Oxide Computer Company's Bryan Cantrill on building rack-scale hardware for on-prem cloud — and why the downside is always zero

Speaker 2: Goodbye. See you. Up next, we have Bryan Cantrill from Oxide, the co founder and CTO. Let's get this gong warmed up. It's time. Nice

Speaker 1: little warm up fit. You're gonna You're

Speaker 2: gonna How much did you raise?

Speaker 1: Let's get another. Oh, man.

Speaker 2: How much did you raise?

Speaker 1: Love it. How much did you raise? We gotta know.

Speaker 6: Okay. Well, so... Hey, guys. I don't know. Is that a is that a gong about how much we raise or about the fact that we're profitable? Is that a different

Speaker 1: Gong for that? Wow. How... Oh, you did the warm up. The mad lads did it.

Speaker 2: We did the They

Speaker 1: did it.

Speaker 2: Fidelity head. Said Now we gotta hit it for how much they raised.

Speaker 1: That's right.

Speaker 2: How much did you raise?

Speaker 6: We raised $4.45.

Speaker 1: And finally, the best hit of all Tell me. For the secondary. No. I'm kidding. No. No. No. No. We're not gonna go there. Tomorrow, man. We're not gonna go there. We're having too much fun. Amazing to have you on. Thank you for matching or exceeding our energy.

Speaker 2: Good day.

Speaker 1: Introduce yourself and the company. What's going on? Oh, man. It's good to be... First of I just...

Speaker 6: I love the Gong. I gotta tell you, the the Gong... So my co founder and I had... There was a Gong at... We were at a cloud computing pioneer joint. And...

Speaker 2: It's actually really loud that we played it there. We got

Speaker 1: the baby gong.... Baby gong.

Speaker 6: The the the founding CEO had this gong that was very awkward. And because, like, the problem with the gong is it, like, then the gong isn't rung very much, and, like, now the gong is just kinda, like, sitting there, kind of, like, foreboding.

Speaker 2: That's crazy. I've never had that.

Speaker 6: Gong ended up, a couple office moves later, ended up on Steve's desk. There's just kind of this inert gong there. I remember as we were contemplating oxide, I looked at the gong. I'm like, Steve, at some point, we are gonna have... We're gonna build this company. We're gonna have our first customer, and we are gonna ring a gong, and it's gonna feel so good. But that gong never made the move. So we don't

Speaker 1: actually have

Speaker 2: a gong on Oxide.

Speaker 1: So the first gong that we've had at Oxide is the gong you guys just write. Amazing. Well, I recommend... I I... We recommend a gong hit for every single dollar of revenue that comes in. So that basically means you need somebody dedicated to the to the craft, you know, and they're pretty much just tapping away all day once you're at, you know, a certain run rate. But, anyways, I still have no idea what you do, and I'm so curious. So... Yeah.

Speaker 6: So we are a computer company. So it's right on the tin. Oxide Computer Company. So we make rack scale machines, hardware and software co designed. And our big observation, Steve and I were together at a cloud computing pioneer joint. Prior to that, I spent fourteen years at Sun Microsystems.

Speaker 1: Do I have to experience There you go.

Speaker 6: Computer company.

Speaker 2: There you go.

Speaker 6: Exactly. The the the OG there. And we... At at Joint, we were building a cloud on top of commodity hardware, on top of Dell HBC Micro. Cool. And when Joint was bought by Samsung in 2016, we went out to this explosive new scale and everything broke. And the stuff that broke in our software, could fix, but there was so much that broke at that hardware software interface that was unresolvable. And I... There's some... In the number of times you heard from Dell, it's like, you're the only customer seeing this problem. And you're like, how the hell is that possible?

Speaker 1: Like That's strange.

Speaker 6: Everyone else is at much greater scale than we are. Then you realize like, oh, wait a minute. Google and Amazon and Facebook are not Dell customers. And so you Dell are kinda speaking your own truth at

Speaker 1: some level.

Speaker 6: Because once someone gets to this kind of scale, they actually can't be a Dell customer anymore. And so as we were beginning to contemplate what was next, we wanted to go... We realized that the hyperscalers had built for themselves their own machines, rack scale machines

Speaker 2: Interesting.

Speaker 6: With hardware, software co design. And the enterprise, the the the... Anyone else didn't have that. You could rent it in the public cloud, but you could not buy it and run your own data center. And our big belief, we are very pro cloud, but we believe that cloud should be available wherever you wanna run your business, and that includes on prem.

Speaker 1: Very cool.

Speaker 6: So the the the the big an iconoclastic belief that we had in 2019 was that cloud computing is going to endure and persist and thrive, is that's gonna happen on prem, and it necessitates this rack scale hardware software co design. And I... You know, I have to tell you how in 2019, was all SaaS all the time, and Yeah. It was very iconoclastic to... And... But part of the reason it is oxide computer company in the name is to make clear to any 2019 DCs that there would be no pivots to pure software, that we are

Speaker 1: No pivots.

Speaker 6: Just this company.

Speaker 1: No pivot. Completely. We just make computers, sir.

Speaker 6: Just make computers. And, you know, actually, I did just because you guys are scholars of history. Digital Equipment Corporation, back in the day, one of the, if not the first venture capital backed company, wanted to make computers straight

Speaker 1: out of the chute.

Speaker 6: And they were told by their VC, is one of the... George Dorio, basically a proto VC in Massachusetts. It's like, you can't call yourself a computer company because IBM is gonna be the only computer company. In 1957.

Speaker 1: It's called

Speaker 6: himself Digital Equipment Corporation, Deck. Yeah. Isn't that amazing? He thought

Speaker 1: there was gonna be a fast a fast takeoff in IBM. Sort of an IBM singularity.

Speaker 6: That's exactly right. Right. Totally. And as it as it turns out, we... And again, you know, many, many, many decades later in 2019, we believe strongly in hardware, software co design and actually delivering... Because that... That's the only way you can really deliver this Elastic infrastructure. But...

Speaker 1: Alright. Give us the big milestones in the business leading up into this point. You started the company in 2019.

Speaker 6: John's... Oh, Doug Doug

Speaker 2: Doug Doug

Speaker 6: Doug Doug's ready. Wow. Okay. Well, on... So honestly, the... I mean, the first big milestone was just raising our seed. So...

Speaker 1: Alright. What happened next? What happened next? Well,

Speaker 6: so we would raise... When we're raising the seed, we raised 20,000,000 out of the chute, which is a very big seed for the team. And the rationale for that, and we feel... As we were talking to VCs in 2019 and only finding these SaaS folks and having to explain that actually the most important products that we have represent hardware software co design. So just finding... And we feel very blessed to have found Eclipse and set that Eclipse having... Being big believers and immediate believers in hardware software co design. And they're... And they saw that, like, actually, the biggest companies ever built are are... So finding them, finding the right investors, raising the right amount of money, and we didn't wanna raise, like, three on 12. Right? Or or what... But that would have been easy, but it would have...

Speaker 1: Well, it would have been beneath you.

Speaker 6: It it would have been a recipe for death because you would... You can't do any of this with three.

Speaker 1: Yeah. Yeah. Yeah. And... Well, the funny thing is the funny thing is that the the $20,000,000 seed became, at least now, is very much standard and companies are now can be capital intensive enough to justify that. It's not just sort of like a, you know, supply and demand equation. But back in 2019, most of the companies that were getting started, like, didn't really need... You know, they they were three on 12 would have would have been more than enough money to, like, at least get to the next round.

Speaker 2: Yeah.

Speaker 6: That's right. And so we... And we knew we needed to raise that much to be able to get to get the team get the team hired. Get... I think, you know, I... In many ways, the most terrifying moment for me in the lifetime of the company was when we got our first term sheet and got that term sheet from Eclipse. Yeah. Because it was like, oh, shit. We bought... I gotta go build this thing, and we don't have the team. Like, I can't go... This is so technically hard. There's so much technical risk in this. One of the things that we heard that drove us bananas in 2019 is like, look, technically, we know you can do this. And I'm like, am I... You wanna stop a VC and be like, no. No. Sorry. Let me stop you right there. Technically, I don't know I can do this. Like, I'm sorry. Like, technically, this scares the shit out of me. There's so much technical risk. We did our own compute sled. We did our own switch. We we we eliminated the UEFI bias. So we did our first instruction on the x 86. We... We've done all of our own software, lowest levels of firmware, all the way up to control plane distributed system, distributed storage. I mean, we are nine startups in one. Unbounded technical. Crazy.

Speaker 1: By the way, Eclipse was actually... Was was absolutely cooking in in the twenty tens.

Speaker 6: Like They were...

Speaker 1: They they have so many bangers that like were just non obvious or like most people just were not that interested in that now look just like genius in hindsight.

Speaker 6: Well, and so they had the advantage of investing in this sector when everyone else was ignoring it. And so as a result, like if you were starting starting oxide in 2019, starting Cerebras a couple years prior

Speaker 2: Yep.

Speaker 6: Yeah. Like, you are so contrarian that you almost have to be good. You know what I mean? Like, those companies are good companies because they were almost tautologically contrarian. And so Eclipse had the luxury of... And what we found... We're we're kind of... We found that we're VC homewreckers because we would find that, like, a single partner... Yeah. Is that... Yeah. Is is that the homewrecker gong? We actually called them pelts on the wall because... We... Well, because... Okay. Here's what happened. You get a... What I learned very quickly is that oxide is very, very contrarian. And obviously, like, everybody loves contrarian.

Speaker 5: Yeah. Yeah.

Speaker 6: But no one actually wants zero. Right? And I think that the the outcomes for Oxide were very bifurcated. It felt like, wow. Generational company or zero? And, you know, VCs are kind of in this quantum superposition of fear and greed, and they would constantly be oscillating in their minds between like, oh my god. In fact, we had a VC being like, I

Speaker 1: love that this is n of one. Or this is n of one. It's like they were scared

Speaker 6: that we didn't have competition.

Speaker 2: Wait. Help me understand how this would be a zero. Because it feels like like even if there's no AI boom, it's like you can still like buy parts, assemble racks. Maybe there's not a ton of innovation, but you just become like a mid market reseller of equipment and like That's zero. Thin margins. Maybe... Oh, okay. Okay.

Speaker 6: So... No. No. That that is a zero because... First of that's not differentiable.

Speaker 2: I know it's not differentiated, so

Speaker 1: you don't

Speaker 6: go anywhere.

Speaker 2: Okay. Yeah. It just like... It just peters out or something?

Speaker 6: It it peters out because the...

Speaker 2: But but when I when I think big VC zero, I think like you you spent a $100,000,000 on like software engineering for a social network that never gets product market fit and it's actually like you ran out of money as opposed to like the downside risk with it with the sector that you're in feels like you become like a small, like, micro center almost like just a place that like assembles... Okay. No?

Speaker 6: No. No. No. No. The the... No. The downside risk is a zero because we're we're not reselling things out of the back of a catalog. We are actually engineering a system. And there are companies that tried to do what Oxide did. Mhmm. You haven't heard of them because they're all zeros.

Speaker 1: Mhmm.

Speaker 6: And some of them are like pretty prominent zeros. Sure. Sure. Sure. And we actually... One of the things that, you know, benefit of of starting a company, you know, later in life and and with a lot of wisdom, and Steve and I both started the company in our forties. Yeah. We were real scholars of how these companies have failed. Yeah. And wanted to understand how every single one of these companies have failed. Yeah. And, of course, like, any founder will tell you, well,

Speaker 1: I failed because I ran out of money. It's like, yeah. Yeah.

Speaker 6: Yeah. Yes. You ran out of money, but also, like, why else? And the... One of the things that we saw over and over again is it... If you have something that is kinda half built, it is very, very hard to get that funding... That that funding round to get to that that real customer, and getting that first customer is really, really hard. And then getting that first customer to to expand is even harder because then the... You need to land this thing, and that there was a company, Nebula, that is pretty similar to Oxide's vision.

Speaker 1: If you guys remember, these are these are... This is

Speaker 6: one of the open... This is during the great OpenStack boom

Speaker 3: Oh, yeah.

Speaker 6: Which you... Maybe you haven't heard of for a reason. Right? The... But so Nebula... And Nebula's vision looks like oxide's vision almost scarily so. And in... And I'm like, I gotta understand every single thing about this company. Best $7 I ever spent my life is on the Stanford Graduate School of Business did a case study on Nebula

Speaker 2: Keep up.

Speaker 6: As Nebula is dying. Woah. So Nebula is, like, gonna go bankrupt. By the way, Nebula shut down on April 1. So 04/01/2015, Nebula shuts down. And I was in the office with my CEO at the time, and he and are looking at this like goodbye cruel world kind of announcement. And we're like, is this like the darkest most fucked up April fool's joke ever? Or is this company literally so incompetent that they can't even go out of business on the right day? Like, go out of business on March 31 or April 2. April 1, I'm sorry, is...

Speaker 1: No. We we we announced we announced an... Our acquisition on April 2. We had the opportunity. I wanted it first,

Speaker 2: but it would have been so much more chaotic than it already was.

Speaker 6: You know what I mean? Exactly. They're excellent. Strong... Good job overlooking... Wait.

Speaker 1: But this company but this company... This company... They had raised a lot of money. It was, like, hot at one point. This is sort of, I think, like, predates my technology It was hot.

Speaker 6: Kleiner had backed up the truck. Kleiner had put a ton of money into it. And it... I was absolutely a zero. And if they... Wow. Now, the the the case study was extraordinary because they failed at everything. They failed at the tech. They they they failed in terms of manufacturing, in terms of supply chain. They failed in terms of software. They failed in terms of of their approach to market. They were trying to get to market too quickly with something... It was kind of that lean idea of like, well, let's just ship garbage. It's like, no. No. You can't ship garbage. You have to actually get something that fully works. And in particular, one of the go to market things that they screwed up was they had this idea that someone will buy one rack and they'll buy one rack and then like a month later, they'll buy 30. It's like, no. No. No. No. No. Have you ever been an enterprise buyer? Sorry. You're gonna buy one rack and you are gonna kick the shit out of that rack for like eighteen months. If you are if you are a bank, or you are a pharma company, or you you you are gonna spend a long time under... Because when you do that 30 rack buy, you are betting your career on

Speaker 1: it. 30

Speaker 6: racks? And the hell

Speaker 1: 30 racks?

Speaker 6: Bet your career...

Speaker 1: What about 30 racks?

Speaker 6: What's that?

Speaker 2: What about 30 racks?

Speaker 1: What about 30 racks? Sorry. I mean, I... We know we know a lot about 30 racks buys. I haven't desperate to buy in 30 racks. Not on this. So I got a little switched up there.

Speaker 2: We're just messing around. Sorry.

Speaker 1: We're just

Speaker 3: messing around.

Speaker 2: But thank you. This is really good history.

Speaker 1: Hope they don't watch this like the team, you know? Because you're like, not only did they have the perfect vision, they had the worst execution across every possible dimension of the business.

Speaker 6: That's not a state secret. It's in the Stanford Graduate School of Business case study.

Speaker 1: Mean, it's like... But it's behind the paywall. It's behind the paywall. Yeah. Yeah. You're breaking open the paywall.

Speaker 6: I'm breaking open the paywall. Yeah. So... But it was it was super helpful to learn from that because you realize that like in your go to market motion

Speaker 2: Yeah.

Speaker 6: You have to rely on the fact. You have to know that that that single rack is gonna be at that single customer. That customer could be a 100 rack customer, but they're not gonna be a 100 rack customer until they are running in production on a smaller deployment for an extended period of time. And that changes the way you gear the business from... It changes the way you raise, it changes the way you kinda set expectations, the way you build a product, the way you set expectations with your own employees, with your customers, because you are... What we are focused on is the... Are are those very large appointments. And what's exciting, and this where you can definitely ring the gong, is that's what we arrived at in the last eighteen months is we

Speaker 1: got those those very large accounts. Got those, like,

Speaker 6: you know, the... Okay. I I

Speaker 1: think it's like asshole. Yep.

Speaker 6: We're we're both pretty fiscally conservative, and, you know, these these companies that, like, celebrate employee number milestones, like, number of employees, I'm like, no. We are never doing that. Yeah. Number of employees that... And it shouldn't be a cause of celebration. Those are all, like... It it should... You should be celebrating revenue and so on.

Speaker 1: Profitability.

Speaker 6: Profitability. And like, can we celebrate when we have more racks shipped than than employees? I'm like, yes. That we can celebrate. And that was a big a big gong banger was when we shipped our... More racks than we have in place. So our our vision from the beginning was how do we run and if... We we have the belief that if you have an extraordinary team, you can run an efficient business. A totally iconoclastic belief along with the iconoclastic.

Speaker 1: Yeah. So we... You know, we're we're we're we're real renegades over here. Yeah. Really out there on the on the fringe. Amazing stuff.

Speaker 2: Yeah. Thanks so much for coming on.

Speaker 1: Great. Yeah. What a what a milestone. This was super You're welcome anytime. Yeah. This was this was a joy.

Speaker 2: Yeah. We'll talk to you soon.

Speaker 1: And congratulations to the team on... Yeah. How many how many racks are you trying to get? What's the new what's the new ratio that you're going for? Oh, looks like a 100 to one? Like, how how...

Speaker 6: I would love that. Exactly. I'm very, again, old school. It's like, what is your revenue per employee? And like, let's get revenue per employee in the many millions of dollars, please.

Speaker 2: That's...

Speaker 1: I love it.

Speaker 6: That's the objective and big healthy generational business. We're not not looking to not looking to flip, we're looking to to actually go build the generational business that Steve and I wanna work at for the rest of our careers, which apparently is also a conoclastic. Yep. I thought that was the point of starting companies is that we build generational ones, but guess not. I guess everyone else is geared for an exit. So... But we're...

Speaker 1: Yeah. Big talk big talk until you get offered a 100 times revenue and then...

Speaker 6: No. Dude, this is like... This is not like a negotiation tactic. Like, is like, honestly, hand on heart. Like, that's not what we're interested in because... Yeah.

Speaker 1: Know. Bigger number. Yeah. I mean, we we are wanting to kick our dent

Speaker 6: in the universe. So that's what we're doing here, docs.

Speaker 2: Love it. The gong. Dent the gong and then dent the universe.

Speaker 1: Great stuff, Bryan.

Speaker 2: Have a great day. We'll talk to you soon.

Speaker 1: Give our best to the team.

Speaker 2: Have a good one. Let me tell you about CrowdStrike. Your business is AI. Their business is securing it. CrowdStrike secures AI and stops breaches. And let me also tell you about Console. Console builds AI agents that automate 70% of IT, HR, and finance support, giving employees instant resolution for access requests and password resets. Our next guest is Daniel Kimminau from the Advanced Spade Company. Welcome to