AI Native Cloud Infrastructure Beyond the Hyperscalers

Welcome to another Techstrong TV interview. I want to introduce you to our guest for today's interview. His name's Kevin Cochrane.

Kevin's the CMO over at Vultr. Kevin, how are you? I'm doing great.

Great to speak with you today. Great to have you on. So we were talking before we started the...

I was going to say the camera's rolling, but I don't know if that even makes sense anymore. But I actually did an interview at Platform Con in New York with someone from Vultr, and I'll be honest with you, although I had heard of the name and I knew the company, I didn't really know the company. And the interview was enlightening to us.

So before we jump into anything else, I want to spend a little time on you and on Vultr, because if I didn't know the company well, people out there didn't know it as well either. So but first, let's start with you. You're CMO, as I mentioned.

Give us a little bit of your journey, how you wound up here. Yeah. So again, great to be here.

And I'm Kevin Cochrane. I'm chief marketing officer here at Vultr. And Vultr's a wonderful little story.

It's what I like to call the best kept secret in tech. It was a platform that was launched in 2014 and built by developers for developers to provide an alternative for traditional hyperscalers, to give them the best, highest performance cloud compute at the lowest cost possible, and to do it globally, and to do it with the highest levels of security and compliance. It's a platform that, without any sales and marketing, grew faster than any other company in history, without any outside capital or anyone in sales or marketing, crossing over 100 million in revenue, profitable since day one in the first three years.

We were also an early pioneer in AI infrastructure, and we had a very simple premise, which is that all enterprises would need to find an alternative platform for building and scaling what we call AI native applications. A new hyperscaler to support the global rollout of AI native applications globally. And so that's really been our mission here at Vultr, and it's actually why we were at Platform Con in New York and earlier, Platform Con in London, and we do all the Platform Con and CNCF events worldwide, is because we're looking to help enterprises as they rethink business processes, as they rethink new tools to power employee productivity, as they rethink new experiences to power their end customer experience.

We want to make sure that we give them the most performant, most cost-efficient, globally available, compliant AI infrastructure possible. Love it. So what you're telling me is you're another, well, in this case, 12-year overnight sensation, huh?

Yeah, that's right. It's funny because in our space, right, there's all these companies that didn't exist two years ago. Yeah, exactly.

Crypto or- They really are overnight, yeah ... and they're literally overnight successes. On the backs of maybe getting one or two massive contracts from a big lab, a big AI lab.

But you got to ask yourself, are those real sustainable businesses? And for some of those businesses, do they really meet the requirements that enterprises have? Enterprises like if you were talking to a platform engineering team, if you're talking to anyone in the iOS ES team, the cloud engineers, the cloud architects, the enterprise architects, there's a list of requirements that did not go away overnight just because suddenly GPUs became the hot thing, right?

Mm-hmm. Yes, you need specialized storage, you need specialized compute, you need specialized networking to support AI native applications, but all the other storage, networking, and compute requirements don't go away. They get layered on top, right?

And so having the operating history that we have globally, working in so many different legal jurisdictions all around the planet, working on security and compliance and trust and safety since day one, that stuff really does matter because for enterprise, security, compliance, global availability, performance, cost efficiency, those things truly do matter. You're preaching to the choir. " And I look at- Oh, interesting.

Tell me more. Yeah, no, it's about... So I looked at railroads, electricity, phones, and then fiber, and the dot com bubble.

When you look, if you're building your whole business on scarcity of, in this case, GPUs, right? A lot of these neo cloud guys, their whole business model's based on scarcity of GPU. There's Jevons paradox.

The more popular something gets, the more it gets used, and so- That's right ... scarcity goes away. And if scarcity- That's right ...

is your moat- That's right ... if scarcity of GPU's your moat, that's not a business. It's not a business.

And I got to read this book. I'm super excited. We're going to do a reverse interview because I'm going to interview you- Okay.

Well, you know what? on your podcast for this book. It comes out next month.

I'm doing it with you. At the end of the day, I mean this, I'm going to interview you. It's going to be awesome, and I'm going to buy this book myself.

But you're totally right. Scarcity is not a defensible value proposition, nor should it ever be a defensible value proposition. So what we're focused on here at Vultr is we're trying to bridge the gap for enterprise adoption and realization of ROI from AI.

So what we focus on is something completely different. We focus on what are the use cases, what are the results enterprises are seeking to achieve, and then what is the cloud stack that they're looking to mobilize to enable all the downstream developers to build and deploy AI native applications? And that cloud stack is a combination of our core compute services and specialized AI services, as well as The third-party service was from a variety of different ecosystem partners, because open ecosystems win whenever you have discontinuous innovation in a tech stack.

And so we're trying to bring the best of open source and open eco innovation and make it fast, easy to deploy for specific application use cases so enterprises can realize value day one. It's a completely different game that we're playing, and- I agree ... it is absolutely unrelated to scarcity because how is that a value proposition?

I don't get it. No, but you know what? These companies you're mentioning, and I'm not naming names necessarily here, but the folks you're talking about, when you boil it down, that's their value prop.

Mm-hmm. And this is why I said when I interviewed the gentleman, I think his last name was Zhang- Okay ... at Platform Cup.

Yeah. This is not a Neo cloud. Sometimes people are dumping Vulture in the Neo cloud with these other folks, but this is a 12-year-old company that's built an enterprise business.

Exactly. So it's really funny because there's a popular industry analyst firm, which I will not name, that recently- You don't have to ... released a very famous report on AI infrastructure, and we did very well on it.

We're super proud of our positioning. But then there was a certain legitimacy that was lended to a lot of these new Neo clouds. And, I really had to question it because when you look at the core requirements for the report, here's some of the things that didn't matter in the report.

Anything related to core cloud, anything related to security, anything related to compliance, anything related to governance, global availability. No, it was a one-trick pony. How do those requirements suddenly disappear?

I'm telling you, if you have an AI native application, where does the work get done? The CPU is indispensable. Indispensable.

Well, look, NVIDIA is acknowledging that with the new, what is it? Right. The Vera Rubin and the other, the one that it teams with, the GPU, CPU.

Exactly. Well, just look on your phone or go on your web browser. You're interacting with agents.

That's an experience. That code is running somewhere. The agent's doing work.

Where is that work getting done? It's getting done on a CPU. I agree.

How did you eliminate all of those requirements, like related to security- Well, because in the rush- ... compliance and global availability in a report? Look, the only thing I can tell you is this, Kevin.

I view it as it's a moment in time. It's a moment in time. That's right.

com? It was a cute sock puppet. Yeah.

com, you were laying fiber, and you thought- Yeah ... fiber was going to be $1,200 a megabit a second- Yeah ... forever before it went to about 50, and that was that.

com. So back to my early day in 1996, I was- Me too ... working with a company called Interoven.

And back then, it was all about getting eyeballs, eyeballs, eyeballs. I was living in San Francisco. I was on Russian Hill.

" And people didn't understand our business model. You know why? Because actually, our business model was really simple.

We sold back then, this is pre-cloud, we sell enterprise licenses where the minimal starting point for an enterprise license was $250,000. Fifteen. Which was a lot of money back in 1997, by the way.

Absolutely. And our target market was CIOs for the Fortune 100. So our targets were like Cisco, GE.

Right. com. " Why are you selling to those guys?

They're dinosaurs. I was like, "That's all going away. None of you are profitable.

com. We went public. We were like- So did we, but not for long.

Right. I helped put together a company called Interliant. We were an application service provider.

Okay. So we're selling Oracle AppSec, Lotus Notes, no cloud, really no virtualization yet. Well, you were early SaaS in the AS world.

We were way too early. Way early SaaS in the AS world. Way too early.

But we did the whole thing, and look, I lived through it, and that's why- Yeah ... I compared that to now and it's different. Hey, but we're going to run out of time here, and I want to make sure I hit something.

Yeah. com in 2014 and followed it with Security Boulevard Cloud Native now, which is our obviously cloud native focused. Yeah.

And one of the things we've seen through the KubeCons and everything else is the migration of Waterfall to- Yeah ... DevOps cloud native infrastructure, right? Yeah.

The multi-threaded microservice based architecture. Yeah. Right.

We've barely crossed the... I don't know if we've crossed the critical mass threshold to make that migration, but now people are being- And now this wave is hitting at the same time. Right.

Now people are, "Wait a second. " Yeah, but it's so much different. So here's the thing- No, they're both on the same stack.

That's the beauty of it. Exactly. They both run the same stack.

Here's the thing. It's like if you look at KubeCon, because we sponsor all the KubeCons. We go to KubeCons worldwide.

Yeah, us too. By the way, good luck seeing Neo cloud there. But your agents, it's just another set of microservices, right?

Yeah. You got to take API first, like your MACH compliance does not go away. It has to be microservices based, API first, cloud native, headless for all of your AI native applications.

So it's like the whole cloud native paradigm just needs to absorb AI development, which hitherto was a specialized team of ML engineers and applied AI people and data scientists on the side project working on side infrastructure. They need to get Absorbed into traditional DevOps, in traditional CI/CD pipelines, in traditional enterprise architecture. And this has been our message at KubeCon for forevermore, which is make the transition from cloud native to AI native, which is just basically taking cloud native engineering principles and applying them to AI development.

That's why we sponsor PlatformCon. It's the platform engineers. Step up, guys, because platform engineers need to industrialize the building and scaling and rollout- Absolutely ...

of AI native services and applications. com in addition to the cloud native. So you're talking my language, my friend.

This is exactly- Well, apparently you wrote a book about it, which I'm going to buy- Yeah ... and interview you. So I get the, hopefully, final edit comes back to me by September 9th, and we're ready to go to presses.

Perfect. I look forward to that. But this is the...

Sometimes things just work out. The fact that the cloud native stack actually is the underlying stack for this new AI native app stack does make it a bit easier for people to get with the program, to move into this modern kind of thing. Yeah.

I think it's such a brilliant observation, which is if we had not had cloud native, would we be able to do what we're doing right now in agentic AI? And the answer is no. Not as quickly.

If your agents didn't have access to... If you didn't already move to an API first architecture, then how are you going to stick an MCP server on it and have your agent do work? Exactly.

Right. So this is a lesson. " Right.

And that is right. This whole AI thing is built on top of this cloud- That's right ... native thing, and the platform engineering and the DevOps, and it's all fits into this nice- Right ...

ball now. Now- And again, this is where we're a little different than Vultr, is we've been through this evolution since 2015. Well, you were there.

Yeah ... and we understand every layer of that evolution. That's how we built the business, and we were pioneers of it, and now we're adding the additional layer.

And so, yeah, are we a Neocloud? Yeah, we're one of the largest providers of GPU from both NVIDIA and AMD, the only global provider for both AMD and NVIDIA, by the way. Well, that's because NVIDIA's making all these guys sign exclusives, right?

Yeah. And they're financing half of them, too, but that's a whole another... We could talk about that on the next interview.

Yeah. It's a bit crazy. You want to talk about circular financing like we lived through in the dot-com.

Yeah. Well, I think the important note is the enterprise buyer is coming, and the- Yes ... enterprise buyer is looking for an alternative to traditional hyperscalers.

Someone that is experienced, that is global, that understands security, understands compliance, understands governance, and has the core services that can integrate a hybrid multi-cloud architecture, and that's Vultr, and it's not a Neocloud. Can't say it any better than that. But let me ask you, for people who want to go get more information, where do they go?

com. That's V-U-L-T-R, right? Vultr.

Right. Yeah. Just the way it shows up.

Exactly. Well, they're going to have a little third on the bottom when this runs with your name and- Oh, perfect ... so it'll all be in there.

Hey, Kevin- Okay, because I've been trying to keep the... I've been trying to keep- No, you don't have to ... keep up with that Through the magic of computers, we'll have your name and company name there the whole time.

Hey- Awesome ... time goes quick on these- Time goes quick ... these.

What a great conversation. It all zooms to September 9th, and- Yes ... the book will be back.

September 9th is the final edit. Hopefully, I'm looking for mid to late September by the time you go through the whole thing. Yeah.

KubeCon, you'll be there, and then that's- Of course ... November 6th, I think. KubeCon Europe, Europe, North America, KubeCon Asia.

We'll be there, too, but- Anywhere KubeCon, we're there ... we'll talk offline, but we'll be in- Yeah ... Salt Lake.

We'll be in Barcelona for Europe as well. It's our favorite event of the year, just because it's just the whole ecosystem's there, and it's like- No, it's nice. We've been doing it for years.

It's nice. We actually do video right on the floor there, and we're doing a bunch of stuff this year. Anyway, though, but it's not about us.

It's about Vultr today. com. Go check it out.

This isn't just a Neocloud. Okay? Kevin, thanks so much.

This better not be the last time you're on here. We'll- Oh, no. We're definitely going to see each other at KubeCon for sure.

All right. Until then, this is Alan Shimel- Okay ... for Techstrong TV.

Thanks for joining us. Thank you, Kevin.