Name: The Future of AI Infrastructure
Uploaded: 2025-12-10
Duration: 19 min 32 s
Description: CoreWeave’s IPO wasone of the defining moments for AI in 2025. CEO Michael Intrator joins us to reflect on the company’s first year as a public company, share insights on the evolving AI infrastructure landscape, and discuss how AI is reshaping the future of the cloud.

CoreWeave’s IPO wasone of the defining moments for AI in 2025. CEO Michael Intrator joins us to reflect on the company’s first year as a public company, share insights on the evolving AI infrastructure landscape, and discuss how AI is reshaping the future of the cloud.

The Future of AI Infrastructure

Conferences

Good morning. Mike, thanks for being with us. Absolutely, um. So, give me, give me a second because I want to explain to everybody what Core Weave does. We've got a live stream on, you know, it's a little complicated. So Core Weave, and please correct me at any point if I get it wrong, Coreweave creates and manages data centers, sells their computing capacity to tech companies. Um, in, in my view, if, if there's a company at the center of the AI boom, it's you. Um, Nvidia owns a 7% stake. That's another company considered to be at the center of the AI boom. You don't own the facilities. You're trying to rapidly diversify your customer base. You have billions in short-term debt and billions more in long-term debt. It is a high wire act. I do not envy your, your, uh, you, sir. And looking at your stock price since your March IPO, it has been seesawing, and to my eyes, I'm seeing the push pull of these guys are the future and these guys are going to be out of business. It just feels like everybody is fighting over that. Do you feel misunderstood? Um I think uh we gotta go, go through that introduction of a few things um so look, um. Core weave. Delivers intelligence. To consumers of compute that require that product to build the most cutting edge, uh, artificial intelligence or even more broadly that require parallelized computing we do that by building. A beautiful software solution to allow for the most performing configuration of GPUs to the market. That is a function of a software layer that we have created um that is singular. It is the best solution in the world for uh providing access to compute. For those consumers, um, we are well known for that, um, and that's your focus. You're not for everybody, you're for this higher end compute. We are for when we, when we started the company, really we started with a blank sheet of paper and say how do you purpose build a software stack? How do you purpose build a company to be able to deliver the best product for consuming compute that is occurring. Um, in a parallellyzed fashion, and that, that is what the company does, um. We do, uh, build our own data centers. We also co-locate our, uh, uh, compute within third party data centers. Um, we are one of the largest, um, um, resources for this type of computational power for the supercomputers that are able to do this in the world, um, so. You know, the, I, I, I really think that, you know, we're not a company that rents GPUs. We're a company that provides a, a, a, a really wonderful solution and you can see us as out there, we're moving up the, the software stack, making a more robust environment with acquisitions like. Weights and biases and you know open pipe and you know like like those are those are acquisitions but in addition to that there's this incredible kind of organic build out of the software that we provide so um that's the way I think of this and and and I think it's a it's a more accurate, um, um. Uh, representation, I probably do it more often than you do. Um, indeed you do. Uh, I believe it's part of the job and, and that, and that I will therefore come back to it, you know, in the run up to the IPO, you obviously had to go on your roadshow, you know, you, you've, you've been out in the market since. Do you, I'll ask you again, do you feel misunderstood? Are you, did, did, does everybody else understand what you just said? So look, um. Let, let, let's, you, you, you're, we're building a company that is challenging. One of the most stable comp uh. Businesses that have existed, right, the cloud business has had these 3 massive players and then you go through this kind of changeover of fundamental technology. You go from sequential computing to parallellyzed computing, um, you have the advent of artificial intelligence. All of these things kind of occur at the same time. The GPU makes these massive strides forward so that you can, uh, uh, engage in computational power at an order of magnitude that the world had never even contemplated before. And all that kind of comes together just as Core Weave rolls out its business, so. Do I feel misunderstood? I feel like it's incumbent upon Core Weave to introduce a new business model on how the cloud is going to be built and run. Into the future and that's what we're doing, um, you know, when you talk about our stock seesawing, um, I, I think people are myopic a lot of times. Yes, it is seesawing. We came out into one of the most challenging environments right around, uh, Liberation Day, um, and in spite of the incredible headwinds we're able to launch a successful IPO. We were, uh, the, the IPO price was 40, the stock is trading at 90. Um, so yeah, it's been incredibly successful and I think more people are beginning to understand it, but when you introduce a new model, when you introduce a new way of doing business, when you, um, disrupt what has been a static, uh, environment, it's gonna take some people some time, um, and I had said that, you know, when I look back in history, um, of the company, it took us. A year. With, with a company, with, with an investor like Fidelity, we had to introduce what we're gonna be doing, we had to explain it, we had to come back and answer questions. It took us a full year before they were like, oh, I get it. And so, look, we've been public for 8 months, um, I couldn't be prouder of what the company has accomplished, um, you know, I feel like. We are leading the innovation within the software stack um in terms of providing uh the environment within which so much of the real cutting edge uh uh artificial intelligence is is occurring uh we have been uh innovative around. The physical components of building this infrastructure we're the first ones out there kind of looking to repurpose power whether it's from crypto or other use cases to access and build really really large scale infrastructure so that that the supercomputing clusters can be built. And uh we are uh the leading innovator around how you go about. Financing this infrastructure whether it was from the IPO to even yesterday when we did, you know, a $22 billion convertible, um, note that was upsized to 2.25%, it was, you know, 5X oversubscribed, you know, it really gives you, uh, insight and understanding to how. Many people from different sides of the balance sheet, both the equity side and the debt side, are interested in joining the journey of Core Weave and, uh, participating in the cap stack in different ways, and you know, I, I, I just think that like that represents. Core, we've once again tapping into, uh, new, deeper, more liquid portions of the, of, of the cap stack to be able to, um, you know, uh, finance and continue to grow, uh, this business that is in such tremendous demand. Thank you for that, by the way, um, I, I wanna ask you another question, but um we do have a question down here so I'll come to you next if you don't mind, um. Let's talk about your customers. Um, everything that you've just articulated is classic AI. Everything's moving very quickly. Everything's changing very quickly. The, the, the, the speed of change is, is extraordinary. What are you seeing in terms of change in terms of what your customers are asking of you? So 8 months is an eternity in AI. Yeah, I do feel old. Uh, so, so look, let, let, let's start with the customers, right? So when, when we came out into the IPO, um, I love how I'm using like this 8 months as if it's some kind. Of like, uh, it's a journey, it's a Lord of the Rings journey, yes, so, so, uh, when we came out we were, uh, incredibly, uh, one of the knocks on us is that we were incredibly dependent upon Microsoft for 85% of our revenue, um, currently at IPO at IPO. Uh, it was a stated goal for our company to drive diversification within our client base, and now none of our customers represent in excess of 30% of our backlog. And so, you know, when, when you say what are our customers, um, looking for, it, it depends, right? There are certain. Um, givens, what are they looking for? The most highly performing infrastructure? What are they looking for? They're looking for incredible speed of the most cutting edge technology to market so that they can gain a relative advantage over the companies they're competing with so that they can deliver infrastructure to the clients that are demanding access to the compute so that they can run artificial intelligence within. Uh, uh, their ecosystem so that they can monetize the investment around, uh, artificial intelligence that they've been engaged in investing in for, for years, um, but there's also new clients that are that are emerging that are, um, looking to, you know, structure deals in creative ways so that they can, uh, have an opportunity to launch a new business, a business that didn't exist in. An environment where it's incredibly difficult for new companies to access compute at the size and scale that will allow them to release products so that they can compete in a world that's dominated by these these foundation models that are so massive and have access to such amazing capital and we've been really thoughtful around how you help that that ecosystem kind of continue to to uh. Um, grow and be resilient and, and scale, and we think that's really important. Yeah, understood. All right, let's go right here. Um, wait for the mic if you don't mind. Good to see you, your name and who you're with for everyone. Yeah, hi, uh, Pankaj, AI investor and early investor in Corbi. We appreciate your support. Um, so, um. You know, the, the market is there, you know, I believe in 5 to 7 trillion worth of data center over the next 5 years. Korvi team is super strong. Your strategy is right. So I think of K we as an execution play at this point. And, um, you know, there was a little bump in the last earnings report and the stock fell like crazy, right? So my question, um, Michael, is, From an execution perspective. What could go wrong if data center deployments get delayed, if energy is not there, if water is not there, if talent is not there, if there is a recession, etc. How do you, what, what keeps you up at night and how should we think about that? Stealing my questions. So please tell us about your nightmares. Yeah, yeah. Um, So look, um, anytime. The world builds. Infrastructure at high velocity. Um, there are shock loads to different parts of the um supply chain, um, that are. Difficult to think about. I, I'll give you an example. I was on a call, uh, yesterday, um, where, um, the, the, the person who was speaking with me was a CEO of a mining company and he was talking through the, um. The, the, the impact of the data center buildout on. Uh, supply chains on metals on the timing of building new capacity, and it was just wild for me because it was like, man, that's like two levels deeper than anyone who is looking at building companies like the one I'm building is really thinking about. We think, hey, you know, where's our damn transformer? We don't think about the. Damn transformers held up because you know there was an issue within, you know, copper or whatever it was and it's just, it goes down levels and levels and you know they're really far downstream and you know as you go further upstream you ultimately end with the bottles and everybody kind of I think has an idea maybe two levels but not the full stack of what's going on. It's fascinating, um, so, so look, um, you know. I, I think it's really important when you're, when you're looking at a growth business, business that's growing and scaling like we are, and you know, I, I don't know that there have ever been infrastructure companies that have gone through the type of scaling that you're seeing at Core Reef, so. If a data center Like from an accounting perspective, there's a day that you mark the, you know, the infrastructure going online, and if it misses by 1 week and crosses a quarter, you know. It's bedlam, but it's not really bedlam on the broader scope of we are out there building and scaling infrastructure every day, you know, um, we added 120 megawatts of, uh, of infrastructure last quarter. Um, we ended the quarter at 590 megawatts. We're going to end the year at 850+ megawatts of infrastructure. Like you gotta. Like, everyone is so incredibly zoomed in on every hiccup and glitch and this and, but like when you pull out a little bit, you gotta, and this is what I do, right? Like, like in running the company is I try to zoom out a little bit and get my arms around these macro trends, these like super cycles that are going on. So yeah, the data center was held up by a couple of weeks, OK. The demand from the most knowledgeable, most sophisticated, largest tech companies in the world is relentless. That's what the trend that matters to me, right, um, as we get larger, and I always refer to this as my denominator problem, right, like if you have Uh, 500 megawatts and you go to 850 megawatts and you missed by a week, it's a big number. If you're, and that's one data center that you miss on, right? If you have. 50,000 megawatts running and you missed by one data center, you guys won't even see the number, right? It'll just be lost in the noise so you're, you're, you just gotta zoom out a little bit and understand like these are macro super cycle trends that are occurring we are deeply ensconced in it we are able to access the information at the level that is picking up. Clients from across the entire space as they plug into our infrastructure and we know where the demand is coming from and we're building as quickly and effectively as possible. Mike, I, I, I have another question for you. Our time is rapidly dwindling, so I, I, um. I'm gonna leave aside the one around how you feel that uh the circular AI economy is a fundamentally fundamentally flawed premise as much as I'm interested in that, um, and I'm just going to simply ask you what do you feel is holding the AI economy back? You were talking about this relentlessness and the challenges of dealing with this absolutely insatiable demand. What is holding us back from being our best AI economy selves? I'm gonna go back to the uh oh you can't. No, no, no, it's just because I'm gonna tie it back to to to the conversation that I was having with uh with the uh the mining uh uh CEO and we're talking about mining a lot more than I anticipated. He, uh, he, he was, he was, you know, and he was just like, look, to, to, to do this, we need to work together as a group, and I said if I do that, I get accused of being in a circular economy like. Companies are trying to address a Violent change in supply demand you do that by working together. So that's all I'll say on the circular economy is like you do that by working together if that's circular, that's the incorrect way of looking at it. It's a lot of companies working to address an imbalance that is distorting, um, you know, the globe and, and this is how you do it and we've seen it before and he was asking to do it on the call, which I thought was fascinating, um, in terms of what's holding it back, you know, look, the, the, um. There are a tremendous number of brilliant entrepreneurs. There are an enormous number of, you know, brilliant data scientists. They are trying to access computes so that they can build these new products and deliver them to market. And, you know, right now there is a physical bottleneck associated with getting. Uh, the, the most performing compute into the hands. The most cutting edge players and, and that's really what's holding us up and we're working like hell across the entire space to ensure that we're going to be able to deliver as much of that infrastructure as possible, um, over the next several years that this demand continues to, to, to rise. Yeah, so not funding, not policy, just, just the physical bottleneck. I mean like. Funding is an issue. It's something we work on continually. Policy is an issue. It's something that you know we're all working on, but ultimately where the rubber hits the road, it's the ability to deliver compute into the hands of these scientists that are going to move the world forward. The reasons that you have challenges in. Delivering that compute is because of policy, is because of funding, is because of physical infrastructure, is because of energy, um, you know, and, and a lot of those themes are, are, are really talked about continuously, but you know, I mean, ultimately, you know, our job is to, to, to allow these scientists to build the world, um, uh, you know, in ways that, that we can't even imagine it. Mike, thank you for taking us into your dreams and nightmares of AI. A round of applause. Thank you, sir.