Hudson.
0:36
Charles, how you doing?
0:37
I'm great.
0:38
I'm excited for Learning Corner.
0:40
I'm excited too.
0:41
It's wildly early.
0:42
Charles's time.
0:43
I've known you for 4 years, which is a 100 years in the venture time warp.
0:47
So I've known that you have 47 hours in the day and everyone else has 24.
0:53
We're excited.
0:54
Some of our initial listeners will be friends of Precursor, but do you mind briefly sharing
0:58
more about the firm and the idea behind it?
1:01
So I started Precursor in 2015 with this idea that pre seed founders needed more capital
1:09
options.
1:09
And I consistently was meeting all these founders who I thought had great ideas and just
1:14
didn't have great access to capital and didn't have really much in the way of product or
1:19
prototype to show people.
1:20
And as you know, Mia, most VCs can engage with someone if they know the person or if there's a
1:24
product to evaluate.
1:25
But if you don't have either of those things, I think it's hard for a lot of people.
1:28
I said, what if we had a firm that just didn't require either of those things and we would as
1:32
I joke with our LPs, we would engage with strangers with data, and we wouldn't be afraid
1:37
to do so.
1:37
And so we've been doing that for the last 10 years, and we've backed over 400 companies.
1:41
10 years in.
1:42
How are you feeling?
1:43
Never been more excited.
1:45
Awesome.
1:46
Love to hear it.
1:47
I know the idea for this podcast came about in a Starbucks in Midtown East, but the title
1:54
Learning Corner already has some pretty deep roots across the fund and our team.
1:59
So for the rest of the listeners, where does the name come from?
2:03
So I've always been maybe a chronic article overshare in our team Discord, and I will drop
2:09
things that I find interesting into our news and good reads channel.
2:14
And one day, I think it was actually Marina, was like, you know what would be more
2:17
interesting than randomly dropping articles into a Discord would be the ability to talk
2:22
about them on a regular basis.
2:23
And so we started doing learning corner every Tuesday morning, and it's just a chance for us
2:27
to talk about a variety of things.
2:28
Sometimes it's an interesting term sheet.
2:30
Sometimes it's something that happened in the ecosystem that I think we should all chat
2:34
about.
2:34
Sometimes somebody found an interesting article, went to an interesting event.
2:38
It's just a chance for us to share some of the interesting things that we're finding on the
2:42
Internet and in real life that impact the world of venture and, and technology.
2:47
That's become one of my favorite meetings we have as a group every week.
2:50
The purpose of the podcast is to offer that to anyone in the venture landscape.
2:55
To just have a couple looks into some articles and Goodreads that our team found and some
2:59
insights if they find it helpful.
3:01
But without further ado, I think we should get it started.
3:03
We can kick it off with the topics and go from there.
3:06
So topic number 1, 24 venture capitalists tell you what one thing did they wish people better
3:13
understood about BC, a series by Hunter Wach with the co founder of both Homebrew and Screen
3:20
Door.
3:21
We've been following this series by Hunter Wach for a bit on our team.
3:24
So essentially he prompts 24 different VCs.
3:27
We're typically GPs on what they wish people better understood about the job and people in
3:34
quotation marks can include anyone from politicians to founders, to LPs, essentially
3:40
anyone outside of the traditional VC group.
3:43
And the most recent piece he put out had some really wonderful quotes.
3:48
And there were 2 that I specifically wanted to bring here to discuss in a little bit more
3:52
detail.
3:53
The first was from Andre at maplebc who discussed how wonderful it is that BC reward
4:00
difference in its foundational form quote, this difference come from the outlier returns driven
4:08
by backing founders who are different.
4:10
And then he goes on to call onto the fact that LPs are really not incentivized to take that
4:15
risk, unfortunately.
4:16
And Andre hopes that LPs were a bit more open to backing GPs who were in fact different.
4:22
So similar to Hunter, I know you've been very involved in Screen Door, which is a fund of
4:27
funds run by VCs focused on investing in new VCs.
4:31
How does Andre's conundrum have downstream effects into the broader venture landscape?
4:38
Yeah.
4:38
I mean, working on screen door in my 25th hour every day is honestly, it's a lot of fun.
4:43
And, you know, when a handful of the the GP advisors and the LPs that that back screen door
4:48
got together, I think there was this belief that GPs had some ability to identify people
4:54
who were maybe doing things that were different.
4:56
And when I think about what can make a VC fund different, it can be the GP's background.
5:01
So, obviously, you know, people spinning out of really established venture platforms are pretty
5:05
well understood.
5:06
But if you're an independent angel, just an operator, if you're coming from some really
5:10
different background, I think that that can make you stand out.
5:13
Sometimes you're different because you're going after a market that's not really well
5:16
understood.
5:17
I know a bunch of people who were doing AI centric funds 5 years ago, and there were a lot
5:21
of questions like, can you build a whole fund just on an AI thesis?
5:25
I think the answer to that is probably, yeah.
5:27
Sometimes it's a geography that you're focused on.
5:29
It's a region that people aren't as familiar with.
5:33
I'd say the other one.
5:34
And the thing I think actually is the hardest for LPs, which is there's something structural
5:38
about your fund that's different.
5:40
It could be like in the case of precursor, like our portfolio construction of 80 to a 100
5:45
companies per fund is challenging for a lot of LPs.
5:45
They're not used to seeing successful funds with challenging for a lot of LPs.
5:47
They're not used to seeing successful funds with portfolio constructions like that.
5:52
And so I've over time become more sympathetic to the plight of LPs.
5:57
As you know, Mia, it takes 7 to 10 years to know if a strategy is gonna work.
6:02
And I think it takes a lot of courage to back somebody who doesn't fit the model.
6:06
And, you know, for the listeners, I think if someone were to come to me and say, what's your
6:10
advice for me?
6:10
I would say you should run a concentrated strategy where you buy a lot of ownership
6:15
because most LPs that you will pitch have very finely tuned decision criteria for evaluating
6:22
funds that fit that criteria.
6:25
Ideally, you have a multi GP partnership of people who trust and like each other and have
6:30
worked together.
6:31
And when you deviate from that platonic ideal, everything about fundraising becomes much more
6:37
difficult.
6:38
So you should only do that if you really have deep conviction that the differences from that
6:43
model are essential to what you wanna do.
6:45
And I've known Andre since he worked at Uber in Toronto.
6:49
Wow.
6:49
Andre is always somebody who I thought has had amazing access to early stage teams and has had
6:55
this really unique ability to build relationships with founders at the earliest
6:59
stage.
7:00
And I've always felt like that's the thing you should be looking for as an LP because,
7:04
ultimately, the business isn't so much about you as a GP choosing which founders to back.
7:09
It's about convincing the very best people to take your money instead of someone else's.
7:13
The Screen Door Investment Committee is an interesting collection.
7:15
We've got 3 super talented people on the staff.
7:19
We've got, you know, Jamie, Lisa, and Lane.
7:21
So everybody on the IC tends to approach these conversations with a different perspective
7:26
based on what they do for a living.
7:28
And I I almost always ask GPs, like, given your strategy, why are the very best people that are
7:33
in spec for you gonna choose you over anybody else?
7:37
I find the answer to that question tends to be super revealing.
7:40
I almost feel like you have to walk this fine line of staying true to your thesis and
7:44
building conviction into an investment area that you feel strongly of while also being
7:48
realistic of how LPs pick GPs to invest in.
7:53
So at what point do you have to change direction, or do you ever have to?
7:58
It's so funny.
7:59
I mean, I I'll never forget this.
8:00
When I was raising our first fund, I went and asked for some advice from one of the most
8:05
respected I won't name this person because I didn't ask for their permission.
8:09
I asked one of the people who I respected the most, who's one of the OG Seed investors who's
8:13
been very successful.
8:14
I went and showed him my deck, and I had all of the, like, really innovative, aggressive, new
8:20
ideas for Precursor at the front of the deck.
8:22
And I pitched him, and he sat there very politely, and he goes, you've got this all
8:26
backwards.
8:27
What do you mean?
8:28
He goes, this is gonna scare people because all of these things that you think are new and
8:33
innovative are risky, and some of the people you're pitching really want vanilla ice cream
8:38
with sprinkles.
8:39
So I would suggest you focus on the parts of your fund that are easy to understand and
8:44
normal and treat these innovations like sprinkles.
8:47
These are things you're gonna do to differentiate yourself or make your fund
8:50
better.
8:50
And he's like, I just worry that if you go out with this pitch, it's gonna overwhelm people
8:53
because it's too different.
8:55
I didn't know what he meant at the time, and now I totally get it.
8:58
And he was absolutely right.
8:59
And I made those changes and it made a pretty significant difference in our fundraise.
9:03
Ken, part of what I like about Screen Door is I think we've taken some smart bets on founders
9:08
who I think are going to turn out to be excellent GPs, but don't exactly fit.
9:13
Either the fund size is too small or their solo GP or they're investing in a theme or thesis
9:18
that, like, isn't well understood or the person doesn't have a tremendously long track record,
9:22
but we think through getting to know them and the portfolio that, like, they're a star in the
9:25
making.
9:26
So I I've really enjoyed it.
9:27
And truthfully, like, as much as advice is useful when you're starting a fund, the thing
9:31
that's way more valuable is money.
9:33
And that's one thing that we can bring to the table screen door that, like, you can't bring
9:36
on your own.
9:36
Awesome.
9:37
Thanks Andre for that.
9:38
That was a great ad.
9:39
The second quote in the piece was by Ryan Hoover from the weekend fund who talked about
9:47
the quote unquote traps of escalating fund sizes and calling out the reality that
9:52
deploying a $10,000,000 fund is wildly different from operating a $100,000,000 fund.
9:58
Charles, I feel like I've had multiple conversations with you about this, but as
10:02
someone who has essentially lived through this, curious to get your perspective here and, you
10:08
know, what's the decision process you went through each subsequent fund that was raised
10:12
and what are the conversations you actually have to have with LPs along the way?
10:16
I think it's really interesting.
10:19
And this, like, tax back to screen door too.
10:20
I always ask people, like, what do you think fund 3 looks like?
10:23
Cause we're usually meeting people for fund 1.
10:24
And it's kind of an unfair question because nobody knows what the future looks like.
10:28
I think if you'd ask most managers in 2020, what does your 3rd fund looks like?
10:32
It'd be 20 times the size of the fund they would take that day.
10:35
And if you ask most people today, they'd be like, flat would be good.
10:38
Flat would be
10:38
fine.
10:39
Right.
10:39
And so and so when I started Free Courser, our first fund was 15.
10:45
And I told people our strategy doesn't really work above $50,000,000 So you should think of
10:50
us as a fund that will always be sub 50.
10:52
There are a couple reasons for that.
10:54
1, I was single GP and didn't really know how big the team would get.
10:57
2nd, most of our companies didn't need that much money from us to get from the place where
11:02
we met them to the place where they tended to engage with us less.
11:05
So I was like, you know, $50,000,000 is enough to give people one pre seed check and one seed
11:09
check, and the good companies will graduate.
11:12
And, like, the other ones, like, they won't.
11:14
And so, like, I I've always approached every fund saying, like, well, what do we need?
11:18
So I always go back and say, like, well, what are we trying to do here?
11:20
We're trying to get people from 0 to 1.
11:24
Like, no product to a product with some data, ideally with seed round, and then from the seed
11:29
round to a series a, and then I find most people don't need us that much, or they don't
11:33
need us in the same way.
11:34
The other thing when I started, I was like, I have no idea if people are gonna let us invest
11:37
in their a's and b's.
11:39
Like, I don't know.
11:40
Like, we're a brand new fund for all I know.
11:41
They're gonna look at the pro rata side letter, and if we don't have one, they're not gonna
11:44
give us an allocation.
11:45
That did
11:46
that didn't turn to be true.
11:47
So over time, a couple things changed, and I think this is true of a lot of managers.
11:51
I think the good reasons to change your fund size and get larger are there something about
11:55
your model you've realized requires more capital.
11:59
You believe you have the ability to deploy more money against the same strategy without
12:04
changing your ability to access the most important deals.
12:07
And for us, what I realized is like, well, very few of our companies were going c to series a
12:11
and one check.
12:12
Most of them needed 2 checks.
12:14
And I was like, well, if we're gonna be a good partner to these companies, we need a bigger
12:16
fund because I wanna be able to write both of those seed checks when it's right.
12:20
The other thing was people were offering us allocations in the a's and b's, which I hadn't
12:24
put into the model.
12:25
And like, I wanna do these selectively, not in every case.
12:28
And that requires more money.
12:30
I do think though the problem with AUM is that some people's ego is very tied to the amount of
12:35
money they have under management.
12:37
Yep.
12:37
And, early in my venture career, someone who I really respect once told me nobody willingly
12:42
raises a smaller fund.
12:43
And, like, that has stuck with me for a long time.
12:45
Like, but what if it's the right thing?
12:46
Like, what if you overshot?
12:48
What if you realize, hey, we don't need and he's like, nobody does it because everybody
12:51
assumes if you raise a smaller fund.
12:53
It was because you had to, not because you chose to.
12:55
And so I've always I always ask myself, honestly, on a weekly basis, like, what's the
13:01
maximum dollars we could deploy given the model and constraints we have?
13:05
And we've talked about this internally.
13:06
I think for us, it's, like, between a $100120,000,000.
13:09
Yeah.
13:09
You get beyond that and, like, a bunch of the core assumptions in our fund break.
13:13
But I think it's really tempting to grow your fund size.
13:16
And I think the biggest mistake I see people make besides simply pursuing bigger fund size
13:20
because it is more fees and more fees make day to day life more enjoyable.
13:24
Right.
13:25
The other mistake I see people make is, well, I wrote a 25 k check-in Fund 1.
13:30
I'm sure I can size my check up to 2.50.
13:32
And then, like, you can size it up.
13:34
You just might not get into the same companies.
13:36
You might go from being the person who was super value add at 25 and produced a 100 ks of
13:42
value to a person who puts in 250 ks and still produces a 100 ks of value.
13:48
And the, and the founder's perception of their relationship with you and your value add will
13:52
change pretty dramatically.
13:53
So I always caution people on like doubling is a big deal.
13:57
Yep.
13:57
Like doubling your fund size is a big deal.
13:59
It changes a lot.
14:00
And so it's very hard to convince managers not to raise the maximum amount of money on offer.
14:05
I found it's very hard.
14:06
It feels like sometimes fund managers forget that while they're raising their subsequent
14:11
funds from fund 1 to fund 4, there have been other people who've entered the space from a
14:16
fund manager perspective and other multistage funds have continued to raise and build
14:20
incubators underneath their platform.
14:23
And so all of that affects you, even though you probably don't wanna do or tell LPs that it
14:28
won't, I assume feeds into the fundraise.
14:31
I always ask people like, how elastic is your strategy?
14:34
I would say series a strategies are actually fairly elastic.
14:39
If you can get into an a and you can maintain, like, if you can get into a good a and you do a
14:43
good job on pro rata rights, you should be able to like deploy a lot of money in the b and c if
14:47
you want to.
14:47
I just don't think pre seed and seed strategies are that elastic.
14:51
I just think if you stretch them too hard, you snap the rubber band.
14:53
So I I always try to get people to think about, like, how much could you really stretch the
14:57
strategy and still do what you say you're gonna do?
14:59
Love it.
15:00
Well, we're gonna move on, but thank you, Hunter, for continuing to share these awesome
15:04
insights.
15:05
We encourage everyone to follow that series.
15:07
We'll include the link in this episode, of course.
15:10
Topic number 2, an article by The Economist titled Open AI's new fundraising is shaking up
15:18
Silicon Valley.
15:19
This article surrounds the rippling of facts across venture from the news that OpenAI, the
15:24
maker of ChatGPT, is on the brinks of reaching a roughly $150,000,000,000 valuation as it
15:31
works towards raising $6,500,000,000 from investors.
15:35
And that valuation is second to only SpaceX in the US.
15:39
And so the article highlights 3 specific challenges that this emergence of generative AI
15:44
funding poses across the traditional venture landscape.
15:48
The first is the amount of funding and therefore check size required to support these
15:53
LLM companies is huge.
15:55
And as a result, it makes it pretty clear who has the ability to be a check writer here.
16:01
Charles, what are the things that you probably need to believe to be true in order to write a
16:07
check into this space as a fund manager?
16:11
I mean, I it's funny.
16:12
A good friend of mine who's a very successful early stage deep tech VC sent me this article
16:19
on a text, and it was like, hey.
16:21
It's like they read your blog post.
16:23
I wrote a post a while ago.
16:24
I think it's very hard right now to be I don't even say contrarian.
16:28
I don't think we're contrarian on AI.
16:30
I'd say to be cautious and conservative on AI.
16:33
I think it's a very hard place to sit as an investor.
16:36
It's very hard to also conversation to have with LPs because they're like, look, all of the
16:40
biggest multi stage funds that they're in are all in on this theme.
16:43
And my view has been there's a lot of awesome AI companies that I think are super promising
16:47
that don't make sense for a sub $100,000,000 pre seed fund
16:50
Sure.
16:50
Because the entry points are, like, 100 of 1,000,000 of dollars, if not 1,000,000,000 of
16:55
dollars from a starting position.
16:57
So you have to believe that these companies can become worth 50 to a $100,000,000,000 in the
17:03
future for those, let alone, can you get access to these deals, to believe that they're good
17:09
investments, and I I'm just not there.
17:11
Yeah.
17:11
And so some combination of reservations about what the technology can do today versus in the
17:17
future, plus entry place, plus our fund size.
17:19
I'm like, these just don't make sense for us.
17:22
Yeah.
17:22
And I think they really only make sense for multibillion dollar.
17:26
If you think about it, I mean, I think for those of you who've listened to, Nick Charles'
17:31
podcast about the launch of Asylum Ventures, I think I maybe don't have quite as extreme of
17:35
you as he does, but I will say this, the only people who I think can play that game are
17:40
people who can say, we could go multiple 1,000,000,000 of dollars of capital invested
17:45
into this company and it doesn't break our fund.
17:49
Sure.
17:49
Maybe it maybe we end up 20% concentrated or maybe we end up 10% concentrated and we have a
17:54
big chunk outside the fund.
17:56
But, you know, being several $1,000,000,000 of committed capital, not like net asset value,
18:01
but like actual dollars in is fine.
18:04
That is not, in my opinion, a game that most seed investors can afford to play.
18:07
And my question is, like, what happens if you're the 3rd, the number 3 or number 4,
18:13
number 5, LLM Yeah.
18:14
And the price to develop the next model is an excess of what you can raise.
18:18
What's the residual value of an LLM company that can't continue to innovate?
18:23
Is it 0?
18:24
Sure.
18:24
Is it greater than 0?
18:25
Is it one of these weird aqua hire license deals where you take the talent?
18:30
It's sort of like being at a poker table, and everyone's like, we're just gonna, like,
18:33
continue to, like
18:34
Pile it.
18:35
Raise pile it and and, you know, the pot's gonna get a lot bigger, but at some point, if
18:39
you fold, you're you might walk away with nothing.
18:41
Sure.
18:41
So Beyond the sovereign wealth funds and these massive $1,000,000,000 funds, I think most
18:46
people forget that most of the funding for these massive LLM companies actually was raised
18:51
by large tech companies.
18:52
So Microsoft and Amazon have collectively invested $17,000,000,000 into either OpenAI or
18:59
Anthropic.
19:00
That pales in comparison to the check side is that most precede and seed funds raise.
19:04
In fact, another good point is PitchBook said that last year, the average fund that was
19:09
raised in America was around $150,000,000 That's it.
19:13
The Economist pointed out that OpenAI is seeking to raise more than 40 x that from
19:18
investors.
19:18
So, like, how does the math math?
19:20
And I think it's really interesting because I remember when Microsoft announced that original
19:23
OpenAI deal, people were like, that's crazy.
19:26
Doesn't seem crazy, but the benefit of hindsight.
19:28
And if you're in the business of selling compute, there's probably no better customer
19:33
for you than these big LOMs.
19:35
So I think if you're those companies, you're like, look, we're investing, like, all
19:38
strategics for a variety of reasons.
19:40
Making money is one of them.
19:42
Getting these companies to take a serious look at our infrastructure and use it is another
19:46
one.
19:47
And 3rd, being close to these companies and ideally getting some insights into, like, what
19:50
they're learning is probably the third one.
19:52
So I think for those companies, it maybe does make sense to to do this because like, if
19:56
you're Microsoft, you're like, hey, if if this investment is part of a bigger BD deal I need
20:02
there's a lot of ways for us to win above and beyond the upside of being in the stock.
20:06
Next one that the article highlighted as another challenge is around the concept that
20:12
investing in these new companies is very different than investing in internet startups,
20:16
where you throw cash at a company to see it grow as quickly as possible.
20:20
With these types of companies, the new game is to invest immeasurable amounts of money into
20:26
the company upfront, to invest in computing power and data to make the models as smart as
20:31
possible.
20:32
It's not a revolutionary thought, but it feels like every VC recently wants to have a stake in
20:38
LLMs and generative AI.
20:40
And I would go out on a limb to say perhaps LPs are also asking this of their fund managers as
20:47
well.
20:48
Wanted to know your thought.
20:49
I I mentioned to an LP.
20:50
I'm like, the amount of dollar exposure you have to AI as an LP in big funds, it doesn't
20:55
matter what you think about AI.
20:58
Like, someone told me, he goes, it doesn't matter.
21:00
The managers that we've chosen to back have already voted for us.
21:03
You give them the capital.
21:04
You trust their judgment.
21:05
They made the decision that it was wise to put a significant amount of that capital to work in
21:10
AI.
21:10
So you are now very long AI, and I hope that that decision is consistent with your beliefs
21:16
about where the opportunities lie.
21:19
And I do think there was like a bit of an LLM craze.
21:23
And I think people said like, I've got to have one of these in my portfolio.
21:27
And I, I kinda get it in some ways, like the business of venture is somewhat bubble driven.
21:33
Like we get these like innovations and waves and the questions like, can you find the
21:37
companies that matter?
21:39
And it all it always seems to turn out there are far fewer companies that really matter.
21:43
Mhmm.
21:44
In the end, I got a chance to spend some time with some of the management at OpenAI.
21:48
It's the first company I've seen that reminds me of Google, not in terms of ambition, but in
21:53
terms of, like, all of the problems that, like I remember when I met Google in the early days,
21:58
they were trying to solve this algorithm problem around search, but also this compute
22:01
problem around how do we solve all these search queries and this data center problem because,
22:05
like, they needed they were growing faster than anyone.
22:08
They were solving all of these problems mostly because they were encountering them in a way
22:12
that others weren't and at a velocity that others weren't.
22:16
And that's, that's a lot of the, like a lot of the internal tooling that's come out of Google
22:20
too.
22:20
Like that was in response to really hard compute.
22:23
And OpenAI is the 1st company I've seen in a long time.
22:25
Or when I talk to the people there, I'm like, Oh, you guys are doing the same thing.
22:29
You're trying to iterate on this reasoning model or whatever the next generation is of the
22:33
core model.
22:34
You're trying to solve a compute and data center problem.
22:36
You're trying to solve a semiconductor Silicon pro.
22:38
You're trying to solve a lot of problems because you're experiencing this new market,
22:42
which is demanding things that tech doesn't provide.
22:45
And I think the history has shown those companies end up being special.
22:49
So as much as I think these AI LLM companies don't make sense for most venture funds, I do
22:55
think if you're like, I have a multibillion dollar fund, what's the one bet I could make
23:00
with asymmetric upside where I could generate a real return on 1,000,000,000 of dollars of
23:05
capital invested?
23:06
Yeah.
23:06
This is like a rational thing to do if your starting position is, I have a 5 to
23:10
$10,000,000,000 venture fund that I'm trying to turn into a 2 to 3 x net fund.
23:16
It's kind of the one place you could I could see how it happens there.
23:19
If you've invested in OpenEye at 8 or $10,000,000,000 on paper, you're up
23:24
Sure.
23:24
15 x?
23:25
Yeah.
23:26
Which is psychotic.
23:27
Like, that's not supposed to happen.
23:29
It ties into the last challenge, which is that the venture ecosystem is now tasked with
23:35
working out how startups that rely on these LLMs can be profitable.
23:41
So to our earlier conversations, since some VCs are being priced out of the LLM funding market,
23:46
some are actually funding startups that can build on top of them.
23:49
So you have your coding tools, cluster support, your digital healthcare, or what have you.
23:54
But to what end does that make sense?
23:58
The other thing that I didn't appreciate that I like, I still haven't worked out what it means.
24:02
There are 2 things I see happening that to me, like, make it very difficult to figure out,
24:07
like, when to start a company in this space.
24:10
And 1 is the cost of compute is coming way, way down.
24:13
I believe both OpenAI and Anthropic believe that, like, if compute is cheaper, people will
24:18
compute more stuff and you will be able to, like, answer harder questions.
24:22
And second, with each successive generation of these models, things that were hard or
24:27
impossible become possible.
24:29
So for me, if you and I started an AI company today, we would have to solve a bunch of
24:34
problems that exist today, and we would have to undertake a use case that makes sense with the
24:39
cost of AI today.
24:41
Yeah.
24:42
6 months from now, there could be something we spent 6 months building and fixing that the
24:46
system just does for you.
24:48
So So you gotta throw that you gotta throw that, like, hack the stuff away.
24:51
And, oh, by the way, the cost of commute of compute is now a quarter of what it was when we
24:56
started.
24:56
Yeah.
24:56
So whatever architecture we built, assuming that compute cost is maybe less relevant, And a
25:02
brand new company could start now and be cool.
25:04
I get to skip solving all the stuff that Charles and Mia solved because, like, it's like
25:08
you look at, like, what, like, the reasoning that you can get
25:11
on a
25:11
new model.
25:12
Yeah.
25:12
People were, like, trying to figure out how to do that, and now you're like, oh, like, the
25:15
model does it.
25:16
So all the time we spent
25:17
And the money.
25:18
Scratching our head and all the money.
25:20
And so, a friend of mine, David Crenshaw, sent me a good article about this.
25:24
He sent it to me on Twitter.
25:25
And it's just sort of like, in the end, all of the specialty hacking doesn't matter.
25:29
It's just like you throw more compute at it and the models get and the models do more of the
25:33
work over time.
25:34
That's a bad summarization of the article, which to me means it seems maddening to me to
25:38
start a company because you know that in the future, everything is gonna be easier, which
25:42
has always been true of software.
25:44
But in this case, we're like 2 quarters from now.
25:47
I remember I met a bunch of people who were trying to voice enable open AI a year and a
25:52
half ago.
25:52
I'm like, isn't open AI going to do this?
25:53
No, they'll never do it.
25:55
Here's what?
25:55
And I was like, I kind of think they're gonna do this and they did it.
25:58
So now I feel like primary value prop was like voice input, like it's offered at the platform
26:03
level.
26:04
You're welcome.
26:05
And that's probably better for people that like the platform does it.
26:09
So I don't know.
26:09
I continue to think these applications are interesting, but I do think like riding the
26:14
wave of these successive model improvements and the things that these companies are gonna throw
26:19
out of left field just makes it really hard.
26:22
Makes it really hard.
26:23
Well, great.
26:24
We gotta move on to the last one.
26:26
The third topic, it was LeadEdge Capital's letter that segmented the type of reporting
26:32
information companies.
26:34
And this was circling on LinkedIn following a post by Better Tomorrow Ventures and Lead Edge
26:38
Capital is a New York City based growth equity firm focused on internet, software and services
26:43
and tech enabled services investments.
26:45
The letter placed pretty concise messaging on company letterhead outlining the reality that
26:51
there is in fact, a hierarchy of information included in decks about traction metrics from
26:57
an investor's POV.
26:58
At the top, we have, if companies have good cash profits, they will report those.
27:04
And at the very bottom, we have, quote, if companies don't have good page views, they
27:10
report on that they were voted the best place to work in X, Y, and Z city.
27:15
Charles, I saw your comment.
27:16
Nobody hides good news.
27:18
Tell me more here.
27:19
Early in my venture career, someone I worked with told me nobody hides good news.
27:23
This was someone who'd been in the, finance department at Disney.
27:27
Know anything about Disney Corp strategy and finance.
27:30
It's a pretty inte it was a pretty intense group back in the nineties.
27:32
And he was just like, nobody has good news.
27:34
He's like, if you're making a bunch of money or profitable, like, that's the lead.
27:37
And you can kinda tell where somebody is by what they decide to lead with.
27:40
And I love one of the things I love about, like, growth equity people is they're kinda no
27:44
BS and they're really direct.
27:46
And I have found what they said to be true.
27:48
Whenever someone tells me about downloads, my immediate question is like, well, let's talk
27:51
about acting.
27:51
Like, I I realized without even ordering them as hierarchy, whenever somebody tells me as a
27:56
metric, I always wanna go, well, what's one what's the one level metric that would be
28:00
better if we knew it?
28:02
So if someone's like, these are these are revenue.
28:04
I'm like, okay.
28:04
What's profits?
28:05
What are gross margins?
28:06
Like, what are all these other things that I wanna know?
28:08
And I think it's like it's one of these things where it's like an obvious tell, it's also
28:11
like, I don't know what else you're supposed to do as a start up.
28:13
If you're like a brand new company and you're like, all we have is downloads, that's all you
28:16
can talk about.
28:17
Right.
28:17
All we have is the best place to work in x, y, and y, city.
28:20
For pre seed and seed companies, what would be the hierarchy of information or maybe even more
28:25
in simpler terms, the what's the one thing that you care about?
28:27
Because to no fault of these super early founders, some of the more palpable news could
28:32
quite literally be more intangible things like awards and media announcements.
28:36
It's funny.
28:36
A lot of what I work with our portfolio companies on is, like, figuring out, well, what
28:39
KPIs do we care about in this business?
28:42
Yeah.
28:42
And how can we, at an early stage, start thinking about tracking and measuring them?
28:47
The results might not be great, but I find if you pick the right things to measure and track,
28:52
you will get clues as to whether you're on the right track.
28:56
And then you can make trade offs.
28:57
Like, I was talking to somebody this weekend who's like, well, you know, our monetization
29:01
and retention is really strong, but our growth is slow.
29:04
I'm like, well, we have to fix growth.
29:06
And, like, I'm actually okay trading some of the success.
29:09
Like, we're probably over delivering in these other areas at the expense of growth.
29:13
And so I think most startups, it's this, like, multi you have to be good at multiple things at
29:18
once.
29:19
Like, you could be profitable and, like, have super awesome retention and not growing and no
29:23
one's gonna care.
29:24
So it's like, how do we balance growth, profitability, good unit economics, all these
29:29
things that I think it's maddening for founders because sometimes the equation for what people
29:33
want changes in 2021.
29:35
It was like, if it costs you $10 to generate an incremental dollar of revenue, spend the $10
29:39
like people wanted growth.
29:41
Right.
29:41
And then it switched and people were like, this is life.
29:43
This is venture.
29:44
People change the balance between those things.
29:47
And I think as a founder, you have to adapt to that, whether it's fair to ask you to do so or
29:52
not.
29:52
That is the job.
29:53
And I will also say, I think there are a lot of investors who also don't know why they're
29:57
asking specific questions.
29:58
I also think sometimes we have portfolio company founders who are like, this person
30:02
asked me for a 5 year financials for a pre seed company.
30:05
Like that tells you more about them than it's about you.
30:07
Tells you like how they like to engage.
30:09
This, this person might not be the right investor for you.
30:12
Right.
30:13
I'm going to deem this as a success for the first episode.
30:17
Thank you, Charles, for taking time in the morning.
30:19
And also thank you to everyone who popped over to the learning corner.
30:23
We'll be running this weekly, so we hope you come back in.
30:26
And in the meantime, please share, subscribe, and tune in next week for another episode.
30:33
Thanks.
30:34
Thanks.