FULL TRANSCRIPT
Slava (00:00)
In this episode of Smart Humans, we talk with John Cadeddu who's founding managing partner at Corner Ventures. Having 25 years of history investing in ventures with big wins in Bloom Energy, Grubhub, and others, we talk about how things are changing with AI and agents. He's gone from investing in B rounds and now also incubation, a exciting breakout with Aurite AI and how he sees the economy evolving and what he thinks about the big three: SpaceX, Anthropic,
and Open AI.
Slava (00:57)
Hello and welcome to the latest episode of Smart Humans. I'm excited for today's discussion. We're gonna be talking about all things venture and AI, very hot topics. We have with us John Cadeddu, who's managing, founding partner of Corner Ventures. John, welcome to the show.
John Cadeddu (01:16)
Great be here. Thanks for making time.
Slava (01:18)
Alright awesome. So you have a ton of experience, all kinds of background. but before we get to the present, we like to figure out how you got here. So take us back as far as much as you like, whether it's, you know, high school, college, work, ten years ago, five years ago, how is it that you end up in the this seat and you are as experienced as you are in venture, AI, et cetera?
John Cadeddu (01:42)
go there. So the seat I'm in right now is in Palo Alto, lovely Northern California. Without going through the whole childhood history thing, let's go to where it starts getting interesting. I made my first investment in venture February of 01, so hit the 25-year mark a couple of months ago, half a year ago or so. And at the time I was with the private equity fund called Duff Ackerman and Goodrich. We were allowed to make a few venture funds. The partnership allowed me to lead them.
that gave us some experience, or you could say scar tissue. enough that by 04 we raised a we spun out a venture fund called Delph Accra DAG, DHE Ventures. We took the first letters of the private equity fund. I co-founded it with my partner Tom Goodrich in 2004. DAG Ventures One was followed by four other funds that I raised. So we raised five venture funds, about a billion six over a period of about 10 years.
Had plenty of good times there, learned a lot, invested across a lot of categories. I am a generalist venture investor. stepped back from DAG Ventures for a few years, let's see, 2014 through 2017. I had a full plate of portfolio companies. Could not look an LP in the eyes and say, hey, I have the time to be a good fiduciary and invest your capital in new companies. Fortunately, the public markets cooperated. We had a lot of MA exits, a lot of
Even you in more wonderful IPO exits during those few years. when I
Slava (03:01)
What what years are this?
John Cadeddu (03:02)
2014 to 2017. And
Slava (03:05)
Okay.
John Cadeddu (03:06)
then in 2018, I started a new venture platform called Corner Ventures, where I am now, with my partner Marvin Tien. and we raised a couple of venture funds there, twenty twenty nineteen, twenty eighteen, twenty twenty one, and have been investing wonderfully through those funds.
Now, interestingly enough, around we all remember late 2022, GPT dropped, we all started paying attention to it. I started talking to my partner Marvin and the team, and I said, look, we need to start incubating some companies to really understand this AI thing pretty well. so we've been incubating companies, investing in companies, going a little bit earlier stage the last few years, and it's been doing it's been very, you know, we've been learning, becoming better investors, and the portfolio companies are doing wonderfully well. So it's
It's been about 25 years where, you know, when I jumped in, it was on the tail end of the dot com bust. And here we are, you know, asking ourselves which part of the, which ending are we in in the AI investment, you know, era? So it's it's it's a pretty interesting journey so far.
Slava (04:04)
We're gonna talk about the inning in just a second. So, you've been investing for such a long time, twenty five years. It's remarkable. most people don't even know venture has been around that long. So give me like three highlights. Give me three companies that maybe somebody's heard of or that have been some of your, you know, bigger results.
John Cadeddu (04:19)
Sure. god, those are those are the ones you remember. Well, you remember two sets of companies, right? You remember the the really lovely ones, and then you remember those like, yikes, what did I learn from that one? And those are the ones you wake up with in two in the morning going, Why did I do it? So, so my first when we raised Ag Ventures One in 04, the first investment I made in close was September 04 was a company the then called Ion America, ION America, renamed Bloom Energy a few years later. Bloom has been went public.
Gosh, almost d a decade ago. when we distributed shares about six or seven years ago, we told LPs, look, this is a special one. If you check the ticker, it's B E. It's been doing wonderfully well the last few years, obviously, helping power data centers and you know, riding the AI CapEx phenomenon. so that's been a wonderful one. some of the ones that folks use every day, Yelp. I had the pleasure of meeting Jeremy Stoppelman back gosh in 08. We invested.
He built an incredible company, good things have happened. we've had a bunch of other ones along the way. I I like to tell people I was early in spotting the great American bet called making it easier to get takeout food when we backed Grubhub in 2014, 2015. I I told my partners and I pitched it, I said, look, we we will never lose money underestimating the American desire to get their pizza more easily. So, you know, Grubhub, Begat, DoorDash, and you know, there there's a whole bunch of others that followed.
So there have been wonderful ones along the way.
Slava (05:43)
Amazing. And then you did skip forward very quickly to the first investments, but what was that catalyst from you were not a V C investor and it's not really a normal job that everybody talks about? And then all of a sudden something clicked in your mind that I'm gonna raise a V C fund. What was that kind of transition?
John Cadeddu (06:04)
That was interesting. So late 90s, I was in private equity. a lot of the activity in private equity at that time, 98, 99, 2000, even though one, was funding the broadband build-out. So you saw a lot of telecom, you know, traditional voice operators in telecom were suddenly saying, Hey, there's this thing called data. In the early 90s, they called it the information superhighway. We knew it as the internet, but you know, you know the internet ate everything.
Mark Andreessen likes to say, you he's right, software ate everything, is eating everything. Before that, broadband was eating everything. So as a private equity investor, we were backing these telecom build-outs, telecom operators, and we noticed that a lot of them were using new startup infrastructure. So, you know, I was the one in the partnership said, hey, look, we're investing in the carriers, the guys who are buying it. We should be looking at also the builders of that infrastructure. And that is qualitatively
Subtle, but it's a it's a materially different thing than looking at the users of the technology. But the team gave me the license to go look at that infrastructure side. We started making some investments infrastructure. There's a lot of upside if you get it right. And that kind of you know led me on, if you will, the under the dark side to look at, you know, builders of speeds and feeds versus the users of speeds and feeds. So that was that moment where I kind of said, this is this is really wonderful.
Slava (07:26)
Very cool. So on our show, we always like to talk about how people are investing their own money because the typical old school portfolio is 60% public equities, 40% bonds. That's kind of the traditional mix, 0% alternative investments. I'm gonna guess you're more than zero percent considering for the last 25 years you've been in the world of alternative investments with VC investments. What is your high level breakdown between three buckets?
Public equities, bonds, and anything alternative.
John Cadeddu (07:59)
Sure. So it I'm I'm unorthodox. my largest bucket is non equities, it's it's farmland. So sure, sure.
Slava (08:09)
So let's not get into the details yet. Let's just start with those three buckets. How
would you split thro a hundred percent into three buckets between public equities, bonds, and alternatives?
John Cadeddu (08:21)
Ten percent public equities, the balance alternatives.
Slava (08:25)
No bonds.
John Cadeddu (08:26)
No bonds.
Slava (08:27)
Okay, this is fascinating. So ten percent public equities, zero percent bonds, and then ninety percent alternatives.
John Cadeddu (08:36)
ninety percent everything else.
Slava (08:38)
I love it. I love it. Okay, so now that we're in the ninety percent alternatives, let's try to break down that bucket into kind of the major categories. So traditionally we think about venture, we're thinking about private credit.
Real estate, crypto, art, collectibles, you already mentioned farmland, potential other real assets. So can you give me a breakdown into some categories? What would be the two or three major chunks if that ninety percent is now a hundred percent? How would you break down your world of a hundred percent alternatives into its major chunks?
John Cadeddu (09:13)
Sure. So half is real estate, meaning farmland, and the other half is the commitment I make as a general partner into our funds and the personal investments I make into startups. So it's half farmland, effectively half venture capital.
Slava (09:29)
Crypto?
John Cadeddu (09:30)
Crypto, yes. I'd say single digit percent within but it's about ten percent of that venture investment. So ten percent of that fifty percent.
Slava (09:42)
Got it. And then outside of farmland, we're gonna double click on that in a second. Are you investing in too much real estate? Let's call it traditional real estate outside of your primary home?
John Cadeddu (09:51)
Excellent question. well, farmland has been very good and it's there's a lot of things going on there.
Slava (09:58)
Yeah, we're gonna have to talk about that. So, art or collectibles?
John Cadeddu (10:02)
Not yet.
Slava (10:03)
So that's not really your thing.
John Cadeddu (10:04)
Well, it th there are great merits to it, but it it hasn't happened yet at you know, there's a price point and and there's a correction going on in the art market that that could change.
Slava (10:16)
private credit.
John Cadeddu (10:17)
Not yet.
Slava (10:19)
Awesome. So basically it would be your hundred percent is let's call it 50% farmland, 40% venture, 10% crypto, or 45% venture, 5% crypto.
John Cadeddu (10:32)
Forty five five, yes.
Slava (10:33)
Yeah. Okay, great. Super interesting. Let's talk about that farmland real estate point. So we've had several people, quite a few people say they like real estate. That's not exactly a novel concept.
But not everybody comes up with farmland. I mean, we've had some of the farmland platforms on here, so it's not the first time our listeners will hear about this. But you're specifically calling out farmland as real estate versus an actual home. Why do you go to farmland? Why is that where where you put your money?
John Cadeddu (11:01)
So this the thesis on that one was to identify, I mean, the thesis, you know, we're collapsing a lot of years into a couple of sentences, but the thesis is to identify areas where there's desirable farmland with water rights, all the things you usually hear from real real asset investors.
that are in the path of prosperity or path of progress. So you're looking at farmland that's likely to become desirable development land in the ten or twenty year horizon. And the area, you know, we started geography in southwestern Idaho and the investment was this is standalone on its own, fantastic farmland. if we are astute in our investment thesis in ten to twenty years that farmland will become suburbs. And
A thesis is played out.
Slava (11:48)
So that's interesting. in the near term, is it a yield product for you?
John Cadeddu (11:53)
It is a yield product. It's it's you know, you're looking at four to six percent yield.
Slava (11:58)
Got it,
but you're really playing more the equity kind of speculation of the appreciation, is that right?
John Cadeddu (12:02)
The appreciation, yes. Yes.
Slava (12:06)
Got it. 50% is quite a bit. Is that like more recent or have you always been thinking kind of in those kind of percentages?
John Cadeddu (12:13)
It's it so it's the first farmland purchases made in late 2012 and
Slava (12:21)
Okay.
John Cadeddu (12:21)
portfolio's been built you know built over you know years. it turned out to be a a well turned out to be a very effective way to diversify as we generated really nice returns on the venture funds and investments did well to diversify. You know, you take some of it, you reinvest it back into your next fund, into the next companies, you find pretty
compelling and then from that you say okay how do I diversify so I'm not correlated with my day job and with the sector of the tech world which you know i is just you know it's a unlevered uncorrelated way to kind of diversify out that asset base.
Slava (12:58)
So first investment into farmland is 2012. So in 2011, were you investing some of your yield-oriented investments or some of your hedging somewhere else? Or were you not doing that? So you were a bit concerned because you were all in on your venture returns and you wanted to figure out how to diversify your basket. Okay.
John Cadeddu (13:18)
The latter, yes.
Slava (13:20)
Got it. Okay. So that's a great takeaway for the listeners, which is you're a little bit too high risk, too concentrated.
Felt like you wanted to diversify and that's where you start leaning in quite a bit into farmland. Is that fair?
John Cadeddu (13:32)
Yes, exactly so.
Slava (13:34)
Amazing. What would be, given that you've been doing this for like fourteen years, give us a lesson or two for people that are new to farmland as to if they want to explore that investment. what's something that you've already learned now fourteen years in?
John Cadeddu (13:49)
So I had a lot of there were a lot of hopes, a lot of I had a lot of ways I thought I could add to the return. and they were all dwarfed by the capital appreciation that came from, you know, towns, communications and towns growing into those farm areas. we explored, you know, moving farmland from you know traditional farming to organic, getting a premium for you know tenant farmers who were farming organic.
we were looking at monetizing water rights, all all marvelous you know ways to to to monetize the the investment. But I'll tell you, if if you know you happen to have farmland in an area where people are moving into that state or in that region, the farmland will go from X to five X in ten years and no amount of inorganic to convert you know organic farming conversion can come close to that. It's it's really
Pretty stunning how fast that development value changes.
Slava (14:46)
So it's all location, location, location.
John Cadeddu (14:48)
It is, you know, you know, I call it the the two rings the two rings concept, right? There's a ring around a com you know current, you know, thriving urban or suburban, you know, region and you look out ten years and that ring, where's that ring gonna grow to? And you kind of point the arrows and then, you know, you talk to you know you talk to the the planners and those communities and they say, no, yeah, we understand this farmland will be you know, it'll be reclassified, you know, we'll have to put schools here or communities here. And if you're patient, it just works.
Slava (15:17)
And what's your entry point? Are you going through a fund? Are you doing it yourself?
John Cadeddu (15:21)
it was
direct per personally buying farms, you know, so a couple times a year I, you know, I'm up talking to the farmers, just making sure that everything they're happy. They're, you know they're the guys who, you know, they they they they do the work, they rent the dirt, they make it work, and you know, you get a four to six percent return annually. you know as as as lease prices go up, you it goes up. and then you start noticing that neighboring
parcels near you are trading at remarkably different prices.
Slava (15:53)
Wow. Okay, great. And in terms of venture, how would you split that up between obviously your own exposure via the fund versus one off investments in pre IPO companies?
John Cadeddu (16:07)
So historically I was all in on the fund. and then that started to change about 10 or 12 years ago. You'd get you'd you'd understand, like, I wish I'd invest in this company. I'd call it Company X. And you know, someone beat you to it, right? The they got a better term sheet, something happened, you missed the deal. Well, you know, with the you know, about 10 or 12 years ago, you started having more and more of this secondary market where employees were leaving, you know, team members leave, there's they're looking for buyers for that product, maybe the
existing investors didn't exercise the right of first refusal. So you can start buying personal stakes in it. So you'd buy a few of those, see how it went. some work very well, some don't. And then you start doing a little bit more of it. So I'd say, today I'm probably seventy, eighty percent fully, you know, invested directly into our fund as a general partner. I might find a couple of opportunities that fall outside the purview of the fund where I happen to know the founder or some of the investors say, Hey, look, John, you
could add some value here and I say it's not for the fund. They say sure, but what about you? It's all invest directly. and every now and then you'll say, gosh, I wish I figured that that was a great one and I didn't pick it up right away. And then you just go buy some secondary shares. So I'd say it's sort of 60, 70% directly through the fund, 10 or 20% secondaries and then maybe 10% just finding something on one of those brokerage platforms where you say, yeah, you know, I didn't understand SpaceX in 06 but it's 2018 and I figured it out.
Slava (17:27)
Perfect. So based on some of your experiences, w back to the farmland, give me two areas in in the country that you think are ripe for farmland investment.
John Cadeddu (17:38)
Well, I we like I mentioned earlier, we we we looked at southwestern Idaho. I still think that's fantastic. It's just the rings have changed. and I think Tennessee. I think if you if you were to pick two coasts or two not coast, two one coast and one near the coast, one mountain region, those are two areas where people are pouring in. I'd say if you want a wild card, look at South Carolina.
Slava (18:00)
Nice. I like the clarity. And then on the pre IPO, since it sounds like every now and then you buy something that's not from your own venture fund, give me one or two that you like in the pre IPO world.
John Cadeddu (18:15)
Let's see here. there there, you know, I I there's a lot of names that come
Slava (18:19)
Well what was the last
one that you did that was a pre IPO, not a not your own fund?
John Cadeddu (18:24)
So there's a
You know, I I I d you know, I I bought it and then I I traded in and I traded up because the market was really high. I loved Pinecone. I loved Pinecone. it was fantastic. I think the investors in Pinecone are world class. they are extraordinary. and then, you know, the market moved really high and I was like, okay, why wait?
Slava (18:49)
Fair enough. Awesome. In terms of the economy and the market, love to give you an open ended question. What do you think of where we're at right now? obviously we're hitting all time highs in the stock market. Economy is hard to tell with K shape. I don't want to lead the witness too much. Very open ended. Give me John's point of view on the market and the economy.
John Cadeddu (19:09)
Economy is well the economy is growing. It's growing nicely. like every economy, every time, ever the always in the American economy, there's you know, cross currents, right? There's growing areas, thriving areas, and there's areas that are struggling or reinventing themselves or you know, just restructuring. So that we we live in that time now. and you've got part of the economy, which I think is going to become much bigger.
Which is, you know, leveraged against AI, leveraged against technology, leveraged against the occupations and the infrastructure of the future. And as we've had for decades, we are also continuing to kind of migrate out of that sort of industrial base, and that you know those those areas are challenged. I think net economy is strong. That that said, I've been around these capital markets enough to understand that we're at a
We're in great fully valued moment. we may well be much higher in five years, but I anticipate we will have very substantial, emotionally gut-wrenching, you know, market corrections between here and there, as we always have. And there's no such thing as easy money. There's no such thing as buy X and it will do well without a roller coaster ride between here and there. We and we will have roller coaster rides.
Slava (20:30)
So let's talk about that roller coaster red. specifically, I have you back on on August sixth, twenty twenty seven, twelve months from now. Do you predict recession or no recession twelve months from now?
John Cadeddu (20:42)
Do not predict a recession over the next twelve months.
Slava (20:46)
How about Fed fund rate up down flat?
By how much
John Cadeddu (20:56)
Fifty to seventy five basis points. Three rate three three hikes.
Slava (20:59)
Okay.
But I guess once they start raising, potentially they're gonna keep raising.
John Cadeddu (21:05)
I agreed. Like the when the Fed r starts raising, they tend to raise a little historically they don't tend to raise early. They tend to raise a little bit later and then when they do they understand that that it's not a one off. They they'll be at least two or three.
Slava (21:19)
Unemployment up, down, or flat twelve months from now.
John Cadeddu (21:23)
unemployment rate will probably be a little bit higher, but I think the labor participation rate will go up.
Slava (21:31)
So unemployment will be a little bit higher, so low fours.
John Cadeddu (21:34)
low fours, I think so th I think it'll be two things. The unemp the measured unemployment rate will be a little bit higher. I think the the the participation rate will be higher, so I think we'll have more work more people working, but there'll be ironically higher unemployment rate just because the participation participation rate will rise.
Slava (21:51)
Great. Inflation up, down, flat, twelve months from now.
John Cadeddu (21:56)
Flat out.
Slava (21:59)
flat up. Okay. So marginally up.
John Cadeddu (22:03)
marginally up. you know, w you know, the the headline changes week to week on on fuel prices. So that's that's such a big delta. But I think we're I think frankly we're gonna find ourselves in a tight labored situation. I don't think labor I think labor prices are going up and I think that's a good thing socially, but I also think that's what will draw people back into the workforce. So that's why the participation rate will rise.
Slava (22:25)
So you mentioned fuel rates. Are we still talking about Iran a year from now?
Okay. That's optimistic, that's good. and what is the impact
John Cadeddu (22:34)
Well, a year from now
now, we're probably still talking about it by the end of this year, but a year from now no.
Slava (22:39)
Okay, great. And what does all this mean for the stock market twelve months from now?
John Cadeddu (22:47)
I think there's a lot of liquidity. I think people are buying the dollar. I think there's a lot of a global flow into the United States. that's why you see the Fed and the Bank of Japan intervening on behalf of the yen. I think that's, you know, it's these are co these are related factors. I I see I believe you'll see continued material inflow into the United States from a capital perspective. I think that will put a lot of desired capital to work into the US, call it equity credit situations, markets. we will have.
increasing volatility. I think when the Fed raises rates, I think you're gonna see reset on small caps. I think that just that's just how it works. I think you'll see headline risk on AI. And I I think we are we haven't really seen how well we scale and monetize the AI AI infrastructure budgets that are currently causing public market investors to pause a little bit on their favorite names.
Slava (23:42)
So stock market traditional eight percent year over year up a year from now, down or up more than that?
John Cadeddu (23:49)
I could see it l up less than eight. I could see it low single digits with a lot of roller coasters
Slava (23:53)
Called four percent.
John Cadeddu (23:54)
along the way. Yeah, let's say four, but you know, you could see a ten or fifteen percent move down between now and then and and back up.
Slava (24:00)
Awesome. And since you know so much about venture and been doing it for so long, I'm gonna ask you a question I've never asked anybody else, but this is gonna interesting. A year from now, what is the enterprise value of SpaceX, OpenAI, and Anthropic? And obviously that's assuming they're all three public a year from now. And if you think not, feel free to correct that.
John Cadeddu (24:24)
I think well so first of all I
So yeah. there's so much going through my head right now. So I love what SpaceX is building. I think there'll be a lot of volatility in that stock as the lockups come off. a lot of hot money went in. I I can see the you know, I could see I can see I I can see breaking under a hundred, I can see breaking under ninety. but I if it does, I'm a buyer.
I love that name. It's a name I want to see in my portfolio in a big way. in a year, I could see it going down a lot and coming right back hard. Don't bet against Elon. The guy is one of one. so I could see a lot of volatility in SpaceX and I could see it being higher. I expect whatever it does in the next 12 months, if I get a buying opportunity, I'll take it. And I can see them five years, 10 years being that name where everybody says I should have bought it when it went down under 100. Now let's go open AI.
Slava (25:18)
Sorry, so what's I you gave me an entire narrative, which is all great.
John Cadeddu (25:21)
Mm-hmm.
Slava (25:22)
Near term, long term, but what do you think it is a year from now?
John Cadeddu (25:27)
I think it's I think it's ten to twenty percent up or more from where it is now, but we will see ninety between now and then.
Slava (25:35)
Okay, awesome. So that's like I don't know, like maybe one point five trillion a year from now.
John Cadeddu (25:41)
Mm-hmm.
Slava (25:42)
And then what is is Anthropic public?
John Cadeddu (25:45)
I expect Anthropic is public and it does extraordinarily well when it goes public. I think if it if there's any pullbacks between it going public in a year from now, people will be buying it.
Slava (25:56)
What do you think is the price of Anthropic as enterprise value a year from now?
John Cadeddu (26:01)
Two trillion.
Slava (26:03)
Bigger than SpaceX
John Cadeddu (26:04)
Yes.
Slava (26:05)
and OpenAI, are they public?
John Cadeddu (26:08)
If they are, they're just they just recently got public in twelve months. I could I could see that being a mid to late twenty seven IPO.
Slava (26:15)
Whoa, that's a hot take. Especially for a guy who said he was gonna go public before Anthropic
John Cadeddu (26:23)
it I just think Anthropic because they locked into the enterprise market and dominated and have monetized it so well, there's so much momentum and so much ownership they have and mind share with the enterprise users that are just loving it. And I understand that, you know, they've got regulatory frameworks and delays and things like that they're dealing with, but I've yet to meet someone who says I'm no longer using cloud code or or
you know, the tools that I've used with them and they might say I'm using other ones too, but they they have the franchise for now and I don't see that that doesn't change in a year.
Slava (26:58)
So
let's assume they just went public a little while ago. what is their valuation in enterprise value a year from now?
John Cadeddu (27:05)
Anthropic or opening eye?
Slava (27:06)
Sorry, open
AI.
John Cadeddu (27:07)
You know, it's it's there's just so much going on in open AI that I don't it's so opaque. I just don't know their numbers and how much they're spending relative to that number. And I think the market will be very discriminating. I I could see that in six, seven, eight hundred billion, but but not not being the billion plus.
Slava (27:28)
Not being a trillion.
John Cadeddu (27:29)
Actually possible.
Slava (27:31)
thank you for playing that game. No one's ever played that with me. That was fun. It's only because you're obviously so smart and experienced. That'd be fun to get your perspective.
So that was really interesting perspective on the market and obviously the fact that you have this unique thought on where the big three SpaceX, OpenAI, and Anthropic are gonna be. I guess my next thought is you just have so much experience around investing and and the and AI. What is happening in AI right now? Can you give me s your thoughts, especially compared to what you've seen in other waves of technology?
John Cadeddu (28:05)
That's a really good question. It's one we end up debating a lot around here. So let's take a look, put in context. So let's look at the waves that we've had, at least since I've been on this on this roller coaster. I worked I'd had some early early jobs before I joined the investment world down in Silicon Valley, you know, tandem computers and Cupertino, Octelecommunications and Milpitas, a couple of other places. and if you went back to the early 90s.
Silicon Valley was transitioning out of the PC era, computer era, and kind of working in the software. And people had used terms like client server, blah, blah, blah. And then we hit the internet and then we really shifted out of hardware into software. And investing went from in the 90s from hardware, mixed hardware, software, and then by the 2000s, almost all software. So
What investors were looking for, what they were pricing risk against was very different in a software context than it was hardware. Hardware, you'd have to hire a bunch of designers, engineers, hardware builders, things like that, put a lot of money in, see if you could build the thing. Software was a very different thing. You had to look for different signals, product market fit. You didn't weren't looking for hardware, could it be invented? You had to understand, does this product sell, does it scale, can can you build a margin on software?
then you had the move into SaaS, you had the move into big data, you had a lot of consumer apps that emerged, social media emerged against mobile devices. So post-2007, you started adding new layers and new nuance to what you were debating around the venture capital table. what risks were you pricing? You started thinking things about like how do you distribute the app? are you going mobile native? what's your social media strategy?
then you start having new business models like Uber, which leverage maps and big data and cloud computing and things like that. So the conversation in venture investing keeps iterating every call it five to ten years. And so here we go, and we're doing our thing during the pandemic, and people are still investing like they are in the late 20, you know, 20 teens. And OpenAI drops GLP. sorry, GF GLP.
I'm starting to think about our friends at Lilly. they start dropping well, they start dropping the first models, right?
Slava (30:17)
GPT. Yeah.
John Cadeddu (30:19)
And you know, and then as the models drop, we start using them. And between the models and what those of us who played with technology for a few decades, we started looking at how powerful these models can be. Anthropic emerges, you see some other models emerge. And so we start talking about models in 2023.
by late twenty-three, you start hearing people talk about agents. And I think, you know, on one hand, in the last 18 months, people have talked a lot about agents and really kind of started breaking out in twenty late twenty-four and and through twenty-five. I think the conversation around this AI as an investment thesis really buckets into a couple of couple of big buckets. One is the models, right? That th there are
Those are huge companies, right? That is immense capital being deployed against immense compute, right? Which is why you see the model builders and the compute infrastructure guys so tightly linked together, all the way, you know, from NVIDIA all over all the way up to you know, Oracle and you know, Bloom Energy we mentioned earlier, and the model builders themselves. so can you get access to compute? That's the you know, that's big boy talk around a lot of big dollars.
from a venturance perspective you're looking and saying okay now what are the derivatives what are what is that first derivative of and the second derivative related to these models getting more and more capable and how does that impact what you should invest in and that's really been driven by a couple of insights one we look at some of the folks who invested in those early AI wrapper call LLM wrapper companies and saying yikes you know that's not a good business to be in you're a thin wrapper around
Whatever model you choose to wrap. And as the model gets more capable, your your thin little wrapper suddenly evaporated. You're a feature on someone on a model that just got released and it's a free feature now. So that conversation then necessarily drives you to ask the question: where is AI changing how we do business or how do businesses use it? And that's where agents get really interesting.
I'll give it a you know, kind of without naming names. one of the more interesting companies that we've invested in recently is one that we found because more and more companies were just telling us, look, we can't live without these guys. Like we have a budget, you know, before AI emerged, we spent a lot of money on this work function X. And because of what these guys can do, we now don't spend that. We spend it on them, and we're able to do a lot more.
now that's really interesting because there's two types of AI companies that grow fast. One is the one that gets a lot of customers who are basically have an experimented and call a budget for experimentation. They're they're you know their bosses gave the the engineering group, here's you know, go go do some AI stuff, right? And then they go and buy a bunch of toys, they play with them, and the guys selling them those toys are like, hey, look, we're blowing up, it's great, it's fantastic. But then the toys never kind of leave the experimentation stage and go get migrated into the business.
And so, you know, eventually there's new toys we bought and you you know you find yourself losing customers. There's a second class of c companies that sell I don't want to say solutions, but they're what they do is they're building product that aren't really being used, or if they're used by the experimenters in the in an enterprise, they're quickly sucked up by the main business. And you know, at some level you say, okay, well that's that's ingenious. And on the other hand, you say, well, that's how that was Oracle, right?
When they started building relational databases and people started saying, I need one of those, and I'm gonna need a lot of them now. So we're looking for that business, and when we find it, we get very excited where you can talk to the customer who says, yeah, I'm using this all the time. It's in my board deck now. Like I tell my board, I can't live without it. Right? It's not it's not in the RD bucket, it's in the this is how we survive and grow and thrive as a business bucket. So if it was only a few of those, you'd say, Okay, well.
Maybe AI is kind of a cool area for venture guys to invest in, but you're seeing it a lot now. So that's you know, back to what we discussed earlier. I can see the American economy once you start seeing a lot more of these businesses, small, medium, and large, adapting and then bringing AI into their functions and not using it to shrink their headcount, but do it to reduce their
waste and and latencies so that they are better able to serve customers, they serve more customers, they grow faster. And when you hear enough of these people talk about how our business is healthier, our margins are growing because of these functions, and you hear them say, yeah, and we're hiring more people, start be a little bit more bullish about the labor market, a little bit more bullish about the American economy, and about you know how can technology, you know, do its job and make businesses better. And so from a venture perspective,
That last AI buckets, I think, will be where most great investments find themselves or come out of in the next five to seven years. Companies building product, not LLM wrappers, there are plenty of great models. The models will only get more, you know, more powerful, more capable. In fact, I'll predict in 12 months, we will look back on the current models and say they're toys, right? It's just that's how fast it's moving. But if you can find and build products that solve
Big expensive problems for enterprises where the as the models get better, your solution improves. You're leveraging that into more and better capabilities for yourself, not because you're just wrapping around the model, but because the model unlocks more things that you can do. That that's that's the bread and butter for venture. And you know, if I'm if I've if I've been a patient LP saying, Where's where's my DPI? Where are where's my cash? Right, it's the Clara Pella thing. Where's the beef? Where's my cash? Where are my returns? I made the capital calls.
Where's my returns? This is where the result returns will come from. It's a really deep, powerful, valuable set of companies that will come out of this category.
Slava (36:21)
How
do you identify an AI company that can be standalone and compete against the anthropics of the world versus the company that's quote unquote the thin layer that is just gonna be consumed by the future release from anthropics?
John Cadeddu (36:37)
So there are a couple of things you want to look for. One is are you building, you know, with the AI tools and solution set and framework that you build around and build with, are you able to unlock, capture and unlock proprietary data that isn't going to be used, you know, accessible or used by the models? So give an example. If you are looking, if you're a health tech investor, there's an immense amount of data.
That people generate, and if you can capture it, you can help them heal, you can help them get better. Right? That's not that's very different than using GPT to say I have these symptoms. You know, GPT for healthcare is like Dr. Google, right? It we're just doing asking, we're asking a model instead of asking Google, you I have these symptoms, and I go tell my doctor I'm dying, and none of it's true. but if you can unlock proprietary, closed data and bring powerful tools to it.
you can then unlock a lot of value for whatever customer is generating that data or the user that's generating it. So that's one. Two, you really are looking for products that are not doing what the models do. So you're not looking for reasoning, you're not looking for compute. What you're looking for is accessing data, being able to
understand the the workflow that the data is coming out of and going back into in a manner that is very valuable to the problem that that data owner has. And you know the the model builders are gonna do a lot of things and amazing things, but that is very much not what they do. And there's a you know you can pick any vertical
You can understand it very deeply. You can use models to help you do things to that data you unlock. but between unlocking the data, authenticating the data, creating, you know, chains of custody that are, you know, auditable, things like that, that's that's not the models world. And that's there's a lot of value there to be had and to deliver for customers.
Slava (38:46)
So does that mean your investments you're looking for are typically vertical specific, like in construction, in aviation, in health, in whatever?
John Cadeddu (38:56)
Increasingly, yes.
Slava (38:58)
Got it. Super interesting.
John Cadeddu (38:59)
Which by the way, sorry, I I don't mean to but but the one thing that also means is that if you find a a founding team where someone has incredible domain expertise in that vertical and they have someone on the team or co founder who's extraordinarily capable from a technology perspective, that's sort of like that magic duo you look for.
Slava (39:15)
What
stage are you typically investing in?
John Cadeddu (39:17)
Historically we were series B to forward, but then when we started incubating companies here in the last three years, we started, you know, basically you know, day one, you know, incubating the office, so seed and forward.
Slava (39:28)
So are you investing into seed outside of incubation or only for incubation?
John Cadeddu (39:32)
At the moment purely in incubation. That's just to keep our interests fully aligned with our LPs.
Slava (39:37)
Otherwise you're investing in B.
John Cadeddu (39:38)
Yes.
Slava (39:39)
What's the what's the prices? What's a B look like these days? Give me give me what's that?
John Cadeddu (39:44)
It's always hot to pay. It's always
high, it's always high. It's but it but it's been high for twenty five years. You know, it's it it
Slava (39:49)
But
John Cadeddu (39:49)
it it's always high.
Slava (39:51)
what's the checklist for a B? Like what do they have to have accomplished? Like revenues, growth, scale, proof, evidence. What's a B look like?
John Cadeddu (39:59)
You know, historically a B look like, you know, three to five million dollar run rate, but that's now an A, you know, because you can build AI being able to build AI product, ship AI product, and meet the demand for AI product often means that you're hitting that threshold eighteen months after your
Slava (40:18)
Right. So what's a B?
John Cadeddu (40:21)
A B today, so let's look at the the the truth is a B today is often A's are where the B's used to be. And the B now is
Generally a company doesn't need the money, but a couple of investors show up and say we'd like to put money in and they they bid it up until someone says, Okay, I'll take money at that price. You know, it's a it's a dilution game. So you're you're talking a hundred million plus value pre market valuation. The pre yeah yeah, pre money valuation.
Slava (40:53)
what's their revenues like?
John Cadeddu (40:56)
You're looking at ten million plus annualize.
Slava (40:59)
Got it.
Okay, great. And you've mentioned you're now incubating. How many have you done of these?
John Cadeddu (41:03)
We've incubated five. we've incubated
Slava (41:05)
And this is for like the last three years.
John Cadeddu (41:06)
over the last three years. The first one we started incubating April twenty-three. we incubated five. my my partner Marvin and I co fund you know, co-funded the the five. the rules are pretty simple. We wouldn't put our we wouldn't ask our LPs to look at these until we had de-risked them, and so we put our personal capital risk and we said, look, we're gonna do them. we started five, I I killed three of them.
That's just
Slava (41:27)
With what within what time horizon?
John Cadeddu (41:30)
all within the any we the the goal was always to kill within twelve months if it wasn't, you know it had j it just has to be spectacular.
Slava (41:39)
So that means how much like investment was put in in those twelve months before you killed it high level on average?
John Cadeddu (41:44)
On average, probably about eight hundred K in twelve
Slava (41:47)
Got it.
John Cadeddu (41:47)
months. Yeah.
Slava (41:49)
And two are alive.
John Cadeddu (41:50)
Two are live. Two are live. Two are alive. one is doing well, one is doing extraordinarily well. It's, you know, it
Slava (41:57)
Which ones do extraordinarily well?
John Cadeddu (41:59)
so that's a company we we created here. We called it Aurite it started life building agents. now it's rolled out a product, it's got rolled out two products. one is for engineers and the development teams that build agents called Aurite ID. And the next product it's just rolled out is it's been called an
They call it, we call it an an agentic control layer, which allows the company's all-rights customers to basically see all the agents running across their infrastructure, control them, stop them from doing things they shouldn't be doing, kill them if they're doing things they really don't want them to be doing, basically control the agents on the infrastructure. And it turns out that enterprises have been building a lot of agents and they really want some, you know, understand what's going on.
control it, report to the board, report to their customers that the agents are safe and there's an audit trail with it.
Slava (42:49)
And are these all agents that are created within Aurite software or are these agents that are being managed outside of Aurite that are being under the Aurite management umbrella?
John Cadeddu (42:59)
Second. So the idea is, you know, while we started building agents for Aurite, just that's what led us to understand that the bigger need was tools like ID that let engineers build their own agents but securely. and then the second to control those agents at the across the infrastructure. And so now with the control layer, enterprises can no matter where the agents come from, as like if they're running on the infrastructure, the control layer gives that.
visibility and control to whether it's the VP of Engineering, Chief Security Officer, you know, whoever whoever needs to know and see it.
Slava (43:32)
and when was that started?
John Cadeddu (43:34)
so Aurite in fact Aurite was a twenty three. It's the it was the second one we created and it's it's out lasted the all but one and it's the one on the highest trajectory.
Slava (43:47)
Nice. So is it soon to be getting a B from you guys?
John Cadeddu (43:52)
you know, if it needs it, we're here for it. That that's what I'll say. It you know, it it we we've we opened it up to friends and family of two months ago because people were asking us, can we invest? You've been talking about Aurite a lot, so we opened it up. we opened it up for up to a million dollars, and it turned out that friends and family found that compelling. we might take on a few more of them, but I don't anticipate, I I wish it could use.
you know, five or ten million dollars to you know to to get to pre keep it but it won't it won't need that much money. So this will be a conversation with the LP saying if it needs it, we're here, but it won't need that kind of capital.
Slava (44:31)
So when you're incubating a company like that, what's kind of the exit you're looking for? Is it is it higher ownership flip of a smaller exit, i.e. let's call it a hundred million dollars, or are you still looking for that five billion dollar public company or our acquisition?
John Cadeddu (44:47)
You're always looking for I mean w we in QA would always want to aim for, you know, you want the home run, you you want the multi-billion dollar exit or or greater. and if you build a company for that, then all sorts of wonderful MA opportunities come along the way and you can entertain them. I've never had good experience working investing in companies or with fund founders who are I hate to say playing small ball, like trying to build something to flip it. You just
You just don't hire the right team to start with. You're you're making short term decisions, trying to hit a window. And that's that's it doesn't usually work. But if you're building something for a big outcome and you're trying to make long term decisions, make the correct decisions, it it lends itself to both getting to the long term outcome, but also to other companies saying, look, you've built something good here, we we just couldn't build it as fast as you or we just missed it. Can we talk about partnership? And that that leads to the happy MA conversation.
Slava (45:41)
Who's the main customer for Aurite?
John Cadeddu (45:43)
So mid to large size enterprises, you know, call it you know
top five or ten thousand you know enterprises in the United States. the target buyer is gonna be the VP of engineering or the security team. it's not per se a security product, but it plays nice with all security products. It you know like I said it'll it'll notice when an agent's trying to make a call or do do something that the policy doesn't allow it to do. So it'll stop it and it'll alert you know pick your, you know, pick your happy security vendor, let them know, and then you can either terminate it or your security tool plays with it.
Slava (46:19)
So agent management I imagine is a competitive space. Obviously good word of mouth is always helpful, but what would you say is one of the let's call it tricks as to how you guys compete beyond just the simple answer is we have a good product and it sells itself?
John Cadeddu (46:36)
Sure. So well first thing, first I I tell the team, look, let we are not a security product. Like convincing an established company that is happy with the security solution that it should trust a startup with its security is is a dead end. So we should always complement security. So first of all, don't put obstacles in your path if you want to get adopted by customers. second,
agent management is absolutely top of mind for you know call it boards anytime I've been on board I'm still I was on a board meeting last week where you know the technical team was briefing the board on their AI strategy and the great progress the team's been making and all the things it's gonna do and and then inevitably there inevitably there's one or two board members who say, but are they safe? Are we you know are we gonna blow up our customers? Are we gonna blow ourselves up? And then there's a lot of, you know, no don't worry we we design these things not to fail.
The truth is there a lot of talk to date. We at Aurite haven't seen any enterprise say, fill in the blank large vendor already does what you're doing. We we actually just I I've sat on a number of the presentations to customers where they see the product working, they see you know agents scrolling and you know all you know every few seconds there's a task called out and agents trying to make a payment or access data or something like that.
You know, and you see it authorized by the Aurite, you green, green, green. every now then there's a red. They go, you actually stopped that. And they look and say, you know, did our other product catch it? No. So there's a lot of attention to the space to date. It doesn't appear that you have a lot of solutions actually delivering on that attention. So so far, fortunately, Aurite has got more work than they can handle.
Slava (48:20)
Great. Being innovative, moving fast. That's awesome. Anything else you want to mention about Aurite?
John Cadeddu (48:26)
think that's it's gonna be a name that people are talking about in a couple of years. It's gonna be a fun one.
Slava (48:30)
Very cool. So what is it? You're obviously a very smart guy. what is it that you like to listen to, read, watch that keeps you smart? Top of mind.
John Cadeddu (48:41)
you know, I spend more time on you know, if I'm if I can, there's more, you so much on Substack. There's just so so many smart and insightful podcasts. you know, I I subscribe, I listen to Lenny. I Lenny's the newsletter, Lenny's you know, it's the there's just so much good stuff going on there. He's just one of them. the the thing that I've noticed in in this turn of the technology, you know, whether it's you know the
whole tech ecosystem is matured or not is just how many experienced, skilled, and smart people are willing to share what they're learning as they you know use use AI. And you know whether it's as a product manager or an architect or salespeople, it's just, you know, you this kind of sharing and learning didn't happen, you know, in in the internet age nobody talked about it. In the, you know, when the when the iPhone came out and mobile mobile ERIC started, people people didn't share this way. It's
It's just a lot there's a lot going on. It's happening very fast. And whether it's podcasts or newsletters, it's it's it's a fire hose and it's amazing.
Slava (49:46)
Awesome. And our final piece, which is we love to put people on the spot and you've already been totally game to be on the spot. So this is almost like a bonus. what will be one company that you would, you know, say is a good investment today for three years out in the public markets and one in the non public markets, i.e. privates or something that's not in the public markets. So give us the two names and why. And this is for three years out.
John Cadeddu (50:13)
So I'll answer with one name to answer both questions. I if even if the price is silly today, I would buy Anthropic in the pre-IPO market. And I think when it goes public, it'll do very well. And I think it will suck a lot of air out of the room relative to its competitors in the next three years. I just think that's a special team. That the it's incredible. The people that are leaving other companies and joining them is, you know, talent is always a signal. And you know, anyone can Google.
Just look at the recent high profile hires going to Anthropic and compare that to the high profile departures at other companies that have their own models. And it's t talent is always the lead a leading signal. And so I would I know the price is absurd, but if I could, I would, and and I'd hold a post IPO and I'd look back three years and say, wow.
Slava (51:01)
So if you could buy right now in the private markets, Anthropic at one point two trillion. You you're buying that?
John Cadeddu (51:07)
Yes. Yes.
Slava (51:10)
Okay, great. I'm not gonna let you get away with that, even though I love your answer and it's totally fair. but I do need a public markets pick.
John Cadeddu (51:17)
SpaceX. I'll take SpaceX. I
you know, I you know I what is it? I don't I've I haven't checked today, but let's say it's it was one hundred fourteen or so yesterday. I jump in, I'd like to l I'd like to buy it under ninety, but you know, put on the spot, that's the one.
Slava (51:33)
But you did say it's gonna go under ninety. And is your opinion that the lockups or is it something else?
John Cadeddu (51:39)
I think it's the lockups. I think
I you know, I'm old enough to remember when Google went public, right? you know, f August 04 goes public and then it sinks. And a couple of years later, 2012, Facebook goes public and it sinks. And I still remember people were criticizing Facebook won't get mobile. I mean literally that was a conversation around the table six months after they had an IPO. And you know, you can argue was was a fa you know the Facebook or now meta IPO, was it properly managed? Whatever. It it it came out, it sank.
People, you know, went negative on it and, you know, come on, you know, they
Slava (52:12)
Five years
from now, which one's more valuable, Anthropic or SpaceX?
John Cadeddu (52:18)
That one I'll give the SpaceX.
Slava (52:20)
Ooh.
John Cadeddu (52:21)
I will give that to SpaceX. You know it it
Slava (52:25)
I'm surprised you just said that. I'm not gonna agree or disagree with you, but it sounds like you're so bullish on Anthropic.
John Cadeddu (52:26)
I know. It's it's it's no, but it it I am
look, I I I am and it sounds inconsistent, but
Slava (52:35)
Does that
mean that Anthropic is going to a plateau somewhat a couple few years out?
John Cadeddu (52:40)
It could, but it
I I just see the
Slava (52:46)
Or is that a statement
of SpaceX is so uniquely positioned in their market that it's harder
John Cadeddu (52:50)
Yes.
Slava (52:51)
to really erode any of their margins or their market?
John Cadeddu (52:54)
I I think it's the latter. I think it's a I think I as we're humans, right? And but but especially in Silicon Valley and in New York, you know, call it the areas where you you get people very very tuned on what's going on in technology and and new technologies and adopting them. That we're humans, we two things happen. We overestimate how fast things will change, and then we get really disappointed when they don't change overnight, and we underestimate how profoundly they will change. So in the you know, call it five years, the long term five to seven years.
I think we we're really enthralled at the with the SpaceX IPO. my gosh, it's opening up space. There's a lot stuff going on there. But boy, the compounding effect of Elon and the team doing what they're doing, we're gonna we're gonna we're gonna we'll probably with a lockup so people get negative on it and they'll be like, okay, whatever. But boy, in five years look back and say, wow, they're they're you know, they're they're at a big number. Anthropic right now, everybody's bullish. I'm bullish, everyone's bullish.
We haven't gone through that valley of uncertainty because there's no uncertainty about AI, right? people can pick on rockets and things like that, but but it's kind of it's it's a great company. I think it's gonna take a lot of air out of room from its competitors. But that SpaceX thing is a wild card. It's it it could be really special.
Slava (54:10)
Amazing. Well, we've covered a lot of ground, John. Thank you very much. From, you know, an origin story going back to two thousand one, twenty-five years of investing, raising many funds, multiple franchises, billions of dollars raised, some really big winners early, Bloom Energy, Yelp, Grubhub always helping to get pizza to us faster. You have a really unique approach to investing, which is great for this audience, only ten percent public, ninety percent into privates, half of that into farmland, which we really talked about quite a bit, which was
Super interesting. You like Southwest Idaho, Tennessee, South Carolina. that was great. And then on in regards to the economy, no real recession predicted 12 months out, up on the Fed funds rate, but maybe two or three turns, inflation kind of flat, stock market up, but not a ton 12 months from now. And really some talking about reset of small caps, maybe the AI risks are coming back up. You played a new game for us, which is predicting what the big three would look like.
three years out, SpaceX maybe one and a half trillion. Maybe it goes under ninety for all those are looking for a deal. Anthropic really coming out the gates fast at two trillion. And open AI maybe not crossing a trillion with its struggles. We talked about all the waves of technology from computers to big data to agents. And of course you talked about how getting access and data for specific workflows seems to be the differentiate opportunity. You gotta find specific verticals and find where they're really getting their own motes there.
the B is now an A, so now a B is much bigger. they're pretty expensive. They're well over a hundred million dollars of valuation. So you started thinking about incubating. You've incubated five times, kill three of them, have one winner, and one that's really breaking out. The breakout is Aurite which sounds really cool with a genetic management really becoming the unique feature. all of us know we like to keep track of our employees, but now you gotta keep track of your agents, especially now one that's going rogue. So green, green, green, red. red's important. We gotta figure out what they're doing wrong.
And of course you talked about the content you liked. You gave us a couple names, but specifically with Lenny's newsletter or others. And then back to predictions. We talked about Anthropic as your pick for private and SpaceX as your pick for public. And surprisingly, you said five years out it's gonna be SpaceX, that's gonna be bigger. Thank you very much, John.
John Cadeddu (56:19)
Appreciate you. Thank you so much.
