Postgres FM - Postgres startup ecosystem

Episode Date: July 12, 2024

Nikolay and Michael discuss the Postgres startup ecosystem — some recent closures, some recent fundraising announcements, and their thoughts on where things are going and what they'd like to see. Here are some links to things they mentioned:Prediction from Dax Raad https://x.com/thdxr/status/1808972166752580039OtterTune shut down https://x.com/andy_pavlo/status/1801687420330770841Snaplet shutting down https://www.snaplet.dev/post/snaplet-is-shutting-downbit.io shut down https://blog.bit.io/whats-next-for-bit-io-joining-databricks-ace9a40bce0d?gi=8ef885454eefTimescale acquired PopSQL https://www.timescale.com/blog/best-postgresql-gui-popsql-joins-timescaleAiven https://aiven.ioHasura https://hasura.ioSupabase https://supabase.comNeon https://neon.techTembo https://tembo.ioFerretDB https://www.ferretdb.comHydra https://www.hydra.sopgEdge https://www.pgedge.comTembo raised $14m https://techcrunch.com/2024/07/08/database-startup-tembo-lands-new-cash-to-expandRy Walker’s Cybertruck with STARTUP license plate https://x.com/rywalker/status/1810061804380557516 Supabase acquired OrioleDB https://supabase.com/blog/supabase-acquires-orioleMichael Stonebraker Turing Award Lecture https://www.youtube.com/watch?v=BbGeKi6T6QIMicrosoft acquired Citus https://blogs.microsoft.com/blog/2019/01/24/microsoft-acquires-citus-data-re-affirming-its-commitment-to-open-source-and-accelerating-azure-postgresql-performance-and-scaleCrunchy Bridge https://www.crunchydata.com/products/crunchy-bridgePeerDB https://www.peerdb.ioParadeDB https://www.paradedb.compganalyze https://pganalyze.comDBeaver https://dbeaver.ioPostico / Egger Apps https://eggerapps.at/postico2Postgres Compare https://www.postgrescompare.comCoroot https://coroot.comokmeter https://okmeter.ioSlides from Nikolay’s talk on monitoring https://bit.ly/pg-monitoring Nile https://www.thenile.devUbicloud https://www.ubicloud.com~~~What did you like or not like? What should we discuss next time? Let us know via a YouTube comment, on social media, or by commenting on our Google doc!~~~Postgres FM is produced by:Michael Christofides, founder of pgMustardNikolay Samokhvalov, founder of Postgres.aiWith special thanks to:Jessie Draws for the elephant artwork 

Hello, hello, this is PostgresFM episode number 105. My name is Nikolai, founder of PostgresAI, and as usual, my co-host is Michael, founder of PgMaster. Hi, Michael. Hello, Nikolai. So you chose an interesting topic. Is it interesting? I'm kidding. But it's not technical at all, right? It depends how far we get into it. Yeah.

So we talk about startups, Postgres startups, ecosystem maybe for Postgres companies and new companies, young companies. We saw many of them launched in previous years. And recently, starting last year, we started observing some of them closing, which is natural for startups. Yeah, let's discuss this. Yeah, sounds great. I thought it might be interesting for people. I think not everybody

pays as close attention to this as yourself and I do. So I think there'll be people out there that won't have heard of a bunch of these companies. I think it'd be really interesting to discuss what some of them are doing, what kind of categories there are, how they run, like what kind of business models are popping up, what problems they're solving, that kind of thing. But the reason I brought this up now, I mean, I'm kind of surprised we haven't done this kind of thing already, but the reason I brought this up now is I saw a tweet, maybe some other, it got a few hundred likes. So I think probably a few people in the Postgres world have seen it already. But I saw, I think it was a semi-serious prediction, which was, oh, and it was from a guy called Dax Rad. I don't know if I'm pronouncing it that right.

He's a bit of a joker online, does quite a few funny tweets. But I think this was semi-serious, and it was a prediction that over the next 12 months, Postgres startups are going to get decimated. And that was the whole tweet, definitely in character for trolling. But I thought it was interesting and funny. And yeah, there were some funny responses too. But that prompted me to think, what is the Postgres startup ecosystem like right now? I think, first of all, it's serious, very serious. And second, I think it's wrong because it talks about next 12 months, not talking about what already happened last 12 months.

I think it already happened. Because if you think like, even if three startups closed, that means they do have 30, right? Because what does this word mean? I did look it up because that was part of the replies. But I think there's two definitions. One is the kind of the strict old school definition, which was from the Roman times, one in ten. Punishment. old-school definition which was you know the from the Roman times one in ten punishment diet dies yeah exactly but I don't think that's what's generally meant when the word is not dies other nine kill this guy this is how how it was it originally it was terrible right like well it was like a group punish one one one random guy is killed by others

right so this why i don't know why we discuss it that's not relevant maybe it's very relevant to the to the tweet right decimated but i think the more normal definition the more normal use of the word decimated means like a whole a lot of them reduction, right reduction of number of reduction. Well, actually, in the world of startups, usually nine of 10 die, not one of 10. Right. It's vice versa, naturally, because lack of product market fit or something like that execution, and so on, which usually looks like we cannot raise next round, right? But back to my idea, what happened in the last 12 months,

I know at least three startups which are dead now. Did we have 30 startups before that? If we did, I mean, did we have more? I think no, actually. Post-GIS related startups. It depends how you define them. And I think even some of the ones you're potentially talking about weren't Postgres only.

So it depends how wide you cast the net. The biggest, Ivan.com from Finland, I consider it a startup because I remember they were super small and then they skyrocketed and raised. They are Postgres-related, because they started, I think, with Postgres and Kafka. It's like alternative to AWS services, but multi-cloud, like single interface to multiple clouds,

but Postgres-including, right? And they had great Postgres materials, and so on. I think I misheard. What was the name of the company? How do you spell it? A-I-V-E-N. It must be a different Ivan to the one I'm thinking of because they seem to still be going strong. No, no, no, no, no. They didn't close. I mean, what I tried to say, I just reflected your words about not of them are not purely posgos and these these guys are not purely posgos of course they have my sequel and a lot of elastic and everything right yeah i wanted to say we have a big range of very big companies and ivan at least according

to crunchbase raised 420 million dollars it's it's insane right with a lot of money yeah i think they were first who reached like kind of unicorn label among postgres let's call it postgres related startups the other one i think timescale db also reached this good mark right like unicorn means valuation exceeding 1 billion by the way speaking of course 420 it's easy to remember sitting in California yeah but it's an impressive amount of money and we have

super based timescale which raised more than 100 million right Neon 2 over 100 million great congrats everyone Hazura as well by the way Neon 2 over 100 million. Great. Congrats, everyone. Hazura as well. Yeah, Hazura. By the way, I stopped hearing about Hazura for a couple of years already somehow.

And by the way, they are also starting from Prosgres. I think they expanded to SQL Server and everything. Yeah, we didn't see collapse of really big startups, but smaller. And I think I would say Autotune managed to be already mid-sized because how much they raised? 16 million? 15? I saw 12, but I don't know what they raised before that. I think 15 or 16 was the total I saw. Honestly, if we say database startups,

this money should be not serious, because database startups require more money, I think, to go serious. And we have a lot of small, very small things. So let's talk about those who closed. Like, what are two of the latest news? Who else? There's only one other that I know of, and that was Snaplet.

Snaplet, yeah. It's not really in the same category, but it's kind of in our space of making tools. Yeah. What about Bit.io? Bit.io, it was a nice idea. You have a spreadsheet, you just throw it to browser uploading, right?

And it immediately creates Postgres node with table and with data. Oh, really? Yeah, it's like super cloud approach, like very, very cloud. You just drop a spreadsheet and start running SQL queries to it. It was a nice idea, but they closed last year, I guess. And similarly to Autotune, they did very, very, very, very, very, very short notice for their customers, which is, I think, very bad behavior. Not good, not good. Like 30 days or so, get your data. Yes. It's difficult, isn't it? So these three startups, like Snaplet, Beta.io, and Autotune are closed.

And my question is, how many startups do we have post-grisly that we're discussing in a broader term in general? Maybe 30, right? But 30 is quite a lot in my head. And it means that maybe we already have the summation complete you know so easiest prediction to make is something that's already happened right well no i mean next 12 months maybe it will happen again what do you think i don't know by way, I somewhere had a list of post-Gastellini startups. I think it didn't exceed 30. Yeah, I've made a bit of a list, but I didn't know what should count. Like how many years does it still count as a startup?

Right. What is startup? of like do they count so it's i think it's really difficult and then especially if you start to include companies that do tools for more than one type of database like we had an acquisition recently in the postgres space where timescale acquired a company called popsicle a cool name by the way but they did tools that were or they did a tool that worked with Postgres, but not only Postgres. And there's been loads of companies like that that I wouldn't have thought to include if I was making this list. So it's a very, very tricky definition. But to be fair, DAX did actually qualify a little bit in some replies, saying it was about the crop of companies that have popped up to often managed Postgres services, which does limit it. It's very different.

There can't be that many of those, maybe five or 10. Right. And yeah, we cannot take existing companies which started doing this because they already cannot be considered startups. Right. So it means we talk about only like very few companies i think not more than 10 because most managed postgres services i can think of they are done by companies which exist many years so superbase neon tambo who else yeah um for sure I don't know how some of the others are monetizing.

For example, there are people like FerretDB offering Mongo compatible Postgres. I don't think they're offering a managed service, though. Is it pgedge as well? Are they doing managed? Should be. I don't follow. I think so. PeerDB as well.

I don't know if they're doing managed. I don't know if they're doing managed. I don't know if you'd count it. They don't have managed Postgres. They have managed replication for Postgres. There you go. I don't think they count indexes prediction. So, yeah, it's a tricky one. I suspect it did its job, and it was pure engagement farming and i've fallen for fallen for

it but in in the grand scheme of things i was most keen just to discuss how do you see postcode startups at the moment and are there any gaps that you'd really like to see maybe new startups pop up or what kind of problems some of these are solving which ones are most important to you, like that kind of thing. Well, I think I see things very differently than a few years ago. I think a lot of new people came, which is great, from different market segments, like even from database,

from different database systems, for example, right? Like Tembo is one of great examples here. And by the way, congrats to them. They just announced the raise. 14 million or something. And special congrats for Cybertruck to Ray. We'll link that up.

With license plate startup. Very relevant to today's discussion, right? True. I think we have very different types of companies and many attempts. And it's interesting time right now because if we talk about VC, maybe it's harder to raise, but at the same time, we know it's harder to find a job, which also means that it's easier to find people.

Yeah. And for those who make startups, it's some additional window of opportunity. So it's a mixed situation right now. You might fail to raise, but you might find good co-founder, for example, technical, right? And build something. Yeah, true. Yeah.

I'm honestly not a standard startup guy at all. I'm slow, bootstrapping. I'm just observing this VC. If you raise, for example, $10-20 million, build something which brings you some significant money, and then you need to shut down your company because you didn't have 10x growth per year. It's weird.

It's weird. For VCs, it's probably like a failure, right? Because they expect hyper growth. But why do we need the hyper growth in Postgres ecosystem? Maybe we should have good companies which have normal growth, like 30% per year. And that's it. I don't know. What do you think? Well, I'm in the same boat as you in terms of my own companies, trying to grow it sustainably with customer revenue, customer funded, bootstrapped at the beginning. So trying to

do things slow and steady but to do that i i felt like i had to pick a very very small niche of a product a very very kind of low surface area compete in like a very very small area but i do think there are problems in databases that do require a lot more investment up front and those things are harder to do self-funded slowly i think they do require a lot of capital and then the only model at the moment for that is either well i guess i guess there are examples even in the postgres ecosystem of companies funding their own development they're like it well back to the hyperscalers but even so obviously companies like aws rds that was all funded through amazon profits and

microsoft as well google but in the postgres space crunchy data for example have been investing in their own platform that presumably largely funded through their own consulting business and then customer revenues but it's like yours right your your business is also mixed consulting revenue so some of these other companies i see are offering kind of service-based business models as well as the kind of more vcfriendly subscriptions or hosting, like usage fees. What is good about this all, it's definitely ecosystem is still growing, expanding, and we see the growth of companies which use Postgres in so many market segments everywhere. It's everywhere. It's great. We see many customers who need help in many aspects. So I agree we must divide companies by what they do, goals. If you want

to change something in database itself, like build new type of Postgres itself or rewrite part of Postgres. It requires a lot of effort and you must have very long vision if you raise like just one million, for example, and next year VC will say either raise or close basically because you don't have customers. This is not how it will work. So fundamental things require a lot of effort and maybe money also. So Neon is a good example. They change.

Yeah, great example. Yeah, and also AudioDB, which is still under development, and Alexander Korotkov joined superbase also good driver probably to continue development of audio db because what is there we as postgres users do need right features which are designed there also i would like to mention here historical things. If maybe you've watched this, Michael Stonebraker, who is father of Postgres, receiving Turing Award 10 years ago or so.

He had a great lecture about startups in database startups. Because he is not only father of Postgres, he's father of database startups as well, participating in doing this a lot. So it was great lecture, I highly recommend it. I remember he was comparing fundamental database startups as riding a bicycle across America, like downhill, uphill, like sometimes it feels feels great, sometimes it's almost killing you. Very good lecture, actually. I enjoyed it. So this is a fundamental thing. But there are other types of Postgres-related startups as well. And I think good things should happen when... We saw in Postgres area area we saw first acquisition by Microsoft, it was Citus, right? Yeah. Which is quite also fundamental I would say, right? It's not Postgres itself, but it's a significant

layer of addition to Postgres. It's actually the one area that I would like to see more deep work done. I think we've got a lot of bases covered in the Postgres ecosystem, but there is, there's this, this kind of like elephant in the room with PlanetScale and Vitesse. I feel like we had, we had great guests on recently and I've since seen somebody describe,

I think it was actually Arca from Notion. He described what Sammy and the Figma team are doing as somewhat starting to build the vtess for postgres and i thought that was a really like i i hadn't clicked that that's what they were doing but that is kind of what they're doing so that feels like a gap at the moment that could be that probably does require a lot of development and could be funded by a vc back startup but that's the one area that I don't see currently being done. Right. But then there are several acquisitions, I think, for smaller startups. And we saw also a few companies which are Postgres related in YC.

Yeah. Like PeerDB. Is that the first time, right? PeerDB and then ParadeDB. Are they the first Postgres-related companies to get into YC? Oh, Superbase was there as well. Oh, great shout. Yeah, good point. Sorry, Superbase. I forgot. Yeah, maybe someone else. But there are bootstrappers as well.

Is Pidgey Analyze bootstrapper? Yeah. I had a list of bootstrappers that I wanted to give a shout out to. Yeah, Pidgey Analyze is a great example, but also dBeaver. These bootstrappers don't get as much, because they don't raise money often, well, never, they don't get as much press. So yeah, the rest of the other ones I had to give a shout out to were Egger Apps, who make Postico, my favorite little Mac client for Postgres. Postgres Compare, which is run by an old friend and colleague of mine, a guy called Neil Anderson. And also there's another monitoring company.

So PG Analyze, of course, but with Lucas and the team. PG Analyze, I see it very often lately. Nice. Do you consider CoreRoot as a Postgres startup? It's very strange to classify, right? Because they have good support for Postgres for self-managed installations, I think, because they do eBPF monitoring, right? I like the idea. If we have many Postgres nodes and not only Postgres,

we don't need to install anything like additional extensions, but we do need to have access to like machine. Right, right. So it's a self managed situation. And they just observe whole traffic and perform a lot of interesting stuff, including like flame graphs. Historically, you choose the period and understand what postgres did in that period in terms of what cpu was doing and also in query analysis yeah query analysis i forgot them but i think it's because they're not purely they're not only postgres right so it's not only postgres but the founder was creating it's the same founder who created OKMeter and it had very good Postgres monitoring,

one of the best I consider still. And you know, last year I had the overview of Postgres monitoring systems at PGCon. And people liked a lot my slide deck, so I know what I'm talking about. And Peter Zaitsev joined him recently as co-founder. Yeah, very cool. So it means serious thoughts about

databases. Of course, maybe not Postgres only, but MySQL, but definitely serious about databases. So we can consider, like, this is the problem. Fundamental, not fundamental, bootstrapped, VC-backed, Postgres only, not postgres only.

So many dimensions for this classification. Well, and that's part of what I wanted to discuss here. It does feel quite rich at the moment. It does feel like we have a lot of options. And I feel like that's really positive. And there's quite a lot of innovation going on, not only at the startups. It feels like there's quite a lot of innovation going on, even at some of the larger, older companies working on Postgres things. Maybe I'm biased and roast into glasses, but it feels really positive to me at the moment.

I think the idea of that tweet was the company who raised a lot are in danger. Interesting. You think? Yeah. Because they need to raise further and if they are not dealing with AI, for example, VCs won't give

them money, right? Maybe. It's a very silly thought, but it's actually in many cases true. Well, it will be interesting to see if any of them do but it also feels like the the ones that i'm aware of they have such good business models like if you if if the business model is working getting companies to host their database on your system the ones that do well will grow with you and pay you more each month,

month on month. And it's very difficult to move off, right? It's expensive, it's risky. So the fundamentals of- Imagine you grow only 50% per year or just half per year. For VCs, it's not enough at some point. Good point. It's tough. But if we're talking about 100 million in the bank, even if you hire a lot of people, that's a serious amount of money. Right, but how many managed Postgres services do we have already?

And RDS is still the king of the hill, right? So I don't know and is it still expanding the user base in terms of managed postgres good question i did look if we're going back to that going back to the tweet like of what if that's the prediction why is it the prediction i personally think the number of databases is growing really quickly the number of startups seems to be exploding at the moment because of the AI boom. A lot of the AI use cases do use vectorized, like they do need a vector database. Postgres is, because of pgVector, positioned really well for vector databases. Do you think so?

Personally, yes. Do you not think so? I have doubts because it's not in core. If you check Postgres documentation, you don't see vector at all. This is a huge chance to lose this battle. Yeah. In my opinion. This is serious. It might look so silly, simple, and so on, but it's a super serious problem Postgres currently has.

And PG Vector and all efforts. It's good that all major managed services quickly added PG Vector, but they usually lag in terms of version. It means that they lag to deliver the best performance, for example, which is quickly improving

in latest versions. And so on and so on. Of course, this is quickly improving in latest versions, and so on and so on. Of course, this is a battle that's happening, but it's a barrier in minds, and some folks will not pass this barrier. I mean, if you're not lazy, you will quickly realize it's kind of a zoo. It's a bazaar. It's open source.

That's why it's a bazaar. So we have many options. You need to quickly understand the options. But I recently had this customer. They said, okay, we are on RDS. I think it was Aurora. And we have PgVector, blah, blah, blah. And I see they use very old version of PgVector,

much older than even already available on Aurora. But they didn't upgrade because to upgrade you need to switch to different... I don't remember some problems were happening. This whole thing is so fragmented. Vector's story is very

fragmented right now. And the only path to make it strong is to put it inside Postgres. But we we know we discussed that there are problems with that well i don't think that's i disagree personally i think if you're really a startup and you can't keep pg vector up to date you have bigger problems than um you know that that that feels to me such an easy thing for you to do if what i'm saying is true if there are lots of companies that want vector databases right now and are willing to go with postgres

i personally think it being out of core is a benefit right now because it means it can get bumps inversion really quickly like we see i know they're not technically major versions but 0.5 0.6 0.7 have all come with quite big functionality improvements and if it was in core we wouldn't have seen those it would have taken years so personally i think there are advantages in that area i've just remembered though i actually think the prediction was made partly because of two other reasons i think it's partly that planet scale is very uh marketing level kind of attractive to kind of medium-sized like non-ai kind of sas businesses i don't know how many non-ai businesses are popping up right now but the ones that are i think that they offer a very attractive proposition

like kind of like mongo did before but with sql and then i think the other driving force for the prediction is probably the number of sqlite or sqlite or sqlite or however you pronounce it there are a lot of hosted sqlite companies popping up saying actually all you need is this like for a long long time you can get by with sq, keep things simple. So I, I think at least that, that part of Twitter that I'm seeing is talking about those topics a bit more, and that might be why they think the Postgres startups might struggle rather than a more nuanced take on vector databases. But I might have totally misunderstood. What do you think? Hard to say. I don't know i just remembered

there's one i completely forgot have you heard of nile n-i-l-e yeah because they popped up quite like relatively recently in the grand scheme of things talking about sass like running a regular sass on postgres and some innovations in that area. So that could be interesting, but they're another one that we're offering a managed service again. Well, yeah, talking about this lack of being in core, of course, I agree with you about the pace of development. It's great, and maybe things will change in the future. But it's just so sad that everyone has everything in documentation, like Tecmunga, Elastic, everyone, like any major database system. They talk about vectors.

Postgres documentation doesn't talk about vectors at all, right? I mean, it talks about arrays, but it's not different, right? So maybe at some point it will go to core. I hope it won't be late, right? But just to push you a little bit, the super-based documentation talks about vectors, the neon documentation... Of course. RDS, Cloud SQL, all of them. Yeah. But people are still then going with Postgres. So what's the problem? The problem is that you need to spend time to choose

and understand all the options here because options are quite different everywhere. But that would be true even if it was in the Postgres docs too. In fact, one more option. Maybe, maybe. Anything you would like to see anything like if anybody's listening wanting to do a startup or like get inspired is there any kind of area you'd like to see more innovation in of course and by the way timber is doing good steps in this direction so i I think, you know, 37signals idea about moving out of cloud. I thought you were going to say, I thought you were going to say moving. So they've also done

this once product that runs on SQLite, like, but no, yeah, the moving out of the cloud one. Yes. Yeah. Moving out of cloud and having greater level of control over your database. Direct access to files, for example, and so on, which is not possible for managed services usually. We discussed it a few times. Good thing that, for example, CrunchyBridge provides a full-fledged super user.

So I expect actually Cytos co-founders, they launched new startup as well, right? And they do Postgres as well. Yeah, good point. UbiCloud, right? That was it, yeah. Forgot them, yeah.

Good point. The idea is great. You can install it anywhere you have full access but it's still, it's like the simplicity of management is similar to you have in cloud. You can do it with Supabase as well, by the way, I think. You can host yourself. It's great. So in this area, I expect good shifts in the future because I think demand will be growing people who are like uh already say like we we've been in cloud enough we want to reduce uh overhead how much we spent and so on of course some people will say cloud is great i understand this but i think this party

will grow as well like an anti-cloud party, let's say. I think it might also be a little bit like stage of company. I can imagine a lot of startups going with the cloud initially and then like 307 Signals, perhaps a bit earlier than 307 Signals did, maybe then moves workloads back off. So I could definitely see both being true. Yes.

So for example, new and fully betting, everything should be cloud. I understand that. But I also understand like them, we pay like five times more than they could. So why do we do this?

Like, let's not... I know small companies which also understand that they don't want to pay a lot and they are good technically, so they are successful

and happy to be not in cloud or in very low-cost cloud, for example. And Postgres management there is interesting because you want all the features but don't want all the overhead. This is interesting. In general, I don't know what will happen,

but it's normal for startups to die, especially if they are VC-backed. Yeah. Because VCs put a lot of pressure, especially after like three to five years of existence of the company. I learned it hard way with my second startup, like 10 years ago, when I realized like three years passed and we need to move on and huge optimization was needed, it was successfully done, but it was very painful. So it's good if you know this in advance. Yeah, true.

That there is some period of time when you will need to raise again or you will just be closed. In my case, it was not closure. It was just a huge optimization to be very profitable at some point. That's why I like the bootstrapped cases as well. Yeah. And I'm glad we got a chance to shout some of them out because they don't always get the attention they deserve. If bootstrapped company doesn't die in first year

and the founders still believe into this idea, it's much more stable company actually because they don't have pressure from vcs but probably they have much less chances to grow significantly right well it depends right like i actually forgot a couple of similar companies that are bootstrapped or mostly bootstrapped or at least they've stopped raising and they're doing really well like at jet brains really really old company going from strength to strength king of bootstrapped companies i think there are some examples of companies that have scaled to the same levels of as some bootstrap as some vc backed companies without needing to. So kudos to them as well. Right.

So my conclusion is I expect more closures, and it's interesting to observe the situation. And unfortunately, I think closures will happen with companies who raised a lot, or they will need to change things a lot. Interesting. Well, I would be surprised, so call me naive, but my prediction would be that none of them do close

at least in the next 12 months. So let's see. I'll be rooting for them anyway. In fact, all the Postgres startups, I guess, will be rooting for. The ecosystem is quite interesting with challenges, but it's still growing, I guess, will be rooting for. The ecosystem is quite interesting and with challenges, but it's still growing, I think. And I think it's interesting to observe the further growth.

Yeah. Should be fun anyway. Thanks so much, Nikolai. Anything else you wanted to add? No, thank you. See you next time. See you next time. Bye.