Latest / The Learning Corner by Precursor / Episode #42: The "Winners" in AI Today, DaaS Is Not Venture Backable, Seed Rounds Grow As Startups Shrink

Transcript

Episode #42: The "Winners" in AI Today, DaaS Is Not Venture Backable, Seed Rounds Grow As Startups Shrink · The Learning Corner by Precursor

• 0:00
  There's this weird circular logic that's been happening probably since 2010, which is why
• 0:05
  round sizes get bigger?
• 0:06
  Is it because funds get bigger, or is it because founders need more capital?
• 0:09
  It's the week of July 28.
• 0:11
  Welcome to the Learning Corner, a Precursor Ventures podcast.
• 0:19
  Welcome to the Learning Corner by Precursor.
• 0:21
  I'm Mia Farnam, and I'm joined by RGP and managing partner Charles Hudson, where we will
• 0:26
  be talking through the articles and Goodreads that caught our team's attention over the past
• 0:30
  week.
• 0:31
  Our first piece is a blog post written by Elan Gill who claims that AI markets are the
• 0:36
  clearest they have been in a few years with new markets finally ripe for disruption.
• 0:41
  Markets like foundational models, code, legal, medical scribing, CX, and search are
• 0:47
  specifically some of the markets where more likely winners have emerged.
• 0:50
  Elad cautions folks that just because some big players have come forth today doesn't mean that
• 0:55
  there won't be newer ones in the future.
• 0:57
  However, it will be rather difficult to fully dethrone some of the companies he mentioned.
• 1:02
  So like the Anthropics, Quad, Mistrals of the world.
• 1:05
  Of course, I agree with Elon saying that there will be many more successful AI companies to
• 1:10
  emerge, some within the more crystallized markets he spoke about in the piece.
• 1:14
  From your perspective, however, if you're running around town saying that you are
• 1:19
  building the best AI fund in the market or plan to, and you're not in one of the big names that
• 1:25
  have already been mentioned, how much will that matter was a question I had.
• 1:30
  Yeah.
• 1:31
  First of all, Elad's obviously been super successful as an investor.
• 1:34
  When I read the post, was like, the capital markets have decided the companies they wanna
• 1:39
  support, and a lot of wood has been put behind some of these arrows.
• 1:44
  I don't know that the capital market level of conviction and the customer level of conviction
• 1:49
  are the same because most of this technology is really new.
• 1:53
  And when I looked at that list, I was like, okay.
• 1:55
  The foundation model piece I understand because capital is an essential input to those
• 2:00
  businesses.
• 2:00
  Like, you cannot build the infrastructure you need to continually develop new models without
• 2:05
  large pools of capital.
• 2:06
  And I think most of the capital has co alesced around, as Ilod said, you know, a couple of
• 2:11
  commercial players than others that are maybe being supported for nationalistic or strategic
• 2:15
  reasons.
• 2:16
  For the more application level stuff, it's unclear to me.
• 2:19
  It's very clear to me that if you were to try to start a legal AI company today, two things
• 2:24
  are likely true.
• 2:25
  One, the first question people are gonna ask you is, what about Harvey?
• 2:28
  And second, if you started looking at what are the venture firms I could actually go pitch to
• 2:32
  raise money, you'd realize most of the venture firms you'd wanna pitch have already made an
• 2:36
  investment of some sort in a company in that category, if not Harvey, one of the others.
• 2:41
  And so this is more, I think, one of the oddities of venture capital is once capital
• 2:45
  lines up behind certain companies, most funds are unlikely to invite a portfolio company
• 2:50
  conflict by investing in a similar company, and they're pretty much wedded to that company.
• 2:55
  And I've seen situations where sometimes the VC market gets it really right early, and the
• 3:00
  company that's anointed ends up going on to be the winner.
• 3:02
  And I've seen also quite a few cases where the capital lines up behind the company that the
• 3:08
  VCs believe will be the winner, and you fast forward five to seven years and the world's
• 3:11
  different.
• 3:11
  And there was a dark horse company or a bootstrap company or a company that wasn't as
• 3:15
  highly regarded that ends up being the winner.
• 3:17
  And I've just seen that play out.
• 3:19
  But usually by the time that happens, the attention has moved on to the new hot thing.
• 3:23
  So I think if you read that article through the lens of these are markets where large venture
• 3:28
  funds and private equity and sort of the capital markets have made a decision about who
• 3:32
  they're going to back.
• 3:33
  I think that's true.
• 3:34
  I think customers have something to say about this too in the end.
• 3:37
  I wanna go back to that piece that you just mentioned about portfolio conflict and that
• 3:40
  typically being a reason why one company will not do another investment in an adjacent or
• 3:45
  direct area.
• 3:46
  I know you wrote a post about how you might predict this to shift and change in this new
• 3:51
  investing area.
• 3:52
  Can you elaborate a little bit about it for people who haven't read the piece?
• 3:55
  Yeah.
• 3:56
  So my argument is in the past, a lot of firms were hesitant to do portfolio company conflicts
• 4:01
  for a bunch of reasons.
• 4:02
  One, you know, most multistage funds take board seats, you can't have the same person on we'll
• 4:07
  say, for example, you probably couldn't be on the Google board and be on the OpenAI board
• 4:10
  without it being a little awkward given how important AI is to both those companies.
• 4:14
  So most venture firms had a policy, hey.
• 4:16
  We're only gonna invest in one company in the category.
• 4:18
  And once we have our bet, that's our bet.
• 4:20
  And it has this nice two sided property in that as the company receiving the money, you have
• 4:24
  some assurances that that investor's not gonna go back your biggest competitor.
• 4:29
  And as a venture fund, it forces you to be honest and say, if we make this investment,
• 4:33
  we're foreclosing the right to make another investment in this category.
• 4:37
  So we better be sure that we're excited about this company before we say yes.
• 4:40
  And that's been a norm.
• 4:41
  Not not not a rule, but a norm in most venture funds.
• 4:44
  And in my article, was like, well, look.
• 4:46
  Venture funds are so big now.
• 4:47
  This might not be a sustainable norm because the need to be in the biggest, best, most
• 4:52
  important companies is becoming existential.
• 4:55
  If you're a big multibillion dollar venture fund and you're not an anthropic or OpenAI,
• 5:00
  it's probably gonna be tough on a relative basis to keep up with your peers.
• 5:03
  And so if the norm of non competition holds, everything Elad said is probably true.
• 5:08
  If people are willing to break that, there could be companies that come out of nowhere or
• 5:11
  are funded differently that become interesting, and venture firms might say, hey.
• 5:15
  I know I'm already in company x in this category, but we need to be in company y from a
• 5:19
  return standpoint.
• 5:19
  I'm very curious to see if this norm persists.
• 5:22
  I can't think of any companies off the top of my head right now, but if you think about, I
• 5:26
  don't know, Facebook to Myspace, WebBand, Instagram, the timeline horizons are a little
• 5:30
  bit stretched out.
• 5:31
  So maybe if there is the same investor wanting to hop into that one, like, maybe that could be
• 5:36
  it.
• 5:36
  But I feel like the difference here is that the timelines are a little bit compressed.
• 5:40
  So folks who maybe are in some of the big winners, if something happens and a big shift
• 5:44
  occurs, they could in, like, two years later create an entire rationale that jumping into
• 5:50
  the next big winner in the same category is worth it if it's gonna return a point of fund.
• 5:54
  Or those that say, hey, we have partner x is on that deal, partner y is on this deal.
• 6:00
  We promise, promise, promise that they don't talk and share and that we have some kind of,
• 6:04
  like, ability to firewall the data and you can trust us.
• 6:07
  I mean, firms already some firms already have this.
• 6:09
  They have a series a and b firm that does, you know, early stage investments, and then a
• 6:13
  growth fund that has the ability to both invest in portfolio companies and non portfolio
• 6:17
  companies.
• 6:18
  And that growth fund presumably has access to the information at the fund about their own
• 6:23
  portfolio, but is also meeting other late stage companies.
• 6:25
  And so, you know, there's some venture friends who go, oh, our growth team is a different set
• 6:28
  of people and Yeah.
• 6:30
  You know, you can trust us.
• 6:31
  The growth team can do different stuff than the main so I think the norm of no conflicts is
• 6:35
  probably gonna look like a relic of the past in a few years.
• 6:38
  Yeah.
• 6:38
  Elad mentioned some other emerging markets like financial tools, sales tooling, security, and
• 6:44
  looks forward to seeing more winners emerge across those industries.
• 6:47
  We talk a lot about the possible reasons why there hasn't been complete clarity on major
• 6:52
  players in every industry.
• 6:53
  And Elon says a lot of that has to do with things like go to market, buyer behavior, which
• 6:57
  we've dissected a lot, teams, and simply the time to build and ship products.
• 7:02
  Anything else you feel like could be a possible reason here that's preventing the anointing of
• 7:06
  key AI winners across every industry?
• 7:08
  I also think people have to remember this stuff as fluid.
• 7:12
  I think there was a period of time where people like, well, Anthropic isn't really that
• 7:16
  competitive OpenAI.
• 7:18
  OpenAI is, like, the clear front runner, and I think that's changed.
• 7:22
  At least my sense is the perception.
• 7:24
  And some of that, I think, is Claude Code.
• 7:25
  I think some of that's MCP.
• 7:27
  But my sense is people's perception that Anthropic is destined to be the number two
• 7:31
  player.
• 7:31
  Like, that is not the case anymore.
• 7:33
  And even if I look at the the coding tools market, we've seen in the last five years,
• 7:38
  Copilot is obviously gonna win because it's got Microsoft and it's got GitHub.
• 7:42
  Yeah.
• 7:42
  That's not gonna work.
• 7:42
  Oh, ChatGPT is really good at code.
• 7:45
  Oh, no.
• 7:45
  There's Cursor and Windsurf and Replit.
• 7:48
  And then lately, everyone's like, know what's really good?
• 7:50
  Cloud Code.
• 7:50
  Cloud Code's really good.
• 7:52
  And so I think when I read the piece, was like, I think we know the circle of companies that
• 7:57
  have the resources and brand to be competitive in all these categories.
• 8:01
  I don't think we know who's gonna win.
• 8:02
  I'd say a lot of people have counted Google out in search and depending on what you believe
• 8:06
  about the numbers they put out recently.
• 8:08
  I don't think Google's gonna roll over and seed the search market to AI upstarts.
• 8:12
  It's just not gonna happen.
• 8:14
  But I think this is an interesting time because these capital allocation decisions are sticky
• 8:19
  because you can't trade out of these positions.
• 8:21
  But we're in a very dynamic market where the facts on the ground change every day, and I
• 8:25
  wouldn't be surprised if later this year, we're talking about companies we haven't heard of
• 8:29
  that have broken through in some of these categories that appeared to be settled.
• 8:32
  Our next topic is a post by Oren Hoffman titled data is actually not a great VC backed
• 8:37
  business, where he shares how his entire tune around investing in pure play data businesses
• 8:42
  has completely shifted over the last few years.
• 8:45
  Most of his original hypothesis was rooted in the prediction that data buyers would grow in
• 8:50
  number and enlist the support of companies whose main value add was data lakes and data
• 8:54
  scientists.
• 8:55
  Turns out, while there were plenty of hedge funds that purchased alternative data, it
• 8:59
  wasn't done at the scale that he predicted.
• 9:01
  And other industries he assumed would be clear buyers like real estate, retailers.
• 9:06
  They just never really invested at the rate he anticipated.
• 9:09
  He goes on to conclude that data businesses are good businesses, perhaps for private equity,
• 9:15
  just given the predictable revenue and high trading multiples, but not venture backed
• 9:19
  businesses, as they don't need tens of millions of dollars in equity financing to be fully
• 9:24
  stood up.
• 9:25
  I felt like this post really landed, specifically because Oren has spent so much of
• 9:28
  his career in this DaaS space from an investor and builder perspective.
• 9:33
  How many investors do you think are optimistically choosing to disagree with this
• 9:37
  article, especially with the added layer of AI right now?
• 9:41
  You know, there are very few people I know who've spent more time on data businesses than
• 9:45
  Oren, and he has no incentive to say this.
• 9:48
  He's made a lot of money and been very successful in these businesses.
• 9:52
  He has no reason to say this other than that he believes it.
• 9:55
  And the number of companies I've met who are like, well, we're gonna have this data, and
• 9:58
  we're gonna sell to a bunch of people.
• 10:00
  And I'm always like, well, how confident are you that the pool of buyers is as deep as you
• 10:04
  say it is?
• 10:05
  And in many cases, it isn't.
• 10:06
  And there have been a bunch of other reasons too.
• 10:08
  Found some of this data doesn't change that often.
• 10:10
  What's the value subscribing to it?
• 10:12
  In a lot of cases, there's a couple of providers.
• 10:14
  But to Ornith's point, my sense has been a lot of data entrepreneurs end up being overly
• 10:18
  optimistic about the universe of people who will see value in their data.
• 10:22
  And some of it feels very like an ex post rationalization.
• 10:25
  It's like, well, because I've gone through all this effort to collect this data, there must be
• 10:29
  people who will find it valuable.
• 10:31
  And if I think about what is what appears to be valuable today, it's the ability to transform
• 10:36
  data into some other useful input.
• 10:38
  Whether that's, hey.
• 10:39
  We have a bunch of training data that you can use to train your LLM.
• 10:42
  We have a bunch of data we can use to train your self driving car.
• 10:46
  It appears that data that enables some other business process is valuable as an input, but
• 10:51
  as an input to another business, not as a business in and of itself.
• 10:54
  And somewhat of a commodity.
• 10:56
  And somewhat of a commodity.
• 10:56
  Like, I have a friend who used to be in the sports score business, and he's like, yeah.
• 11:00
  At one point, was really hard to get sports scores.
• 11:02
  Then it was pretty easy, and then you had box scores.
• 11:03
  And he's like, the incremental things you can do to differentiate yourself, other than for
• 11:07
  gambling, is, like, at some point, you get to the asymptote of this is good enough for most
• 11:11
  use cases.
• 11:11
  And I still hear people tell me, oh, we're gonna build this data business.
• 11:15
  We're gonna accumulate this data asset, and we're gonna license it.
• 11:17
  I'm like, to Orin's point, I'm like, you probably won't build a big business unless this
• 11:20
  data is gonna dramatically increase the market size.
• 11:24
  And I just think about when all the payment for order flow stuff happened with Robinhood and
• 11:29
  people were really upset about it.
• 11:31
  And I'm like, well, this isn't a new thing.
• 11:32
  Right.
• 11:33
  But, like, the existence of Robinhood expanded the sort of aperture for selling that payment
• 11:37
  for order flow data because you brought a bunch of new traders into the market and Robinhood
• 11:42
  users who hadn't been present before.
• 11:44
  So occasionally, yes, you can have a data business that gets where, like, something else
• 11:48
  have and it expands.
• 11:49
  That just hasn't been my experience.
• 11:50
  Yeah.
• 11:50
  I mean, I think Orin shared it.
• 11:52
  There are only 60 publicly traded DaaS companies in the world, and most of them were
• 11:55
  started forty years ago.
• 11:57
  And so not like founders who wanna build in data can wake up the next day and within a year
• 12:01
  amass so much data that's automatically useful some for multiple use cases.
• 12:06
  Again, are exceptions, and I think that was my last question I was gonna ask you, which I
• 12:09
  think you were half answering.
• 12:10
  What is that bar of which a company can surpass, which makes it very applicable to the
• 12:15
  needs of businesses today?
• 12:16
  I think Scale AI has done this.
• 12:17
  We did an investment in a company called Secuity.
• 12:19
  Anything else that comes in in mind for you?
• 12:21
  I'd say, like, I've met a bunch of people who are trying to build either custom underwriting
• 12:26
  or, like, a new credit score.
• 12:27
  And I'm like, well, we can all have qualms with TransUnion or Experian, the big three.
• 12:32
  And I'm like, part of the reason those businesses are valuable is because other people
• 12:36
  put faith in the score Yeah.
• 12:38
  And because there's an economy built around credit scores.
• 12:41
  If you're gonna build a more accurate credit score based on data you collect, collecting the
• 12:46
  data isn't the hard part.
• 12:47
  Convincing other people that your score has value and merit and that they should care about
• 12:51
  it is actually a much harder problem to solve.
• 12:54
  And so I would say, like, one of the fallacies I see in a lot of data businesses is people
• 12:57
  like, oh, there's this data source.
• 12:58
  If we use this and mix it with this other thing, you can make more accurate predictions.
• 13:03
  I'm like, great.
• 13:03
  Who's going to trust you enough to actually run their business based on the data that you
• 13:08
  provide?
• 13:08
  Yep.
• 13:09
  And they're like, oh, I'm like, well, if it's more accurate and nobody uses it, A, how do you
• 13:13
  know?
• 13:13
  And b, who cares?
• 13:14
  Right.
• 13:14
  Accuracy almost becomes completely subjective.
• 13:17
  Mhmm.
• 13:18
  Our last topic is another piece by Tomas Chungas titled Why Seed Rounds Are Growing As
• 13:24
  Startups Shrink, focused on the paradox of growth across seed rounds while the actual size
• 13:29
  of companies are shrinking.
• 13:30
  The first part of this piece is the growth in seed rounds.
• 13:34
  While nearly ten years ago, a sub $5,000,000 seed round accounted for nearly 70% of all seed
• 13:40
  deals.
• 13:41
  Recent data showed that we're sitting at nearly half of that figure today.
• 13:45
  Tomasz breaks down three forces that drive the seed round growth, VC funds becoming larger
• 13:50
  than ever, inflation, although that's a pretty small contributor, and then broader market
• 13:56
  forces like heightened competition for deals, increased pre c valuations, and a proliferation
• 14:01
  of seed funds and beyond.
• 14:03
  The irony of this is that despite the larger rounds taking hold, the size of startups are
• 14:07
  shrinking.
• 14:08
  Carta data showed that SaaS companies are 20% smaller at series a than in 2020.
• 14:14
  Just to start off the discussion here, Charles, were you as surprised about this dynamic after
• 14:19
  reading through the data?
• 14:20
  No.
• 14:21
  Because, you know, even internally as a team, we've always had set the bar.
• 14:25
  K.
• 14:25
  This is the maximum fund size that makes sense for precursor to get involved.
• 14:29
  You know, last time we chatted, we're like, maybe we need to raise the bar by 25% based on
• 14:33
  what we're leaving on the cutting room floor with round size.
• 14:36
  I I've always felt round size is there's this weird circular logic, Mia, that's been
• 14:41
  happening probably since 2010, which is that why round sizes get bigger?
• 14:45
  Is it because funds get bigger, or is it because founders need more capital?
• 14:48
  And what I've seen that happens is I remember when seed rounds used to be one, one and a
• 14:52
  half, and two was the ceiling, and then two and a half became the norm.
• 14:55
  And everyone's like, okay.
• 14:56
  Well, now seed rounds are two and a half.
• 14:58
  And because I have an ownership target, the price of the seed round has gone up because the
• 15:01
  founders aren't gonna take more dilution.
• 15:03
  So now I have to write a bigger check, and now I have to raise a bigger fund.
• 15:06
  Oh, now that I have a bigger fund, I can be a part of a $3,000,000 seed round because it
• 15:10
  doesn't break the model.
• 15:12
  Okay.
• 15:12
  Now seed rounds are $3,000,000.
• 15:14
  Oh, if I want extra cushion, now I need a 3 and a half million dollar.
• 15:17
  And and I think we've been in this feedback loop where round sizes have gotten slightly
• 15:20
  larger.
• 15:21
  In in response, people raise bigger funds, which has the effect of pushing round sizes
• 15:24
  even higher.
• 15:25
  And eventually, you get to a point where seed rounds are so big that multistage funds are
• 15:29
  like, well, I could write a check-in a round of that size.
• 15:31
  And I really do think this is the tail wagging the dog.
• 15:34
  I think seed rounds have gotten bigger because the people writing seed checks want to deploy
• 15:39
  capital in bigger chunks for two reasons.
• 15:41
  One is they want companies to be capitalized to do the work that they think needs to get done.
• 15:46
  And two, you can't deploy a $500,000,000 seed fund a million dollars at a time.
• 15:50
  It just doesn't work.
• 15:51
  And so as a result, you know, we do pre seed.
• 15:54
  And when I tell people sometimes, hey.
• 15:55
  We write million dollar checks.
• 15:56
  They're like, to whom?
• 15:57
  I'm like, there are people a million dollar rounds.
• 15:59
  I'm like, look.
• 15:59
  There are people out there who want smaller rounds.
• 16:02
  They are not the norm, and they are not the dominant force in the market.
• 16:07
  And I meet a lot of companies.
• 16:08
  I'm like, this is a business that I think would do better with less capital.
• 16:11
  It would force more constraints.
• 16:13
  They don't need it.
• 16:14
  To your point, they're not trying to hire a ton of people.
• 16:16
  But most founders, if you said, hey, you have two choices.
• 16:18
  You can take 2 and a half million dollars or you can take $5,000,000.
• 16:21
  Most of them would say, I would rather take $5,000,000 because that is more insurance, more
• 16:25
  time to live.
• 16:26
  But also, think what people forget is the flip side is a company that's had $5,000,000 and
• 16:30
  goes out for a series a, the expectations are different because you've had $5,000,000.
• 16:35
  Yep.
• 16:35
  And they're like, show me your work questions are gonna be much more pointed, and I think
• 16:39
  they should be.
• 16:40
  And I think that's one thing that we don't talk about enough is the expectations and benchmarks
• 16:45
  for success have continued to rise as we see rounds have become larger.
• 16:49
  And it's funny, I was having a conversation with one of my friends who works at, a larger
• 16:53
  multistage firm the other day, and she was sharing how talented founders can actually do
• 16:57
  so much more today with way less capital.
• 17:00
  Yep.
• 17:00
  And then my response to that was, well, that's completely indifferent to the strategy of you
• 17:06
  jumping into, like, larger seed rounds and and building My question is, what aspects of this
• 17:11
  dynamic might fuel positive outcomes for Venture, and then where could it create
• 17:15
  friction and unintended consequences?
• 17:18
  I think there's one primary beneficiary of these large seed rounds, which are talented
• 17:23
  teams that maybe have mediocre to okay ideas as a starting point.
• 17:28
  Because if you raise less money, the idea actually matters more because you don't have as
• 17:33
  much room to navigate.
• 17:35
  You're sort of stuck with the thing that you have.
• 17:37
  So the idea the initial starting point has to be better.
• 17:39
  If you raise $5,000,000, you have more time to explore.
• 17:43
  You could actually do two companies that require 2 and a half million dollars to get off
• 17:46
  the ground with one $5,000,000, Hedron.
• 17:48
  And I hate to say it.
• 17:49
  I think as an investor, if you're like, that company has $5,000,000.
• 17:52
  I cannot worry about them, but let's call it a year.
• 17:55
  For a year, if they don't do it, like, don't like, mentally, I can take that company and put
• 17:59
  it in the box of companies that is well resourced.
• 18:02
  And, of course, I care, but I don't have to have high anxiety.
• 18:05
  A company has fifteen, eighteen months of runway, you have to care about every month from
• 18:10
  the immediate month you give them money, which I actually think is better.
• 18:13
  Yeah.
• 18:13
  It's more stressful and it's harder, but I actually think it's better.
• 18:15
  But the flip side of that, I guess, is the folks who have so much more money, they
• 18:19
  probably burn at higher rates and then kind of feel empowered, as they should, to start
• 18:24
  doubling down in projects on the side just because.
• 18:27
  So do you actually get to save yourself on heartburn on those ones?
• 18:30
  Yeah.
• 18:31
  I think sometimes, like, nobody's paying attention to the companies that have
• 18:33
  experienced founders and raise a of money.
• 18:35
  Everyone's like, oh oh, they'll figure it out.
• 18:36
  But I'm like, well, somebody should be checking in and making sure that they're actually
• 18:40
  figuring it out.
• 18:41
  Well, we are all out of time today.
• 18:42
  Thank you, Charles, for jumping in.
• 18:44
  Thank you to our listeners for joining.
• 18:46
  Please like and subscribe, and we will talk with you all next week.
• 18:48
  Thank you.
• 18:49
  Bye.