Auren Hoffman (00:01.474)
All right. Hello, fellow data nerds. My guest today is Spencer Kimball. Spencer is the founder and CEO of Cockroach Labs. Cockroach Labs manages CockroachDB, a global distributed SQL database. He also created the GIMP open source image editor while in college at UC Berkeley and has spent over 10 years working on major infrastructure projects at Google. Spencer, welcome to World of Desk.
Spencer Kimball (00:24.957)
Thank you, Auren. It's a pleasure to be here.
Auren Hoffman (00:26.962)
I'm really excited. Now, what are the different models that different open source projects monetize? Seems like there's like maybe four go-to things that people are out there. Like, how do you think of it when you're thinking of like monetizing open source?
Spencer Kimball (00:44.329)
Well, when we started, it was a different world and the model was what's called OpenCore. Before that, probably Red Hat was the first. So that would have been, you know, the turn of the century, a long ago. And they were providing support on open source software, primarily.
Auren Hoffman (00:56.407)
Auren Hoffman (01:02.006)
Yep. Yeah. It was like a services model in a way.
Spencer Kimball (01:05.177)
Exactly. And they expanded way beyond that. And they actually started, I think, doing more than just support. They have quite a number of different products under their portfolio. Then you had Cloudera come along and you had Hortonworks and those two were locked in quite a bit of competitive pressure. And actually Hortonworks, they were both selling MapRedos. They took a very different approach than Cloudera.
Cloud Arrow had started to, I think, really pioneer the open core model. What important works was continuing to sell the pure open source product and provide support for it. So that was kind of an early split in the business model.
Auren Hoffman (01:47.106)
And both of those guys were essentially building things off of Hadoop. And so, like in some ways, the product they were selling was kind of the same. Right. Or how do you how do you think? Yeah. OK.
Spencer Kimball (01:57.933)
Oh yeah, it's very much the same. And that was actually created some bitter rivalry there because Cloudera was first to market and then the open source developers were sort of rounded up and brought into the, to create the Hortonworks as a company in direct competition. And I think it was an example of a lot of competition that ultimately made it more difficult for either company to do well with Hadoop in that whole.
Auren Hoffman (02:27.554)
And that seems to rarely happen usually when there is, I mean, back in the day, maybe with Linux and stuff, but nowadays it seems like when there is an open source project, you know, some person has commit rights over it. And then that is the person usually who ends up like starting the company around.
Spencer Kimball (02:47.057)
That does seem to be the normal model. I think MySQL is an interesting, really interesting. The complexities there are definitely worthy of, I think their own long discussion, but you had obviously Oracle being involved after Sun acquired it, after Oracle acquired Sun and Sun had acquired MySQL. I think that was the order that things went in. So that.
stewardship of that project, which is extremely popular at the time, the most popular open source database. I think Postgres has eclipsed it now. That spurred, I think, a lot of thinking of we can't have this open source project under the stewardship of Oracle. And so it split off into a number of different variants. And I think that took a little bit of the wind out of its sails, to be honest. By the time we started Cockroach in 2015, the open core model was in full swing.
very viable. And I'll just give a quick description of what the open core model entails. You have your core, which is the open source project. And that's really what leads to ubiquitous adoption. And typically, a company would start around that when you already had good adoption of the open source project, and it's starting to move up market considerably. And in order for a bigger company to adopt that open source project, they do need the support. So in that sense, it's very much like the early
Red Hat model, you're selling support on the open source product. But what people quickly realize is that, especially at the high end of the market with an open source product that worked pretty well, people might wonder after year one or year two, if they look at their budgetary line items and there's something for $500,000 or a million dollars a year that's support costs only for an open source project that they haven't used any support on in the last year, they start to wonder whether they need to pay a million dollars a year on the support. That's natural.
Auren Hoffman (04:37.314)
Spencer Kimball (04:42.577)
So that led to some of the thinking, I think, at Cloudera. It's like, hey, let's think about how we can extend the core and create a constellation of enterprise features that are licensed differently. And ideally, and this is the model Cockroach initially started with, you really kind of find the distinction between features that should be core, they're very important for the adoption and the usage of anyone that really just doesn't want to or really have the capability to pay for.
software, you want to make sure that you're not creating any blockers for them. But at the same time, you don't want to be too permissive with too powerful of new capabilities or else nobody's ever going to pay you. You don't, you don't create an, like a big company where the thing works really well and it does everything they need also won't necessarily pay you, especially if it's not something very, very critical in the stack. So you kind of want, you want to create enough of a, of an obstacle to using something, uh, that the
folks that should pay you, that really should pay you, they can afford to, and they're using, they're deriving immense value from what you're offering and you need to keep developing that immense value and someone's got to pay for it, you want them to be happy to pay you. All right, so that's kind of where these, if you draw the line properly and you create additional capabilities that are really useful in say, an enterprise company setting, like you integrate with LDAP, that's kind of a classic example. Just something that, you know, a startup definitely doesn't need.
Auren Hoffman (05:45.772)
Spencer Kimball (06:10.725)
but a big bank, absolutely, they can't even think about using it unless they have that. So that's a perfect thing to add because it's going to be
Auren Hoffman (06:17.362)
Yeah, or they can, they can, they could start using it as like a project, like an internal project for free using, and then, and then they are like, Oh wow, now we need this other feature. And then they essentially upgrade. They go from free to up.
Spencer Kimball (06:33.357)
Exactly. And that's the beauty of the open core model, right? What you're getting.
Auren Hoffman (06:38.598)
Why is open core really any different from, you know, kind of a freemium thing? It's like, hey, you get this cool product for free. Um, and, you know, classically maybe much more on like the, the SMB or consumer. There's this cool product you're using for free. And then now you want to have a slightly better product or you need these new features or whatever, and then you get up sold.
these other features that you pay for. Like in some ways it's just kind of the same thing, except you can see the code or something. Like what's the big difference?
Spencer Kimball (07:08.969)
Exactly. I mean, you just you just pointed out the big difference, right? The difference is open source and closed source. But otherwise, the economic model, the business model is the same. You have a free version that leads to, you hope, more adoption and therefore a lower cost of customer acquisition. People call that CAC. And that actually fills the top of your sales funnel. And you you're kind of taking orders, right? Assuming that your freemium product is powerful and people use it.
They get to a point of usage, the ones that ought to pay you, and they pay you. So that is exactly the same. And I think there was no new ground broken in that open core model in that sense. But the difference is open source and closed source. And open source has these other amazing benefits. You can create these communities that actually build your long tail features and integrations. You can create communities that provide support for your users, especially a super widely adopted, ubiquitously adopted
product. You really want to have people believe in the idea of a project that they're all contributors to that actually greatly increases the quality of the contributions and the esprit de corps of the community. It's very different from a freemium product. A freemium product in the traditional closed source ecosystem was one that you kind of threw over the wall. It was a binary. People would probably happily use it.
why would they really have a strong investment in it? Open source is just different. Like the quality of the commitment of the people in those communities is vastly different.
Auren Hoffman (08:44.298)
Yep, because they could contribute to it in some sort of way.
Spencer Kimball (08:48.433)
Yeah, and also, you know, there's a different dynamic and adoption at the high end of the market, especially tech companies. People love the idea of being able to fix their own problems if they come up and they're exigent. Like this, these closed source products, it's kind of like you're at the you're at the mercy of your vendors. And the reality is that, you know, even when you're, let's say, one of the biggest banks in the world, trying to get one of these big vendors, Oracle, AWS, whoever.
who has some closed source product that they're offering, whether it's a service or it's a piece of software that you're paying subscription for, trying to get them to listen to your concerns, you're not always gonna do it. Doesn't matter if you pay a billion dollars a year to them, you still might not get what you need, right? So from a fast moving tech adoption, tech company adoption perspective, having the option to go in and fix something yourself, figure out what's wrong, make a suggestion is valuable.
And it depends on just how valuable that is. I think that people want to be in control of their own destiny. And they have a number of great engineers at their disposal. And being able to debug something is very valuable. When I was at Google, you mentioned that from my bio, there were a number of instances where, Google used a lot of open source software, where they actually got down into the Linux core and solved bugs.
Like using Microsoft, you know, NT products or whatever for the server, that, that would have not given them that flexibility and the cost would have been enormous. So, you know, open source just, it has so many benefits that have played out for the last three decades. I don't think that you should underestimate just how, how different, you know, the open core is from the freemium for that reason. People just love open source.
Auren Hoffman (10:35.646)
And so we had like the Red Hat model, as mentioned, turn of the century, which was support. Then, you know, cloud error comes in, you know, 15 years later, it's kind of this open core. We've seen other 10 years later. Then we see then we see like other types of models like Databricks is kind of like running like a compute kind of model. So it's a in some ways, it's just like it's like a cloud service on top of.
Spencer Kimball (10:47.488)
It might have been like more like 10 years later.
Auren Hoffman (11:05.826)
the system. Like how do you and how do you see these like? Yep.
Spencer Kimball (11:09.641)
That's the new innovation, right? And so it's kind of like open source ate the world of software because the time to value and the, I guess, value delivered for the cost assumed was dramatically better. I think that time to value is really the key metric. And how quickly can you stand something up using open source software? Like if you're doing it from scratch, the difference could be months, right? And that was...
I think the fundamental impetus for open source is popularity. It's just like, hey, we can just get started, we can move. They're not talking to sales reps, we're not having to do procurement. All of the, negotiate a contract and then we're paying all this money, we're not even using it yet. And you get, there's no community and you get like a set of printed manuals. That old world, that sucked, right? And so when open source came along, it was a slam dunk.
But now you're starting to see, okay, well, there's a faster time to value, even if you have to pay. And by the way, you always pay for open source, right? Because you have to run it on something. So there wasn't a cost that was associated with it. Yeah, absolutely. And that's a huge one, right? How much time does it take to learn how to run the thing? There's the day zero getting it set up, and then the real cost comes in day one plus, right? With services delivered on these cloud
Auren Hoffman (12:13.902)
Of course, yeah. Yeah, and your people's time of implementing it, yeah. Yep.
Spencer Kimball (12:38.309)
the degree to which complexity can be managed because someone else is running it for you and is responsible for all of the, you know, SRE type support necessary to keep something up, properly alerting, you know, has the right security posture. And that stuff is really expensive. You have to do that for every piece of open source infrastructure that you're gonna chain together to actually release a new product or service. So the time to value if you have a good business idea.
in today's world is substantially reduced with these cloud services. So how does open source fit into that? I think there's still this potential to have a huge amount of adoption and community and community contributions. If you can figure out how to tease out a part of your ultimate product, which is just generally useful.
often times in really big companies and things where they're not really ready to use managed services yet, things like that. They can still use a chunk of what you're building and do it themselves, build their own internal service or whatever. But what you're going to do is you're going to say, okay, all those big companies will eventually consume my service. Meanwhile, all the smaller companies and things that are born in the cloud, they definitely only want to use the service. So I'm going to use the open source thing to create top of funnel opportunities and build a community and so forth. And I'm going to...
spend the bulk of what we're doing beyond that in building a managed service where we are the true experts at running the managed service for this now popular open source product. There's a lot of this stuff in AI. I think that's kind of like the hottest area now.
Auren Hoffman (14:14.346)
Yeah. When you think of like an AWS, on one end, they both supported and promoted a lot of open source projects. On the other side, some of the people in the open source projects say that, well, they've taken advantage of the project because they're monetizing it in some sort of way. Like, how do you think they've done? They've walked that line. And obviously, a lot of open source projects change their terms.
to make it more difficult for Amazon to take advantage of it? And where do you think that's going over time?
Spencer Kimball (14:49.853)
Well, it may answer the last part of that question first. I think that it may become increasingly irrelevant.
because what Amazon exploited for a relatively brief period because people, as you say, started changing their licenses and so forth was that original open core model. And I'll talk about what that exploitation is because I don't think there's anything wrong. If you have an open source product, anyone should be able to monetize it and however they want. That's the part of the beauty of open source, right? So I didn't think that Amazon was doing anything wrong from that sort of strict, how do you use open source perspective? They were totally within the license rights.
and you can't really blame them, they're very good at that. And they also from their, you know, but I guess maybe their underlying ethos for their whole company is obsess about the customer, which I think is fair. And they always did that, you know, how do we provide the lowest cost services? Well, you know, using open source software is a great way to do it because we're not having to pay some vendor like Oracle to put their, you know, put Oracle and RDS, right? They do that as well, but it's much cheaper for them to put Postgres or MySQL and RDS.
So what I think the word exploit can be properly introduced into their behavior is around this idea of stewardship, right? AWS is a gigantic concern, commercial concern, and they have many, many products, and they do a good job of integrating those into their platform and providing a lot of value to their customers. The problem is that building these open source
projects to the point where they're really good and they continue to get better and they have all these long tail capabilities and surface area that's necessary to encompass or accommodate more and more use cases at higher and higher mission criticality. You get the idea, right? In order to do that, you have to keep building and building and you need more and more engineers. And these open core companies like Confluent, for example, with Kafka.
Spencer Kimball (16:58.053)
in Elastic with Elasticsearch, Mongo with MongoDB. They were taking an open source core that got a lot of good adoption, and then they started building their business in a way where they said, okay, we're using the open source for top of funnel, and then we get these customers, and then we're using the revenue they're paying us in order to make the product better, and that helps the core, and it helps us get more customers and allows us to stay in business and keep making this product better.
Auren Hoffman (17:16.812)
Spencer Kimball (17:27.665)
trillion dollar company, and they said, okay, we're just going to ride on your stewardship. And we're going to actually start pushing you out of the market as you struggle to survive and make this product better, and we're going to keep benefiting from your efforts. And we're not going to give anything back. And they kind of gave some stuff back, but that was not their central mission at all. And so that felt like, okay, well, they just came and dropped a bomb on this open core model which was working for a number of companies. And so then...
people realize, okay, we have to change licenses to make this, to make it so that, you know, no matter what you believe about open source and how ideological you are about it, like everyone tried to find their own workable solutions to support the benefits and the longevity and the, just the, I guess, the practical existence of open source in these advanced projects. They just keep getting better year after year because there's enough, you know, engineering attention devoted to them.
balancing that with trying to stay in business so you can keep doing that without having someone come in that is Really just going to be a free rider on that process. So BSL was one that's what coppers labs did MongoDB went to this you know expansive version of what the Uh, the what was it? I can't remember what their license is called, but it was a you know, a pure open source thing that actually
Auren Hoffman (18:34.091)
Spencer Kimball (18:51.573)
required that if you were going to offer something as service, you had to open source all of the things that were necessary to build the service, which is what everyone holds as closed source in this world where we're trying to build cloud services for everything. So there was kind of a diaspora from the open core model into these different kind of offshoots of versions of that.
that had different kinds of protection. And also everyone at the same time was saying, hey, we have to build this as a cloud service. And so naturally, there's so much intellectual property being put into competently running a cloud service and doing things like supporting multiple tenants in a price performant architecture, like virtualization, all these things, huge amounts of IP, but you don't really have to open source that stuff with the same mandate.
Because people really want that core product that you had, but they don't necessarily want to run their own database as a service or their own Kafka message queue as a service. So nobody's expecting you to open source those things. So these companies almost uniformly concurrently move to this world where they're trying to change their business model to be cloud services providers. And all of that new IP in order to suit that new kind of approach.
product, which is a better time to value all these things that we just discussed, that didn't need to be open source. Nobody expected it. So we have a ton of closed source being written out. Nothing wrong with that. But that is a very interesting observation that open source is no longer eating the world the way that it was before.
Auren Hoffman (20:29.622)
Well, and if you think of in some ways, maybe not exactly a database project, or depending on how you think about it, but if you think of like a snowflake or something, like they were not open source, but incredibly successful company with a great product. A lots of companies love their product is there. And they went a very different route than they went on kind of like an old school traditional route. But at the time that they were doing it was almost non-traditional.
because at that time it was very, very open source heavy.
Spencer Kimball (21:00.817)
Yeah, I mean, Google has Spanner, Cloud Spanner, AWS has DynamoDB and, you know, other things, Aurora has got a large non-open source component. So, yeah, Snowflake is absolutely, it's even further, for example, than just building a managed service around an open source product, like open source doesn't exist there. I mean, they may open source things and contribute to open source products and things. Everyone still uses open source. I mean, it's, I'm not saying that it's dying.
But there is a very healthy amount of the reintroduction of closed source ecosystem.
Auren Hoffman (21:35.262)
One of the things, there are different types of open source projects and there's kind of like the open source project like Cockroach, which is really now a company. And it lives past the couple of original contributors. Like if they get hit by a bus, it still works and it's a company, right? And so it's something you can rely on. But there's also a ton of open source projects where it's there.
there might be some people contributing to it to like pay the salaries of the people working on it, but it's very dependent on one or two people. There was one particular service that I was using for quite a while. Great, great little service. And at some point the engineer who is, who is the main person doing it just lost interest and decided to do something completely different. And then it just kind of died. So there's something about like having a company there at some point.
where you know you can rely on it, you know it will survive past the whim of like a one particular individual.
Spencer Kimball (22:42.289)
Yeah, I mean, there's.
I think open source nowadays is, the way I see it is it's getting pushed further and further into things that are harder to make money on. So it's kind of a natural aspect, right? The open core thing didn't quite work out. So that had to evolve. Meanwhile, cloud was ascendant. And so you naturally wanted to build these services where open source absolutely thrives and every company in the world uses it almost exclusively is all of those libraries and like dev stuff, like dev.
Auren Hoffman (23:13.633)
Spencer Kimball (23:16.457)
productivity and things. And some of that also, it's obviously a very complex landscape out there, but you do see that there are these frameworks and toolkits, certainly front-end stuff, and it's still the perfect environment in which open source flourishes and really adds tremendous value. But areas where...
Auren Hoffman (23:37.002)
But even there, you've got like New JS, but you also have Versel.
Spencer Kimball (23:40.729)
Yes. Yeah. So, you know, there's, there's exceptions to every rule and maybe they're not even really rules, but there's a sort of interesting observations. It does seem the world's in tremendous flux, right? So it did seem like just yesterday that, that we were, you know, struggling with the idea that we couldn't just build an open core company and oh, we need to not just have pure Apache, we need to have this BSL thing, we need to have an enterprise license. Oh, boy.
The concerns that we have in 2023 are, you know, it feels like several generations past that. So, you know, it's just, I do sometimes, listen, I do a lot of angel investments and many of them just because of my background, they come my way because they have some open source component. And so I get asked a lot, you know, how do we go to market with this? What should we be worried about and so forth? And, you know, I'll be honest, in a lot of these cases,
I caution them about what they, to really look deeply at what they want to accomplish, what they think their end state's gonna look like. Because it is a tremendous distraction to try to run a successful open source project and build a community, and also build
that same capability as a managed service. And it's like, I think a lot of people just want to believe that the open source thing kind of happens for free and that's just a separate part. And they're just going to bundle that up into the managed service. But all kinds of like difficult dilemmas present themselves if you succeed, you know, both in starting to build your cloud service, but also having a, oh, the people really seem to be interested in this open source project. And that comes in the form of these, you know,
Wolf and sheep's clothing, right? Which are these big companies that love your open source product. Meanwhile, you're trying to build your business with your cloud product. They do not want your cloud product. They absolutely want your open source project and they're gonna pay you $2 million a year. And for you, that's like, oh my God, we're trying to get 200,000 in our cloud products for these small companies. This big company is offering this and we can get a bunch more of these big companies. Should we build an enterprise sales force? But we have to do the self-hosted. We've got to support it in their environment.
Spencer Kimball (26:06.565)
And then if you do take that plunge, you start to realize that like, okay, there's a whole bunch of things. One, they immediately become the tail wagging the dog. They think they're the dog, in some ways they are. Exactly, so you think that too, you can't help but conclude that. And so they're asking you for all these long tail features that nobody is asking for in your cloud product. And then you start to realize that to get more of these, you need a whole different marketing message.
Auren Hoffman (26:15.574)
Right, because they represent 90% of your revenue or something, right? So then you have to do it. Yeah.
Spencer Kimball (26:33.197)
To talk to enterprise companies is vastly different from talking to the general market for your cloud service. And you're, you know, meanwhile, you're, you're trying to convince yourself that these two things, you know, converge that the enterprise will be using your cloud product and there's truth to that. It's, and so it's a seductive path. But.
Auren Hoffman (26:51.514)
You mentioned it as an investor, like there was this point in like 2019 to 2021, somewhere in that point where, you know, it was really how many stars do you have on GitHub? And based on that was your valuation and you're able to raise really, really fast. That seems to have subsided quite a bit. Like where do you think that market is and kind of startups in open source?
Spencer Kimball (27:18.217)
That's a good question. I still think that VCs pay attention to the number of stars you get and how quickly you get them. It's a great signal. Like is there just general interest in the, I mean, who stars a GitHub repo, right? It's like, it's your early adopter.
Auren Hoffman (27:33.738)
Yeah. I mean, it can be manufactured, of course, like all these things, like any type of rating. You know, you can buy 200 stars for a few hundred bucks if you want to.
Spencer Kimball (27:40.423)
It's a GitHub.
Spencer Kimball (27:45.673)
Oh man, that's depressing. I'm not surprised to hear it, but I think that a savvy VC, like the ones you really do want to raise money from ideally, probably would, yeah, I think so. I mean, they'd look at the stars and then they'd talk to some of your users and if there's some kind of strange mismatch, which if you've been buying your stars, I imagine there would be, that is depressing. But unsurprising too.
Auren Hoffman (27:58.166)
They look at real users. Yep.
Auren Hoffman (28:07.178)
Auren Hoffman (28:14.018)
What if you think of like the database market, okay, we have like Oracle, which has been kind of the giant of the market, but their market share has declined pretty significantly in the past few years. Like, how do you just see that market evolve?
Spencer Kimball (28:31.677)
for databases specifically? Yeah, it's, well, first of all, you know, the cloud platforms have a tremendous advantage for things that are born in the cloud. And...
Auren Hoffman (28:33.428)
Auren Hoffman (28:46.27)
Yeah, because you have to move the data out of there. It's already there. And.
Spencer Kimball (28:49.969)
Yeah, it's all integrated into a platform and the billings integrated and, you know, just having a one-stop shop.
Auren Hoffman (28:52.011)
Auren Hoffman (28:55.306)
Yeah, you just, your billing just goes up a little bit and it's kind of simple.
Spencer Kimball (28:59.057)
Yeah. And they also offer like a million different choices. That's an exaggeration, but it's truly a vast array.
Auren Hoffman (29:01.803)
I don't know if it is an exaggeration, like all the different configs. It, it's so complicated. Like if you log in your editor or Azure console and you try to like click around, like you get even the best developer can gain, get lost really, really fast.
Spencer Kimball (29:20.069)
Yeah, I think that's part of why digital ocean became as popular as it did so quickly. You know, our sense of the database space, right, is that it's competitive, it's crowded, and yet there are incredible opportunities. And the reason is just that it's growing massively and every use case in the world needs an operational database underneath it. So, you know, they expect it to be.
Auren Hoffman (29:25.1)
Spencer Kimball (29:49.137)
you know, somewhere close to 200 billion a year by 2030. That's, I mean, I got that a little bit wrong, but it's growing at like double digit compound annual growth rates, right? That's pretty tremendous for a market that's already 80 billion. That obviously is a big pie, right? There's lots of different kinds of operational databases and that's the crucial insight, right? Where do you sit in that solution space? What are your differentiators?
Auren Hoffman (29:59.978)
Sounds about right. Yeah.
Auren Hoffman (30:05.471)
Spencer Kimball (30:18.149)
Is that a relatively well defended niche? Is that niche expanding? Is it ascended in terms of people's concerns for let's say the decade ahead? And then what does your customer look like? Who needs that capability? Who plays in that part of the solution space and how does that work for what they think of in terms of how they wanna buy databases and run their business, right? So companies are looking to make money, save money or eliminate risk. They're not gonna buy if those things aren't.
Hopefully in conjunction, but at least one or more. So.
When you look at something like Oracle, they own a vast plurality, I imagine, maybe a majority of legacy use cases that are in production. So 90% or something like that of the world's IT spend is in a relatively small number of enterprise companies, I mean, relatively small compared to the total universe of companies out there. So there's just this enormous enterprise spend
Frankly, a lot of it's going 20 billion or something into Oracle's flagship database. That is a very defensible revenue stream because it's hard to move off databases. So you can see it.
Auren Hoffman (31:35.574)
And we can see that because I mean, you know, some of these databases haven't changed that much in 20 years. If you think of, you know, there's still tons of people using MySQL. Like it's really hard to move from one to another. And once it gets going, like it's just too big of a burden, even if it's sometimes like better to move or cheaper to move or something like that.
Spencer Kimball (31:56.893)
I mean, that's the thing, like when you actually look at how many legacy use cases are running out there, if you, if you wanted to try to move them all at once, like you would take probably, I don't know exactly, but maybe 20 years with all the developers in the world working on it. I said, it's like, it doesn't matter whether people would have the appetite or have the money, like the, like literally the resources don't exist through it. Maybe AI that's far more advanced can just move everything automatically. I don't know. So I don't want to speak too much about some.
future that's hard to imagine at this point. However, these legacy use cases for the most part are gonna stay where they are. However, there's a lot of new things being built, which is why this market's growing so quickly. And there are legacy use cases with acute pain points, or there can be these company-wide mandates to become cloud first and things like that. So people are moving. And the question there is just like, okay, are we gonna move from...
Everyone needs to place themselves in that solution space and really understand what their niche demands and really service that niche. So when you look at Cockroach specifically, we look at this crowded space out there and we realize that we have a globally distributed SQL database, right? Which is very unusual. But it's also very unusual to have that need. The good news for us is that need is increasing, right?
It's not always global, obviously, you know, have something in every continent. But you often have multiple regions, like most businesses do, even if they have most of their architecture in a single region right now. They are worried about surviving a regional failure. They're worried even about surviving a cloud failure.
Auren Hoffman (33:35.05)
Yeah, I mean, syncing data is not, it's always been a super hard problem and it continues to be a very big concern for, for anyone using any type of data, right? If you're trying to just read and write, um, you want to make sure that the, these data sets are synced.
Spencer Kimball (33:46.898)
Spencer Kimball (33:53.145)
Exactly. These big, so what we've discovered in terms of our product differentiation is that it's very well differentiated. That's good news, right? We've built a defensible product category. Uh, and you know, there's still competitors, believe me, because it's a very valuable space. And that's the, that's the critical thing. The companies that do need cockroach. We got a little confused about this over our history as we built a cloud product and the big companies weren't ready for it, but the companies that really need our differentiators are big companies. They're often in financial services. They're often in.
big tech, right? They're in retail, they're in gaming. There's some other ones, those are the big ones that where we consistently land. And this is a good space to play in, right? Because just in financial services, the value of those mission critical use cases that actually demand cockroaches differentiators over the next 10 years is astronomical, just in that one vertical. So you start to realize that, man, we're playing in a big space.
And we are not the cheapest general purpose thing out there like Postgres. We look like Postgres, but Cockroach has new complexity. We kind of got confused. I'll double click on that in a little bit. When we actually built virtualization and we provided a serverless capacity, we were like, oh my god, we can actually give away a very complex, sophisticated database, Cockroach. We can make a free tier, because this virtualization makes it so inexpensive to host a bunch of new starts.
And we thought, wow, maybe every developer will want to use this because we would use it as developers, right? That's kind of our mindset. That's why we built the product in the first place. But most developers look at Cockroach like, okay, it's like Postgres, but I already use Postgres and I understand that there's a lot of complexity here. I don't really need what Cockroach is offering right now. It's good that it's Postgres compatible. Maybe I'll use it in the future, but do I want to, to pay the, the cost in terms of what I need to learn?
about building for a distributed system. And that's a fair realization on the part of each of these developers. And so what we've realized is that, OK, you can spend your time trying to talk to these developers, or you can spend your time talking to these enterprises. But boy, the things that you have to say to these two different audiences could not be more different. And you can't do both effectively. If you try to talk to everyone, you're talking to no one.
Spencer Kimball (36:17.169)
So it's very important when you have your defensible niche to really talk to the people that want to buy that product. And just be very clear about that and just laser focused on it. So what we're playing at is like an ascendant niche for a very valuable target, ideal customer profile. And there's all kinds of interesting aspects of trying to formulate these strategies.
I mean, one is just like, how do you compete with the cloud vendors? And again, when you have a narrow. Absolutely.
Auren Hoffman (36:53.182)
And by the way, everyone has to figure this out, right? Like every single software company, both has to figure out how do I compete with the cloud vendors and how do I interact? How do I partner? They're kind of both competitors and friends at the same time. And this is a dance. And sometimes even these cloud vendors have to figure out how they compete and interact with one another.
because it's not uncommon nowadays for a big enterprise to have multiple cloud vendors.
Spencer Kimball (37:26.885)
Exactly right. And you're touching on, you know, part of the realization of the right strategy when you have a narrow conception of who your ideal customer is, that actually can suggest you the path forward for your strategy for all these kinds of customers we're focused on. Think of them as very, very large enterprises with global customer bases, you know, incredible needs around data sovereignty, sometimes latency, sometimes scale, always, always IT resilience, right?
Auren Hoffman (37:47.158)
financial services. Yeah.
Spencer Kimball (37:56.657)
These companies are very interested in multi-cloud strategy for reasons of pricing leverage, for reasons of cloud portability in the case of a systemic risk from a cloud vendor that actually can be baked into regulations in certain industries like power, utilities, banking. And they sometimes actively want to run across clouds, which is really cool, right? When you start thinking like, okay, this is, when you understand your customer, you can understand how your particular product
either can be talked about in a way that's more resonant or expanded in terms of its capabilities to get better at serving these customers. And so, yeah, how we talk to a customer that's potentially like a Fortune 50 bank about cockroach is we're stressing the idea that, hey, you know what, you can run it anywhere. All these companies have hybrid realities, all of them through acquisitions or permissive policies have multiple cloud.
Auren Hoffman (38:31.531)
Yeah, your roadmap.
Spencer Kimball (38:54.817)
They have a footprint or a foot in multiple clouds, even if they have a center of gravity in one. And many of them are starting to think, how do we build things that float above all the cloud vendors? It's a very hard thing to do. But if you have enough foresight and a desire to not go from a frying pan, like think about the vendor concentration all these companies continue to have with databases and Oracle into the fire, which would be like, we're gonna put all of our eggs.
not just our database spend, but our entire IT spend into one cloud vendor. That is kind of a mind blowing vendor concentration risk. That people regretted everything that happened in decades previous with some of these big vendors they had. Now to even consider the idea that your billions of dollars of IT budget are going to go to one cloud vendor, it's preposterous. So.
Auren Hoffman (39:44.318)
A couple of just a couple of other questions. You're, you're, as you mentioned earlier, prolific angel investor. I kind of call you a dual threat CEO. You're a CEO running a company, but you're also, um, you know, a, uh, a super successful investor as well. Now I can understand that being a CEO makes you a significantly better investor because you know what trends are happening.
you're much more likely to get into the deal because a lot of people would want the CEO of CockroachDB in their deal, right? And so getting into the deal is probably easier. How does making an investor make you a better CEO?
Spencer Kimball (40:22.685)
That is a good question. Let me think about that. You know, part of it is mentoring. So people come to me on the, you know what, nothing, and I'm sure you've heard this aphorism before, but you know, you don't really understand something until you have to teach it. And so being able to verbalize what it is, like I'm doing in this conversation with you,
Auren Hoffman (40:32.47)
by teaching them it helps you be better or something or?
Spencer Kimball (40:52.401)
I've been asked these questions before, which helps. I'm not just thinking what's top of mind without any preparation here. And that same thing applies in the process of talking to people about, for example, how they should formulate their go-to-market approach with an open source product and a cloud product in the off and how they should think about a big customer coming to them.
that might be a significant derailment of their roadmap strategy. So it's super helpful to be able to furnish this advice to people. And so just having those interactions somewhat broadly, it does help develop the skills that I need just to operate because I'm constantly doing that internally as well, that kind of thing. It's not always the same topics by any means, but this is not something I was ever very good at.
Auren Hoffman (41:32.695)
Spencer Kimball (41:51.065)
I was, I love building since I was 12. I just programmed, I programmed all the time. I programmed nights and weekends and through all the things I was supposed to be doing, I was always working on some other product and I loved that. And I kind of had to give that up and learn other skills in the process of this job evolving the way it has over the last decade. And I've never been a manager. And you know, as a CEO, it's not quite the same thing as traditional people management.
Auren Hoffman (41:59.478)
Spencer Kimball (42:19.057)
Because the people who are my direct reports, they do their jobs way better than I ever could. I didn't know how to advise them on their careers, but there is...
Auren Hoffman (42:27.07)
Yeah, it's not like you know that much about being a CFO or something.
Spencer Kimball (42:31.454)
I mean, I'm learning all the time, which is another great aspect of this. But I think being an investor really does help stretch me in ways that allow me to accommodate the responsibilities of my job more professionally.
Auren Hoffman (42:32.65)
Auren Hoffman (42:44.618)
Now you, you're named after your great grandfather, who I believe was the 12th president of the Church of Latter-day Saints. Do you think that like, that just being that, having that same name has influenced you in any way?
Spencer Kimball (42:57.201)
Well, it's hard to say, you know, where, what's the chicken and what's the egg? What came first, right? You know, there's clearly something to be said for some kind of hereditary strains of behavior. And obviously anyone, I mean, I'll give you a little bit more family history.
Auren Hoffman (43:03.263)
Auren Hoffman (43:17.262)
Because you could have been named Robert Kimball instead of Spencer Kimball, right? Yeah.
Spencer Kimball (43:20.625)
That's right. A little bit more family history, if you go back great, I had Heber C. Kimball. If you've ever been to Park City in Utah, you've probably seen Heber Junction. He actually came with the original Mormons, Brigham Young and Joseph Smith, across the country and he actually was the most married man in U.S. history because Mormons were polygamists back in those days. So he had 43 wives and 67 children. So yeah, he was quite a character, Heber C.
Auren Hoffman (43:41.862)
Oh wow. Okay. Oh my gosh.
Spencer Kimball (43:48.613)
you know, ancestors not direct in the line. There's someone named Golden Kimball, who was like quite a preacher. But all of these people, if you think about what Mormons continue to do today, they proselytize successfully, I might add. They have a lot of belief. They have the capability of belief. And belief is in the hierarchy of success, the very most important, right? There's belief and then there's an intellectual capability.
Auren Hoffman (44:01.822)
Spencer Kimball (44:16.945)
that allows you to plan and conceive.
Auren Hoffman (44:18.562)
Sherry, what do you mean belief in the hierarchy of success is the most important? What do you mean by that?
Spencer Kimball (44:23.377)
Well, so belief sits on top, right? And then there's intellect and the ability to conceive of a solution and plan. And then there's the sweat and the effort, right, that goes into actually executing.
Auren Hoffman (44:36.098)
So it's like, I have to, like, if you think of a startup, I have to believe that I could build this product or something like that. I have to, I'm creating this, you know, supersonic rocket that goes to Mars. I have to believe in that, okay. Yeah.
Spencer Kimball (44:50.057)
Exactly. Think about Elon Musk. That's a good suggestion there, right? His belief that he can change the world is absolute. And so he changes the world. But if you don't have the belief, the intellect is not going to be steered productively. It's going to find dissolution and confusion and doubt. Right? And if you don't have the intellect to plan, you're clearly not gonna direct your efforts productively. You're just going to...
Auren Hoffman (45:03.135)
Auren Hoffman (45:14.422)
But part of that belief is really what you're saying. There has to be some sort of optimism there too, right? Coupled with it, because I find that so many super smart people have a very pessimistic bent. And the founder is that smart person who also, in the Venn diagram, is also optimistic, which is somewhat of a rare Venn diagram overlap.
Spencer Kimball (45:20.786)
Spencer Kimball (45:41.529)
Yeah, I think that's exactly right. The people that I've seen that reach great success, they have a, yeah, there's a combination of unbridled and optimism that simply can't be tamped down because of defeat. Like everyone runs into roadblocks. If you don't maintain that belief that there's a way forward, you can give up, right? You can pivot too quickly.
Auren Hoffman (46:07.5)
Spencer Kimball (46:10.029)
You got to be able to break through these roadblocks. And if you don't have the belief, you don't have the willpower. Like, do those two things go together? So I think that whatever I inherited and I got the name to boot, there is definitely, I have a great, I never get tired of talking about CockroachDB, for example, never, even though I do it way too much.
Auren Hoffman (46:35.057)
Interesting. And do you think like, is that something like, is that something that was somehow impressed upon you as a young kid? And is that something you're trying to impress upon your children? How do you think about that?
Spencer Kimball (46:57.009)
I don't, yeah, I mean, it was definitely.
Impressed is a little bit of a hard word. My parents, I think, were always very careful in terms of how they put any of their own hopes and dreams onto us. They really let us find our own way, which in hindsight was very impressive because it's always tempting as a parent to have a vision for your kid. And you're like, you're doing the wrong thing. You gotta do this other thing.
Auren Hoffman (47:03.828)
Auren Hoffman (47:21.198)
To steer them. Yeah.
Spencer Kimball (47:27.693)
I was doing a lot of the wrong things over the course of my life. My parents never said a word. You know, they helped me when I needed it. And they were remarkably hands-off, right? Yeah, I mean, I can too, honestly. I look back at that as a parent now, and I'm really impressed at how they managed that. But they made it very clear to me how much they respected my great-grandfather and that I had his name. So there was definitely some impressing.
Auren Hoffman (47:33.294)
I'm sure they said a few words here and there, right? OK, maybe I can learn from them.
Spencer Kimball (47:56.677)
there. And, you know, I never took to Mormonism, even though I went to church until I went to college. But I have a great deal of respect for my forebears because that was a... They built a religion that survives today and brings, I think, a lot of meaning to many people's
Spencer Kimball (48:27.867)
built the current state of Utah. That was a pretty unforgiving.
place to actually try to build a civilization effectively. So there's plenty of things that are fraught in there, given our current understanding of the past. But I have a great deal of respect for that kind of belief. And honestly, this is also a fairly unpopular thing to bring up these days. But there's something that is true for these.
Auren Hoffman (48:35.083)
Spencer Kimball (49:02.785)
ancestors of mine, which is they were always seeking God, right, which is, I think, something that we, to our great peril, are generally avoiding in this materialistic world of ours today.
Auren Hoffman (49:18.34)
What is a conspiracy theory that you believe?
Spencer Kimball (49:24.688)
Well then I wouldn't call it a conspiracy theory if I didn't believe it. Let's see.
Auren Hoffman (49:28.674)
Well, one that maybe other people might label it as such.
Spencer Kimball (49:31.369)
That's a really random question. Hmm.
Spencer Kimball (49:41.317)
I don't know if I have a good answer for you. Maybe I can ask you that, I'd be curious.
Auren Hoffman (49:46.037)
Well, I've got a ton of them. All right. Well, last question we ask all of our guests, what conventional wisdom or advice do you think is generally bad advice?
Spencer Kimball (50:11.377)
I've got a number of them around starting companies. I think typically people are encouraged to find co-founders.
Spencer Kimball (50:24.793)
And I'd rather, you know, I advise people never to start a company with somebody that they haven't worked with for years, through thick and thin. Much better just to hire somebody as your first employee and give them some more equity or whatever, but don't, don't get that, you know, don't wedge yourself to an unknown because doing a startup
Auren Hoffman (50:37.823)
Auren Hoffman (50:44.706)
So even if you know them, if you haven't worked with them for years, you shouldn't start a company with them. Yeah.
Spencer Kimball (50:50.141)
probably sound, but you know, I've also found that people's stripes don't change very quickly. So if someone was amazing five years ago, they're probably still amazing. But you know, to your point, life can intervene and people can have a lot of other concerns and it, you know, you might not get the same capability because they're distracted. So obviously recency would be something I'd consider there.
Auren Hoffman (51:16.878)
Okay, well this is great. Thank you Spencer Kimball for joining us World of Dazs. I really appreciate it, this has been really great.
Spencer Kimball (51:24.273)
My pleasure, Oren. It's always a pleasure to talk to you.
Auren Hoffman (51:28.874)
All right. Amazing. Thank you very much.
Spencer Kimball (51:31.505)
Spencer Kimball is the founder and CEO of Cockroach Labs, a $5 billion company that makes the CockroachDB database used by Netflix, Nubank, Shipt, and many other leading tech companies. Spencer was building and contributing to major open source projects before he graduated from college.
In this episode of World of DaaS, Spencer and Auren do a deep dive on open source companies: How they make money, what the ecosystem looks like in 2023, and competing with the giants like AWS and Oracle.
Auren and Spencer open with an in-depth history lesson on the software industry and open source’s role in it. Spencer explains how open source and open core companies evolved and how the rise of cloud giants like AWS has changed the industry. They also discuss how belief and optimism affects founders, Spencer’s “hierarchy of success,” and essential advice for first time founders.
Godard Abel, CEO of G2, talks with World of DaaS host Auren Hoffman. G2 is a software marketplace, valued at $1.1B that helps people make software decisions based on peer reviews. Like Yelp for Software.
Auren and Godard cover how G2 took on the very hard task of creating software categories, unintentionally built a hard-to-replicate, revenue-generating data product, and mastered the art of SEO. They also discuss why Venture Capital wasn't ready to invest in G2 at the time of its founding, the advantages of being a multi-time founder, how to identify and grow leaders, and more.
Amjad Masad is the founder and CEO of Replit, an online coding environment.
Amjad and Auren discuss what the rise of AI means for coding, developers, and tech companies. Amjad shares his thoughts on consumer vs creator computer culture, why development hasn’t moved to the cloud faster, and why everyone should be able to call themselves a programmer.
They also discuss Amjad’s experience coming from Jordan to start a company in the US. Amjad shares whether he still thinks Silicon Valley is the best place for startups. They also discuss unique considerations for CEOs that aren’t talked about often: free speech as an executive, assessing expert opinion, and how investing makes you a better CEO.