How does one of the most established venture capital firms in the world think about the so called “SaaS apocalypse”?
Jeremy Levine of Bessemer Venture Partners joins the Newcomer Podcast to discuss the SaaS repricing, the acceleration of AI, and why venture capital remains a long game.We unpack whether SaaS is broken or simply reset after years of excess, and why AI companies are scaling faster than anything we have seen before.
Jeremy shares his perspective on foundation model giants like Anthropic, the coming wave of robotics, and the unsolved manipulation problem that could define the next decade.We also discuss scale in venture capital, how AI is changing investing, and why, in Jeremy’s words, this is ultimately a patient person’s game.
00:00:00
And if, as an investor in venture capital, you're looking
00:00:03
for returns every quarter or every year, you're in the wrong
00:00:05
asset class. This is a patient person's game.
00:00:07
How does one of the best software investment firms in the
00:00:10
world feel about the SAS pocalypse?
00:00:12
Today's guest is Jeremy Levine of Bessemer Venture Partners.
00:00:15
He has a long history of early prescient investments in
00:00:18
companies like LinkedIn, Pinterest, Yelp, and Shopify.
00:00:21
We sit down to discuss his firm's over 100 year history,
00:00:24
how he lives in reality and keeps his principles amidst
00:00:27
everything happening in AI. He's the guy I turn to when the
00:00:30
world is losing its mind. This is the NEWCOMER PODCAST.
00:00:41
Jeremy Levine, welcome to the NEWCOMER PODCAST.
00:00:44
Thank you for having me. Let's let's start big picture.
00:00:46
You know, you see it all at Bessemer and have watched a
00:00:50
number of hype cycles, you know, in venture capital, you know, it
00:00:55
feels like the haves and have nots.
00:00:57
You know, there's the story of obviously all the AI mania, but
00:01:02
then there also sort of has quietly been this downturn, the
00:01:06
recovery from 2021 that's hit, you know, some venture fund
00:01:10
managers and then also a bunch of start-ups.
00:01:12
I think we talked a couple years ago now in light of that 2021
00:01:18
euphoria and the recovery, what how do you see where are we
00:01:21
right now in sort of the venture capital cycle?
00:01:25
I mean, it's hard to know exactly where we are until we
00:01:27
have perspective of being able to step back and look at it a
00:01:30
number of years later. But that said, I now I think
00:01:34
it's pretty clear that 2021 was kind of the the culmination of
00:01:37
this incredible period of growth in SAS software as a service.
00:01:43
The companies grew, the industry grew, the investors around the
00:01:46
companies grew like everything went up into the right for many,
00:01:49
many years, more than a decade. And then all of a sudden the
00:01:53
market thought for a minute, maybe these really nice
00:01:57
businesses aren't quite as valuable as as as we thought
00:02:00
they were. And that was the adjustment that
00:02:03
we went through from 2021 to 22 or 23.
00:02:07
And I think that's generally in the rearview mirror, say for the
00:02:10
SAS occur that has recently happened, which we should talk
00:02:12
about it in a minute. But I think the, the industry,
00:02:16
the venture industry, the entrepreneurial, like I said,
00:02:19
all shifted to AI obviously. And, and the growth of these AI
00:02:24
businesses is making the SAS stuff look like child's play.
00:02:29
It's extraordinary. And so before software as a
00:02:32
service, when a company had to roll out software, it had to
00:02:34
build infrastructure servers and, and, and run all the stuff
00:02:37
on premise. And SAS made it so much easier
00:02:40
that you saw SAS companies go from zero to 100 million of
00:02:44
revenue in five years, whereas traditional software companies
00:02:47
took 10 or more years. And that was this massive
00:02:49
compression and acceleration. And now with AI, what used to
00:02:52
take five years is now taking one year.
00:02:55
And so it's an incredible speeding up of everything which
00:02:58
I think has the the industry wildly excited and it's fun to
00:03:01
watch all. Right.
00:03:01
SAS occur a word I cannot turn away from.
00:03:05
I mean, you know, you're a firm that had what the Cloud 100 has
00:03:09
been very loud on software. A big believer vertical SAS in
00:03:12
particular. You guys were extremely strong.
00:03:16
Yeah. What do you make of the public
00:03:18
market's total turn away from software as a service?
00:03:22
Yeah, I so. At a high level, it's
00:03:24
complicated and nuanced and you really have to think about it on
00:03:27
a, on a name by name basis. But I'll I'll nonetheless get
00:03:30
myself in trouble by generalizing.
00:03:32
And that is to say, I think the public markets have overreacted
00:03:37
very significantly in some cases, but there's also probably
00:03:41
a little bit of valuation or price compression that's was
00:03:45
meant to be. And So what do I mean by that?
00:03:47
So for starters, I, I think the meme, if you will, is that with
00:03:52
AI, anyone can create software. You don't need to be an
00:03:56
engineer, let alone a really good engineer.
00:03:58
And that drives the cost of creating software to near 0.
00:04:01
And so why as a company or a customer, should I pay SAS
00:04:06
company to provide software for me when I can just build it
00:04:09
myself almost for free? And I think that oversimplifies
00:04:14
a lot of what a SAS company does.
00:04:16
But even if you take all that at face value, I think that's not
00:04:20
what makes the most valuable SAS companies really valuable.
00:04:24
What the software itself, that is what makes the most valuable
00:04:27
SAS companies really valuable is the fact that they grew a
00:04:30
network effect. And so I'll give you a couple of
00:04:33
examples. And and the reason why the
00:04:35
network effect is so important is because you can't replace it
00:04:38
just by building the same piece of software for free.
00:04:42
Perhaps my favorite example of Shopify was a Alex Ferrara at
00:04:45
Bessemer led the investment for us about 15 years ago.
00:04:49
And at first, the software that Shopify made was just really
00:04:53
good software and that was their competitive edge.
00:04:56
If that were their only competitive edge, I would be
00:04:58
nervous for Shopify in this Assiger.
00:05:00
I'm not. The reason why I'm not is
00:05:02
because along the way Shopify built multiple network effects
00:05:06
that insulate its business from someone else just building the
00:05:09
same software and giving it away for free at a cheaper price.
00:05:12
Right. But Shopify, yeah, feels like a
00:05:15
unique company. When I think of software
00:05:17
companies, yeah, I think of them as sort of more than the
00:05:19
aberration than the representative case.
00:05:22
Like do you have a view on Salesforce or a more classic
00:05:26
software company? So, so I think that the the way
00:05:28
I would analyze any given name and we can go through a whole
00:05:31
bunch of them, but Salesforce for starters is like, is there a
00:05:34
real network effect there? And I think in Salesforce's
00:05:37
case, there are real network effects.
00:05:40
I don't think they're quite as strong as the, the shop pay
00:05:42
thing I was, I was about to share with you, but they're
00:05:45
real. And so a couple examples.
00:05:47
One is Salesforce has built or an entire ecosystem has been
00:05:50
built around Salesforce of third party consultants and
00:05:54
integrators who've built products that plug into
00:05:57
Salesforce. And so you can't just replicate
00:06:00
Salesforce itself to get the value of Salesforce.
00:06:03
You then have to replicate all the third party capabilities
00:06:06
that exist around Salesforce. And the reason why they exist
00:06:08
around Salesforce and not other platforms is because Salesforce
00:06:12
is by far the biggest. And so that's a situation where
00:06:15
the the the first player, the the first mover has the massive
00:06:18
advantage because the ecosystem develops around it.
00:06:21
I think the second potential network effect around Salesforce
00:06:24
is among its users, its users are so familiar with it.
00:06:29
And to the extent human salespeople remain a thing,
00:06:33
that's that's a second, a second source of value for Salesforce.
00:06:37
There's there's so much comfort with it.
00:06:39
Absolutely. Then it makes them less
00:06:41
efficient. It makes them less happy.
00:06:43
Now, that said, that is a network effect that could easily
00:06:46
disappear if in fact, salespeople writ large or
00:06:49
replaced by AI, because AI doesn't care what the interface
00:06:52
looks like. But I think that's what you have
00:06:54
to do for each of these companies, understand, is it
00:06:57
just good software or in some cases mediocre software?
00:07:00
Or is there something else going on?
00:07:02
And if you ignore the something else, I think you're going to
00:07:04
make the wrong call. Do you have a view on what the
00:07:07
valuation multiple of software companies should be?
00:07:11
I know that's like a market derived.
00:07:13
Yeah. And it really depends company by
00:07:14
company, but I think ultimately, you know, companies are worth
00:07:20
some some aggregation of their future cash flows.
00:07:24
And so when these companies that have no profits are being traded
00:07:28
for say, 10 times revenue, I think what someone is really
00:07:32
saying is I believe this company will become profitable or quite
00:07:36
profitable. And when I discount those future
00:07:38
cash flows back to today, it happens to equal 10 times
00:07:41
today's revenue. But I don't think they're saying
00:07:43
there's something inherently valuable about the revenue
00:07:46
itself that they're willing to pay 10 times for it.
00:07:48
I think the map is implicit. And then people get lazy and
00:07:51
they stop thinking about the future cash flows and just start
00:07:54
thinking about revenue multiples.
00:07:56
But I think as we, you know, in 2021, we saw fast growing
00:08:00
software companies regularly trading for 40 and 50 times
00:08:03
revenue, which implied extraordinary future growth and
00:08:07
high profit margins. In some cases that actually
00:08:10
materialized, in some cases it didn't.
00:08:12
I think today people are willing to be a little less bullish
00:08:15
about the future and therefore that translates just to a lower
00:08:19
revenue multiple. Are you still investing in
00:08:21
software companies at early stages, or is that thesis out
00:08:25
the door now? No, no, absolutely.
00:08:27
And in fact, one of the reasons why is because what you're able
00:08:32
to do now with AI embedded in your software is just so
00:08:35
powerful. The functionality is, is
00:08:38
fantastic and I'll give you some examples.
00:08:39
So we invested them. It's probably been about 18
00:08:42
months now in a company called Relevance.
00:08:45
Relevance is essentially a software platform for managing
00:08:48
AI agents. And so as companies adopt or try
00:08:52
to adopt AI and and want agents to do the work of often times
00:08:58
the grunt work that humans don't want to do, or supplement humans
00:09:01
with additional capability where these agents can run in the
00:09:04
background 24/7 doing work to to make people more productive.
00:09:10
They need a platform to manage all this.
00:09:12
And, and so that's what relevance builds.
00:09:14
I'd say they were a little bit ahead of their their time in
00:09:16
that most people weren't really talking about AI agents 18
00:09:19
months ago or even 36 months ago when the company first got
00:09:22
started. And today they are, and they're
00:09:24
looking for solutions. And companies like Databricks
00:09:27
and Confluent, as they try to embrace AI in their internal
00:09:30
business operations, are all using relevance for it.
00:09:33
And so, yeah, there's certainly opportunities for software.
00:09:36
Most of them are, I think in today's world about leveraging
00:09:40
AI and making it productive in the enterprise.
00:09:43
And to what extent are those like, oh, we're going to have
00:09:45
sort of a chat bot feature or like where, where do you see
00:09:48
them? I mean, there are a bunch of
00:09:49
different strategies for replacing a software category
00:09:53
with AI. Or do you have buckets that
00:09:55
you're starting to see where it's like, OK, there's this
00:09:58
approach, this approach, and this approach?
00:10:00
So we are seeing folks simply try to staple a chat bot feature
00:10:04
onto a product. It's not going to work because
00:10:06
the the the, the foundation labs, the foundation model
00:10:09
companies, they have really great chat bots and you don't
00:10:11
need much else. You can just append them or open
00:10:14
them in another browser. So I call that a very thin value
00:10:17
proposition. The thicker propositions are
00:10:19
ones that builds real capabilities into the software.
00:10:23
And so for, for example, we have another portfolio company,
00:10:27
Hockey Stack. Hockey Stack is a software
00:10:29
company, also a relatively new investment, maybe 18 months old
00:10:32
for us now. And Hockey Stack started off by
00:10:35
building a data platform that you can use as a marketer to
00:10:40
understand where your better leads are coming from, your
00:10:43
worst leads are coming from. And they, when they first
00:10:45
started the company, I don't think they had thought a whole
00:10:47
lot about AI. Today, the entire company is
00:10:51
geared around AI and, and in in many ways, it's sort of back to
00:10:54
our Salesforce conversation flipped Salesforce on his head.
00:10:57
And so you can think of Salesforce as a tool to make
00:11:00
human salespeople more productive, where the human
00:11:03
salesperson is the the engine or the brain of the operation.
00:11:07
And it's the that human is leveraging tools like Salesforce
00:11:09
to make their job better. Hockey Stack has flipped the
00:11:12
thing upside down. And Hockey stack is saying,
00:11:14
well, actually there's so much data in sales and marketing that
00:11:17
it's hard for a human to intuit. And so wouldn't it be better if
00:11:21
AI were the brain sifting through all the data to make
00:11:24
decisions? And then when helpful, it will
00:11:27
ask a human to do something. So the AI is researching the
00:11:30
prospects, evaluating which prospects are the most ripe for
00:11:34
potential sale, and then deciding what's the action that
00:11:38
my company should take with respect to this prospect.
00:11:41
And if it's reach out with an e-mail, it will draft the e-mail
00:11:45
and at the human center, maybe even send the e-mail itself if
00:11:48
it's have a conversation or make a phone call, it will tell the
00:11:52
human like, hey, this is a good phone call to make.
00:11:54
And so the AI essentially becomes the quarterback and the
00:11:57
human is a tool that the AI is using to, to further the
00:12:00
initiative. And so I think there's so many
00:12:03
categories of software that need to be rethought or re imagined
00:12:06
because when they were designed, these AI capabilities didn't
00:12:08
exist. And now they're they're wildly
00:12:10
powerful. What's what's your view on the
00:12:13
big foundation model companies today?
00:12:15
I have sort of conflicting intuitions.
00:12:17
On the one hand, feels like they keep doing better than we
00:12:21
expect, like the revenue is growing more.
00:12:24
I think investors keep marking them up more and more the the
00:12:27
sense that wow, they're really dominant and highly valuable.
00:12:31
I think that sentiment has continued to accelerate.
00:12:34
I think people have seen some progress, you, your investors, I
00:12:38
think in Anthropic. And then on the other hand, I
00:12:42
don't know, it feels very like thin and tenuous.
00:12:45
You know, it's like, can open AI get all the money that it wants?
00:12:48
It seems sort of desperate for more money.
00:12:50
We saw XAI sort of clumsily merged into SpaceX in a way that
00:12:55
didn't make me feel from the outside, I guess, as optimistic
00:12:58
about how that business was doing.
00:13:00
What's your read? And I know they're all
00:13:01
particular businesses with different strategies, but what's
00:13:04
your read on sort of the value of foundation models and how
00:13:07
much of the sort of the business opportunity here they're going
00:13:10
to gobble up versus leave to other companies?
00:13:13
So I think that it's an interesting conundrum because on
00:13:16
the one hand, I heard somebody describe a foundation model as
00:13:20
the fastest depreciating asset in human history.
00:13:23
It goes down in value faster than a new car that's driven off
00:13:26
of a new car dealership lot, which would suggest it's a
00:13:30
really rapidly moving treadmill on your as a model builder,
00:13:34
you're sort of pissing money down the drain building today's
00:13:37
model when you know it's going to be outdated in a minute.
00:13:39
And I think that's largely true. However, and kind of amazingly,
00:13:45
I also believe these the good foundation model companies, of
00:13:50
course, Anthropic among them. And you know, admittedly talking
00:13:52
my, my own book or own book a little bit on it is going to be
00:13:55
one of the most valuable companies in the world because
00:13:58
the stuff is so damn powerful. And, and you know, credit to my,
00:14:03
my partners who invested in Anthropic for us first a couple
00:14:06
of years ago. And obviously that investment
00:14:09
has improved in value more than our very recent investment.
00:14:13
But I actually think the best investment in Anthropic on a
00:14:16
risk adjusted basis is the one that just recently happened at a
00:14:19
$350 billion valuation. Because it's become even more
00:14:23
clear or wildly more clear today how successful this company has
00:14:27
become and how much momentum there is for it to grow because
00:14:31
people just can't get enough of this stuff.
00:14:32
And it's I'd say it's meaningfully penetrated certain
00:14:35
sectors like software development, but it's barely
00:14:38
scratching the surface of dozens of massive other sectors of the
00:14:41
economy, including sales and marketing or, or, or market
00:14:45
research or frankly, a lot of white collar, you know, behind a
00:14:50
computer office labour where people are using it a little
00:14:53
bit. But there's just so far for it
00:14:56
to go and, and you can just see that, I mean, I, I can't
00:14:59
remember what's public and what I know in my head as as private,
00:15:02
so I won't share the data. But the financial performance
00:15:06
of, of Anthropic is just, it's astounding.
00:15:09
And so it turns out that while yes, yesterday's model is much
00:15:16
less valuable today than it was yesterday, a very small number
00:15:20
of companies have proven the capability with the talent
00:15:25
required to keep innovating and coming up with the next model.
00:15:29
And it turns out that there actually are some pretty good
00:15:31
uses for yesterday's model. It's maybe a little bit cheaper
00:15:35
and a little bit less capable, but there's still plenty of
00:15:38
people who are happy to use the slightly cheaper model for
00:15:40
slightly less capability. And it's a, it's a really great
00:15:43
business. In terms of venture strategy and
00:15:46
portfolio construction, like I have you said how much money you
00:15:50
invested in Anthropic? I mean, there are firms I know
00:15:53
like the Menlo story a little bit better where it feels like
00:15:55
they put a huge percentage of their fund into Anthropic.
00:15:59
Like how have you thought about how much to say, all right, this
00:16:02
is the thing like put a bunch of money in it.
00:16:05
Now we're gross stage investors by old school standards versus
00:16:09
saying no, we're we believe in portfolio theory and we should
00:16:12
have a bunch of different bets. So we, so we, we invest in the
00:16:16
full across the full spectrum of, of companies.
00:16:21
So we, we sometimes we write $100 check in a person with
00:16:24
an ID on a napkin and sometimes you write $100 million or even a
00:16:28
$200 million check in what's obviously already a really
00:16:32
compelling business. But what's in common across both
00:16:35
is that we think the company that either barely exists or
00:16:39
exists meaningfully can be 10 to 100 times larger in the future.
00:16:43
And so, you know, one can debate is it easier to go from a
00:16:48
million of revenue to 100 million of revenue or to go from
00:16:52
100 million of revenue to whatever that would be 10
00:16:56
billion of revenue, Is that right, 100 * 100?
00:16:59
I think that's right. I think that's debatable.
00:17:03
And so and and in fact $100 million revenue business
00:17:06
actually looks like an early stage company relative to a $10
00:17:09
billion revenue business. And so we think it's pretty much
00:17:11
the same skill or the same puzzle that we're trying to
00:17:14
solve. But if we're much less likely to
00:17:17
invest in a, a quote, UN quote growth stage company, which may
00:17:21
be $100 million or even a billion dollar revenue business,
00:17:24
if we don't think it can go 10 to 100 times bigger, we're not
00:17:27
looking to make, you know, a AA2X return.
00:17:30
That's that's much less compelling.
00:17:32
And so I think what's remarkable about Anthropic in particular is
00:17:37
that it's already at a very significant scale and still
00:17:40
growing as fast or faster than any company we've ever seen.
00:17:44
And so these things don't come around all that often.
00:17:46
And so hats off to Menlo for making what I think was a a very
00:17:50
prescient early stage investment, but also I think a
00:17:53
really smart later stage investment because I don't see
00:17:56
the ceiling on Anthropic even remotely on the horizon.
00:17:59
Which is part of why I think actually it's a better risk
00:18:01
adjusted bet at a $350 billion valuation, given all they've
00:18:04
accomplished and all the risks that have been eliminated than
00:18:07
it was even a year ago when the price was much lower, but it had
00:18:11
achieved much less. But I want to talk about agent
00:18:14
marketplaces, like are you bullish on the idea that we
00:18:18
should be building marketplace businesses for agents to
00:18:20
interact with each other? We saw what the malt book sort
00:18:24
of thing, which seems somewhat fake, somewhat real, somewhat
00:18:29
fake, I don't know, still still to be seen.
00:18:30
But I don't know what are you betting on.
00:18:32
Any companies that are building like purely on a sort of agent
00:18:35
marketplace thesis? No.
00:18:38
Well, I mean, I mentioned relevance.
00:18:40
Relevance is helping enterprises deploy agents to do white collar
00:18:43
work. And so yes, but a system for
00:18:47
agents to talk to each other, we haven't invested in any pureplay
00:18:51
companies doing that, but we see it happening in some of our
00:18:54
companies. And so for example, if you want
00:18:56
to buy a ticket to an event, you, you probably find your way
00:19:00
to StubHub at some point. And StubHub historically found
00:19:05
that most people who were starting the event search
00:19:06
process would start it on Google or maybe on StubHub directly.
00:19:11
And increasingly they may start that process on Claude or on
00:19:15
ChatGPT. And so they want to allow third
00:19:21
party agents to complete transactions.
00:19:23
Same thing for Shopify across all its stores.
00:19:25
If you're, if you're starting your shopping search process
00:19:30
with a bot, you every retailer wants to make sure its inventory
00:19:36
is known to the bot. That said, I think it's going to
00:19:39
depend a lot on what kind of transaction you're looking to do
00:19:43
because humans like choosing that.
00:19:46
You know, while I might prefer to have a bot reorder my paper
00:19:51
towels and toilet paper, I'm not really shopping around when I
00:19:54
make that purchase. If I'm trying to decide which
00:19:58
seat I might want to sit in in the stadium, I want to see like
00:20:01
what are the views? And I want to, I want to make
00:20:03
the choice myself. And so I think depending on the
00:20:06
category, we're going to find that in some instances.
00:20:09
We have the agents go all the way, and it's literally an agent
00:20:12
to an agent transaction. And I think a purveyor of
00:20:16
commodity goods like Amazon is going to find Asian commerce
00:20:19
really important. But in other places where taste
00:20:23
and choice matters much more to the human, I think we'll have
00:20:26
the agents do a bunch of work for us.
00:20:28
But ultimately, we're going to be the one who clicks the final.
00:20:30
Like, yes, that's what I want. I think these things are all
00:20:33
emerging. I'll give you one other example,
00:20:36
not to take this conversation in a random direction, but we
00:20:40
invested in, I won't talk too much about it.
00:20:42
It's a stealth investment and a wildly talented entrepreneur
00:20:46
who's reimagining travel. And, and, and so in the way he
00:20:52
described to me, the, the pitch was so compelling because he
00:20:55
said, you know, the Internet when it came along that reformed
00:20:58
travel and it basically eliminated hundreds of thousands
00:21:03
of people whose jobs were as a travel agent.
00:21:07
But the Internet didn't actually come up with a new travel agent.
00:21:10
It just made you and me into travel agents.
00:21:12
And we're now the ones checking what the flight options are,
00:21:15
checking what the hotel options are, and kind of orchestrating
00:21:17
it as a travel agent, which kind of sucks.
00:21:20
But AI has the promise to actually be the travel agent and
00:21:25
maybe live up to the original moniker, right?
00:21:28
You know, the Expedia and the bookings of the world, they're
00:21:30
called OTAs. That's the moniker which stands
00:21:33
for online travel agent. But in fact, they aren't online
00:21:35
travel agents. They're booking engines.
00:21:37
And maybe with AI, we can actually build an online travel
00:21:39
agent. So I think the other thing
00:21:40
that's interesting about all these these sort of consumer
00:21:42
agent technologies in particular is that the interfaces right now
00:21:46
are all chat bots. But I think that's likely to be
00:21:49
step one. And we're going to see wildly
00:21:52
more creative interfaces that almost certainly will
00:21:55
incorporate typing to a bot or even speaking to a bot, but also
00:21:59
other elements that we're not yet seeing.
00:22:01
It will be amazing if they can build actual brands for
00:22:03
delivering, you know, the service successfully.
00:22:06
Because I think right now a lot of this agent stuff, it feels
00:22:09
like, OK, it's going to put this in your checkout box and then
00:22:14
you're going to check out and say yes.
00:22:15
And it's still like, maybe it gets you closer to the final
00:22:17
decision, but you're still like pulling the trigger When we get
00:22:20
to an era where like, okay, I've trusted this travel bot, It's
00:22:24
booked it now. I'm willing to just sort of show
00:22:25
up at the places it says. And I have a level of deference
00:22:28
to it. I think that yeah, will be.
00:22:31
It's sort of an amazing. Maybe, although it depends, like
00:22:32
if it's a business trip and I know I need to be in, you know,
00:22:35
San Francisco at 2:00 PM on Thursday, like, sure, let the
00:22:38
agent do the whole thing. But if it's a vacation, like
00:22:40
part of the fun view is like is looking at the photos and
00:22:43
deciding where to go. And, and so like I said, I think
00:22:46
the use cases are going to vary a lot.
00:22:47
I think the simplistic idea that I'm going to do everything
00:22:50
inside of a chat bot and I'm never going to go to a, a
00:22:52
website anymore. That's ridiculous.
00:22:55
But there will be definitely gradations.
00:22:57
All right. Is this a consumer company like
00:22:59
it's for? It is, yeah.
00:23:00
Yeah. What what is the state of
00:23:02
consumer today? I know that's another big
00:23:04
question, but obviously the foundation models, you know, we
00:23:08
sort of just talk through how in some ways, maybe you're going to
00:23:11
anthropic or Chachibuti Claude and talking through the problem.
00:23:16
And then they have to, you know, have a relationship with StubHub
00:23:18
or something and you're the consumer experiences all in the
00:23:22
chat bot. Is that the world you see or you
00:23:23
see? Do you see opportunity for AI
00:23:26
businesses that build sort of direct relationships with
00:23:29
consumers themselves? I definitely see the
00:23:31
opportunity. So I so to step back, I would
00:23:33
say there was an explosion of compelling new consumer
00:23:38
companies between 1996 and 2005. The first couple generations of
00:23:43
the Internet and companies like Google and Pinterest and
00:23:46
Facebook and and the like all emerged.
00:23:48
And then there was a second slightly smaller but still
00:23:51
significant explosion of companies that happened around
00:23:54
the development of the Internet in your pocket on mobile phones
00:23:58
and, and, and companies like Uber and and DoorDash all
00:24:03
emerged. People want a platform shift.
00:24:06
Some new things happen. It's not that they want it.
00:24:08
It's not that they want. It's that once a platform is
00:24:10
established, the incumbents amass the power and it becomes
00:24:14
very difficult to get distribution.
00:24:16
And so I think what when people think about consumer, they
00:24:18
assume it was like some new whiz bang capability that built or
00:24:23
was the foundation for a company.
00:24:25
But in fact, while there often is a new whiz bang capability
00:24:29
like being able to summon a car to you anywhere you are in your
00:24:32
phone that that's Uber. What actually is much more
00:24:35
important is how do you get that capability into the hands of
00:24:39
millions or hundreds of millions or even today billions of
00:24:42
consumers without having to pay to market to each one.
00:24:45
Because if you need to pay to market to each one, your company
00:24:48
will never be worth anything. But Google who controls the
00:24:52
Internet platform alongside Facebook or or Apple, who
00:24:56
controls the mobile platform, they'll make all the money and,
00:25:00
and you'll be stuck. And in fact, you'll have to
00:25:01
raise tons and tons of money. And so if you, if you, I think
00:25:06
most of those early generation Internet companies, Facebook
00:25:09
printers, Google, they never spent any money to acquire
00:25:11
consumers. They got all the consumers for
00:25:13
free. The second generation around
00:25:15
mobile, it was a mix. Some did spend a lot of money.
00:25:19
But actually I also think that they got a little lucky 'cause
00:25:22
they were built in this zerp era, the the zero interest rate
00:25:27
policy era where they were able to get and raise billions of
00:25:30
dollars very cheaply. If that hadn't existed, I think
00:25:33
many of those companies wouldn't exist, but they do and they did
00:25:36
and they took advantage of it, which was quite smart.
00:25:40
And and maybe the best and last example of a consumer platform
00:25:44
that grew in the United States was TikTok.
00:25:48
And they did it with billions of dollars of financing from a
00:25:51
Chinese parent. They spent the money, but most
00:25:53
start-ups can't realistically access that capital.
00:25:56
And so what's consumer becomes interesting when there's a new
00:26:00
platform, because a new platform means there's a new distribution
00:26:02
technique. And so I think right now there's
00:26:06
a confluence of two things happening at once that make me
00:26:08
quite optimistic that we will see some new consumer companies
00:26:11
for the first time in a decade, maybe more.
00:26:14
And those two things are one, the way people are starting
00:26:18
their Internet experience was almost exclusively on Google or
00:26:25
one of the Meta properties. And now there's like a new front
00:26:29
door and the new front door is Claude or ChatGPT or Gemini, the
00:26:34
AI portion of Google search. And, and so the companies that
00:26:37
had massively infiltrated the old front doors of the Internet,
00:26:43
like in travel, as an example, it was booking like if you do
00:26:45
any search on Google, like you see booking all over the
00:26:47
results, 'cause they optimized the platform and it, they became
00:26:51
$100 billion company that really dominates consumer travel.
00:26:54
But for this new front door of the Internet, which is a
00:26:57
chatbot, it's like open seasoned.
00:26:59
And so I'm sure there's a team at booking trying to figure out
00:27:02
aggressively like how do we make sure Booking is as present in
00:27:05
these chat bot results as we as they are in the Google search
00:27:09
results. But it's like a brand new game.
00:27:11
Who? And are start-ups doing well at
00:27:13
that or what's even the strategy?
00:27:15
Well, and so, you know, in the early days of, of, of search
00:27:18
engines, it was how do we manipulate the search engines to
00:27:21
show our stuff high up in the results.
00:27:24
And there were companies like Yelp, which did it really well.
00:27:26
And anytime you search for a restaurant in any city, the Yelp
00:27:29
results were among the 1st results.
00:27:31
That was essentially the Kickstarter for how, how Yelp
00:27:33
grew. And so, and, and it became an
00:27:38
entire art or almost a discipline called search engine
00:27:41
optimization. And there's an equivalent of
00:27:43
that discipline now emerging for this new platform.
00:27:45
And like I said, we've seen lots of people trying to innovate or
00:27:50
reverse engineer or think about how to build their business in
00:27:54
media. It's like one of the things that
00:27:56
does well, I think in these spots is media.
00:27:59
So yeah, happy to have that leverage, I guess at this.
00:28:02
Point, yeah, but it the point is it's a new channel and it's very
00:28:05
immature and it's a Sprint to go figure out how to crack this new
00:28:08
distribution channel. So that's one.
00:28:11
The second and compounding force is that there are new
00:28:15
capabilities that you can build with AI that get consumers
00:28:19
interested. And so now just that the latter
00:28:23
is sort of more exciting from a principled perspective because
00:28:26
like do we need another travel booking engine?
00:28:31
We don't. They work just fine, but there
00:28:33
may be an opportunity for one just by by dint of this new
00:28:38
front door on the Internet. If you can be the one who is
00:28:40
right behind the new front door of the Internet, you can build a
00:28:42
replacement travel booking engine for booking.
00:28:44
But the idea that you can build this wildly compelling new
00:28:47
functionality because of AI makes it exciting.
00:28:50
So now the consumers are also, in addition to discovering
00:28:53
what's behind this new front door, are bumping into these new
00:28:56
capabilities that I think will be sort of, you know, eye
00:28:59
opening or potentially jaw-dropping.
00:29:01
And when you put those two things together, I think you
00:29:03
have the conditions for new consumer companies.
00:29:07
The manipulation of the chat bots to produce sort of a, a
00:29:11
company. I mean, it sounds somewhat
00:29:14
depressing, right? I mean, I think what's great
00:29:16
about the chat bots is like, if I'm like, I want the best pan,
00:29:19
like what's the best pan? Sort of like, well, here's the
00:29:21
best expensive one, Here's the best sort of medium, you know,
00:29:25
price one. Obviously it's rooted in reviews
00:29:28
and sort of different websites that trust probably Reddit to a
00:29:31
large degree. How much do you see these
00:29:33
businesses as needing to sort of manipulate the chat bots versus
00:29:38
accept sort of a paradigm where they're thinking about things in
00:29:41
sort of a different way? I mean, we sort of saw the
00:29:44
degradation of search over time, which has allowed some of this
00:29:47
opportunity, which is just over time it got more and more sort
00:29:50
of games and, and sort of, I don't know, degraded in my view.
00:29:54
If you want to think about it cynically, you can get pretty
00:29:56
far just manipulating someone else's front door, and there are
00:29:59
a lot of SEO companies that did that.
00:30:01
I can't remember the the the name of the one that was.
00:30:05
It was just like a massive farm of crappy content that would
00:30:07
show up. It was called demand something,
00:30:10
demand media. Eventually the the the front
00:30:13
door the search engine. One day they're like, no.
00:30:15
More figured it out. I'm like, this is crap and they
00:30:17
just killed it. So you can get pretty far, but
00:30:19
you probably can't build a great business.
00:30:21
What makes it great business is that you, you, you, you may do
00:30:25
some manipulation, if you will, these front doors to get get
00:30:29
your product in front of consumers, millions of them or
00:30:31
billions of them for free. But then what makes your
00:30:34
business special is that the product is awesome and the
00:30:37
consumer's like, wow, this is great.
00:30:39
And next time when they want that experience, they don't go
00:30:42
back through the front door. They come directly to your
00:30:44
service. And that's that's like the Holy
00:30:46
Grail of consumer and so. Are there any early businesses
00:30:49
that have had success? I mean, I, I think Guillermo and
00:30:52
Versal tweeted at one point, you know, that's not a broad
00:30:55
consumer business, but tweeted that he was getting a fair bit
00:30:57
of his business through people sort of deciding what to build
00:31:00
in, in the foundation model zone.
00:31:02
Are you seeing companies where they're getting a lot of
00:31:05
customers through through these chat bots?
00:31:07
It's just beginning, but we are and we're just trying to invest
00:31:12
in them. So I'm not going to share the
00:31:13
names with you because the last thing I need is even more
00:31:16
competitors in a competitive world.
00:31:18
But I'll give you an example of a company that we we haven't
00:31:19
invested in that I think has done this.
00:31:22
It's a business called Suno. I don't know if you know Suno
00:31:24
and. Music Creation.
00:31:26
Exactly. And so like there, the product
00:31:28
is wildly compelling and it allows you to do things that you
00:31:31
could never do before, or at least most people couldn't do
00:31:33
before. And and I think they're also
00:31:36
getting distribution through some of these chat bots.
00:31:39
I think they're probably getting more distribution now because
00:31:41
the product is so compelling, people are telling each other
00:31:43
about it within the music world. But as an example, like you have
00:31:46
to build something that's good or you end up going the ways of
00:31:49
of demand media. But like I said, if you have to
00:31:53
buy all your consumers one at a time, you need to have a, a
00:31:57
massive sugar daddy with billions of dollars to to fund
00:31:59
you or you're never going to get there.
00:32:00
And that's that's tough to do. One, you know, investor craze
00:32:04
and honestly, I, I can't remember if you made a bet in
00:32:07
this category, they'll like QVC, you know, the online shopping
00:32:12
for video category. Did you make a bet in that
00:32:15
space? If we did, it was really small
00:32:19
material. I don't know is that over or
00:32:21
TikTok wanted to the extent exists, I guess what not is
00:32:23
still sort of fighting it out. Do you ever read on that space?
00:32:27
Because that was such like a hot category?
00:32:29
Yeah, You know, it's funny because I think that, that I'm
00:32:32
not a expert on the Chinese Internet, but, but my sense is
00:32:37
that that's a much bigger, it's, it's much bigger thing for
00:32:43
Chinese consumers than it has been for consumers in the West.
00:32:46
And there I essentially think of there being 2 internets, There's
00:32:49
a Chinese Internet and then there's the, the rest of the
00:32:51
Internet and they don't mix very much.
00:32:54
And so, and I don't know why, for whatever reason, no one has
00:32:58
been able to, to quite make it work.
00:33:01
But I, I, I wouldn't rule it out yet.
00:33:02
I think we've seen a bunch of people try.
00:33:05
Nothing has really stuck. I think Whatnot is interesting.
00:33:09
I think it's a scaled business. My sense is a lot of what's
00:33:12
happening on Whatnot are people buying things speculatively so
00:33:16
they can resell them and and as opposed to buying things they
00:33:21
necessarily want for themselves. So that's like a different, a
00:33:23
different phenomenon altogether. But but you know, I haven't
00:33:27
given up on it. But you're right to point out it
00:33:29
hasn't happened yet. Yeah, I guess the gambling
00:33:32
economy thesis, are you staying away from that?
00:33:34
I mean, it's clearly, I mean, prediction markets loom super
00:33:38
large. I mean, yeah, I think you're
00:33:40
correct that what not is fuelled in part by some of that same
00:33:44
behavior. Yeah.
00:33:46
Yeah. Is that a thesis you've embraced
00:33:48
so? I think there are a lot of, if
00:33:51
you're setting out to build a company, there's lots of
00:33:53
companies you could set up to build.
00:33:55
If you're setting out to invest in a company, there's a giant
00:33:57
menu of companies you can invest in.
00:33:59
And I think one of the criteria that I apply that not everybody
00:34:02
does is I want to be proud of my association with the company.
00:34:06
And and so I'll give you like a classic example of a company
00:34:09
that I admire the financial metrics on, but would never
00:34:13
invest in because I just didn't want to be associated.
00:34:16
And that was a company called Jewel.
00:34:18
You may remember that, you know, they made they made vaping pens
00:34:20
by the way people invest in it really made phenomenal returns.
00:34:24
It it wasn't for me. And I think that when we look
00:34:29
back on some of these prediction markets or gambling sites, there
00:34:33
are a bunch of parallels that make me somewhat uncomfortable.
00:34:36
That's not to say they aren't great businesses, and that's not
00:34:37
to say someone else may not be a genius for investing in them,
00:34:41
but but they're not for me. And, and, and by the way, I
00:34:43
happen to have three teenage kids, two boys and one girl.
00:34:47
And I can see among particularly my sons and their friends, it,
00:34:53
it sort of reminds me of Jewel. It's like Jewel is purportedly
00:34:57
an aunt a, a smoking cessation device for adults, but he
00:35:00
actually looked at who was actually using Jewel with all
00:35:02
its various flavors. It was like teenage kids.
00:35:04
They're essentially introducing teenage kids to a new form of
00:35:07
smoking, which was not good for those kids.
00:35:10
And I think in many respects these new prediction markets
00:35:14
which are allowing betting on anything all the time, I suspect
00:35:19
one of their big customer bases are teenage kids who are
00:35:23
particularly susceptible to the interesting characteristics
00:35:26
about the. Children angle.
00:35:28
Yeah, they're not like gating it.
00:35:30
They, so they, I think very recently they've introduced some
00:35:33
KYC things, but which require you to prove that you're of a
00:35:37
certain age. But you know, if you ask any
00:35:40
teenager about, you know, some restriction that was imposed on
00:35:43
them by someone or something, I can pretty pretty much guarantee
00:35:46
they have like three ways around it.
00:35:49
And so I'm not sure how effective the KYC stuff is.
00:35:51
I I, I confess to my ignorance. I haven't looked into it
00:35:53
carefully, but even if it's a modest percentage of their
00:35:56
business, it's just not something I would necessarily
00:35:58
want to be associated with. So you're staying away from like
00:36:01
the whole theme? I mean, I'm watching it and and
00:36:03
by the way, maybe they find a way to eliminate that stuff
00:36:06
entirely. And, you know, like if there
00:36:08
were a jewel. You agree prediction markets,
00:36:11
which I agree in sort of quotes, which are becoming sort of
00:36:15
sports gambling sites are one of the major sort of consumer
00:36:19
themes of the. Year oh it it's they've grown
00:36:21
enormously yes no question about it I mean they they I mean I
00:36:24
think what what held gambling back was all the regulation and
00:36:29
a couple of really sharp entrepreneurs found a loophole
00:36:32
around the regulation where they can make it available to
00:36:34
everybody instantly without having to go through any of the
00:36:36
regulatory processes or reviews and like lo and behold it's huge
00:36:39
so yeah not surprising do. You think gambling was always in
00:36:42
the crosshairs? I guess they did a good job of
00:36:44
making it seem like, you know, elections and everything before
00:36:47
it became about gambling. It I mean, sorry, I mean sports,
00:36:53
sports in particular, yeah. I mean, look, the difference, of
00:36:56
course, as I understand it is that there's they they don't
00:36:59
take a side of the bet. It's it's you know, they're just
00:37:01
matching better. So it's it's a marketplace and
00:37:03
that you're going to. See the utility of like
00:37:06
predictions in a way that predictions of sports don't?
00:37:10
It doesn't feel. Like there's any utility like.
00:37:12
What repricing the Vegas line, you know?
00:37:15
Yeah, I, I, I'm, I, I suppose in in certain instances,
00:37:19
understanding what the world thinks on average about
00:37:22
something is useful, but to whom?
00:37:25
Other than a speculator, maybe not so clear.
00:37:30
Any any other categories or theses that you guys are really
00:37:33
chasing at the moment within consumer specifically or no
00:37:38
consumer or more? Broadly.
00:37:41
So the other one that I'm really excited about is robotics.
00:37:45
And so we, as I've talked about with you and others in the past,
00:37:51
like we tend to be very road map oriented where we try to build a
00:37:53
hypothesis and test the hypothesis through experts about
00:37:57
some change happening in the world.
00:37:59
And often times the most compelling ones are the simplest
00:38:01
ones. And so, you know, by way of
00:38:04
example, 20 years ago, as we saw the consumer Internet emerging,
00:38:08
we thought like, oh, consumers are going to contribute content
00:38:11
to media properties online, where the media properties have
00:38:14
the benefit of being media companies without having to
00:38:16
actually spend money on content. And that led to investments like
00:38:18
Yelp and LinkedIn and Pinterest. And it was like a really basic
00:38:21
idea. And once you have the basic
00:38:24
idea, often an insight shared with an entrepreneur, you try to
00:38:26
go find all the companies that are not competing with each
00:38:28
other that are consistent with this idea.
00:38:30
A really simple minded, like one sentence view on robotics is
00:38:34
that there will be probably 10 times, maybe even 100
00:38:39
times more robots on planet Earth 5 to 10 years from now
00:38:42
than there are today. And, and there are essentially 3
00:38:46
big challenges in robotics. 1 is can you make hardware that
00:38:50
functions as a robot. 2 is can you make the hardware move.
00:38:55
They call that the locomotion problem.
00:38:57
And three is can you make the hardware manipulate things in
00:39:00
the real world and the hardware itself essentially works.
00:39:04
There's tons of hardware. A lot of it is being made in
00:39:07
China. Increasingly companies are
00:39:08
trying to make it in the US. So that was problem one.
00:39:11
Problem 2 is the locomotion problem.
00:39:13
And now as evidenced by by Waymo, another recent Bessemer
00:39:16
investment, like we can make hardware move without issue.
00:39:21
It's it's solved. And you see it in robot dogs,
00:39:24
you see it in autonomous cars in all categories.
00:39:28
And that was a machine learning problem, which is essentially
00:39:30
solved. It isn't perfect, but it's it's
00:39:32
close enough that it's usable. And the unsolved problem for
00:39:36
robotics is around manipulation. So we have these things in the
00:39:39
real world, but we need them to interact with objects to perform
00:39:44
functions that in today's world, many humans do.
00:39:48
But they're pretty crappy functions.
00:39:50
Fold laundry. Yeah.
00:39:51
I mean that's the consumer one. But there there's an enormous
00:39:54
number of, I mean there are many companies working on it and and
00:39:57
they have very various approaches to it.
00:40:00
And I'm quite convinced that the the ironically, the LLMS of
00:40:05
today have solved a bunch of what I think of as white collar
00:40:09
type jobs before they've solved the blue collar type jobs.
00:40:13
And in fact, an MIT professor I met with said to me the the hard
00:40:17
things are actually relatively easy in AI.
00:40:19
Like AI can do PhD level statistics hell of a better than
00:40:23
I can. Yeah, the easy things like
00:40:26
picking up this glass, moving it here and moving it back without
00:40:28
spilling the water, which is so easy for any human to do, is
00:40:31
actually hard to do with AII believe it will be solved.
00:40:35
And I, I I can't tell you exactly when and I can't tell
00:40:37
you exactly what the path is and so.
00:40:39
That is the basic thing. You know, I have a four month
00:40:41
old right now and obviously babies spend so much time
00:40:44
figuring out like, how do I hold on to something that I think
00:40:47
experiencing the human brain form, you're like, Oh yeah,
00:40:50
obviously it's it's a lot of work navigating the physical
00:40:53
world. The Are you bullish on humanoid
00:40:56
robotics? I think eventually.
00:41:00
We wrote a little barricade, just you don't have to agree
00:41:03
with it, but we were somewhat skeptical and newcomer about the
00:41:06
humanoid form factor. So I think it's it's fun for
00:41:09
humans because it reminds us of ourselves.
00:41:12
I think the the the autonomy of robotic form factors will be
00:41:16
wildly varied, and there may be certain instances where the
00:41:19
humanoid form factor is particularly powerful.
00:41:21
I think it will come later because there are a bunch of
00:41:24
challenges with humanoids that don't exist and other form
00:41:25
factors of robots, but that it's not going to get in the way of
00:41:29
these robots being wildly helpful or productive for, for
00:41:32
humankind. And so anyway, so, so then the
00:41:34
question is like, how do you invest in it?
00:41:36
And, and we, we've started, we made an investment in a company
00:41:39
called Foxglove. And if you believe there will be
00:41:42
100 times more robots, then we're going to need the same
00:41:45
kind of infrastructure tooling for robots that we have for
00:41:48
computer servers. And there's companies like
00:41:51
Datadog that make some of the infrastructure tooling for
00:41:55
computer servers, and we're going to want that same kind of
00:41:58
capability for robots. And that's essentially what
00:42:00
Foxglove does. Makes it very easy to.
00:42:02
And so then you avoid the particulars of which robot
00:42:05
format. Exactly, It doesn't matter which
00:42:07
robot companies or, or how they work, but if if they have the
00:42:10
leading solution for the underlying infrastructure so
00:42:12
that the robotic makers don't have to go build this
00:42:16
infrastructure software, they can just take it off the shelf.
00:42:19
Boxcliff should be a winner. So that's exactly right.
00:42:21
So that's one area of investing. And then the second one is, OK,
00:42:24
how do we figure out how to solve this manipulation problem?
00:42:27
Once manipulation is solved, these robots become wildly more
00:42:30
valuable to all of us. And so those are generally even
00:42:34
earlier stage investing often times in academicians or
00:42:39
inventors who are trying to solve this, this unsolved
00:42:41
problem and a bunch of companies who've raised billions of
00:42:44
dollars. My general view is it is nice to
00:42:47
have raised lots and lots of money so you can, you know, pay
00:42:50
people in Mexico to, you know, mimic the behavior of a robot
00:42:54
and record it and collect all this data.
00:42:56
I suspect the winner will be much more innovative than these
00:43:01
companies using brute force to collect data.
00:43:04
But it's early. And obviously, I don't know for
00:43:07
sure, but we're trying to find really smart innovators in this
00:43:09
category. I want to talk about venture
00:43:12
capital as an asset class for a second.
00:43:14
Bessemer, you Bessemer as a firm has been investing for more than
00:43:18
a century, right? Is that?
00:43:20
Yes, we came out of a family office which which has been,
00:43:24
it's now I think since 1911, so 115 years.
00:43:27
I don't know, I think Andreessen Horowitz just published a piece
00:43:30
like defending sort of the mega firm or scale and venture
00:43:33
capital. What?
00:43:35
Yeah. What is your sense of sort of, I
00:43:37
don't know, the the rise of like the mega fund and how venture
00:43:40
capital is being practiced today?
00:43:43
Do you think there is just sort of, you know, you're a huge,
00:43:45
you're a big multi stage fund yourself.
00:43:47
Like how do you think about scale and venture capital?
00:43:50
I think scale is critical in venture capital.
00:43:54
Does that mean it's required to have a successful fund?
00:43:57
No. Do I think it's required to have
00:43:59
a successful firm and a collection of funds?
00:44:02
Yes. A bunch of things have changed
00:44:05
even since I've been doing it. So I've been doing venture
00:44:07
capital for 25 years and 25 years ago, if you were an
00:44:11
entrepreneur, you found out about venture capital and
00:44:14
venture capital firms through word of mouth and some friends.
00:44:17
And it was a little bit of an old boys network and it was
00:44:20
literally boys at the time. We've gotten wildly more
00:44:24
diverse, thankfully, although it's still the predominantly an
00:44:29
old boys network. And and therefore having scale
00:44:33
and, and building brand and marketing awareness and and
00:44:36
content like none of that mattered back then because it
00:44:39
was it. Was Now you're forced to come
00:44:40
slum it on my podcast I. Mean so, so now there's the
00:44:45
Internet and, and, and people discover and learn in so many
00:44:49
ways. And, and if your, if your firm
00:44:52
or your brand is only in the conversation 1% of the time and
00:44:57
someone else's is in the conversation 10% of the time,
00:44:59
that's an advantage. And, and so I think there's this
00:45:03
counter narrative, which I think is wrong.
00:45:06
And that is if you want to have a 10X fund, it has to be a small
00:45:10
fund because it's virtually impossible to have a 10X fund on
00:45:12
a $3 billion fund. But if it's a $200 million fund,
00:45:15
you could actually end up with a 10X fund.
00:45:18
I think that's right, except what it misses is all the $200
00:45:22
million funds that are going to be a, you know, a .1 X fund.
00:45:26
And that if you have somehow the luck or oppressions to invest in
00:45:30
the $200 million fund, that does become a 10X fund.
00:45:33
Good on you. But that's maybe as much luck as
00:45:36
it is skill. Whereas if you want repeated
00:45:39
sources of competitive advantage, you need scale and
00:45:42
you need to be in the conversation all the time.
00:45:45
And it's impossible to do that as a boutique firm with four or
00:45:48
five people who who do all the work themselves.
00:45:52
Or you're leaving money on the table, right?
00:45:54
I mean, there's like I sort of recently gave Benchmark a hard
00:45:57
time in a story, which is they sort of have the brand and
00:45:59
they're, they're leaving it some of on the table by not investing
00:46:03
these later stage rounds. Like you, there are a couple
00:46:05
firms, you know, USV or whatever that have built brands that if,
00:46:09
if it's if the reason you need scale is because you need to
00:46:13
stand out in people's minds. There are firms that built
00:46:16
brands that are smaller. They they, there are, but I
00:46:19
would argue that the brands of the small boutique firms
00:46:23
relative to the rest of the industry 20 years ago were much
00:46:25
bigger and they're increasingly smaller, getting crowded out by
00:46:28
by by larger firms. But that's just one reason why
00:46:31
scale matters. The second reason why scale
00:46:33
matters is because time frames to evaluate opportunities have
00:46:37
shrunk substantially. 25 years ago, I might meet a company and
00:46:40
we might get to know each other over months before they were
00:46:45
ready to raise money or we were ready to invest in the company.
00:46:49
Increasingly, that number of months is shrinking to days.
00:46:53
And if you want to make good decisions, you have to arm
00:46:56
yourself with meaningful data. And even for an early stage
00:46:59
company that has 16 early trial customers, like that's the whole
00:47:05
business. You want to talk to 10 of the 16
00:47:10
to really know what's going on. And if you're in a small firm
00:47:14
with limited resources, it's kind of impossible to do that in
00:47:18
three days time. Whereas a a larger firm with
00:47:22
more resources is going to have talked to every customer and ten
00:47:25
others to understand like, is this an easy thing to sell and
00:47:28
what a prospects think? And that's just an incredible
00:47:32
information edge. And, and I've seen it recently
00:47:35
and I don't want to name names, but there was one instance where
00:47:38
we were looking at it was a classic Series A investment and
00:47:41
I was working on it with several colleagues.
00:47:43
And so we actually called every company that had its logo on the
00:47:48
this prospect website and we found out that not a single one
00:47:52
liked the product. And these were the logos of on
00:47:55
the company's website. And another really well known
00:47:59
respected firm at a, at a smaller scale than us end up
00:48:03
investing in the company. Because I think it was
00:48:06
impossible that they had to sort of go on more faith and to kind
00:48:09
of make a judgement, a snap judgement.
00:48:11
Like is this entrepreneur talented, truthful and so forth.
00:48:16
And I think they made the wrong call.
00:48:18
I think if they had the same data that we had, they would
00:48:20
have made a different decision. And so I think there are other,
00:48:23
it's not just brand and being in the conversation that's that
00:48:28
scale gets you, it also gets you information.
00:48:31
Does honesty still matter or I feel like there's a nihilism.
00:48:33
I mean you, I mean, we're it came up sort of in the
00:48:37
prediction market piece, but you know, you only recommend in my
00:48:41
last podcast was sort of or in a recent podcast described venture
00:48:46
capitalist as sort of amoral financiers.
00:48:48
Like, I mean, I guess you could be amoral and still think sort
00:48:52
of honesty matters, but how how much in this sort of world
00:48:55
where, I don't know, people are playing pretty fast and loose.
00:48:58
Do you think it matters to back honest founders?
00:49:01
Hi. I mean, look, do you want to be
00:49:04
friends with honest people? Do you want to be married to an
00:49:06
honest person? Like that's a value judgement.
00:49:08
I mean, I do, but like, you can have your own values there.
00:49:12
That said, I think more people than not resonate with people
00:49:17
who do what they say, say what they do, are honest.
00:49:19
And so I think is it impossible to build a company as like a
00:49:22
pathological liar? It's not impossible, but I think
00:49:25
it's probably harder because it's harder to have loyalty and
00:49:29
have people bought into the mission and so forth.
00:49:31
But like I will say it can you, there's also this, this age-old
00:49:36
Eddy in Silicon Valley of faking it till you make it.
00:49:38
And so could you, could you fake it and have 15 fake ish
00:49:41
customers raise a bunch of money and then with that money turn
00:49:44
things around? It is possible.
00:49:46
It has been done, but it's not the kind of gamble that I want
00:49:50
to take. And so it may well work out for
00:49:53
this company that we didn't invest in that now that they've
00:49:56
raised money from a a really well known investor, they're
00:49:59
going to use that to reform their ways and become a great
00:50:01
company. But I'd bet against it.
00:50:04
You know, one advantage you're saying you get with venture
00:50:07
capital scales you have, you know, all these people running
00:50:09
around and doing all this diligence, do do you still need
00:50:12
all the people or how much are you relying on analysts today
00:50:15
versus using? AII just published some research
00:50:20
that Venture 5 had done, and they saw that analysts and
00:50:24
venture capital firms saw their salary fall.
00:50:26
Like, do you, do you think you're hiring fewer of them
00:50:28
because of what AI can do? In the short term, no, but
00:50:34
advantage three of a venture capital firm of scale is that we
00:50:37
have the resources to invest substantially in AI tooling and
00:50:42
AI tooling is both a source of data.
00:50:45
As a, as a, as a silly example, imagine if you knew any time
00:50:50
someone of a certain seniority level left Google and, and
00:50:54
changed their LinkedIn profile to be, you know, stealth
00:50:56
founder. That's really valuable
00:50:58
information for a venture capital firm to have.
00:51:01
We have tooling that tells us that for every single person.
00:51:04
If you're a little boutique venture firm with no resources
00:51:08
to spend on tech and innovation yourself, you're not going to
00:51:11
have that information. You're depending on your network
00:51:15
to get you close to those people at exactly the right time.
00:51:18
That worked really well when that's what everybody was doing.
00:51:21
But now when there's a group of of competitors who have access
00:51:24
to this incredibly rich information generated by their
00:51:27
own AI tools, you're at a disadvantage because they also
00:51:30
have the same personal networks, but they have tooling on top.
00:51:33
And so to date, we've used our tooling to make all of us more
00:51:37
productive. And I would say we're in inning
00:51:40
one or two in a nine inning baseball game that probably even
00:51:43
goes into overtime. So it's still early, but I have
00:51:48
a hard time imagining in the short term, the people in
00:51:51
venture capital mattering less because fundamentally, companies
00:51:55
are still run and founded by humans, and humans don't really
00:51:59
want to interact with or build a partnership with an AI bot.
00:52:02
They want another human. And so having people who are
00:52:06
smart, ambitious, hard working and compelling as humans within
00:52:10
our firm is really important because those are the people who
00:52:12
have to build relationships with today's founders and the
00:52:16
founders of the future. Because fundamentally what we do
00:52:18
is we become, we become partners.
00:52:20
Like it's why it's why it's in our name, Bessemer Venture
00:52:22
Partners, not Bessemer Capital. Because we're looking to partner
00:52:26
with these people in a very human partnership to help them
00:52:29
through the UPS and the downs, solve problems along the way, be
00:52:33
the shoulder to cry on, be the person to bring them back down a
00:52:36
little bit when they're celebrating a little bit too
00:52:38
much. And those are all very human
00:52:40
things. If the venture capital industry
00:52:43
is, you know, inevitably booms and busts over hype and under
00:52:48
under hype sometimes, where do you think we are in sort of, I
00:52:52
don't know the, the, I think we're still rooted in the.com
00:52:55
story, but like, where do you think we are today in that sort
00:52:58
of hype journey? I think we are, well, I mean it
00:53:02
is cyclical and it the growth of AI and the emergence of these
00:53:08
companies growing at faster rates and reaching bigger scale
00:53:11
than anything we've ever seen before echoes to me the early
00:53:15
days of the, the Internet, the consumer Internet.
00:53:18
But the scale and speed is an order of magnitude or more
00:53:22
bigger and faster. And so I think we are on our way
00:53:26
up a a well justified hype curve now, much like in the early days
00:53:32
of the Internet, there were hundreds if not thousands of
00:53:35
Internet companies that failed. They had the wrong model, they
00:53:38
had the wrong ideas, they were focused on the wrong things, but
00:53:41
got funded in what was objectively a gold rush of
00:53:45
sorts. We will see the same thing
00:53:47
happening here, but I think much like some phenomenal companies
00:53:51
emerged from the advent of the Internet like Google and Meta
00:53:57
and many others, we will see massive companies emerge from
00:54:00
this and I think they will be even bigger than the ones that
00:54:02
we saw in the Internet. And so I think it's a great time
00:54:06
to be investing. It's an amazing time to be
00:54:08
alive. And I'm so curious what some of
00:54:09
these things will actually evolve into.
00:54:13
But but nonetheless, venture remains the same, which is that
00:54:17
it's a long game. And when we make an early stage
00:54:19
investment today in on an idea like there'll be 100 times
00:54:23
more robots in the future than there are now, that may not
00:54:26
fully come to fruition for 7-8 or nine years.
00:54:29
And if as an investor in venture capital, you're looking for
00:54:32
returns every quarter or every year, you're in the wrong asset
00:54:35
class. This is a, this is a patient
00:54:36
person's game, even with these incredible momentum ideas like
00:54:42
AI emerging. And so there'll be some
00:54:44
instances where things just work immediately and it goes up to
00:54:47
the right and you can actually see and taste the value right
00:54:49
away. Most of the time it's a slow
00:54:52
burn process until suddenly something appears to be really
00:54:55
large when in fact it was growing a little bit along the
00:54:57
way and just no one was paying attention to it.
00:55:00
Jeremy, thank you for coming on the Newcomer Podcast.
00:55:03
My pleasure. Thanks for having me.
00:55:05
Thanks so much for listening to the Newcomer Podcast.
00:55:08
I'm Eric Newcomer. Follow us on Substack at
00:55:10
newcomer.co. Hope you enjoyed the episode.
00:55:13
See you next week.