Is non-consensus investing overrated—or the secret to venture returns?
a16z General Partner
is joined by Martin Casado (General Partner, a16z) and (General Partner, ) to unpack the debate that lit up venture Twitter/X: should founders and VCs chase consensus, or run from it?They explore what “consensus” really means in practice, how market efficiency shapes venture outcomes, why most companies fail from indigestion, not starvation, and the risks founders face when they’re too far outside consensus.
Timecodes:
00:00:30 - Consensus vs non-consensus investing
00:06:38 - Measuring market efficiency
00:11:04 - How consensus affects founders
00:15:22 - Is venture getting smarter?
00:23:19 - Defining “consensus” at seed stage
00:43:06 - How to frame consensus
00:49:02 - How to think about incentive alignment
00:53:06 - Has multi-stage won seed investing?
Transcript
00:00:30 - Consensus vs non-consensus investing
Erik Torenberg 00:00:30
So Martin it looks like you've helped spark a little bit of an existential crisis on venture Twitter, and I thought we'd all come here to talk about it.
Martin Casado 00:00:39
Great. Super. Let’s do it. I’m excited to be here.
Erik Torenberg 00:00:42
Why don’t we recap, Martin, from your perspective, what were you saying in that tweet?
What were you trying to say in that tweet? And then we can get into the great back and forth that you and Leo had and get into the conversation.
Martin Casado 00:00:54
So, let me paraphrase the tweet. The paraphrased version of the tweet is, it's dangerous to do non-consensus investing. Like that's a dangerous idea. The impetus of the tweet, which, by the way, wasn't well thought out, which I think a lot of the viral tweets happen to be not well thought out, is, you know, listen, I've been an investor for 10 years, I've done almost 200 investments, either as, you know, running the fund or being directly involved.
And it seems being blinkered to how VCs view companies is actually quite dangerous because you're so dependent on follow-on capital. And actually it reminds me a lot of being an academic. You know, I used to write a lot of papers, and like, you do all of this great research, but when you wrote the paper, if you didn't actually think about how the program committee would view it, like it wouldn't get accepted, right? It felt very similar to that.
And so that was the origins, but I wanna be very clear. I did not say, and I would never say consensus investing is a good idea. I'm just saying not being aware of consensus is a bad idea. And I think the underlying, the last thing I'll say on this, I think that my underlying belief is early markets are actually pretty darn efficient, a lot more efficient than people realize.
And so if you're alone in your view, you may just be missing something.
Erik Torenberg 00:02:20
Leo, we're stoked to have you join us as a friend, fellow venture nerd. What was your reaction?
Leo Polovets 00:02:28
Yeah, I mean, I actually agree with a lot of what Martin just said, which is, eventually you have to get to consensus, whether it's when you're investing or later because otherwise, you know, if you're dependent on capital markets, it's very hard to keep the company alive if nobody wants to fund it. I would say, like for me, and maybe we invest like a tick earlier, more like towards pre-seed and seed, but for me a lot of my best investments have been more on the non-consensus side.
Not in terms of, I had some crazy good insight, and nobody else had it, and like I'm just brilliant, but more like, you know, these companies often struggled in the early days because before there's proof points, it's not obvious that it'll be a good idea. And then once they get good, the valuation skyrocket so fast that like, you could still get good multiples, but they're just much lower than at early stages.
Erik Torenberg 00:03:17
Yeah. And then there's sort of broader commentary on looking at a list of big winners over the last, you know, 15, 20 years and saying, “Hey, what was consensus, you know, which were consensus, which were non-consensus. And then Martin, what were your reactions to that sort of broader commentary?
Martin Casado 00:03:36
Well, listen, I mean, again, it wasn't meant to be a technical tweet where like the wording was exact. On the face of it, it’s almost an ill-defined statement because we don't know what consensus means, right? And so then everybody picks apart the consensus. But here's my reaction to the list of like the Airbnbs and this and that, which is, I think we need to be very careful not to conflate a company having a hard round with market consensus, right? Like if you look at the list Keith Rabois put out, which is great, and I love Keith, I mean these are like MIT founders, known spaces. Like I'll bet if you took like the median value of their raises over the lifecycle of the company, I'll bet they're way above market.
Many of the companies were YC companies, and so I just think it's so easy to come up with these anecdotal, “Oh, this one company had a tough raise,” when that's definitely not within the spirit of what I was trying to say, which is, markets are actually quite efficient. If the market is efficient, and it's a good company, the price is going to be high.
And if you don't recognize that, then you're probably beating yourself as opposed to the market, right? And so it really comes down to, you shouldn't be looking for good deals with respect to other investors. You should be looking for good companies, and price shouldn't sway you from that. I mean, that's really at the heart of this, and so I just don't think that list, unless you actually run the numbers, which we haven't done, I just don't think it demonstrates that the idea that consensus is important is wrong at all.
Erik Torenberg 00:05:22
There are a few quibbles I had with some of the names on that list, like some people put Anduril, and it certainly was a controversial investment. But you know, Palmer Luckey, you know…
Martin Casado 00:05:32
I mean, second time founder, billion-dollar exit, Trey, who's phenomenal.
You know, Elon, this is in the shadow of Elon, who shows that you can already create these defense tech companies. I mean, if that's our definition of “non-consensus,” it just shows how insular we are as a community. I mean, it's almost an indictment of us that we even make this list.
Erik Torenberg 00:05:52
And wasn't the seed round at like a hundred or something? Like there was a very expensive early round.
Martin Casado 00:05:56
Every round was super expensive.
Leo Polovets 00:05:59
I'm not sure an ex-unicorn founder would ever be non-consensus really.
Erik Torenberg 00:06:04
Yeah. It is interesting because there's also sort of, you know, there are rounds that are maybe, you know, non-consensus at 10 million or something, but then become, you know, super hot rounds at 50 or 100 and then become $10 billion companies or $100 billion companies.
And even if you invested at that, you know, consensus round you 10x’d or 100x’d. And so it's sort in the face of, “Hey, if it's a hot deal, that must mean it’s not good.” Peter Thiel once had a line, which was like, “The faster and higher the up round, the more you should invest because it's like, you know, working.”
00:06:38 - Measuring market efficiency
Martin Casado 00:06:38
Yeah. I would love to do a correlation analysis. Actually, Leo and I had, I thought, a very interesting discussion on trying to figure out how you'd actually measure this, how you'd actually throw some data at it. We actually have an analyst working on it now, like the data isn't ready yet.
So I have a new one actually. I want to test this with you, Leo, on a good thing to test. So I’ll bet the best prediction, the best correlate of a high up round outside of the business is the fact that the previous round was hot.
Leo Polovets 00:07:08
I think that's probably true.
Martin Casado 00:07:11
And if that's the case, it would suggest that the market is actually pretty efficient because it's almost inductive that like the previous round knew that the next round was going to be hot.
Leo Polovets 00:07:21
So, I do agree with that. I think the question for me is like, “Where is there more opportunity,” right?
Because if there is, you know, like the 5 hot companies keep having great rounds, and then there's like 10,000 not hot companies, but a 100 of them will become hot over time, even though the odds of becoming hot are low, most of the hot companies end up coming from the not hot batch, right?
Martin Casado 00:07:45
Right, so the question comes down to is it easier to spot the company nobody sees or get into the company that's obviously good? And maybe even further than that, which is, to what extent do even high price rounds under-price hot deals? Because if I'm right, if the view is correct, that hot deals are hot because they're good companies, not, like actually the market is very efficient, and that drives the most of returns, then I think the next obvious question is, well, if that's the case, then the market isn't that efficient because it's underpriced the company, right? If the majority of returns are in high-priced rounds, and the market has underpriced it, but I think risk-adjusted, that's not necessarily true.
Which is, it could be still priced right because there's still chances it goes to zero. So I guess my sense is until we run the numbers, we're not going to quite know the answer. But I think a lot of these theories prove out pretty anecdotally, and I think maybe that's the problem is there's kind of an anecdote for every theory.
Leo Polovets 00:09:08
Yeah. I think the basket analysis is probably the most interesting one, right? Of like, not how did this one company do, but how did this, you know, portfolio of companies that raised like really quick follow-on or had like 10 term sheets at this series A, like how did those end up doing over time?
Martin Casado 00:09:23
There are even cases in my portfolio where a super hot company, from an investor standpoint, so, many term sheets, the business didn't work out, you know, at the level that you would kind of expect, but the outcome was still really good. And so in some level, even independent of the productive asset, like human opinion about it matters.
So there's almost two ways you can slice this conversation. One of them is like, the asset is what's productive and produces the value, right, and the market will determine if that's valuable or not, right? So that's kind of this productive asset view, and that's kind of the one that I hold, which I think that actually investors are very smart. I think that they know which companies are good, and then they pay for those. That's kind of my view. But that's the productive asset view. But there's another view which is, independent of whether the company is good or not, there are things that people think are good, and so you're almost like playing to like the human perception of the company, independent of the underlying business.
And I would say, again, anecdotally, until we run the numbers we won’t know, that also seems to be a bit true.
Leo Polovets 00:10:36
So I've been in Venture for like 12, 13 years now. I've definitely seen this in sectors where like sectors have fallen in and out of favor, right? We had like e-commerce was hot and then it was dead, and then like Dollar Shave Club got acquired, and it was hot again.
And it's like e-commerce didn't, I think the fundamentals didn't change that much year to year, but like the valuations and the like appetite for investing and maybe starting companies changed a lot year to year. And so that to me is sort of an indicator, like it's not just the fundamentals. There are all these other like forces as you mentioned.
00:11:04 - How consensus affects founders
Erik Torenberg 00:11:04
One other part to your tweet, Martin, that I think was underappreciated was sort of the risk to founders of being seen as non-consensus. Because founders need to raise money, and they need to raise follow-on funding, you know, within 18 to 24 months, you know, sometimes even sooner.
And so if everyone is passing on you or people are bragging about how, you know, other investors don't want to do your deal, that’s not going to be super helpful to you in your next round.
Martin Casado 00:11:36
I actually think the most interesting aspect of the tweet was like the sociological study that followed of like how different people interpreted it, right?
Like the tweet itself was pretty banal, right? It's just kind of a non-statement. It's almost tautological. But like different constituencies viewed it very differently. So like, I would say relatively inexperienced investors kind of used it as an opportunity to be like, “Oh, Andreessen Horowitz consensus invests,” which anybody that knows anything about our investments knows that’s just totally not true. I mean like even my own portfolio, many of the top deals I've done, nobody else was in the deal, you know, etc, right? So like this is a statement about consensus investing. So that was one cohort. There was another cohort like Leo and Keith who have a lot of data and they have a lot of really interesting good things to say, and that ended up being great discussion.
I think there's still a lot more to do there, but most of the founders, and I got a bazillion DMs who’re like, “You're totally right.” So the founders clearly feel this tension that it’s dangerous to be non-consensus because, you know, they have to cater to VCs, and they know it, and they see the pattern match responses.
They deal with this all of the time. And so from a founder perspective, it's like, you almost have to be non-consensus to have alpha in the actual product market, but you have to look consensus when you’re raising. And I think that's actually probably right.
Leo Polovets 00:13:08
I think this is probably one area where I differ a bit. I think there's benefits to being non-consensus because from the company side, I think when the money is hard to raise, you tend to be more frugal with it. And then also, if the next round is less certain, like I think there's less of a sort of like, “it could crumble at any moment” aspect, right?
Because I think when it is hot, and you're raising subsequent rounds very quickly, unless the assumption of things will go perfectly, if anything slows down, it's like, “Hey, now everything, like now you can't raise any more capital all of a sudden.” And I think if you're in the mentality of, you know, growing quickly and spending, I think that's pretty hard.
On the flip side, if you're non-consensus, it tends to be like, you tend to be more cash efficient, tend to be more frugal out of necessity. I think the other side is, it depends on the form of like consensus, but sometimes there's also like much softer diligence. Like I think the worst form of consensus I've seen is like, “Oh, Sequoia, Andreessen did this round. Like, let me just do 2x markup in two weeks because I want to be in the same company.” And then like, there's no diligence there, right? It's just like, oh, this is hot, let me do it. But I think then like maybe you're overlooking like, is it actually a good business? Like, you know, Sequoia and Andreessen and we all make good investments and bad investments and, you know, so it's like maybe this is one of the bad ones and you're just marking it up because you want to be in the hot deal and like that ends up not being good for anyone.
Martin Casado 00:14:25
I think this is a tremendously important and good point. I tend to believe now that most companies fail from indigestion, not starvation, which is they just raise too much money too easily. They don't listen to the actual market, which is, you know, the customer base. And as a result they just have a bunch of bad practices and end up running out of money.
And I think that there's a lot to that. I actually think in 2021, if you look at, if you just did a study of that cohort, the companies that had these billion dollar Bs. If you remember that time, it was totally crazy. I'll bet that's probably one of the biggest wipeouts of capital.
So I definitely think like consensus investing is definitely very dangerous, and only leaning into this for a founder is definitely dangerous. But I also think the flip side is true, which is if you're totally blinkered to it, I think your life is pretty tough.
00:15:22 - Is venture getting smarter?
Erik Torenberg 00:15:22
And so there's a broad question as to like, you know, of the companies that do win, you know, how many of them are sort of competitive rounds versus not competitive rounds.
And sort of what is the duration between them being non-competitive rounds and then becoming not non-competitive, and what percentage are really able to sort of… And one question I have is like, “Is the market getting more efficient over time?” A lot more investors, you know, we should be getting smarter as an asset class on how to evaluate these companies, a lot more capital.
Are we getting better? And if so, what does that mean?
Martin Casado 00:16:02
Well, I'd love to hear Leo's view on this.
Leo Polovets 00:16:05
It’s something I've been thinking about for a while. My take would be that for non-consensus companies, it's getting more efficient because the more investors there are, the more likely you are to find at least one or two that like what you're doing.
I think for the consensus companies, it’s starting to get more inefficient, right? Which is like when you have 10 term sheets, you get, you know, 5x the market, like what maybe the fair value should be. And then like, it's great for the founder and maybe again, it’s a little bit more of a house of cards if things go south at all.
But it's also like, it's not necessarily great for investors, right? Because you might have to pay 2, 3, 4x over like the actual intrinsic value of a company, or like the likely future value of a company in order to get in.
Martin Casado 00:16:48
But that would be actually efficient, right?
The price is actually approaching the return profile. From a market standpoint, that'd be efficient. I mean, it sucks from an investor standpoint because prices go up.
Leo Polovets 00:16:59
Yeah, that's what I'm saying, right? Like for founders it's getting hyper-efficient. Or maybe there’s such an imbalance for like, the really hot companies that, you know, maybe your price gets bid up way past where it should be.
And similarly, for non-consensus companies, it's the opposite, right? Where like, there's not enough investors, so your price is lower than it should be, perhaps, right? But I think for me, like those two are kind of opposite ends of the spectrum.
Martin Casado 00:17:25
I totally agree. This is a great question. So I think we can all acknowledge that there's a failure mode where the consensus gets bubbly. And then companies raise too much capital and then there's a bunch of wipeouts, right? So that has always happened. That will always happen. So that’s just part of the market.
We, I think, can also all agree that there's parts of the market where there's probably unnecessary pessimism. So for example, right now during this AI craze, like, you know in my area of infra, a lot of the traditional companies that, you know, two years ago would be great, like can't even raise right now, just because they're not in the sweet spot.
And so I think that will always be an aspect of the market too. But in general, for the mean investment, I do feel like the market over time has gotten a lot more efficient. Meaning, you know, we can deploy more dollars with more regularity, and the price is converging on what will ultimately be a fair price.
You know, this is acknowledging both of these failure modes on either side of this.
I mean, we're seeing one right now. I mean, this is the reality, there's AI companies that clearly are raising, you know, speculative money where nobody even really understands the business model, and there's great companies that can’t get investments. So we're seeing this right now, but I will still say the reality is, OpenAI has grown tremendously, and Anthropic has grown tremendously, and Cursor has grown tremendously. And so like there is some underlying market signals to fuel the chaos.
Leo Polovets 00:19:14
Yeah, I think part of it's like if you ever look at vintage year data for venture funds, it's probably a good way to see how consensus and non-consensus do over time. Because when you look at like the dot-com bubble years, like, I think the median fund was terrible, and I think it's just like, “Hey, everyone overpaid, and then the companies weren't worth that.”
And so like, you know, even though everything was hot, like it didn't do well, and then a lot of the funds didn't do well. If you look at the Airbnb, you know, Uber like 2010-ish era, it's kind of the opposite area where like, I think the top quartile funds like crush it, and it's because the market was pessimistic.
And so if you were willing to invest and like you had a different opinion, you did really well. And now it's probably kind of somewhere in the middle.
Martin Casado 00:19:54
Maybe I'll just go through like kind of my own startup just as kind of a single anecdote to frame the conversation a little bit.
Right, so, you know, I did my PhD at Stanford. I was a classic, you know, take the research, do a startup. You know, we had so many term sheets before we had any idea of what we were doing. And it was like the hottest thing ever, and it was great. And so we did a seed fundraise, actually Andy Rachleff, you know, Benchmark, Andy Rachleff joined my board and, you know, we raised at the time, which would've been a super, super high price kind of seed round, which is 10 million post. This is in 2007.
Then the market tanked in 2008, and we still didn't know what we were doing. And you know, it was just a bunch of researchers. And so we couldn't raise any money at all. I mean, Sequoia very famously, you know, gave us a black eye and, you know, we couldn't raise. And then, you know, as we started to come out of the recession, Andreessen Horowitz, NEA, Lightspeed, a few got very interested. And then we had a pretty hot round again. With Andreessen, it was actually over the market price, even though the business wasn't quite working, but it was starting to see signs of life. Then we had an incredibly hot round because the business started working, and then when we actually sold the company, I mean it, you know, it returned a fund, you know, it was one of the highest acquisitions on multiples of revenue at the time in enterprise software.
And so you kind of ask the question, “Was the initial flurry of interest warranted or not” because it turns out like we were probably a month from going bankrupt and we actually didn't know what we were doing. And the company definitely wasn't working. And then actually what we had pitched at that time didn't make any sense.
We were going to change, you know, switch hardware, which didn't make any sense. And so there's one view that's like, the market was over-exuberant. You were lucky. There's another view that says actually the initial conditions were there to do it. I just feel like if you run the data, it just seems like the companies that have good outcomes did have sufficient interest along the way because there are enough signals to do it.
Leo Polovets 00:22:13
I think, at least on my side, for a lot of pre-seeds and seeds I’ve done… I went back, I think over my top like 10 investments, maybe 6 or 7 or 8, took a month to raise a seed round. And a lot of times, like a lot of passes, like they were all down to the wire, but then they ended up doing better over time.
I think that transition from like non-consensus to consensus ended up being really important because if you never transition, it's really hard. And if you're always, you know, if you're always consensus, that's great for you. But like, one thing I noticed that was interesting is a lot of the companies that struggled, obviously some of them just go to zero, right?
Because they struggled because the business isn't that great and people recognize it. But the ones that did well, a lot of times the gap between like the C and A or the A and B was literally like 20x or 50x, right? And so I think part of it's like, as an investor you can still get good returns at like the series A or B in those companies, but it's just, I think it's so different to invest at the seed where there's like a 1000x versus like the A at a billion, where now maybe there's still like a 10x or 20x, which is very different.
00:23:19 - Defining “consensus” at seed stage
Martin Casado 00:23:19
So I've got a question for you, Leo, because I think that you play a bit of a different game than we do, which is, so if you have a seed, which is, let's call it non-consensus, and again, we're using this very vague definition of a consensus, but like, you know, they're having a tough time raising, you’re the only person putting money in.
Do you have a theory on how it will be consensus, or is your belief that the underlying productive asset is going to do very well, and that by definition is consensus? Do you see the question? So the question is, is it, this is just true belief in the underlying business? I mean the ultimate sign of success is just the business is really working.
So are you like “For the next raise, the business will definitely be working?” Or do you have some other theory on what will attract the investors?
Leo Polovets 00:24:07
I would say it's often the latter. I would say, and especially true these days because I'm investing more in deep tech companies, and so at seed it's very rare to see like, “Oh, there's an asset that's gonna be working here at the series A because usually the asset is still going be like being developed at the series A or maybe series B.” I think what I'm looking for is like, there's maybe not enough here for somebody to write a $5 or $10 or $20 million check, but the company has milestones that I think if they hit them, then it would become, you know, sort of consensus enough to merit a check of that size. And then I'm basically trying to evaluate like, okay, the company has these milestones. Do I think they could hit them or not? And also if they hit them, are they compelling enough? But I think that's sort of the big, you know, investment wager.
Martin Casado 00:24:49
Yeah. So in this case you do think about like what the follow-on thing is going to want to see. You have reached a conclusion for the current round that is non-consensus.
Leo Polovets 00:25:03
And I would say like the consensus piece is part of it in that I definitely meet companies where they're like, “We're raising 3 right now, it'll help us do these milestones, and then we think we can raise 10.” And then there's other ones where, you know, it's like, “We're raising three now. We're gonna hit these milestones, and then we wanna raise like a $50-100 million series A.” And that's actually a much harder bet, right?
Because you have to assume they're going to be consensus by the time they raise the next round, and it's going to be like a top 5% series A, and that's a hard bet to take. For the companies where the capital needs are more modest, or they have like a more traunched roadmap planned, I think it's a little bit easier to, you know, to predict like, “Hey, would these milestones be enough to raise 10?”
Like a lot of times, I don't know if it'll be enough to raise 100, like probably not, but 10 feels like pretty feasible if you do the things you think you're gonna do with this 3.
Martin Casado 00:25:49
Has your kind of view on this shifted in the last… like do you find this AI wave to be different than previous waves or fairly similar?
Leo Polovets 00:26:00
I'm probably a bad person to ask actually. I haven't invested much in AI because of the deep tech angle, so I see maybe like 10, 15% of my companies are pure AI. Others obviously use it in some way, but that's not the product.
Martin Casado 00:26:13
Well how about deep tech then? Because I think that's also, you know, like pretty different than what we were all investing in 5 years ago.
Leo Polovets 00:26:19
So maybe on the AI side, and I’ll touch deep tech next. I think AI is interesting to me because on the one hand, I've never seen faster growth, right? Like people talked about the like triple, triple, double, double, double thing for a while of getting from a million ARR to 100 in 5 years. And that seems so antiquated now, right?
Like most companies are doing it in like one or two years.
I think on the flip side, the endurance or like how long those companies endure, last, and grow feels like much more of a question mark. Because in the triple, triple, double, double, double ARR, like if you hit 100 million ARR, and there was no one close to you, you probably just keep growing.
And now it feels like you could hit 100 and then you drop to 50 because someone else came out with a better product. Yeah. So, you know, I think there's like, the growth is amazing, and then the moats are weaker. And so I think there's a counterbalance there, and I'm not sure how I'd evaluate it because I haven't invested that much of that stuff.
Martin Casado 00:27:08
I agree. Yeah.
Leo Polovets 00:27:09
On the deep tech side, I definitely see areas with a lot of hype from time to time. For example, we invested in defense a lot, 3, 4 years ago. And then we kept looking, we basically paused for a year and a half or 2 because after the Ukraine and Israel thing, you know, prices just went up like 2, 3, 4 times, but the company fundamentals didn't change. And then it started being an opportunity cost of like, should I invest in this defense company at 40 when there's this really great, you know, energy company at 15? And so I think defense was kind of like that. I think bio has had a lot of ups and downs.
I think in robotics, like humanoids are probably one of the most hyped areas where the valuations just get crazy before there's any, any revenue. I feel like I kinda lost the thread in the original question, but…
Martin Casado 00:27:55
No, this is great. I mean, I was honestly just wondering how you thought about this current wave, and you did a great survey of the set of the waves and I actually agree.
Leo Polovets 00:28:02
I would say like for the consensus areas, like humanoids, like we end up not explicitly, but implicitly avoiding them because once you have a few companies that have raised like hundreds of millions, whether they end up being great outcomes or not, I think it's pretty hard for someone to start something new with like, you know, near zero resources and team.
Martin Casado 00:28:20
Yeah. You know, it's interesting when you do the humanoid, so I think there's all sorts of types of investing, and they're all pretty valid. One type of investing is humanoids are clearly interesting. Big companies are clearly interested in it, so why don't you back a bunch of good teams? And worst case they get acquired, and I think that's totally legitimate, but that's not how I think at all. Like for me, like the company has to make sense as a standalone business at scale. So things like humanoids are tough for that just because the unit economics right now are just so unknown. Like competing with a human body is a very, very hard thing to do.
And then of course you can be like, “Okay, well, you know, we'll put it where human beings can't go, like a car factory.” But then all of a sudden now, you know, you're building a manufacturing company, you know, so you verticalize heavily, and the company has to look at kind of whatever sector that the robot is going into, and it's more constrained and I don't understand the competitive set and, you know, yada yada yada.
So I just feel like, from my standpoint, the idea that this is very buzzy and hot, you know, in the industry for big companies, and it may have an M&A, I don't know how to invest that way. I just don't know how to handicap that. And so the way that I tend to view these things, I mean like for AI, for better or for worse, like you have great unit economics.
I mean, you know, everybody knows kind of like, and we always talk about the OpenAIs and the Anthropics, but take like ElevenLabs, for example, or Midjourney. I mean these are just famously model companies where the unit economics are great. They're grown, you know, very quickly. And so I understand that.
But you know, I think there's kind of been this weird, and this happens a lot where people take the example of these model companies, and they apply it to totally different spaces where you don't have the proof points, you don't have the economic case, and they kind of apply it. And that's one thing I don't know how to do.
So certainly I don't believe, you know, we should all just follow the common consensus around areas to invest in. But I do think that like, there's going to be a pool of capital and it's going to want companies to look a certain way. And if you don't consider that when you're investing, I think life will be a lot more difficult.
Leo Polovets 00:30:34
Yeah, I agree. Sort of an aside here on the humanoid stuff. I think what I've seen over the last, like 10, 15 years is if the market is big enough, it really distorts like VC investing because, you know, it used to be that. You look at a market, you're like, “Oh, it's a 2 billion a year market. If there's a 1% chance they could capture it, they'll be worth this much. So let me justify a seed price.” If the market's like $5 trillion of human labor or something, like any price makes sense, right? But then, but then I think that really just starts with like “How much value is there in this?”
Martin Casado 00:31:05
The most boneheaded partner meetings were like, “Well, yes, it is cold fusion, but this is the largest market ever. So on the off chance it works…” I'm like, this isn't engineering, man. This is like laws of physics. I'm not sure that like, you know, a good, you know, software founder is going to bend the laws of physics. But yeah I think I totally agree. I also feel like, I don't want to harp on this too much, but like unit economics is so important.
I mean, like, what is the story for autonomous vehicles, right? The story for autonomous vehicles is that even after the industry has put $100 billion in it, 100 billion, the unit economics are still, you know, let's call it on par with Uber. Let's just call it that, right? And so does that make sense for venture investment?
It’s really hard to build a standalone business with those types of economics. I mean, Google can do it, sure. And Tesla can do it, sure. But can startup X do it? No. And so you're either playing for, this is a great company that got acquired, which a lot of that happened, and people made good money.
But that’s, again, that’s not saying that, you know, the startup… Or you're building picks and shovels. Like Applied Intuition, where you're like, you're building software for this market. But I do think that a lot of investment dollars do follow these spaces where there really is no thesis on the ultimate unit economics.
And I think you're exactly right. I just think that there's this kind of Market TAM sloppiness that says, “Well, if the market's infinite, then the expected payout is high.”
Leo Polovets 00:32:41
Also infinite.
Martin Casado 00:32:44
That’s right, also infinite, exactly right.
00:32:46 - Bigger outcomes
Erik Torenberg 00:32:46
When I look at my portfolio, I see both… Some of the winners, Pave and Scale, were non-consensus, non-competitive, you know, unproven, but very talented founders. And then on the more consensus, competitive Jack Altman and Qasar were…
Martin Casado 00:33:04
Wait, how is Scale non-consensus?
Erik Torenberg 00:33:07
At seed. You know, Alexandr Wang was 18.
Martin Casado 00:33:13
It's a total known space. He's phenomenal.
The A was done by Volpi, who's amazing. I mean, I just feel like this is a very narrow definition of “non-consensus.”
Erik Torenberg 00:33:25
Sure. For nearly most of the rounds, it was competitive. So yeah, I can agree.
Martin Casado 00:33:33
I mean Dan Levine was… I mean, come on, these are like the best investors in the world.
Erik Torenberg 00:33:35
I brought the example to say that Qasar’s round was almost an order of magnitude more expensive. And I think what people have been late to really internalize and what a16z was super early to internalize, was just the outcomes are an order of magnitude bigger, maybe two orders of magnitude bigger.
And so you can get seed-like returns at, you know, order of magnitude or even two orders of magnitude more expensive. I mean, remember, YouTube, Instagram were considered, you know, very expensive acquisitions at, you know, just a few billion dollars. And you know, in a few years, we're gonna have more trillion dollar companies. And so, once we truly internalize sort of the outcome expansion on the order of magnitude, I think it makes sense to Leo's earlier point, that then it would beg the argument of like, okay, but can you have, you know, 1000x like returns at not just what we used to consider seed-like pricing, but maybe at series A or maybe even series B?
Martin Casado 00:34:29
Well this is a very interesting question because you actually do run into fund mechanics as an actual, you know, price modulator in this discussion. So you're exactly right. So again, I'll go back to my company. So my company was acquired for $1.2 billion. We had, let's call it, you know, less than 10 million ARR, right? So does that make any sense? No. And then a lot of people are like, this is totally crazy. This makes no sense. Except for when I left, you know, three and a half years later, like the run rate was, you know, $600 million within VMware, who acquired the company.
And then right now it's, you know, let's call it 2 billion, right? It was actually, at one point in time, it was, I think it was 40% of the growth of VMware, like the business unit that I ran, that was part of the acquisition. So clearly it made sense to VMware. So as a result, you should say all the check sizes should be high for the winners because the outcome was so good.
And this actually returned a lot of money to a lot of investors. The problem with that is I just think that that would mean fund sizes would be too large and you'd have to unlock different pools of capital, which by the way did start to happen during the SoftBank and the Tiger and the Coatue era.
So you could argue that all of their theses were correct, right? Like SoftBank was actually right and Tiger was right, and it was actually a macro issue that caused all the pullback, and that's gonna come back again. I mean, that's, I think, a very legitimate thesis. But I really feel the reason that prices don't continue to go up is more just access to LP capital.
So, Leo, let me just try to, to make this a bit more concrete, which is, I think what Erik said is correct, which is the outcomes are so big, it suggests the prices are too low that we actually pay. You know, the fact that we get the returns we do suggest the prices are too low.
So the question is, why are the prices too low? And I think the answer is like, we just don't have the dollars to place all of those bets. And a number of people have actually questioned exactly this. Very famously SoftBank questioned this. Tiger questioned this, and so they raised these, you know, Insight questioned this, they raise these huge funds, and they deploy a lot of capital.
And those experiments had very mixed success, but it's not obvious to me that the reason they had mixed success is because the prices were too high. I mean, there's a lot of reasons why those could not have worked, including kind of macro cycles. And also the fact that none of them were Silicon Valley insiders, none of them were, you know, traditional early stage investors, etc.
So there's a very reasonable question, which is, you know, maybe someone should just go run the Tiger strategy again, but as a Silicon Valley insider.
Erik Torenberg 00:37:10
There's the failure cases just to some degree. You know, I mean, Thrive raised bigger funds, Founders Fund raised bigger funds. We raise bigger funds. You know, the winners have also, you know, multi-stage, have raised bigger funds.
Martin Casado 00:37:24
It just could be that this is just the market being efficient. Like actually the reason that more money is going into this and the funds are getting larger is because the opportunity set is larger and this is just a market working its way out.
But Leo, you're very quiet. This is actually a pretty controversial statement, so I wanna make sure that like…
Leo Polovets 00:37:41
I'm not sure what you mean by “We should be paying more.” Do you mean that, like, you think the current prices are still like well below where they should be and I guess if so, like…
Martin Casado 00:37:56
I'm riffing off of Erik’s statement, which I thought was right.
Which is venture capital has been a top returning asset class. And you can look at individual investments. If you just take the top 10 percentile of funds, you know, they return so much money. So there is an argument that even with these high prices, they're still underpriced
Erik Torenberg 00:38:18
And to put it differently, Leo, it's like, a seed fund may say, “Oh, I'm not gonna invest in something at 50 post or 100 posts because I don't think there's a 1000x, you know, potential. I don't think Anthropic is gonna be a $100 billion company, or you know, OpenAI is gonna be a $100 billion company” or whatever it is.
But it turns out it is.
Leo Polovets 00:38:31
I think it would be contrarian now to say OpenAI is going to be a $100 billion company.
Erik Torenberg 00:38:41
Right, exactly. I mean a few years ago. And so, it doesn't seem like we've sort of truly internalized that this is the norm, that there's gonna continuously be a $100 billion outcomes.
Martin Casado 00:38:59
Or that the market just continues to grow and therefore it necessitates larger fund sizes. I mean, I would say that probably the venture market was what, a hundredth the size, 20 years ago.
Leo Polovets 00:39:10
Probably something like that. It's kind of wild to think about.
Erik Torenberg 00:39:13
Yeah. And we did think a few years ago that there'd be a great contraction in the asset class. That 2021 was a blip and that, you know, it would sort of right size back to where it used to be.
And it doesn't seem to be the case that it's going to 2010 levels. I’m not sure if you guys have the data on you, but when, when I talk to our team, when I talk to Thrive, it seems that people think, no, more capital is just gonna keep entering.
Leo Polovets 00:39:39
I think that's just because companies stay private longer too, right?
But I think the actual number of $100 billion plus companies in the last 20 years is pretty small. Like, I don't know the exact number, but I bet it's like 10 or 15 or maybe 20 or something. So it's like you're really betting you can get like the one every year or two that gets there.
Let's say you're doing a series A at like a billion post or something, right? And you want a 100x, even ignoring dilution.
Erik Torenberg 00:40:09
You'd have to bet that there's more of them, more of them are going to happen, and that there are also more ways of getting liquidity from them.
Martin Casado 00:40:19
But that also kind of suggests purely by the numbers that the most important thing is, if you can, the most important thing is being in one of those independent of price.
I mean, that's the high order bit.
Leo Polovets 00:40:34
I mean, I generally agree, right? Like if you're in like the best company of the year, I don't think ownership matters that much. I don't think like the price matters that much. If it's gonna be the best company, it's like 10 years forward.
I guess to your earlier point where, you know, if venture funds had more money, they like do higher valuations. I mean, it sounds like then you could do the higher valuation today too, though, right? You could just be like, “Hey, if we just want to get in this one, we'll pay twice the price and get half the ownership” or something. Right?
Erik Torenberg 00:41:02
But you also need a diversified portfolio. You need enough companies to get…
Martin Casado 00:41:05
You need the fund size to run that strategy. This is why I think a lot of this comes back to fund size. I mean, even in the Andreessen portfolio, I was just thinking off the top of my head, we have 3 companies that are—4 companies at the $100 billion mark, right?
I mean, there's Stripe, Databricks, Coinbase, OpenAI, and so they're not that rare.
Leo Polovets 00:41:26
You guys have awesome coverage. I guess the question is like, how many more could you name though from the last like, you know, 15 years? My guess is 10, 15, probably not like a hundred, right?
Martin Casado 00:41:36
Yeah. I mean 20 billion plus, there's a lot. And that used to be so rare. And enterprise software, it used to be an adage that nobody ever broke, you know, 20 billion or 10 billion, right. And Palo Alto networks was at 15, and we were like, “This is crazy.” Now there's so many of them that have, and so maybe with a hundred billion, you're right.
But in the world that I live in, the amount of like decacorns is probably an order of magnitude more than what it was 10 years ago. And on the face of it, that would argue for an order of magnitude higher fund size, if you wanna play the strategy of being in the winner. I mean, there's clearly multiple strategies. For me, the key question, I don't know the answer, I want to run the numbers, is if you take a dollar of earnings, like a dollar of earnings for a venture capitalist, did that come from a company that raised at high prices or not?
I would guess the answer is “yes,” just because the winners are so outsized.
Leo Polovets 00:42:36
I mean, I'll say there's like multiple ways to play it, right? Which is if the outcomes are 10x bigger, you can have a 10x bigger fund and basically run the same playbook, keep the same ownership, and like a big outcome still returns the same amount of the fund.
You can also do more investments at, you know, a fraction of the ownership, and then, each investment maybe moves the needle less, but you have a higher chance of hitting like, you know, the Stripe of the year, the Uber of the year. So I think there's definitely different models that could work here.
Martin Casado 00:43:04
Yeah, that's a good point.
00:43:06 - How to frame consensus
Erik Torenberg 00:43:06
I want to make a few related points here. So one is, I remember someone quote tweeted Martin’s tweet and said, “This is a sign that the asset class is dead” or something. The idea of a more efficient market.
And I think what that really means is more that that individual's firm is… If an individual firm can't compete and win deals in an efficient market, they're going to lose. And so, my second point, which is I think a lot of venture capitalist identity is tied in being non-consensus in being able to see things that others can't see because it's hard to win against all these other much bigger, much more well-funded players. And for that reason, I less want to use the term “consensus,” “non-consensus” because it's so core to people's identity, and more want to use the term like either it's a hot round or it's not a hot round, you know, it was a competitive round or not competitive.
And I think another way of framing that, it's not perfect, is, is the company working or is the company not working at the point of investment? And let me add some nuance to it, which is, if something is working, then it's like, you know, what is the price, and what is sort of the, you know, potential return multiple and how does that work with your threshold, etc.
There's some things that are competitive and not working, but have an incredible founder, whatever. It's early enough that people believe the vision and so you're still paying that price based on what you think. And then there's lots of things that are not working, or not obviously working. And we’ve chosen to do less, I believe, consumer things that are pre traction. So it's basically, it's like, do you want to invest in things that have traction or no traction. And there's failure modes with both. And not every hype thing, not every competitive thing has, has traction of course, but it's just another way of framing this. I'm curious, feel free to quibble with my framing.
Leo Polovets 00:45:03
I think I saw the same quote tweet. I'm probably somewhere in between. Like I don't think venture is dead. I think it gets a lot more fun if it's purely consensus. The reason is, I think in a purely consensus world, like it all just comes down to the cost of capital.
And so if my LPs want 5x and yours want 2x, you could pay two and a half times higher prices, and the company's not better. It’s just like, “Oh, like your cost of capital is lower, so you're going to win all the time.” But also it’s like we all see the same value. Everyone sees the same value of just like, who wants the smallest return that can still stay in business.
And that just feels less exciting to me.
Erik Torenberg 00:45:42
That’s exactly right.
Martin Casado 00:45:45
I'll get a little bit philosophical on this, but like, the thing that's always bugged me about PE investing and public market investing is it just doesn't care about productivity really.
I mean, it does to some degree, but I just like, you know, if you're in a large public company, like I was, you realize that the public markets really care about predictability over innovation. And so innovation has stifled so much, and in fact it kind of causes large companies to protect themselves through kind of incumbency and monopolistic practices and everything else, just because they're not allowed to be aggressive on growth.
So I feel like it's almost this negative force on progress and innovation, and I don't want to be too dramatic about it, but I just feel like… I'll bet if you draw a dollar at random, that gets invested, you know, 90 cents of that dollar goes into like keeping incumbents alive and/or, you know, predictability, and not to growth.
And I'm a huge believer in creative destruction, man. I'm like, “Fuck, man, get him out of the way. Let's invest in growth.” And so I love the idea of venture as an asset class getting more efficient. And I love the idea of more money going into it. Because the entire thesis is growth. You never invest on, I don't, I mean, I'm sure you don't Leo, I never invest on downside loss. I don't care. You only invest on upside. And so to me, more dollars going into ventures is only a positive for humanity. And again, I don't mean to sound too grandiose, but I do feel it's just a net positive.
Leo Polovets 00:47:30
Well, so maybe on that front, like, I think that's a really interesting perspective. More from a company perspective than investor perspective, I feel like a lot of the most disruptive products were maybe non-consensus at the time.
Martin Casado 00:47:42
Totally.
Leo Polovets 00:47:43
Where you start with, you know, like no buttons on the iPhone or you got like Uber, instead of taxi, it's a stranger driving. And those are the ones where I think if you were like, “I'm gonna build a taxi company, but it's like 20% more efficient,” like probably could be a big business, but not quite the same level of disruption and growth as like, you know, you take a big bet and very high chance you're wrong, but if you're right, like you're gonna be, you know, in a really good position.
Martin Casado 00:48:07
Yeah. And this is so critical. I'm glad you brought it out. I really believe the best companies themselves are non-consensus to customers. I just think that the investing market is different than that. Like they kind of understand that. And therefore, a comment on investors being consensus is very different than a product being consensus. Does that make sense? Like. Investor sentiment, I think, is actually much smarter than people think. Like the adage is, VCs are dumb. Like they just, you know, chase trends, and all of that is true.
But the reality is, as a group, we have identified a cohort of companies that are quite disruptive and invested in them and priced them. And the companies themselves tend to be actually quite non-consensus to the actual consumer or to the market.
00:49:02 - How to think about incentive alignment
Erik Torenberg 00:49:02
I do wanna build, Martin, on your point because I think it's so interesting just to comment on how not everyone's incentives are totally aligned here. Especially between sort of the, what's good for the individual and what's good for the ecosystem. And so in the sense that, yeah, you know, if you're an individual VC, you don't want more capital. Or if you're a founder, you don't want more founders in your space. And so people were saying, competition is bad. You don't want competition. But competition is what fuels incredible product. It's like the Darwinian process. Like this is how, you know, we get, you know, bigger startup outcomes, a startup ecosystem, having more value, incredible products for customers and users.
Martin Casado 00:49:41
This is how we solve cancer, man. More money goes into VC, and we invest in companies. And then as opposed to investing in dying companies' ability to retain their place. 100%. Like all the finance needs to change.
Erik Torenberg 00:49:52
I think VCs are trying to straddle sort of, you know, LP incentives, founder incentives, their own incentives. And there is some overlap, and there's magic there, but it's also just worth acknowledging that not every individual person is aligned, and that's okay.
I also do still very much believe in the barbell, that there will be, you know, these big, massive funds that continue to win, and invest, and compound value. And also these smaller, focused, concentrated, experts, the boutiques who absolutely crush it. And we all work together.
Martin Casado 00:50:25
So Leo, we’re gonna run the numbers. I was trying to get it done by now, but there's a lot to do. The numbers are fuzzy. I just want to walk through what we’re going to be looking at, and then maybe we will schedule another podcast once the numbers are out to actually discuss it.
So one of the numbers we're going to look at is, if you cohort companies into winners and not winners, call it, looking to whether, on average, for that company, the rounds were priced above or below median for other companies at a similar stage, right? So this will say whether, you know, is it relatively high priced for winners or not?
And then the other one, which is even more difficult to determine is, given actual returns, are the bulk of the returns from companies that were, on average, high priced or not? And I think these two numbers will give us a sense to whether the market is actually pretty smart about the value and the price. You should not look for price arbitrage if you're looking for returns. Does that sound fair?
Leo Polovets 00:51:41
Yeah, I think that sounds fair. I definitely agree with the not looking for price arbitrage piece because I'll say, for me personally, my best investments have been ones on average that took a while to raise their seed round.
A lot of people didn't get or didn't like it. But on the flip side, some of the biggest misses are also the ones where it's like, “Oh, we liked everything except the price.” And like we thought it should be a 10, and some big fund gave them a term sheet at 20, and we passed, and then now it's a $10 billion company.
So maybe that was like, not a good pass.
Martin Casado 00:52:10
Yeah. You know, you know, Leo, honestly, as we go through this conversation, it does strike me that I think a lot of this is honestly, we have a bit different perspectives. Like I have to deploy a lot more money than you do, right? Like I’m a series A investor who needs to basically cap out 30 to 40 million in order to have a significant position.
And so I may have to be a bit more concerned about this than you do at the early stage. And I’m sure stage does color this conversation quite a bit. Because everything you're saying is totally sensible to me, so I don't think there's any disagreement.
Leo Polovets 00:52:42
It was definitely something I was thinking about, which is, I think if every check you write has to be at least 100 million, I think it's actually very hard to do non-consensus.
Martin Casado 00:52:50
Yeah. Right.
Leo Polovets 00:52:51
Because there's not a lot of companies that hit a stage where you'd invest 100 million, but it's still not clear if it's a good company or not.
And I think the earlier you go, if it's like $30 million checks, 10, 5, 1, I think you get more and more of a category where like you have the option and you could do the one assuming you have access to the consensus opportunities.
00:53:06 - Has multi-stage won seed investing?
Erik Torenberg 00:53:06
Leo, I'm curious what, and you know, you guys have absolutely crushed it at seed with, you know, Robinhood and Flexport, etc. But I'm curious what you think of Ramtin’s sort of thesis that multi-stage has won seed more or less in the last like 10 years. That when you look at a lot of the big winners, they were done from multi-stage firms, you know, at seed.
I'm, one if you agree with that sort of, you know, reading of history and then two, if you think that’s, well, definitionally you probably don't, think if it’s likely going forward.
Leo Polovets 00:53:37
Can I join you guys? I actually thought this was an interview. Sorry, what's the second part of the question?
Erik Torenberg 00:53:44
Did multi-stage win seed or more than seed firms win seed? Obviously there's, you know, First Round, Susa, you know, like lots of great seed firms. But when you look at the aggregate of winners. Do they have a multi-stage with seed, or not?
Ramtin’s argument is they had a multi-stage seed, and that's why he co-invests with multi-stage as his whole strategy. And then just, you know, the past isn’t, the future necessarily. What do we think about the future?
Leo Polovets 00:54:10
I haven't rigorously analyzed like the $10 billion, $50 billion outcomes. Over the course of Susa, I think we've invested in like 10 or 12 unicorns roughly, maybe like a third of those or quarter of those had a series A investor at seed. I’m not really counting, like sometimes it was like, “Oh, this series A investor did a 50k check in the YC round or something.” I mean like actually like took half the round or more.
So most of them still were seed only or were like, seed funds dominated the early round and then they went to multi-stage very quickly after that. But, so in my experience, I think there's a subset of seed where, I don’t know if I'd say multi-stage funds one, but they have like a very strong advantage.
Where if it is a founder that previously built a business that exited for 100 million, and they're like in the space that they know super well, that's gonna get done at like 40 instead of 20 or 80 instead of 20, post. And chances are it's gonna be a multi-stage and not like a boutique seed firm.
So I think for that segment, like multi-stage hasn't won, but I think it's probably like the predominant, like the majority of the time, they have a big leg up. I think for the other ones where it's less obvious, it tends to be much more seed fund dominated. Yeah.
Resources:
Find Leo on X: https://x.com/lpolovets
Find Martin on X: https://x.com/martin_casado
Stay Updated:
Find a16z on Twitter: https://x.com/a16z
Find a16z on LinkedIn: https://www.linkedin.com/company/a16z
Subscribe on your favorite podcast app: https://a16z.simplecast.com/
Follow our host: https://x.com/eriktorenberg
Please note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details, please see a16z.com/disclosures.
