Daytona hit $3M ARR in 95 days. CEO Ivan Burazin explains how (10 Minutes or Less)

INTRO (00:00)

Ali Rohde (00:00) Ivan Burazin is the co-founder and CEO of Daytona, which builds sandboxes — i.e. composable computers — for AI agents. Before Daytona, he built Codeanywhere, one of the first browser IDEs. The company most recently raised a $24M Series A led by FirstMark Capital and has been growing like crazy, reaching $1M ARR about 50 days after launch, and then $3M ARR 45 days later. Ivan is from Croatia and the company is split across Zagreb, Split, and San Francisco. Welcome, Ivan.

Ivan Burazin (00:30) Great to be here. Thanks for having me.

Q1 — Where are you based? (00:45)

Ali Rohde (00:45) So where are you actually based these days?

Ivan Burazin (00:45) In an airplane — that is the truth. Right now I'm in San Francisco, but back and forth between Croatia, San Francisco, and whatever events or conferences we have anywhere else in the world.

Q2 — How is the team split? (01:00)

Ali Rohde (01:00) And the team, how is that split up?

Ivan Burazin (01:00) Roughly it's about ten in Split, ten in Zagreb, and five in San Francisco.

Q3 — How does that shift over time? (01:08)

Ali Rohde (01:08) And how do you see that shifting over time?

Ivan Burazin (01:10) Hopefully we don't add more offices soon — ideally you want one office, but we can't do that. So we're at three right now and hopefully we can keep it at three as long as we can. But we already have multiple, so it's only logical that depending on geos and job descriptions, we'll have to add more.

Q4 — Local, event-based go-to-market (01:36)

Ali Rohde (01:36) It seems like your go-to-market strategy is very local, very event-based. Is that right?

Ivan Burazin (01:41) Our go-to-market is almost entirely in-real-life, in-person events. Almost entirely.

Ali Rohde (01:48) I think I saw zero salespeople. Is that still the case?

Ivan Burazin (01:52) That is still the case. It's mostly me. But we're trying to change that.

Ali Rohde (01:56) So you will hire salespeople.

Ivan Burazin (02:00) Yes, we are trying to do that at the moment.

Q5 — You used to organize conferences (02:02)

Ali Rohde (02:02) But still very local — and you actually used to organize conferences.

Ivan Burazin (02:08) Yeah, I had a conference called Shift, based in Croatia. We ended up doing it in multiple cities, and Dublin as well. It was a series of developer conferences that still exist — vendor-agnostic — but it got acquired by one of the competitors of Twilio. So that's where my experience organizing in-real-life events comes from.

Q6 — Stage fright and getting forced to MC (02:36)

Ali Rohde (02:36) I read that you used to have terrible stage fright, and then you were forced to host and MC all these conferences.

Ivan Burazin (02:42) Absolutely terrifying. I was very bad at it — I still get a little bit nervous beforehand. But at our first conference, the MC bailed a day before. So it was either I MC the whole thing, or it doesn't happen. And I basically had this switch in my head — like a different person did it. I'd done it once, so I could do it again, and again. I was forcefully made to lose my stage fright.

Q7 — Still doing conferences for Daytona (03:12)

Ali Rohde (03:12) Aversion therapy — just do it. And I think you're still doing conferences for Daytona as well.

Ivan Burazin (03:21) The idea was never to do conferences again. I said I'm never, ever doing it again. It's one of the most stressful jobs in the world — I'm half joking, but it's like firefighter, policeman… and then conference organizer. Everything happens at one point in time. There's no pause, no update, no patch. I remember one conference where, because of how many people we were hosting in that city, no one could procure that much food for lunch. So we had food made somewhere else and transported in — and the truck that was transporting it died. We're talking about three and a half thousand people with no food for lunch. You can't fix that. Luckily we had cocktails and drinks, so we just gave it to everyone to get them tipsy and happy. But you constantly have these things — a speaker doesn't arrive, or falls asleep — and you have to change the agenda. People get angry fast if things aren't as planned. You're constantly aligning the speakers, the audience, the sponsors, the vendors — and it all has to work together at a single point in time.

So I said I'd never do it again, and I didn't, for a year or two. Then one of the people who used to run the conferences with me came on board to do the in-real-life events — the meetups, hackathons, dinners, the small ones. And he said, "You know what, it's our unique value that we can do these conferences, we should do that." I said no, no, no. He said yes, yes, yes. And we ended up doing one at the Chase Center, in March of this year — our first Daytona conference.

Ali Rohde (05:21) How'd it go?

Ivan Burazin (05:22) Fantastic. Really, really well. About twelve hundred people were there, the venue was great, and everyone who pulled it off did a really good job. It was the least stressed I've ever been at any conference. Good team.

Q8 — No Fire Festival outcomes (05:37)

Ali Rohde (05:37) Amazing. No Fire Festival outcomes there.

Ivan Burazin (05:42) No — but I have to tell you, as someone who's organized conferences, Fire Festival was deeply hard for me to watch. If you watch the Fire Festival documentaries, up until about halfway in, everyone doing a conference or event is in the same boat. You're not purposefully lying, but you're trying to sell the vision. You have to sell the entertainment and the speakers on the idea that this is going to happen, and you have to sell the sponsors that the people are going to come — it's a chicken-and-egg thing you do at the same time. With Fire Festival, at some point he knew whether he was doing it or not. There's a difference.

Q9 — Theranos: fraud vs. building the plane as you fly it (06:28)

Ali Rohde (06:28) That kind of reminds me of Elizabeth Holmes and Theranos, and the debate around her — how much is straight-up fraud versus just being a founder and building the plane as you fly it.

Ivan Burazin (06:41) It's a very hard line to draw. At one point you're pretty clearly committing fraud, and at another point you're not — so where's the line, and where do you cross it and come back? That's why watching Fire Festival, I'd think, "Wow, I kind of feel for him." And then — no, I wouldn't have people fly to an island without a place to sleep. You can't do those things. I can understand where it comes from, and why some founders end up going in these directions: instead of acknowledging that you haven't succeeded at something, you keep pushing it, and you push it too far.

Q10 — Billboards before a product (07:28)

Ali Rohde (07:28) You, I think, have mastered where that line is. I read that you got a billboard in SF before you had a product.

Ivan Burazin (07:37) Yeah — we got eleven of them before we had a product.

Q11 — The pivot story (07:44)

Ali Rohde (07:44) Let's start digging into what it is. Let's start with the pivot story — what it started as.

Ivan Burazin (07:50) We started Daytona as an on-prem product for automating dev environments for human engineers. If you've worked at any of the large tech companies, it's trivial to set up a new dev environment — usually running remotely, not locally, and it's all automated and easy. If you work at any other company in the world, Fortune 500 or otherwise, you don't have that. It's a pain to get up and running. So we decided to bring that tooling into the enterprise — companies that aren't tech companies. We built it out, we had customers, we had revenue. But it was a grind. It was growing, but slowly. Enterprise customers wanted it, but procurement took a while.

While we were grinding on that, we had some inbound interest for running these environments for AI agents. This was very new — autumn of the year before last. People didn't understand yet that agents would need these computers or runtimes. We gave them our old tool; they said it wasn't what they needed. We kept talking to them and tried to figure out what it should be. And when we built this new version and gave it to them, they said, "This is the most amazing thing I've ever seen" — there were some swear words in there, a lot of excitement. That excitement made us decide this is the future. This is pull. I'd never felt pull like that, where everyone you show a demo to says, "I need this right now." Because of that, we decided to hard-pivot — we're not doing the old product anymore, we're doing the new thing.

Q12 — How did you get conviction to pivot? (09:45)

Ali Rohde (09:45) That's tough. How did you get to that point of conviction?

Ivan Burazin (09:49) A couple of things. The product we had, we had two competitors. They were doing okay, but none of them were breaking out. So when you talk to investors, your comps aren't that exciting and there are already comparables to you — maybe they already have the market and you don't have a play there.

The other thing: selling to enterprise as a startup is very hard, especially pre-AI. I remember being on a call with one of the biggest banks, and they asked, "How big are you guys?" I said fifteen. They said, "Five-zero?" I said, "No — one, five." That was the end. Nothing else made sense after that — "you're too small, you can't service us."

Also, even if we'd acquired all those companies, that's a great business, but you still have to think about your second and third act. Versus agents — once we got pilled that there will be vastly more agents than humans, the TAM is just vastly bigger. There were competitors, but no one owned the market, and the market was growing and could be vastly bigger. So this made more sense. And one thing we'd learned the hard way across multiple companies is that not focusing is the worst thing for us. It's very hard to let go of revenue and some success. So this time I told my co-founder: if we do it, we do it all the way — a pure cut. So we cut it off and sent any customer we had to one of our competitors. Here you go.

Q13 — Early adopters of Daytona 2.0 (12:04)

Ali Rohde (11:57) See you later. Who were those early adopters of Daytona 2.0?

Ivan Burazin (12:04) Early adopters were companies that were essentially Manus or ChatGPT variations of those. There's a lot of vibe-coding and general-purpose agents now — there were even more a year ago.

Ali Rohde (12:22) There were more a year ago?

Ivan Burazin (12:33) More that believed they could compete at a large scale. Now, the vibe-coders are basically Lovable and Replit — that's the market. Wix is pushing base44, Figma has theirs, but it's not their bread and butter. Those two have basically won the market. There could be a thousand more, but no one really considers them competitors or puts them on market maps. A year ago there were a bunch of companies that had the aspiration and the belief that they could actually compete — which I think is less true today than it was last year.

Q14 — Customers today (13:14)

Ali Rohde (13:14) Tell me about your customers today.

Ivan Burazin (13:18) Our customers range from the YC startup all the way up to Fortune 500 enterprises and some of the largest tech companies in the world. The nice logos I can't actually name — we have a bunch on our website, those we can say; the others we can't.

Ali Rohde (13:36) Give some names.

Ivan Burazin (13:47) What I can say for sure: among the top five most valuable companies, two are our customers. Then in the Fortune 500 we have a bunch of household names that aren't tech companies, using us internally — and then all the way down through Series D, C, B, A, seed, and inception. As for use cases, we have two major ones and one minor, newer one. The two major ones are background agents — an agent or service where a human interacts with the agent, and the agent calls our system. Think Harvey, Gamma, Lovable — not saying those are our customers — where a human interacts with an agent and the agent goes off and does something. The second is researchers, for reinforcement learning or evals, who use us instead of a Kubernetes cluster to run their CPU workloads.

The newer, interesting use case is that we've started running compute for CI vendors. We don't offer CI ourselves, but we offer compute to CI vendors. The reason: the number of coding agents that now create PRs means you have to run all that through CI, and you get a queue, so you can't get the throughput you want. Because of how we built our infrastructure for agents, it can be very spiky — if you need to jump from, say, ten thousand to fifty thousand concurrent CPUs, we can do that instantaneously, which is very hard even on the hyperscalers.

Q15 — The third category emerging (15:36)

Ali Rohde (15:36) That's really interesting, this third category emerging.

Ivan Burazin (15:40) We didn't expect it. If you look at consumption from a background agent, the usage patterns are very similar to human patterns — follow-the-sun. Peak usage during the day, least on weekends, down on holidays and summers, because those agents are kicked off by humans during work hours. If you look at RL and evals, it's basically flat at 100% of their quota until the load is done, then up again. It's a very different usage pattern.

Ali Rohde (16:22) Is it useful for you, then, for them not to be in parallel?

Ivan Burazin (16:27) For us it's interesting — the more different patterns we have, the better it is for utilization. CI is between the two: bursts up to a hundred percent, but shorter, and during follow-the-sun times, because PRs mostly happen during work days. RL is just flat — some of our biggest customers run all day long, it just runs and runs until it tapers off and they spin up another one. The more different shapes of utilization we have, the higher utilization we can get from the compute.

Q16 — Another use case on the horizon? (17:16)

Ali Rohde (17:16) Is there another use case you see on the horizon, or suspect will come?

Ivan Burazin (17:22) I think there will be different products we build on top of our infrastructure. One thing I learned at our conference: everyone serving infrastructure for AI agents has similar problems — these new workload patterns are very spiky. To handle spikiness you need a lot of idle compute. There are two ways: either you load up on a lot of compute so you get instantaneous speed, or you offset the spikiness to a cloud provider and spin up VMs or bare metal on the fly. But even then, spinning up bare metal takes two minutes, provisioning takes another minute or two — there's a delay, it's not instantaneous. We can spin up fifty thousand in about seventy-five seconds. So we're betting on the way where we're a cloud provider — we'll handle the empty compute and figure out the utilization patterns to fill it up over time.

Q17 — Becoming a neo-cloud (18:43)

Ali Rohde (18:43) That's interesting. So you're becoming a neo-cloud of your own?

Ivan Burazin (18:46) Exactly. We have a bunch of machines that are ours and a bunch that we rent from other people. The way we structured our software stack lets us run on any bare-metal machine — CPU, GPU, doesn't matter where it is — and taper that together into one cohesive interface. It's similar to a Baseten or Fireworks, but they have a different product — we don't do inference, nor do we plan to. In the sense that we sit on top of multiple clouds and our own, and present it as a single interface or experience for the customer, that is what we do.

Q18 — Why not do inference? (19:24)

Ali Rohde (19:24) Why not do inference?

Ivan Burazin (19:26) There are just a lot of people doing it. Whenever people see revenue in something, everyone wants a piece — that's true for everything from vibe-coding to coding agents to CLI agents to inference to GPUs to sandboxes. There are now like a hundred sandbox providers. The pattern I see is that the people who start at the beginning — not necessarily first, but early — end up at the top and get the vast majority of the mindshare and revenue. So being the two-hundredth company doing inference, I don't think there's much alpha for us. But I do think there are new products and services that need to be built for AI agents that haven't been built yet, and that's the direction we want.

Q19 — Competitors and Daytona's alpha (23:39)

Ali Rohde (23:39) Who do you see as your biggest competitors?

Ivan Burazin (23:46) There are obviously a lot of companies doing sandboxes — I think there are like a hundred of them today. But the number-one competitor is Kubernetes. We see this from startups, but more from large companies and enterprise. When a team or engineer says, "I need a machine for my agent," the most logical thing is to spin up a Kubernetes cluster and give it a pod. But Kubernetes was made for something different.

What we've seen in the last thirteen or fourteen months is that people usually come back and say, "My infrastructure is now hindering my feature set and product growth." We've heard this multiple times, and a lot of people have come over to us, or I assume to other competitors as well, because of that. The sandbox is still an evolving term, because the feature set agents need is evolving, but we break it down into three things: performance, the isolation layer — the container, VM, sandbox itself — and tooling. On performance, that's everything from millisecond spin-up times to concurrency, like fifty thousand in seventy-five seconds. On tooling, it's guardrails for the agent: firewall, secure secrets manager, all the things an agent needs to work really well. And then the isolation layer itself: is it just CPUs? Can you have a GPU? Can I have Windows? Can it be Linux, Android, can it resize on the fly? There are so many things that are not available in Kubernetes out of the box.

Q20 — Modal specifically (25:30)

Ali Rohde (25:30) From a startup perspective, I'm curious what other providers come up and what you see as the difference. My guess is Modal is one people mention a lot — they're taking up a lot of energy in the room, and I think they just raised at a $4.5B valuation.

Ivan Burazin (26:04) They've been great — one of the competitors we keep running into. They do a lot of things; as far as the reporting goes, sandboxes are about a third of their revenue. I'd say they're probably the most competent — branding, product, style, stability, all very good. It's good to have a strong competitor, so I respect them deeply.

What we started with originally, and still believe today, is that Daytona sandboxes are fully stateful and long-running — there's no timeout, it will never be preempted, never shut off. Very much like a human's computer: it doesn't turn off unless the battery dies or you pull the plug. That's very unique to us. Our native support for Docker, and Docker-in-Docker, and Kubernetes or K3s inside these, is also quite unique. When we built Daytona, we always thought of an agent as a user — but one that has to run everything a human runs on their computer. That made us decide specific things in the architecture that let us serve some customers very uniquely.

Q21 — Attention on competitors vs. heads-down (27:19)

Ali Rohde (27:19) More philosophically — how do you think about how much attention to pay to competitors versus being heads-down on your own product?

Ivan Burazin (27:29) You always keep an eye on what they're doing. But I used to assume that when a competitor did something, they knew what they were doing — they're smart, they understand. I no longer think that. Now it's: that might be interesting, that might be useful, or it might not. What we try to do — and it's how we built the entire product — is just talk to our users. One of the co-founders is always on the sales or customer calls. That lets us figure out what customers actually need. If we then see competitors building the same thing, that makes sense. But we've seen competitors build things no one ever asked for — not once — and it turned out to be correct, because I never heard of anyone using that feature afterward. So watch what they're doing, especially if it logically makes sense in the market — but treat it as additional signal to verify what to build, not as a directive to follow suit.

Q22 — Primitives for agents you want others to build (28:39)

Ali Rohde (28:39) What other primitives for agents do you want other people — maybe people listening to this — to build around you?

Ivan Burazin (28:48) Whatever your agent has problems with. I think agents should be able to do everything we do, end to end — and that's not true today. If you give an agent a task — say, "set up this podcast, turn on Riverside, invite Ivan, send the script" — I'd argue you still can't. So what does the agent need? Tools, access, communication channels — whatever it needs to get that done end to end. Whatever's missing in that stream is what needs to be built. Some of it is app-layer — maybe Riverside has to expose something so an agent can do that. But maybe it's the environment the agent runs in that needs something pre-installed. What we believe is that every single agent will need a computer to get its job done, so we — or a competitor — should be at the center of that. But we won't build everything. Anyone building should ask why they can't give their agent a task to do automatically, figure out what's missing, and build that thing.

Q23 — "If you're taking a break these holidays, you're not gonna make it" (30:13)

Ali Rohde (30:13) That's the what — I want to transition into the how. I'm going to quote you: "If you're taking a break these holidays, you're not gonna make it."

Ivan Burazin (30:25) The reason I tweeted that — which exploded and surprised me, and kicked off my Twitter career — is that last summer, no one took a break. No VCs, no startups, no one. Everyone was working all the time. Then I heard from multiple people, "I worked so hard this year, I'm taking a break, I'm going skiing." That's totally legit. But when I tweet these things, I'm talking to myself — and to founders or investors, people for whom it can materially make sense to be working at that time. If we're at the hype of this supercycle, every moment you're not working, you're seeding it to someone else, because someone else is going to work. This wasn't "you should take a break on Sundays or weekends" — it's that this is a point in time where you should take advantage of it. There were even VCs from outside the US getting term sheets signed during the Christmas holidays, and people said that's not fair — but it's a global market, everyone takes advantage of what they can.

Ali Rohde (32:11) I love that someone responded, "Bro, I've never heard of your company," and you responded to that.

Ivan Burazin (32:15) Yeah — I said that's exactly why I can't take a break, because I had to keep working. And I haven't taken a break — maybe two days total since. It's not a flex, it's just what we have to do right now. I do look a bit tired, I could've used some sleep, but it's very hard to stop working when everything is moving so fast. You feel like you're losing if you're not.

Q24 — Hours of sleep (32:47)

Ali Rohde (32:47) How many hours a night do you sleep?

Ivan Burazin (32:49) Depends on the flight. Last night was about five and a half. That's bad — I need sleep.

Q25 — How do you recover? (32:58)

Ali Rohde (32:58) You seem like a pretty intense guy. How do you recover, keep yourself going?

Ivan Burazin (33:08) I don't do much for recovery — I probably need to. It's basically just go, go, go. I do IVs quite a bit — about once a week. They're a lifesaver, especially when you fly a lot, because you dehydrate. Getting a liter of water into your blood helps a lot. People ask why I don't just drink water — you'd have to drink the equivalent of eight liters, which isn't possible. Even the plain one is just water and salt, nothing crazy, and it helps a lot. And the adrenaline is great — the hype of the journey.

Q26 — Does this let up soon? (33:56)

Ali Rohde (33:56) Does this let up any time soon?

Ivan Burazin (34:01) I don't know the answer to that.

Ali Rohde (34:08) I agree this is the time — but it's going to be the time for a while, I suspect.

Ivan Burazin (34:16) We hope so, but we don't know.

Q27 — Codex or Claude Code? (34:21)

Ali Rohde (34:21) Codex or Claude Code? How open are you to changing — do you watch what's best, or are you Claude Code for life?

Ivan Burazin (34:35) No, I change it. Even ChatGPT versus Claude — I liked ChatGPT for a while, then went to Claude, and vice versa. Not too religious about it.

Q28 — What founder or company inspires you? (34:45)

Ali Rohde (34:45) What founder or company inspires you?

Ivan Burazin (34:49) Different people for different things. Amazon is an amazing company — a lot of things we emulate there. Stripe is an amazing company — a lot of things we emulate there too.

Q29 — What you emulate from Amazon and Stripe (35:03)

Ali Rohde (35:03) What do you emulate from Amazon and Stripe?

Ivan Burazin (35:06) From Amazon, everything they did at a technical level — being able to build AWS, and getting such position and power and ubiquitous usage. From Stripe, the product is about as good as an infrastructure product can get. Everything is meticulously well done and thought through, and they've done it consecutively for a long time. There's a lot of craftsmanship in that company and its products — even though it's hard to think of technology as a craft.

Q30 — Is the future of SaaS headless? (35:53)

Ali Rohde (35:53) Is the future of SaaS headless?

Ivan Burazin (35:55) Yes.

Ali Rohde (35:57) Everything?

Ivan Burazin (35:58) Everything.

Ali Rohde (36:00) Are you even building a UI, then?

Ivan Burazin (36:03) We do now. You need a UI because people have muscle memory — humans don't want the UI to change every time. But if you're going to succeed and exist at scale, it has to be headless. That's my firm belief.

Q31 — Headless for all apps? (36:28)

Ali Rohde (36:28) For everything — all the apps people use?

Ivan Burazin (36:24) Everything, yes.

Ali Rohde (36:28) This is interesting — I'm not so sure. I was talking to a friend at a DevTools company. They put out their MCP and CLI, and initially he was worried, because they charge for seats for access to their premium tools, including dashboards. But he says no one's using the CLI.

Ivan Burazin (36:48) Because those agents aren't autonomously doing things yet — the vast majority aren't. I actually think that before that happens, there will be a long period — a year, two years — where agents will use computer-use to operate the seats of humans, because they can't access everything headless; most things don't exist headless yet. So first we'll have agents that use computers, sandboxes, logging into everything by hand. Eventually it migrates to headless, because computer-use is slow and costs more — the screenshots, the slower path. So long term, yes, headless; mid-term, it's more seats first, then headless.

Q32 — How many times do you check Twitter a day? (37:47)

Ali Rohde (37:47) How many times do you check Twitter a day?

Ivan Burazin (37:50) A lot. I don't know — a lot.

Q33 — Founder failure mode you avoid (37:56)

Ali Rohde (37:56) What's the founder failure mode you're most careful to avoid?

Ivan Burazin (38:01) This is a good question. I try to avoid the self-doubt days — they happen quite a bit. You have days that are shitty, where you think you're going to fail, it's going to die, everything's worthless. And then two days later it's fantastic. Every founder goes through this. I try to base myself, survive the day, keep computing, and not get into that downward spiral. That's probably the thing I work on the most.

Q34 — What do you do in those down moments? (38:35)

Ali Rohde (38:35) When you feel those down emotions, what do you do?

Ivan Burazin (38:38) Nothing — just push through, go through the motions. If you have to wake up, wake up. If you have to do work, do work. Brush your teeth, do what you have to do, go to sleep, wake up tomorrow.

Ali Rohde (38:53) So you just keep pushing through, knowing it'll get better.

Ivan Burazin (38:56) Yeah — the motions are key, because if you stop, that's the problem. It's like in the army, where you have to make your bed in the morning. I've never been in the army, but you get one thing done and you feel better that day. So you go through the motions: wake up, do cardio, take a shower. It's usually less velocity, less energy, but you get it done. Then the next thing, and the next — it compounds, and you don't feel that bad. I've gone through years of shitty stuff like that, and you build that muscle to continue through it.

Q35 — Routines (39:43)

Ali Rohde (39:43) It sounds like routines have been very supportive for you.

Ivan Burazin (39:46) Yeah, routine is probably the most supportive part. I'd agree.

Q36 — Croatian vs. US engineers (39:52)

Ali Rohde (39:52) What's the difference between Croatian engineers and US engineers?

Ivan Burazin (40:01) Imagination. Really — imagination. In Eastern Europe, people generally aren't very good at sales and marketing. They're very good engineers — technically, mathematically — but in the US everyone's good at sales, marketing, talking, speaking, and that trickles down to the engineering. So with Eastern European engineers, it's hard to sell them that you're going to change the world or build something enormous, because they haven't seen it done, and they weren't taught to have that imagination. It's "keep your head down and do your work" — historically that's what was taught. So when you say "we're going to do this thing, and the entire world is going to use it," they say, "No, this won't happen, we don't need this scale." It's funny — some of our engineers will look at our compute bills, not the ones we charge, and say, "This is an insane amount of money." Sure, but here's the revenue side — you can't grow the revenue without growing this. So I'd say: imagination.

Q37 — How Croatia shaped you as a founder (41:19)

Ali Rohde (41:19) How has growing up in Croatia shaped you as a founder?

Ivan Burazin (41:25) Here's the caveat — I was born and raised in the suburbs of Toronto until I was fourteen, then thirty years in Croatia. So I'm a mix of North American and Eastern European Croatian. Someone told me a long time ago, half-joking, half-true: I'm nice like a Canadian, but resourceful like an Eastern European. Those are the two things I do pretty well.

Ali Rohde (41:56) Best of both worlds.

Ivan Burazin (41:58) I guess we try. Not all the best, but yeah.

CLOSE (42:01)

Ali Rohde (42:01) Ivan, thank you so much for joining.

Ivan Burazin (42:04) Thank you so much for having me, Ali.