Turbopuffer's Simon Eskildsen on adding Thrive Capital, reaching trillions of objects, and building a concentrated cap table like a team

here in the studio with us. Can we go to the Can we go to Mr. Puff?

Right here.

Will you set up the the puffer fish?

There we go.

There he is. [laughter]

We're puffing.

Here we go.

Simon, good to see you. Welcome to the stream.

Uh good to see you guys, too. I'm happy to you give uh Puffy a prime location there. I'll I'll speak on Puffy's behalf tonight. Yes, please do. Please do.

Amazing. Uh anyway, uh have we told the story of where how you landed on the the the name for the company Puffer Fish Expand? I don't know. Tell it.

Say it again because it's a good it's a good story.

There's there's there's three stories. Do you want one, two, or three?

All of them.

But in time,

I want the truth and I can handle it. Give me the truth.

The truth is the least interesting one of the stories though. Um I was in

I've changed my opinion. I want the interesting one. Lie to me.

You [sighs]

It's It's because you know you the all the data is stored in object storage cheap and slow

and then we puff it up into RAM into the disk and so it's the puffer architecture. Okay.

That was uh that's sort of the the backwards facing one. The real story was just, oh, this is an emoji that doesn't have Yeah, this is it. [laughter]

This guy,

nobody nobody claimed Nobody claimed this. This

I I stumbled upon him outside of a fire station and there we were.

There we go. That's good. [laughter]

Anyways, uh it's great to have you. It's great to have you on the show. Uh we caught up at 5:30 a.m. this morning. You were sharing.

I didn't I didn't know that people in the LA got up that early.

We have to America operates on the same market hours. You got to be up early. A lot of a lot of people in in LA who are on something that ties to market hours at all are up super early. Not

Yeah. I I like to I like to I live on the West Coast, but I operate in Eastern

West Coast Mindset.

EST.

You like to be there on the floor of the NYSE?

We do. Spiritually, I like to be conscious, you know, while while the opening bell is rung.

Exactly.

Yeah, it was great. It was

We're going to see you out there soon or are you going to jump no funding pre IPO round, right?

I mean,

I like I like the NYC vibe, you know, if we're talking about the stock exchange, like I like the vibe. I think NASDAQ has like a tech thing going for it. I just, you know, as someone who grew up in the old world,

Yeah.

Um, the New York Stock Exchange has a has a certain appeal, but NASDAQ's also good. You know, I'll I'll let them fight it out.

Okay. Okay. [laughter] Um,

yeah. You can't you can't you can't be too obvious. If you're going into a negotiation, you can't make it too obvious who which side you prefer, right? You need two two horses in a race.

Exactly.

Uh, anyways, uh, break down the news. It's been uh I don't know a couple months since you were on.

Yeah, it's been a couple months uh when I was in LA and and and spent a bit of time with you guys. Um what's the news? We we got a new investor on board this morning. Uh so we announced that and doubled with Locky Groom, uh the legendary one and only.

Um but most of all, we're just we're just puffing up. We're trying to get everything to puff. We now have trillions and trillions of objects in Turbo Puffer that are all puffing. I was waiting. I was waiting. I was like, he's not going to say what how much he raised. He's not going to say the valuation. Will he give us any numbers?

This was this was the perfect way. This was just I when you told me you're that you were going to announce the raise and announce no information other than you just add [laughter] added one added one investor to the cap table. I thought that is just a perfect way to just keep keep mogging the capital markets that are so desperate. I know you have so many I know you have so many fans out there that uh that want to be a part of uh Turbopuffer but is a great opportunity to say you can become a part of Turppuffer just shut wind down your fund and start applying for jobs. Turbuffer careers. I would [snorts] I would I say I would say that uh you know there's a there's a brilliant mind at at Insight and he ended up just joining. Um and I think that was a smart move.

Uh and I'm trying to I'm trying to convince more of the of the capital allocators including the one who made the puffer fish that's on your table to just join the company. It's a great way to get on the cap table. Um but most of all I mean the way that I think about this is that they are part of the team and I think the exam for for for them is who's brought the most value and you know

you you actually keep a list right of of of value ad and this these are people these are people like you you've kept track of block

on your nice list.

Well no it's just it's just a tier list. It's like how much value have you added? I think every every person that's reached out to you invest or or you know you basically just put them in a spreadsheet and then you just whenever they do something for the company you just mark it down and then you rank them and that's how you decide who gets

you check it twice. It's a very amirtocratic process. But you know the way that I think about it is that I have to go and convince my leadership team of like who's the right person to add. And we like to compose the cap table the same way that we compose the team, right? It's it's few people, highly concentrated and people we want to build a long relationship with. Um and Thrive has shown us over the past two years the kinds of mountains that they can move for us. Um and so that's why we decided to to add them um and um and allow Locky to also double down. um who's also in the absolute S tier league. Um so that's that's really what happened. And there was it was very simple.

Let's give it up for the absolute S tier league. [applause]

People are always talking about tier one. They're not they're not they got to be thinking about the the S tier. Um what about on the customer side? I saw Ivan from Notion uh shared uh earlier.

We have Slack. We have a shared Slack channel with you, Simon, and the team. every message within minutes.

Pull it up.

Every feature request is always answered.

It feels like these guys work at Notion.

This is the level of care and support we want from vendors. Bullish of Turbo Puffer. That's from Ivan at Notion. Uh

it was it was yeah so heartwarming. The team just absolutely loved it. And I mean this is why that's what we want people to focus on. What do our customers say? What are they doing with Turppuffer? That's what we care about. Not not numbers like you know to me the fundraising round would be like you showing you my office lease. These are the things that we need to get things done with but it's not the important thing. The important things are tweets like the one from Ivan. That's what we want to be measured on. And I think I'm just old school in the way that you know my my entrepreneurial example when I was a kid was going to my my grandmother's store every second weekend and just sitting and playing on the floor. And every day when she came home, she's doing all her own accountant and saying, "This is how much money we made today and just talking about her employees and customers. Those are the things that matter and those are the things that matter to me and to us and to the culture that we have here." So the tweet from Ivan and many of our other customers today, that's that's what I want to be measured on.

So you're saying making making money and customer satisfaction are sort of underrated in and venture.

I I think they might be. [laughter] uh what what uh are are there any new interesting case studies that you've noticed this year? How are you kind of uh describing the actual like benefits, workflows, the implementations like the value that the companies that you've worked with, Anthropic, Cursor, Notion, etc. like uh how are you framing the position of the actual product value that they're that they're as they deploy it? Yeah, I think there's two things, two big things we're seeing. The first is that the first wave of adoption, stage one of adopting AI in in in SAS is essentially okay, let's connect these LLMs to a bunch of data and try to reason over it, right? The, you know, chat with a document, chat with a document plus. But I think what we're seeing from linear, from notion, from Atlassian, from from many of these customers is that they can bring more value if they can connect more data to AI. And so we're just seeing an explosion in data where they're they have to earn a return on the pricing of the search underneath them. And that's what we can provide with much much better economics than the alternatives just because of the way that it's been architected. So more data into context is what the frontier AI companies are doing. The second thing that we're seeing is that customers want to read reason over extremely large data sets. We're getting more and more demands of people who are essentially building their own Google, right? indexing tens or hundreds of billions of documents and wanting to search over them. That's very difficult to pull off. There's, you know, at Meta, at Google, everyone has built custom indexes, but you can pull this off today by dropping Turopuffer into a Slack channel and just puffing real hard. We can get you to 100 billion. And so, we've had to do a lot of R&D for this um to do um vector search over 100 billion documents at once um with very very low latency. um and it's taken an enormous amount of engineering to do that with as little compute as possible. Um and that's been a lot of what we focused on in the past few months. The second thing is that it's not just about searching vectors, right? Vectors are sort of you give a bunch of data to a model and spits out a number in a coordinate system and that's really useful. But what's also useful is just searching over text and we've gotten very good at that. And the state-of-the-art is a library called lucine. underpins elastic and open search and many of these solutions and we're now starting to see uh similar or better performance and one of the reasons we can do that is because LLMs which are often the ones that are querying write very long queries like humans write you know a couple words when we go on Google but LLMs can write very very long very precise queries

and those are fundamentally different with a different set of trade-offs uh to process and we can hyper optimize for some of those trade-offs which align really well with other trade-offs we made inside a turbo puffer to serve them at a pace that is is very difficult to do on some of the tech that's otherwise out there. you said that uh you uh can you tell me a little bit more about the the hardware constraints that I mean you're maybe not fully bumping up against but uh everyone is laser focused on uh GPUs GPUs GPUs maybe TPUs maybe uh tranium but uh talk to me about the memory market the CPU market um uh there's this article in the Wall Street Journal today microns blowout results are bad news for anyone buying a new phone or PC next year. Um, do you have any context on the shape of the AI buildout and where it's going to impact different hardware subsectors?

I think we have very intentionally tried to steer clear of of of GPUs because if you think there's a lot to get out of CPUs and they're easy to procure, they're in more data centers. U when we work at vendors that need GPUs, they often have very tight restrictions on what regions they can run in, which might not necessarily be the region that customers are in.

Um so we work, we were talking to one vendor today where we're trying to get better performance and they only have GPUs right now on the west coast and so we're waiting for them to get in the east coast, right? These things really really matter and we are starting to get to the point where we have to go to Google, we have to go to Amazon and we have to request the SKS ahead of time which is similar to when I was at Shopify and we would have to do the capacity planning a couple months uh before because the cloud is not infinite when you start to get to turbo scale to Shopify scale and you have to do this ahead of time but it is still infinitely better than back in the day when we had to request these things 6 months in advance and someone had to go and drill the holes in the walls at the m last minute to make things work. Um, but certainly there are also compute restraints on the on the CPU side. We can't always get the SKs that we want, the newest CPUs. So, we've also optimized to be able to run on many different generations and types of CPUs in different environments to just stay nimble.

Yeah.

How big is the team?

Team is now 22. Started the year at five. So, big growth.

Wow. Yeah.

Wow.

It's amazing.

Revenue. Revenue growing faster.

Love it.

Uh, what's the biggest fish you've ever caught? biggest fish I've ever caught was uh uh this this winter in uh in Florida where we were out fishing for barracudas. I was that's that's a big fish.

You got you you guys got one.

Talking of fishing, by the way, uh when we were renegotiating the contract, I really wanted to get you guys up to Canada to go for some ice fishing.

Someone in the chat, we got to do a Canada meet up. We got to make it happen.

Can we address this? Can we address this? So So can we just negotiate this on air here? Like I will pay probably 10% more.

Okay.

To go

if you come up

and do a live ice fishing on the lake. Yeah.

Ice fishing show.

We'll do a fire. We'll do a fire.

What are you going to pay in?

Uh

shares.

What are we going to pay in

shares? Canadian.

Oh, you want turbo puffer sharers?

Yeah. Yeah. Yeah.

As much as I love as much as much as I love ice fishing. Uh, no, we we um we're we're working out the travel uh schedule next year and we would love there's there's a lot of events.

Chad loves a live debate.

There's a lot of events.

Invited and the the thing too about the puffer jacket, right, that we're working on um is that I've described this as this is the um this is the LA guys idea of a Canadian compatible jacket. [laughter]

So, I can't wait to see you guys out there on the in this jacket. Yeah, we're fine.

Well, credit to me. I got the original the original version of the jacket that we both have.

Uh I did feel the fabric and I was like, "This is not fit for ice fishing. This is going to this is going to get ripped up immediately." So, we're coming back.

It needs more puff.

More puff. More more uh like rip stop, something like that.

Um but uh but yeah, watching watching you guys the whole team build this year has been amazing. the uh the the the the puffer jacket is merely one piece of clothing that is it's all about layering. This is what I learned when I was in Boston. I'm not although I am an LA guy.

Oh, it's so cold in Boston, John.

It is cold in Boston. It is cold in Boston. And you got to wear layers. So, you got to layer the TBPN rugby with the Turbo Puffer puffer jacket. Maybe throw another TBPN jacket on top. TBPN hat. Once you get the whole merch, every layer, then you'll be warm. Honestly, the thing we've been maybe more focused on is uh procuring an actual puffer fish for the studio and uh you can get a mud skipper for 20 bucks. That's a kind of uh fajaka.

So, this is also like,

you know, Jordy, in our chat, there's a lot of ideas and I need to see more action in the studio because we were talking about the Mabu puffer fish, right? That can grow to two and a half uh feet and needs about a thousand. the Maboo that needs about $1,000 of feed every month in shell [laughter] and and and and the reason it needs so much so much so many shells is apparently because it has really long teeth and if they get too long I think it kind of can end up waling itself. I'm not sure.

Wow. You need the shells to sort of like grind it up. Yeah,

grind it up.

Absolutely insane.

But we're working on a lot of stuff.

Yeah.

But uh

there's a lot of VCs that have huge fish tanks. I know one that has a million dollar fish tank, I believe. Yeah. You had to go That's your next Maybe that's the next round. Maybe that's the next round.

Yeah. Yeah. Find a VC. Pick the VC to do the next round based on their experience with fish tanks.

That's if you have if you're out there and you have a Maboo puffer fish and you're in the Bay Area, like I'm coming. Okay.

I want to see it.

You want to see it? Oh, this is the way to get

at feeding time

with with Simon.

You could do kind of a one for one. It's like if you for every dollar you you put into puffer fish conservation and restoration, you can put a dollar in to turbo puffer at a $10 billion valuation. It seems like a good trade. [snorts]

Yeah. 10 billion. I like that. We could do that.

Yeah. There you go. [laughter] So, there's a price. There is a price.

He's not saying no to VCs.

Hey, we're talking charity here. We're talking charity. It's about the charity angle.

That's great.

Uh anyways, uh you guys have had a massive year and and it's been it's been an honor to watch you guys cook and puff and can't wait to see what the team does in 2027.

Yeah.

Honor to puff with you guys, too.

It's been an honor.

It's been an honor to puff. Talk

goodbye. Goodbye, Puffy.

Goodbye. Always with us.

Take care.

In spirit.

Take care.

See you. Goodbye. Well, um let's go back to do you know uh how elite the Puck News readership is? They're so elite that [music] uh when you go to subscribe to Puck News and they ask you to select your income level, the options are less than 100,000, then 100,000 to 500,000,