Built for the Future of AI: Inside Nebius’ Play to Redefine the Cloud

How are new cloud platforms transforming enterprise AI infrastructure as organizations shift from experimentation to production-ready solutions?

‍

Hosts Patrick Moorhead and Daniel Newman are joined by Nebius’ CRO, Marc Boroditsky, for a conversation on how Nebius leverages its Yandex engineering roots and vertically integrated cloud platform to respond to enterprise AI adoption and the evolving demands for sustainable, AI-native infrastructure.

‍

Key Takeaways Include:

‍

🔹Vertical Integration & Engineering DNA: A deep dive into how Nebius’s unique Yandex engineering heritage enables it to build a vertically integrated cloud infrastructure that is purpose-built for AI workloads, providing direct access to the latest NVIDIA GPUs and end-to-end optimization for demanding applications.

‍

🔹The Shift to AI-Native Infrastructure: A critical look at the industry trend where organizations—including major players like Shopify—are transitioning from general-purpose hyperscalers to specialized cloud platforms to meet the rigorous demands of production-level AI.

‍

🔹Redefining Sustainability for the AI Era: Insights into how the conversation around sustainability is expanding beyond standard metrics like PUE to include a holistic focus on supply chain and lifecycle considerations for AI-native infrastructure.

‍

🔹A Look Ahead at AI Trends: A forward-looking perspective on the ongoing AI trends, including infrastructure evolution and changing customer demands, that are poised to shape future innovations in the cloud and AI landscape.

‍

Scale faster with Nebius: https://nebius.com/b200

‍

Watch the full video at sixfivemedia.com, and be sure to subscribe to our YouTube channel, so you never miss an episode.

‍

Or listen to the audio here:

Disclaimer: Six Five Media is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded, and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors, and we ask that you do not treat us as such.

‍

Patrick Moorhead: We are back with another episode of the Six Five podcast. Daniel, how are you my friend?

‍

Daniel Newman: It's good to be back. It's, you know, a lot going on, Pat, and it's been an incredible couple of months. You know they used to say in this business that there was something called summer and we used to, you know, not travel. None of that has been true for me. I've been on the road almost every week now because this massive trend of AI has not given us the breath. But it's got to be one of the most exciting times to be alive, especially in our business.

‍

Patrick Moorhead: It is, it's crazy in the tip of the spear. I mean you've got Nvidia GPUs and others. You have Capex and Hyperscaler Capex, you have Neo Clouds and you've got a ton of competition out there delivering GPUs as a service. Nebbys is one of those players out there and this is great. First time talking to Nebius. Marc, I want to introduce you to the crew here. Thank you so much for joining us. Let's talk about what you guys are doing, how you're winning business. My gosh, you guys are crushing it out there.

‍

Marc Boroditsky: Well, I want to start by saying thank you for the privilege to join you gentlemen today. It's an honor to have this opportunity. And you're both right. By the way, this is an extraordinary time in the world. The way the pace of activity around AI. I've joined Nebius just three months ago and I've had the privilege of being a part of a number of organizations that were super high growth in the past. Twilio and Cloudflare as examples. Nothing compares to what's going on right now. The amount of interest, the amount of activity, the caliber and kind of things that people are trying to accomplish is just thrilling.

‍

Daniel Newman: Yeah, it is great. I was having a good laugh today Marc, that company, one of the companies that a lot of that Neb has partners with very deeply Nvidia. I said, you know, used to be that Nvidia was just a company in the chip space. Then they became kind of the name in AI and then now every quarter there's a, there's a celebration that's their earnings day and now it's about a two week long phenomenon. I said I've had like Pat and I have been fielding media and press calls now for two weeks and of course nebbyous itself has become quite the celebrated name. You know, we're both very avid across the media. We both follow X the Fin twit, I call it community. And in your new three months in digs there, Marc, that company has a passionate fan base. There's a lot of people out there rooting for you and it's been really incredible to see this, Cloud 2.0 as I like to call it. You know, there was the CPU era of clouds and now we're in the GPU era of clouds. And people are definitely attached to the Nebia story. But let's talk about, you know, the background a little bit. Let's go a little bit into history. So Nebius came from Yandex. I don't know if you want to share a little more about that when you answer the rest of this question, which had a lot of really great engineering work. Talk a little bit about how that has influenced the approach because I think it's important that everyone understands not all Neo clouds are the same. I think there's this idea that, oh, it's just access to Nvidia GPUs and it's, or AMD or something. It's like there's so much more. And I think that's why Nebus's story has taken off. People are seeing that it's really doing a lot more than just offering capacity.

‍

Marc Boroditsky: It is really a unique origin story. I have to tell you, the team here is extraordinary. Super gritty, scrappy, highly innovative. Okay, this is a group of hundreds of engineers that have worked together for a long time. Having built Yandex, they built an extraordinarily scale cloud service, consumer app company on the level of Google, the Google of Russia, as they were known. Very successful in all the categories. They entered by all measures a company to be reckoned with in the European and Russian markets. To take the talent that had already built an extraordinarily successful scale organization that was based on applying machine learning and AI to make their services possible and pleasurable and successful in the market. To take that team, forklift them out and start an entirely new cloud company was unique. So this is not a traditional startup where you've got four humans in a garage trying to figure out how to make something work, you know, trying to get their Series A or their seed stage or their friends and family round completed. This company started with scale in its DNA. You know, the humans that could build and innovate the experience of already building at scale and the scrappy grittiness to do what it takes to build an extraordinary company and conquer a category. Now you did bring up the critical question regarding neoclouds and boy, we can go on for a long time about this. It's early going and we're already seeing a wide spectrum of different kinds of companies out there. On one end of the spectrum you have the bare metal players that are operating on a bring me a deal, I'll build it for you and months from now you're going to be ready to go. On the other end of the spectrum there are the companies like Nebias that are coming with a full stack offering with a strong point of view and opinion about the kinds of users that we want to service. We want to be able to give them ML engineer all of the capabilities they need to imagine, build and scale their apps.

‍

Patrick Moorhead: Very exciting because I was going to ask you, AWS is 19 years old and Azure is 15 years old. How is it possible that a new platform like Nebius can emerge on the scene and be successful?

‍

Marc Boroditsky: Well, those are amazing and extraordinarily valuable platforms that people have built great things on. Hyperscalers make money by building things at scale. I mean, hence the name hyperscalers. They want things to be cookie cutter and uniform. They incorporate AI as one of the things you can use their platform for. We are specifically focused on delivering AI capabilities to AI engineers and AI companies. The depth and breadth of our functionality and capabilities is highly specialized to the kind of requirements that an AI application or service or whatever you're trying to build would need. So a very different type of experience compared to the hyperscalers. In some respects you can say that the hyperscalers mission is to commoditize things. Yeah, our mission is to specialize and continually deliver innovation that matches or surpasses the demands of the market. So we're diverging in some respects in the fullness of time. I think in many respects you can think of us as the next generation of hyperscalers that are going to be GPU based. Think of us as the Neo scalers in some respects.

‍

Patrick Moorhead: Interesting. Neoscaler. I hadn't heard that before. Did you make that up? Did you make that up on the fly Marc?

‍

Daniel Newman: I don't know if somebody in Marc, you just gave your marketing team a whole new job night.  I do think though as we sort of look at how proud we and you know, in the lab and evaluating these things, you know, I, I sort of, when I asked you the question Marc, I called it the GPU era. You know, one of the things I said is there was a complete rerating of clouds that happened when we flipped the switch. The biggest cloud provider in the CPU era did not necessarily immediately or has not necessarily become the most successful AI hyperscaler in the GPU era. And of course it has brought this inflection that has opened the door. Enterprises for instance, are a good reason why I think this might happen. And so I would like to ask you about this. I mean, you know, for us to all adopt an LLM, it's not a hard thing, right? You build it into your app on your phone, you use chat GPT for placity. Google Gemini sits inside of our mail tools or inside of search and it gives you AI responses. But doing it for business is hard, you know, incorporating a mix of private and sensitive data compliant requirements, governance, of course, you have all kinds of business applications running your business and then you have all kinds of data and logic that's off, you know, at the edge on devices. I mean, you know, you have to be thinking a lot about these challenges. So as enterprise AI becomes more of the thing, like what are some of the things that are indicative to you that enterprises are more ready? Because as we see it, it's still very early innings for enterprise adoption.

‍

Marc Boroditsky: Great question. And actually I think it's, it's the, it's the big question, right? Because if you look at total tech spend, right, two thirds of tech spend is by enterprises. And it's, I mean it's no surprise that you know, the AWS's and the azures spend a lot of time focusing on enterprise success because that's where the, the all the upside is, the AI, the demands that the enterprise has of AI are going to take time to fully satisfy. I mean I've had customers tell me why I should trust my crown jewels, my critical business processes to your capabilities. They need to get to a place where they're 100% satisfied that they're not going to risk their business. Okay. And that's going to take time. And what I'm talking about there is the full scale implementation of AI enabling capabilities across all the business functions in an organization. At this stage it's very focused on extremely high value creating use cases. So think of things like high frequency trading, you know, that has been utilizing high performance computing for a long time now. They're actually looking at how we apply this stuff to the next generation of their models to be able to outperform the markets, et cetera. Or think about pharma. That's doing drug discovery. Drug discovery is costly and complicated. If you can digitally twin the lab and run simultaneously a thousand permutations of some drug discovery process, you accelerate and cut the cost of the drug discovery process. So very focused high value enterprise applications are already taking place. I think over the next several years you'll see that expand out as confidence and trust is built. We'll be able to reach into all the different pockets of enterprises where there's opportunity to help them to be more efficient, also help them to innovate the offerings that they have to the market, their consumers, etc.

‍

Patrick Moorhead: Yeah. So Marc, it's interesting, you know, if it's not Sam Altman talking about melting GPUs and you know, I, I forget which exact date was it, but nuclear reactors became grain a lot of the hyperscalers and even Apple said, hey, these are all green now. And then you've got hyperscalers firing up Three Mile island again. It takes a lot of energy to drive these amazing AI use cases. And one of the things as it relates to sustainability, Sustainability AI is at an oxymoron. I don't think so is pue. So power usage effectiveness is a metric that has been in place for a long time. You actually have Nebius has a very refined, a different way to equate or pull together the different variables for sustainability. Can you, can you walk us through that? It's more than pue.

‍

Marc Boroditsky: Well, first of all, something that I mentioned earlier in terms of the unique characteristic of this team. They have been building scale data centers for a long time and they've been doing it inside of Russia. So in some respects they had to be really scrappy. They're not in the middle of the Western world's supply chains. So they've been building their own racks, their own power systems, their own cooling systems. They've been designing specking their own machines for a long time. We're an OEM by the way of Nvidia because of that, you know, we build from the ground floor up.

‍

Patrick Moorhead: Interesting.

‍

Marc Boroditsky: So, the opportunity. Yes, you're right, it is to make sure that we're optimizing power usage, but we go beyond that. We're trying to figure out how to optimize our unit economics, power of which is one of the inputs. You can also think about, you know, where we place our, our facilities and what we do with the surplus heat that we generate, which for our facilities we do pipe pump the surplus heat to other applications. You can also think about the choices that we're making in terms of locations, knowing that we're going to start with a low cost location in order to ensure that we're maintaining superior economics to the alternatives everywhere we can. We're trying to make sure that we're managing towards lower unit economics the way we're funded. You know, we're not floating tons of debt that we're going to ultimately be burdened by. We're trying to manage a responsible equity based approach to the way that we're building the company.

Daniel Newman: Yeah, I appreciate that. Well, we don't have to name names, but there are definitely some depreciation and schedule issues that I've struggled mightily to try to understand.

‍

Patrick Moorhead: Oh, you mean, CoreWeave.

‍

Daniel Newman: Okay, I didn't say any names, Pat, that you said names. All I'm saying is that there is an incredibly rapid cycle of technology innovation. And what I know is, you know, I talked to a young student at Rice University just yesterday who had bought a, a recent, roughly recent, nearly recent generation mainframe for his own garage and he bought it for a few hundred dollars. My point is that like technology depreciates, you know, that thing was hundreds of thousands of dollars and it, you know, this stuff. So I don't understand how this stuff works. When you have two cycles a year of new technology, how in six or 10 years, you know, you're going to still put a lot of value on it. But you got your car guy, you got cars behind you. Some cars hold their value. Most Marc do not. All right, let's, let's, let's do a little future forward. Not just Nebius, but in general, like the AI proliferation. I just published a Forbes op ed, you know, challenging Sam Altman's $500 billion bubble comments and saying we're not in a bubble. And a lot of the proof points come from companies like yours, the demand and by the way, their demand, like he's melting GPUs. But he says we're in a bubble. I don't know what he's talking about, but you know, what are the AI and infrastructure trends that kind of have you excited that we know this is going to keep going, this train is going to keep rolling. Marc.

‍

Marc Boroditsky: Well, I think the first one is what you mentioned earlier, which is enterprise. Okay. Just thinking about how big enterprises have lots of opportunities to increase efficiency, to reduce costs, to deliver more, let's call it human valued experience. Okay. Think of every enterprise experience being simplified with a voice interaction and knowing that you're going to get the result that you're looking for. And there's a high degree of ease and pleasure coming from the experience. Then start thinking about industries that are getting creative because of AI. I mean, physical AI is just starting robotics. You know, the device interacting with an object, being able to pick it up and do something with it. That's all inference driven, that's all model based. Having that computer vision capability to recognize the physical world, to put that layer of interaction on top of it. So you're talking to the device instead of a keyboard or an app. All of that is going to be AI driven. And think of all the times that those interactions are being updated and improved. It's going to be tons and tons more data and tons and tons more inference. And physical AI is not just going to be robotics. It's the autonomous experience we have in our cars and our planes and drones. All of that is going to be governed by more and more AI applications. So I think physical AI is just getting started. You know, we're all going to have fully automated capabilities in our homes, in our workplaces, in the, I don't know, the entertainment experiences that we're, that we're exploiting. Beyond physical AI. Think of healthcare as an amazing realm of opportunity. We talked a little bit ago about uncovering the opportunity in pharma to accelerate drug discovery. All aspects of healthcare are going to get improved. You know, everybody's disappointed and frustrated these days with how hard it is to get somebody on the phone, how hard it is to get something serviced, how hard it is to get an answer to a simple question. All of that's going to get enhanced with AI.

‍

Daniel Newman: Yeah, that's super exciting. And I'll be, you know, Pat, you've heard me say this endlessly, but the ability to sort of take all these little pieces of AI that we're all able to do every day, you know, whether it's the stuff that's been layered into our business applications, whether it's the tools like developer tools like cursor and cloud code that have enabled all of us to build stuff like. But then there's so much plumbing and tying it all together and making it work to scale. So again, we're all developers in our own little caves. And then, you know, Marc, we try to become, you know, we try to scale to market. And this is where it seems that a company like Nebius has such an opportunity to really help put all the pieces together, to help an enterprise execute on this. So really interesting stuff. Really excited to see this story and the continuation here and to track the journey. It's going to be a challenge because everyone's after the same meal you are. But the growth has been impressive and I think even better than there's really nothing better than having a market where everybody's chasing capacity and if you can add value on it, that creates margin, that creates stickiness and that creates a great company. So, Marc, great to chat to you. Love to have you back on the show. Let's do this again sometime soon. Thank you, bro.

‍

Marc Boroditsky: It was a fantastic discussion. 

‍

Daniel Newman: And thank you everybody for being part of this episode of the Six Five. This is our virtual webcast edition. Talking to Nebius. What an exciting company. Hit subscribe. Join us for all of our content, all of our shows. Patrick and I speak to so many great executives. We are following and tracking the trends of technology over this episode. For this show, it's time to say goodbye.

‍