theCUBE + NYSE Wired: Marketing Leaders | Kevin Cochrane, Vultr

Clips
More from theCUBE + NYSE Wired: Marketing Leaders

Speakers

Kevin Cochrane

CMO

Vultr

play_circle_outline John Furrier introduces Kevin Cochrane, CMO of Vultr at RAISE Summit 2025 in Paris.

play_circle_outline Vultr scaled profitably beyond $100 million without outside capital or sales/marketing teams.

play_circle_outline Vultr CMO Kevin Cochrane credits platform engineering and AI excellence for performance gains and cost efficiency.

play_circle_outline Driving Innovation: Vultr's Commitment to Customer Success, AI Partnerships, and Developer Engagement Through Hosting Hackathons

Info
Transcript

Kevin Cochrane, Vultr

Kevin Cochrane

CMO Vultr

In this interview from the RAISE Summit 2025 at the Carrousel du Louvre, Kevin Cochrane, chief marketing officer at Vultr, sits down with theCUBE’s John Furrier to reveal how the AI-first hyperscaler crossed the $100 million mark with no outside capital, no sales team and a cost model that slashes cloud spend by up to 90 percent. Cochrane explains why a founder-led culture and word-of-mouth adoption put Vultr on the global stage as an engine for AI-native innovation.

The conversation dives into the company’s next growth vectors: composable AI infrastru... Read more

explore Keep Exploring

What event is taking place in Paris, France, and who is being interviewed? add

What makes Vultr's business model and growth story unique in the tech industry? add

What factors contributed to the widespread adoption of the Vultr platform by developers? add

What are the primary vectors that fuel growth in the context of AI deployment and innovation? add

bolt Powered by CUBE AI

Kevin Cochrane, Vultr

search

>> Hello, I'm John Furrier here with theCUBE. We are in Paris, France for the RAISE Summit 2025. It's an international event. We're here with the NYSC and The Wire team. It's full here, SiliconANGLE, theCUBE, and The Wire making it happen, getting all the data, sharing it with you. Of course the innovation story is front and center on the global stage, Kevin Cochrane, the CMO of Vultr is here. Big booth and generous supporter of theCUBE. Thank you very much for supporting the community. We really appreciate it. Thanks for coming on theCUBE.

>> Thank you so much, John. So happy to be here and love everything that you do and honored to support.

>> Yeah, really appreciate it. The community loves it and the content's going to be flowing. Parade of startups, parade of innovation. Talk about Vultr. You guys have a great booth. You got a lot going on. The business is phenomenal. Take a minute to explain what Vultr's doing and some of the momentum. I think a lot of people need to know the stuff that you guys are doing. It's really doing well.

>> Yeah, so Vultr is an AI-first hyperscaler. So we have a ten-year operating history as a public cloud platform operating in over 32 data center regions around the globe. We are specialists in integrating CPU compute and GPU compute to enable enterprises to build, train, and deploy AI-native applications globally.

>> And you guys are founder-led as well.

>> Correct. - Talk about that because this is a huge milestone.

>> Yeah, it's one of the most interesting stories I think in tech because to the best of my knowledge, Vultr is the only company in tech history that has successfully scaled north of $100 million in revenue profitably without a single dime of outside capital and without a single person in sales or a single person in marketing. So Vultr was absolutely founder-led. It was a group of engineers that just wanted to build a better platform for developers to build and scale cloud-native applications and simply put, because it actually was so cost-efficient and so resilient and provided such incredible performance, developers adopted it in droves.

>> No fluff. - No fluff.

>> Word of mouth. I mean basically word gets around.

>> Word got around. It literally... the economics are such that if you deploy your workloads on Vultr, you have better performance at cost structure that's between 50 and 90% lower than the hyperscalers. So it literally does what it says on the tin. There is no fluff. In one hand, they didn't need a single person in sales and marketing because the platform just works. Now that said, it's nice to have a little marketing to be able to broadcast to the world how great you are and it's more specifically how you can help customers, which is essentially what we're doing here at RAISE Paris.

>> And so growth kicks in. By the way, theCUBE is right behind you. We have no outside capital. We're nipping at you here. Not a hundred million. We'll break the record.

>> We'll break it. Go for it.

>> Yeah, we're going to try hard. Let's talk about what the enterprise... because I love that story. I wanted to get that out because I think personally companies that can scale without taking outside capital is the alpha position in entrepreneurship. And a lot of people think raising money is the thing. Doing it without it is more of a king position. In my mind, it is the top of the mountain, but now you've got growth. Okay, so great. The better product sells itself. Check.

>> Yeah. - No problem. Now you've got growth. So this is where now capital comes in, go to market.

>> Correct. - Talk us through what your plans are and what's happening now because now you've got the good product, product-led growth is a great thing. What's happening now with your customers? What are the things that you guys are doing?

>> Yeah, so there's three primary vectors that basically fuel growth, which is number one is we need to lead with customer success. Every single person, every single enterprise is now looking to deploy AI at scale and they're looking to actually get ROI from their AI initiative. So what we need to do as a vendor is partner with our customers, help them down that golden path to success, document how they achieve that success with all of the technical architecture and help show the rest of the world how they can follow that example. So we are 100% focused on identifying key industry use cases, industry by industry, and getting lighthouse customers for each of those use cases up and running on our platform, delivering ROI so we can help others understand how they could do similar. That's vector number one. Vector number two is it's all about an open ecosystem. The world is moving fast, everyone is innovating and we need to make certain that we provide a platform that enables people to compose a best-of-breed cloud stack, quickly, easy at the lowest possible cost and hard as an entire ecosystem of innovation. This is why we're here at RAISE. It's amazing. So some of the people here are our customers, the rest of them are all our partners. We love it. It's an open ecosystem. It's absolutely phenomenal. And then the third growth vector is you can't actually forget that developers matter more than ever. Developers are the innovators. So this is why we're running a hackathon here at RAISE. So we're one of four track sessions. We got over 110 teams, over 400 people literally hacking away right now in the next room building agentic AI systems. So we're doubling down our focus on developers and helping to provide all of the educational content and hands-on training sessions and labs to help them unlock a new future with our AI infrastructure.

>> These builders are not just developers, it's platform engineering, the persona of developers has opened up. Of course, I saw Jim Zemlin last night, the Linux Foundation. We've been partnering with them since we started theCUBE 16 years ago, but now it's opened up because you got the systems game, you guys have been playing in the systems game from day one. Now you have an end-to-end systems mindset, but you still got that ease of use going on. I could be just building apps, vibe coding one day, trying to put in production, scaling with a team to I need to innovate my high bandwidth memory.

>> Right. No, I mean you totally nailed it and it's an important concept that... and important word that we need to start using. We need to take a systems approach to this. This isn't about a point innovation in just this one. It's an entire system, it's entire architecture, and you do need to have a platform engineering mindset. Enterprises need to develop their AI center of excellence to systematize not only the building and scaling of new AI initiatives, but also put in place all the governance. It's funny because after me, you're going to be interviewing George Kurian, the CEO of NetApp, and I had the privilege of having breakfast with George earlier this morning. And one of the things that George highlighted is, and you'll probably highlight it as well, is that enterprises need to build up their AI center of excellence. They need to systematize a systems approach to AI native applications and core to that is a platform engineering team. We love platform engineering teams. We think they are the future of scaling AI globally. So yeah, we essentially-

>> It's interesting. The platform engineering teams, and I love the fact that Kubernetes is becoming standardized de facto or whatever you call it, people now run their operations. Now they move to the next level, which is agents and data. So of course George Kurian's excited because they're in the storage vertical, which is not a sexy word anymore. It's data platform. So I mean storage is data, but data platforms now are fabrics.

>> Yes, correct. - You got compute fabrics, you got storage fabrics.

>> That's right. - You got networking fabrics. You got database fabrics.

>> Right. Exactly. - So the hyperconvergence of the future is database, storage, networking, and compute.

>> Oh my god. You completely nailed it, John. Completely nailed it. And I think in particular when you talk about NetApp, and I don't want to steal George's thunder, is at the end of the day we finally get to unlock decades of enterprise data. And we often think about code artifacts, we think about model artifacts, and we think about how we can compose these things very rapidly using low-code, no-code tooling as part of the integrated DevOps, SecOps pipeline to deploy new applications. But we got to remember that in the day and age of AI, the datasets are themselves artifacts that you need to have a pipeline to push them through to different training clusters and tuning clusters and vector databases. So I think the whole world of storage and data fabrics becomes absolutely fascinating because it gets elevated to this new type of pipeline and reusable datasets and application artifacts that I think is going to be transformative for the world of storage.

>> Well, let's take that to the next level on two points. I want to take two tracks with you on this because this is really the biggest story in my mind in the market. One is enterprise AI where there's a lot of backlog, my word, of POCs.

>> Yes. - Production is a very thin road right now. And since we're globally... since we're here in Paris, I got to bring up the sovereignty challenge. So you have sovereignty, my country, I call it on-prem, on- country kind of scenarios where I need cloud on demand basically with policy wrapped around it, which means inside a country I'm going to have governance issues.

>> Right. - Not just like GDPR stuff, old school. We're talking about this agent can't cross this line or this database has to sit over here.

>> And no one's even thinking about that right now.

>> Yeah, it's going to be- - problem.

>> It's going to be a big story going into this year and then probably dominate in 2026, very buzzy. It'll get buzzy quick, but let's talk about first.. We'll get to that in a second because I think that's super important because we're global, but the enterprise AI, they need faster time to production.

>> Yes, correct. - And there's a lot of backlog and some of it's mechanisms like, "Hey, we're used to dealing with IT projects this way," to, "Do I have the skill sets to evaluate the startup. " Give me some Vultr and integrate it. I mean, what breaks the backlog?

>> Yeah. So there's two aspects I'm going to actually have to hit on both of those actually. So first, how do we break out the backlog? So for Vultr, it's all about pioneering a concept of composable AI infrastructure and core to composable infrastructure is having pre- composed cloud stacks. These pre-composed cloud stacks are your infrastructure's code. It's like your Terraform templates. It's the things that your platform engineering team can build, certify, and it provides the golden path for how to build and deploy an AI native application. These things are product managed. So these are proper products. It's an infrastructure as code template. And so what we do is use case by use case industry by industry, we actually pre-compose default cloud stacks integrating all of our services, integrating all of our third party services from our ecosystem partners and say, here is an example of how to get up and running quickly. Wouldn't it be wonderful if you wanted to deploy a healthcare agent that was HIPAA compliant, GDPR compliant in all the different regions that you're operating here in Europe? And it was as easy as taking a Terraform template and then just clicking a button and just spinning up all the infrastructure for pill test and deployment within 10 seconds. That is literally possible. That helps break the bottleneck of going from POC to production if you actually show people a golden path. Now we're doing more than that. We're actually providing a whole composition environment to basically use low-code, no-code tooling to basically build these pre-composed templates and opening up a marketplace so people can share ideas. Wouldn't it be amazing if someone who built something amazing could actually share with you their infrastructure's code template and you could use that as a starting point for your own initiative? That would be amazing. Now I have to talk about the sovereignty-

>> Hold on one quick... I want to double down on this. So an enterprise, you can offer them the ability for them to compose.

>> Correct. - Or a startup can say, "Hey, I did the work for you."

>> Exactly. 100%. 100%. So our goal is everyone before you... Think about it, you have an AI center of excellence and they're basically pre-building application artifacts, they're pre-tuning models, they're pre- defining the data pipelines that are necessary to support different application initiatives. All the downstream teams actually fetch all of those resources from the AI center of excellence. One of the things the platform engineering team and that AI center of excellence does is they pre- compose the entire cloud stack. So the downstream team just needs to take a template and they don't need to think about the storage. They don't need to think about the compute. They're consumers, they shouldn't think about it. It should be instant on for them. Now, what happens if you take that further and us as a vendor say, "You can build your own private templates to share with your internal teams," what if we had a public registry of thousands of these things?

>> You're like TSA PreClear. - Right. Exactly.

>> Go right through. - Go right through.

>> I'm just saying that we need this product. This is a huge problem. It's not talked about publicly, but it's a public secret. A lot of startups have one year contracts and they book them as ARR, so they're missing an R, it's not recurring.

>> Right, exactly. - It's AR, A one R.

>> It's actually quite a good point.

>> And so that will fail the ecosystem if they don't have this solved, and it's not the startup's fault, they're building this stack, but they've got to deal with the enterprise pre-existing conditions. I got a Dell rack, I got AI factories from NVIDIA, and then they say, "Okay, how do I go outside here? Do I go to Vultr? the speed test in? " So I think this is where I see what you guys are doing is very relevant because you can just boom, go to you because the cost.

>> It's fast, it's easy, it's most cost effective and it's also the most compliant platform because at the end of the day when you're deploying agentic AI systems, you have all sorts of data residency, data sovereignty issues, like you said, agents aren't going to be able to cross boundaries, so on and so forth. And this is .

>> Let's get through the sovereignty thing.

>> Very strong. - Let's go to the sovereignty thing because I don't want to miss that out. So you guys have a great... We're going to come back to that at another segment another time. So love that. We'll continue to talk about the backlog on the AI side on enterprise. Sovereignty, since we're in Paris.

>> Sovereignty, right. - Talk about that.

>> And there's actually two vectors for sovereignty. There's basically from a national government perspective, ensuring that there is sufficient AI infrastructure, GPU compute, so that digital startups, universities, research labs have the capacity that they need in order to innovate and build the future. That's good for the economy, that's good for societal welfare. But there's another aspect of sovereignty as well, which is every single enterprise that's building and deploying a global application also needs to be adherent to local market regulations. The datasets that they're using to train and tune an infer model are going to be resident within a geographic boundary. The agents that they're deploying are only going to be able to communicate in many cases within that geographic boundary. So there's all sorts of controls. So the issue of sovereignty actually also matters for the enterprise right?

>> Yeah, because they got to run on those countries. .

>> 100%. - It's on-country.

>> Correct. - On a prem. Kind of the same mindset.

>> Exactly. So when you're looking at highly regulated industries like WeServe, like healthcare financial services, so on and so forth, these issues come to the forefront. So if for example, I wanted to enable an AI native application where a patient could ask questions of a local position, the questions that they're asking, the actual model that they're inferring, it cannot be informed by patient data from the UK. If you're here in France, it can only be informed by patient data here in France. That is just a rule. So at the end of the day, you need to be able to set up your systems where you can not only have the compute capacity available here in France, but all of the data flows, the data isolation, the data residency, that all needs to be contained within these geographic boundaries.

>> This ties into the stacks. You mentioned these templates because you can have agents understand... I mean, first you have to set up the distributed computing architecture, one. What's your hardware... hardware. Compute requirements, the supercomputer and then the software layer all have to be in play.

>> They all have to be in play. - And that's not easy. Just scope-

>> Well, unless you're working with Vultr, then it's a lot easier.

>> Scope the challenge for that to pull it off without Vultr. Scope the pain point.

>> Right. Well, I mean essentially without Vultr, you're going to wind up having to essentially build your own public cloud platform. So that's a non-starter for-

>> Yeah. What are they going to do? Open stack?

>> Yeah, exactly. - Like what?

>> But there's a lot to operationalize around that and that's where the complexity... unless you're willing to staff a team of hundreds of engineers and build and maintain it, and even then you still have all the problems of all of the on-demand infrastructure. I mean, just think about it. The product cycle of GPUs is incredibly short. There's always a new generation of GPU that's coming available in the market. This month it's NVIDIA, next month it's AMD. And how do you keep up with the fast pace of change in storage, in network, in compute at the same time that you're trying to build and maintain separate sovereign cloud infrastructure? So with something like a Vultr, because we're sovereign by design, because we have privacy by design, ten- year operating history on the compliance fund, we make it super simple to stay up to date with the latest generations of network and compute and storage with all of the enterprise requirements for data residency, data sovereignty all built in.

>> So you guys are essentially standing up sovereign clouds as a solution.

>> On demand, correct. - On demand. Versus build your own essentially or VCF at VMware, but that's cut if you have VMware or private cloud, all that. And what's the risk there? What are the gaps that would trigger them to like, "Okay, I got two choices now. I do my own, roll my own or Vultr." Where's the tripwires on the-

>> I think the only risk is for people to think that sovereign cloud infrastructure can be stood up by any one of 148 different neoclouds that have popped up over the past two years. And I say that because when you're talking about sovereign cloud, it's not just about having isolated compute resources in a geographic region. The whole notion of sovereign cloud, it's all around security, privacy, and compliance. And so when you have these new startups that got a lot of capital, bought a lot of GPUs, and they're standing up in a given region, they don't have all of the security governance compliance, they don't have the ISO certification, they don't have the SOC certification, there's risk there. That's the risk. People say, "Well, this company can provide me GPUs in this region and they can give me an isolated tenant in this region, that's my sovereign cloud. " The answer is you're forgetting what sovereign cloud's all about. At the end of the day, it's around the security compliance and they don't have that, so there's risk. There's risk there.

>> There's no compression algorithm for experience. That's the line that we hear a lot.

>> You are coming out with all of the amazing one-liners today.

>> I give you Andy Jassy. - I might have to steal that. I will give you proper credit if I ever use it.

>> I stole it from Andy Jassy, he says it at Amazon because that was Amazon's whole thing at the beginning because they had the experience, but now the world's changed.

>> That's right. - We hear a lot of people, like I hear Dell, HPE, everyone's like, "We have VMware. " OpenStack says they have... I mean, all kinds of sovereign washing going on. Now, what do you say to that when people get confused? I'm enthusiastic about sovereign cloud because I have to do it.

>> Yeah. I mean, at the end of the day, I think it's a good thing for one reason and one reason only. People are finally taking this seriously. We launched our sovereign cloud offering 18 months ago. 18 months prior to that, we talked about the need for sovereign cloud. And so only now in the past few months is it starting to become an imperative and starting to take primacy in the conversations that we're having as an industry. So I think it's a great thing. More voices get more focus on it. It's a critical, critical thing that needs to happen. So no, I'm happy about it. More people talking about it. That's good.

>> I mean, I hear you talking. I want to just run this by and get your reaction because the way I've been thinking about it, and the way you just described it is I like this stand-up a sovereign solution that you guys have, I think that's a great direction. Where I see people thinking about this, they're like, "Okay, I got all this AI and agents stuff. I got the stack," and they go, "All right, we got it. All the bells and whistles and the switches, we've got the data," and they go, "Great. No wait, what are we measuring that against? " And then they go, "Oh, the long list of compliance, security," and they didn't design it in from day one, which makes it look harder. It looks easier if you're like, "Okay, let's do the tech stack. "

>> Right.

>> It's not that hard, not that easy. But still we can solve it. But then if you look at both of them together, it's compounded.

>> That's correct. Correct. - And I think that's what you're getting at.

>> 100%. And they need to do that. And we always have this motto, which is like, let's not do the traditional thing and move fast and break things because in the day and age of AI, if we move fast, we won't break things, we'll shatter things and we won't be able to necessarily put them back together again. So we actually do think that being very intentional about how we're standing up new AI infrastructure really matters, and we need to get the chief compliance officer involved, we need to get the CISO involved. They need to really think through how to secure these systems and make sure that these systems can run with good compliance because... and governance because at the end of the day, otherwise we're going to have Pandora's box.

>> Well, Kevin, I'm super excited for Vultr and I love the history. I'm glad that you got that out there because I think it's important. I also really appreciate your community approach and really appreciate you sponsoring theCUBE and making it available to the community. Really appreciate that. I guess I'll end with saying, put a plug in for why you're winning. What are people saying about Vultr? Obviously the word of mouth is just on its own, incredible, just the success of it. What are some of the comments from customers? "Oh my God, this has saved my butt through our business. " What are some of the highlights, highlight reel, commentary on why you're winning and what value and what excites the customers?

>> Yeah, so thank you for the opportunity for that. So I mean, I think there's three things there, which is, number one is we deeply partner with our customers. Our goal is to lead with customer success. So we have white glove treatment of our customers. We are side by side with them. We know their problems. We live and breathe their systems every single day, and at the end of the day, we want to bring them to events like RAISE Paris and put them on stage. In our booth are our customers telling their story about how we made them successful. So we love that. We're more than an infrastructure provider. We're literally their partner.

>> What are some of those stories? Which customers? Can you share their names that are here?

>> Some of the ones that are here are like, we have Base 10, we have Clarify, they actually have their own booth presence here, but they're also in our stage. Then we have some of our customers in healthcare life sciences, like Athos and IPA Therapeutics, and then we have some of our partners like NetApp as well. So we love our customer stories because at the end of the day, customers are going to learn from other customers. So we want to help them do that. The other thing is people love our cute little plushies. Our plushie is named Morty, so it's everyone's favorite plushie. So if you want the cutest little Himalayan vulture that also doubles as a pickleball, come by our booth and people love that.

>> Kevin Cochrane, the CMO of Vultr. Again, very successful startup. Over a hundred million without any outside capital, continuing to thunder away on a growth trajectory. And also a generous supporter who made theCUBE possible to be here to get all this great content. I want to thank Vultr, and thank you for watching theCUBE.