From the New York Stock Exchange: Financing the AI Revolution

0:13Welcome everyone to the information's TI TV. My name is Akash Pasricha. It is Monday, April 27th, and we are broadcasting here from the New York Stock Exchange. The information is hosting our flagship finance conference here today. We're going to be talking about financing the AI revolution. We've got some big names coming through for the event, some of whom will be joining me on the show this morning, including Glenn Hutchins, the co-founder of Silver Lake, I should say. We've also got EQT's co-head of technology private equity. We're going to talk Sasspocalypse. We're going to talk about the global AI race.

0:48We're going to talk IPOs and exits, circular financing. But I want to start by bringing on the information senior finance editor, Ken Brown, and deputy bureau chief of finance, Corey Weinberg, to walk us through what they're hoping to talk about on stage at the conference today. Welcome to the both of you. So great to have you here. Corey, I'm going to start with you. Tell me, what are the big questions you're hoping to get answered today from the event? I want to better understand from our esteemed speakers, where is this AI data center build out going from here? I think last year, this time when we had this event, the question was more about is this bubble going to pop next month?

1:27Is this demand really going to come? I feel like the question now is more, who are the winners and losers going to be? I think this is a true infrastructure build out now and the space is maturing, but it's not clear exactly where it goes from here. Ken, what do you think? Well, there's no doubt that the money is pouring into this build out. I mean, just Friday, Google says 40 billion for open AI. There was another a deal for$16 billion for a data center in Michigan. The money is just pouring in. CoreWeave is getting a ton of money, all this stuff. So the question is, is this the beginning of this huge boom or are there going to be more bumps?

2:05And yeah, like Corey said, execution is huge. If people don't deliver, it's going to be a harsh penalty on them. Right. The big news story this morning that I want to get both your reactions to. So Microsoft and OpenAI have renegotiated their agreement once again. They've done away with the AGI clause. I'm sure we'll see more iterations to come. I mean, you know, this was a partnership that started very early in the lifespan of OpenAI. But Corey, I mean, just walk me through, you know, the way I'm thinking about it is Microsoft has a lot of the power here in terms of if they no longer need to only sell OpenAI models and they can sort of expand their purview of products, it seems like this is going to be an uphill road for OpenAI.

2:49I I don't know, what do you think? I mean, OpenAI has seen this coming though a mile away. Their relationship with Microsoft has been a bit of a, not a slow motion train wreck, but it's definitely been souring over the last few years. And OpenAI has been running to grab, you know, sort of deeper relationships with companies like Amazon. And so they've really, they have entrenched themselves like more broadly into the sort of tech giant hyperscaler universe. I think your question is a good one though. I see them having maneuvered this fairly well, but we'll see. Ken, what's your reaction to this?

3:23I mean, that agreement was made in the Stone Age, right? It's just so long ago in AI time, right? And so, yeah, it had to end. It had to change dramatically. I mean, I think OpenAI has actually been pretty good. They've gone out and gotten a lot of investment from a lot of different folks, and they're going to go public at some point, right? So I think it probably would have been better for both of them if they managed it a little better or if the agreement was a little smarter. But I think they're both now going to go their own ways. And Microsoft has plenty of business. And now they can sell these other models more easily.

3:58Corey, last question for you. I mean, we are here at the New York Stock Exchange. I mean, the IPO market you've covered for nearly a decade here at the Information. Sure, yeah, exactly. I've seen them all. You've seen them all, but the biggest ones are still yet to come. What are the big questions around these IPOs that you're hoping to get answers to today at the event. Yeah, well, that's the two piles of money going into AI right now that are huge is this data center infrastructure debt. And it's going to be these massive IPOs, which could raise, I don't know, upwards of$200 billion this year if SpaceX, Anthropic, and OpenAI all go public.

4:37But the first question is, is the SpaceX deal going to go well? That's the deal we know is going to be on the road in about a month starting to sell. And look, I think investors are having to get their head wrapped around this sort of Frankenstein of a company. But the story is going to be AI. And that's not something we expected when we started hearing about SpaceX IPO about six months ago. We thought rocket company IPO. Actually, Elon Musk is saying, yeah, you can look at it as a rocket company, but I'm going to be selling you the AI dream. And there's a lot of reasons to be skeptical. Right.

5:10Well, there's a lot to watch. I'm excited for the event today. That is Corey Weinberg, our Deputy Bureau Chief of Finance, and Ken Brown, our Senior Finance Editor here on TITV.

5:24I'm joined now by Glenn Hutchins, the Chairman of North Island and North Island Ventures, also the co-founder of Silver Lake. Glenn, welcome to TITV. It's great to have you here. My pleasure. Nice to be here. I want to get your reaction to a couple of the big headlines making news this morning. So the first headline is Microsoft and OpenAI have renegotiated their deal yet again. Looks like they've taken out the AGI clause. Microsoft can go with various AI models to resell to its customers. OpenAI can do with many different cloud providers. I mean, just broadly speaking, how do you think about these agreements changing so quickly, you know, the AGI clause coming out of it?

6:04What's your reaction? Look, Microsoft and OpenAI were real pioneers. And they were building, if you look back on it, they had the foresight to both build the models and then figure out how to finance their production. But as the industry changed, when you're a forerunner in something, you have to adapt to changes. We now have many other competitive models, not just LLMs, but other kinds of applications that Microsoft obviously wants access to. And OpenAI has grown up and become a real company. Right. It has its own balance sheet, its own set of financing, and its own prospects of going public.

6:39So it's just time for them to find a more stable commercial relationship in light of the current phase of development of AI. Right. Seems natural to me. Just growing up, really. The other headline that I was really excited to ask you about is you just came back from Singapore. And we got the news this morning that, or over the weekend, I should say. China is blocking the Meta Manus acquisition. This was the$2 billion deal for the sensational AI agent company that really captivated people. Did you expect this coming at all? I wasn't really thinking about it. I didn't know what to expect. It doesn't surprise me when I look at it.

7:21Why is that? Well, when you get over to, I mean, I don't know if there's any real antitrust or other kind of implication for it. But when you spend time in Asia, you understand that there are now two different technology ecosystems, one organized around the United States and one organized around China. And there's a fair amount of friction between them. They're all, by the way, eagerly anticipating what might come out of Trump's visit in late in mid May. That's a big deal for that region. Right. Right. And so nothing is done now in the absence of thinking about the superpower competition between the United States and China.

7:57That's expressed in this case at the technology level. So my reaction is that this was setting the stage for that kind of discussion. It could be the digital equipment of rare earth minerals. You can't have them until you give us something we want type of thing. That's one piece of it. But the other more interesting piece is right now, there's an interesting debate in that part of the world right now about whether the Chinese are better or worse off as a consequence of not having access to the most advanced GPUs. Right. The conventional wisdom is that it slows down their development, allows the United States and its Western allies to maintain a strategic competitive technological lead.

8:43The emerging view in China is that it's teaching them how to do more with less. They're becoming more efficient. They're being compelled to develop their own internal industries. That they're accelerating a GPU chip produced by Huawei, and they're accelerating the development of a competitor to ASML. And as a result of which, this might in the long run be better for China because it spurs them to competitiveness and efficiency. Interesting question. Do you think the West might be underestimating all of the AI and the infrastructure coming out of China right now? No, I think it's not an easy choice to make strategically.

9:23In other words, on the one hand, if you look at it from just a narrowly defined but important national security point of view, you want to be able to have access to the most advanced GPUs and associated models in order to protect your country. So there's a reason why you'd want to sort of separate the two. On the other hand, the other argument, and then you can just make an argument about economic security, access for entrepreneurs, that there's a cascading set of advantages that comes from keeping the development for yourself. On the other hand, that historically has not worked well, even in the early development of the United States industries where the Brits tried to keep us from getting the steam technologies.

10:02It didn't work. We figured out how to do it, right? And the secondary argument is that we'd be better off getting them hooked and dependent upon our tech stack. Right. And have no incentives to do anything else. And then continue to lead there and have that advantage. And there's no easy answer to that. There's no one clear answer to it. You make one decision, there's tradeoffs. So we've made the decision largely to wall it off. And the consequence is they're racing to adapt to that. and they're no pretty pleased about kind of where they are and how what they're doing right right now so if we you know we get our so i don't think you can look at that manis deal in isolation from that big power struggle kind of set up dynamics that's my point right right one part of a larger story another part of the story i wanted to ask you about our software valuations which is a space you know very well i mean they continue to fall have they fallen too far you know i i so let me put this in context for you when we created silver lake one of the insights people said it was tech buyouts one of the insights was the software companies which was in a brand this was 1999 were a brand new type of entity were actually better businesses than the ones they replaced you know the more traditional companies they replaced because you had no capex you made your product once you sold a million of the times it became very sticky inside the customer enterprise is very hard to rip out and you could sell all sorts of upgrades and serve and training and services associated with it it's one of the greatest businesses in the history of capitalism and that was actually leverageable you could actually put debt against it so we would buy software companies for six eight nine times even done put three to four times even leverage on right you know imitation is the purest form of flattery 25 years later, people are buying software companies for 20 times, even though I'm putting 15 times leverage on it.

11:55It's a very different thing than when we do. So point one is there's a big chunk of leverage on these companies relative to its cash flow. That's sort of stunning to me in terms of thinking about how you manage a company with that. That's kind of one. Two is that if you think about the software business in general, you have to make a judgment, which maybe we can come back to later, but i think i think i used to think that ai was going to change the world as we know it and i've recently concluded i way underestimated its impact why we come back to that but i want to know now no we'll come back yes but i think it's going to change not just disrupt but transform the way we do almost everything in our economy uh and claude code uh is the first example of that But imagine a massive tsunami that's hitting the beach.

12:48The software companies are the first sunbathers on the beach. So they're the first ones taken out. But the wave is going to continue inland and take out lots of other kind of enterprises as well. So software is just a precursor of what's to come across the broader economic landscape. Right. And then I think the third point I make is if you imagine software on leverage, that's a very dangerous place to be. software on leverage. You take these leverage balance sheets with no cash flow to reinvest in a company. They're now becoming a melting ice cube. And so all you're doing is, in many cases, not able only to service your debt.

13:25You can't reinvest in rebuilding the company for the AI age. That's a very dangerous place to be. I want to ask you about the AI piece in a second here. But I've got to get your take on these moonshot IPOs that are sort of dominating the conversation now. I'm talking, of course, about orbital data centers. What's your view on that? I really don't have a developed view of it. I mean, I think it's not relevant in any time frame that I'm thinking about investing. So I haven't really developed a point of view about it. But as I've just looked at it very superficially, there are a lot of challenges associated with basic physics, getting it done.

14:05And I think I'd rather focus on the next data center in Abilene than one on Mars. So then a little more near term then, I mean, you know, as you look at the private equity landscape at large, the venture landscape at large, and then how big tech companies are implementing AI. I mean, the idea of layoffs, you know, being a very real reality for some of these companies is something that people are concerned about. But should we expect a wave of AI-related layoffs, not just in big tech companies, but even private equity-owned companies that are very used to cutting costs? That's a really good question.

14:44So if you put this in a historical perspective, over the course of the last roughly 45 years, the value of American manufacturing output has roughly tripled. Most people would tell you they thought it had been cut in half. but it's roughly tripled. So we make roughly about three times as much as we did in 1980, but we employ 60 % of the people to do that. Right? So we're, you know, 12 times more efficient, right? And so in that kind of way, that's what, and that's technology. And that's why we have these weak employment markets today. and we have all this political ferment around the cost of living and whatnot.

15:33And that's just about, not because of imports, but because of technology into manufacturing, right? Right. And that's about to accelerate enormously with AI, for sure, right? So the output is really, it's the output. If you put it in the increase, that's kind of point one. Point two is, right now in, technology enterprises are a little bit different, But the conundrum for the broad economy, we'll come back to technology enterprise in a second. The conundrum for the broader economy is a really good article about just a couple of months ago by a very famous economist. It says, how can we have strong economic performance, weak employment performance, and executives reporting across the broader economy, no efficiencies yet from AI?

16:19Which is the case. So what's happened is, in my view, is if you go to the big companies across the world, we'll come back to the technology economy in a minute. They are just now beginning to reap the benefits of 10 years investing in cloud-based and other technologies. Just now? Just now. They really put, because those investments are starting to generate benefit for them. And most importantly, there's a new generation of digital native CEOs running the companies who know how to use these technologies for their benefit. And the older generation of management that dragged their heels on the implementation of cloud technology are now retired.

16:56So we haven't seen yet the benefits of AI expressed in the broader economy yet. In fact, if you look at it, we've actually had people are hiring up in order to have the talent for AI. They're in the bottom part of the J curve. From the cloud-based technology around the accelerating part of the J curve, for AI technology around the bottom part of the J curve because people are investing in this stuff. Right. But this new generation of executives are not going to make the mistake their predecessors did by dragging their heels on AI. They're going to go fast. They're investing very, very aggressively in it.

17:30So you're going to see the J-curve accelerate. I believe that AI has great impact. Now, if you go to the technology world, they're just a precursor of that. Right. Right. And I think you add, they also did a little bit of what economists have a term for it, labor hoarding. right they grabbed all this talent they could and now they realize they don't need it but that's just a precursor what's going to happen in the rest of the economy let me ask you what we got a minute left here last question for you you are on the board of core weave it's true circular financing deals why should they not concern people because i don't really think they're circular in other words um one uh gigawatt of financing one gigawatt costs about 50 billion dollars okay so this deal with nvidia where there's like a gigawatt contemplated a gigawatt two billion is the is the is the deal the size that they normally do but it's a 50 billion dollar enterprise yeah yeah right so i don't see it as circular financing it's not like it's meaningfully supporting the actual spending you're doing in the last this year 2026 core weave has raised almost 20 billion dollars worth of financing from the markets people don't focus on that they focus on the little bit that came from nvidia and say that's circular i just mean yeah the interconnectedness not just corley by the way i mean you know all these i know but look i think if i was in the shoes of those guys look i'm an investor i'm doing the same kind of thing i would invest in those companies too if i knew i was creating huge amounts of value and they were going to grow in terms of in terms of enterprise value and i can make an investment of them right but i don't i see it more as an equity piece of an overall relationship as opposed to some financing without which you wouldn't be able to get right this implementation done right i just think that that's what circular financing implies circular financing is something that journalists like to talk about i you know we we just tried we try to track the track the the flow of the money that's that's all we like to do but i take your point it's a small small drop in a larger bucket and i think that it's very very different than in the past where you were literally um buying paying companies to buy your product yeah and it's a one-to-one relation in terms of quantity how much you put in them, what they spent with you.

19:44Right, right. It's a very different kind of thing. Well, Glenn, I want to thank you for joining us. That is Glenn Hutchins, the chairman of North Island and North Island Ventures here on TIT.

19:59I'm joined now by Arvind Kumar, the co-head of technology, private equity and a partner at EQT. Arvind, welcome to the show. It's great to have you here. Thanks for having me, Alpesh. So we had this story last week talking about how General Catalyst, of all firms, you know, they are taking this approach where they are helping their portfolio companies access compute. And this is something that I wanted to ask you about because, look, venture capital, private equity, different ballgames, but similarities in the sense that you have a portfolio of companies. and I imagine your companies as well are trying to access, compute, manage AI.

20:37Are you taking that approach at all at your firm, helping your portfolio? Yeah, well we have a number of different strategies internally. We have a venture capital fund, we have a growth equity fund, we have the big buyouts fund, which is where I'm a part of. Then we have an infrastructure fund, which actually owns data centers like Edge Connects. And so we are across the entire AI value chain and supply chain. So we try to help our portfolio companies however we can, whether it's getting better pricing from some of the big cloud service providers, better pricing on the LLM side, sharing best practices across the portfolio, and then having dedicated forward deployed engineers to go help our portfolio companies innovate at a much faster scale.

21:17That's some version of what General Catalyst and others are doing, and we're no exception. So forward deployed engineers, so do you have forward deployed engineers on staff at EQT that you are then deploying out to different portfolio companies? Yes, it's a newer initiative, but yes, we do. And so that's only at the beginning now, right, where we will scale that up over time. But that's a big area of acceleration of innovation. So give us a flavor of some of the challenges then that your portfolio companies are having implementing AI. I hear about these consultants now that have become all over age because they need help.

21:53They don't know where to start. What are the big challenges? Yeah, it's not turnkey at all. I mean, just like the mobile revolution, the cloud revolution, doing anything new, right, at scale with no mistakes and pleasing, continuing to please your customers, which is key in the enterprise, requires really good execution. And really good execution means you need a lot of education, a lot of training, need to avoid mistakes. And AI is changing so rapidly that mistakes are very easy to be made. It's not necessarily intuitive for everyone. What's a common mistake that you've seen people, companies maybe run into?

22:28Well, you know, AI models are probabilistic, generally not deterministic, right? So accepting a probabilistic solution as deterministic could be fatal if you're managing people's money. You know, 2 plus 2 has to equal 4, not 3.8. Otherwise, that's billions of dollars of lost money. And then you can lose all your customers if you do things like that. So I think you basically need to keep training the models, right? Having human intervention to check, right? That's human in the loop. So that would be an example of not accepting inaccuracies as fact and continuing to train the models. You mentioned you have an infrastructure fund as well at EQT.

23:06And so, look, we had this story last week talking about the GPU crunch right now that, I mean, it's starting to feel a little bit like three years ago insofar as maybe it's not a supply chain crunch that's going on on a macro basis, but there's just not enough compute. Right. How are your portfolio companies managing that? Are you negotiating these deals en masse for your companies? Yes. I mean, we have a lot of scale at EQT. We have over 300 billion euros under management across really all the regions of the world and a lot of sectors. So we have a lot of scale to be able to drive, you know, favorable pricing.

23:43So we do that where we can. The infrastructure fund is in the, you know, providing the capacity side. So that area is on fire, right? There's insatiable demand. Even Anthropik doesn't have enough compute relative to its growth rate. So there's a lot of demand everywhere. As it relates to our portfolio companies that are consumers of tokens, right? Token costs have gone up a lot as there's more usage. We think that that will amortize over time. There's a term called Jevons Paradox, right? When more usage, actually costs come down, usage goes up. So we see that happening. Yes, costs are going up in the portfolio in this area.

24:19But the ROI they're getting on the usage of tools like Claude and others far surpasses any sort of cost increase through higher revenue growth, more margin, innovating at much faster cycles. So we're making our money back in spades relative to the increase in cost. Now, software valuations have taken a hit across the board. Is this a buying opportunity for EQT? are you looking to make more purchases? Or in an environment where it's harder to find exits, do you have to keep your balance sheet a little lighter? Yeah, like everything, it's not black and white, right? And we're always trying to drive exits.

24:59We've been very good at driving liquidity where we can, and we have to be more creative at times these days. On the buy side, there's winners and losers emerging, right? And certainly the public markets are assigning that sort of winner versus loser, as are the debt markets. and there's been a lot of articles around private credit, right? We see that to be fair. Some companies are undervalued in the public markets. Others are properly valued or may even go down further. And it's our job as specialists in software and technology to know which ones are actually quite attractive. And so we have a framework we've developed to evaluate all of our companies in this AI world around who are extremely defensive in terms of their business model attributes and market characteristics and then who are not.

25:45And then on the offensive side, what is the AI opportunity? How massive is it? And then separately, how good is the management team? On the execution side, they go take advantage of the opportunities that AI can afford. So we have a pretty rigorous framework. And then through that, we identify the winners versus the losers. So how do you think about this balance here then? You're implementing AI at your portfolio companies. at the same time there is this fear that AI will sort of eat software altogether. We'll get to cybersecurity in a second because I know you're involved in that space. But so is AI going to kill enterprise software the way we know it?

26:23Well, I think parts of it, yes, and parts of it are only going to get stronger. I think if you don't, if you, well, I'll answer this way. If you're a business that has high regulatory complexity, like having proprietary data, let's say dealing with financial information sitting here at the New York Stock Exchange, right? Dealing with HIPAA in healthcare, right? These are, it's hard to replicate that, right? In a, with AI de novo, right? So high regulatory complexity, proprietary data, where all the data is coming through your system directly, not sitting on top of something else, right? Where the workflow is extremely embedded in your platform, where people are using your platform every single day to do their jobs.

27:02Yeah, a tool to get through their day, right? If you're a plumber using a software tool to book all your jobs, to get paid, right? To understand customers, to find customers and so on and so forth, it's gonna be hard for you to displace that tool and use something unproven, right, as an example. However, on the other side, if you're the opposite is where there's more vulnerabilities. If you're a software coding, that's very at risk. If you're just a BI tool, that sits on - CRM. Well, you know, this is where, you know, rules-based, logic-based, right? That's what LLMs and AI labs are very good at, right?

27:40So if you just have that, right, that's more easy to be replicated. So if you're just providing a BI tool sitting on top of something else, a software coding platform, just a CRM platform, those areas are more disruption. Last question for you. So you sit on the board of cybersecurity companies as well. And I wanted to ask you about this tradeoff here where the more AI you implement, as we've seen the last couple of weeks, I mean, there are cybersecurity concerns. So that's one way of looking. The other way of looking at it is that cybersecurity businesses like your own, I mean, there's a lot of opportunity to play.