Robert F. Smith on the Future of AI Value Creation | Vista Equity Partners

Join us right now to discuss the markets, artificial intelligence, deal making, and so much more. Robert Smith is here, he's the founder and CEO of Vista Equity Partners. He's also the author of Lead Boldly, Seven Principles from Dr. Martin Luther King, Jr., and I wanna talk about this book. We previewed this a little bit in Davos when we saw you earlier this year. Before we get into that, I wanna just get into the whole AI situation and how you're thinking about it. Because we've got NVIDIA with their earnings later this afternoon. It seems like everyone's keying off of that. A lot of people are worried that we're living in some kind of AI bubble. What do you think is really going on here when you look at this cycle? A few things, I always say history doesn't necessarily repeat itself, but it rhymes. What we're seeing again is another general purpose technology just like the Internet. If you looked at late 90s, early 2000, you saw this massive expansion of capital spend. Of course, this is orders of magnitude greater. Some early players in that environment obviously flushed out, and some other later players actually captured massive economic rent and massive economic returns. In this environment today, the interesting thing about it is this new technology has a chance to really transform every industry globally. The compute dynamics require a massive amount of power. And it's interesting, a lot of the early phases of this are all being directed towards one company. So the way that technology usually plays out in these markets is there is a first wave, which is called the silicon providers. In this case, NVIDIA is getting the vast majority of that economic rent back well over 60 plus, 70% of that activity is driven towards them. The second wave typically are those who are providing infrastructure and door services, the hyperscalers fall into that category. You have new entrants in that case. And the third wave is usually what I call the application providers. That's you guys. That's the world I live in. And so we're now starting to see as we've launched our agentic factory, a number of our companies starting to install these systems. So with all that said, when you start to unpack the math around this thing, what it says is you actually have to now start diversifying and cutting the, call it the cost of compute. Because is that sustainable? You take some of the words that Sam is talking about where he's going and saying, all right, over the next 10 years, I need 50 gigawatts of power. And it's about $60 billion to make that per gigawatt. And so you do the math, that's $3 trillion of investments that you need. Long term, I think you're going to see some really interesting returns on those sort of investments. But short term, that is not necessarily sustainable when you take open AI and multiply that times Gemini and multiply that times, all right, every one of those. So if it's not sustainable, then what's the investor supposed to think? So what is now happening is now you see the efficiencies of ecosystems start to play, right? So now you see people moving towards, what are more efficient ways to drive this compute? Now you're seeing the AMDs and the Broadcoms now becoming beneficiaries of that, because they've got to bring that cost down. They've got to bring that $60 billion per gigawatt down to something closer to 20 or 15 or whatever it might be. So there's efficiencies in hardware that's going to need to happen. It's going to have to be efficiencies in utilization of the software, in PUDA systems, all those sort of things. You have to see efficiencies of the utilization at the edge. You actually need that level of compute for all applications. And in the enterprise, you actually don't, is what we're finding. So those are the dynamics that are going to- What about the economics? Because you, look, you live in the private equity world, which includes debt. You lever some of your investments. That's what you do. Right. One of the things that we're seeing right now is both some of the big hyperscalers and others, including Oracle and others, are taking out loans for the first time, basically ever, to finance this stuff. Right. You've got private credit vehicles and companies like Blue Owl that are on the hook for some of this. People are wondering whether that all makes sense. And then we're seeing these deals that some people are describing as circular. Right. Where you'll see Microsoft, for example, invest in Anthropic. And Anthropic is going to then turn around and run its business through Azure. Right. Is this a red flag to you? Is this normal course of business? Is this what? Yeah. Again, these ecosystems are evolving. And the efficacy of them, the amount of power required for the compute for ultimately what will be the inference, not only the foundational models, but the inference infrastructure, there's going to be this evolution. There are going to be some huge winners, and there are going to be a number of losers in this context. Just like in the Internet. Right. When we first did the first build out in the late 90s, early 2000s, you remember the global crossing. Right. A bunch of fiber we had to lay, and then all of a sudden, companies have become very challenged. And then later, that became the commodity that everyone needed and became profitable once it went through its restructuring dynamic. So I think we're really in that dynamic motion of these different equilibrium states. The difference here at this moment in time, from what I can see, is a vast majority of the energy and effort is being driven right now towards one hardware vendor that will get diversified over time. Now, that hardware vendor has figured out how to take advantage of their pricing, their capability, and their scale to deliver returns. How much time have you spent on this question about depreciation for the value of all these chips? Because that seems to underlie sort of the whole economics. There's a renewal cycle that has to occur, right? And that's going to be some efficiencies. You know, Elon takes a different approach and is like, listen, brute force it, right, in some respects, and get as many chips in the configurations in the cluster so that you can be more effective and just deal with what I call the replacement cycles. These are all part, you know, it's all math, right? But it's all part of the equation that certain of these aspects of infant mortality on these chips are known and certain are going to be new in these different cluster environments. You're in the app business or the enterprise level of this. We were talking about the agentic side of this, which you're involved in now, and how you actually use this technology. I mean, you have an agent for yourself. I do. So what does this agent do? How does it work? So my agent has a number of aspects to help me run our business and run Vista. There are agents that actually do things like report the news around the different theaters, what's happening in Asia, what's happening in the Middle East, what's happening in North America, Latin America, etc. That has relevancy for, you know, the business that I'm in. One part of the agent is actually focused on what we just talked about. What are the innovations in AI that are occurring? What are the papers that are being published and what are the implications? This is like having a research assistant for you, that you wake up in the morning and you get a dashboard, or you're asking it questions, that you've given it a prompt, it's doing this for you. I have a dashboard and some are pre-prompted in essence, because I'm asking the same sort of questions. What's happening in the portfolio? For instance, we launched now our agentic factory. You and I chatted about that. And now we've got 19 companies as of this morning that have an agentic solution, an enterprise class agentic solution being sold into the marketplace, real revenues and productivity in the companies that they are offering. I track the KPIs of that. What is the actual, you know, what's the economic rent pickup we're getting from what was a SaaS model into an agentic model? What is the most successful of the 19 right now? What's the killer? What's most interesting is one of our companies, Gainsight. And if you look at what Gainsight is doing, they are actually thinking about a customer success model. And so when you think about, you know, let me give you some implications. If you're using a tool like this as a customer success agent, we might have been getting $2,000 plus, you know, per person that's using it. We now have this installed in certain environments, we didn't need to know people to do that work. And so all of a sudden- What does a success agent mean exactly? So if you're a large customer or company, you have people who cover your largest customers, and they have access to all the data around trouble tickets, you know, what's the environment and how it's being installed, used, and helps manage, I'll call it the relationship, in ways that the product is being successful for them. So every large company has these. There's also a long tail of these companies that actually for renewals, et cetera, that no one covers because they're not necessarily large clients. And so those so often are outsourced. We've been able to use this solution in an outsourced environment, but now you no longer need anyone to do that work. So the implications are, you know, 20X, 30X productivity from using an agentic solution installed versus having a outsourced infrastructure. We've got to get you an agentic solution, Joe Carlin. I want one, but I want to call them Q. And you already, you took that? I did take it. How did you, that was a James Bond- You just, you just go take it. Was that available? No, it's a letter, it's a letter in the alphabet. That's what I used. So you have this new book. You can ask questions from Q. Yeah, absolutely. Yeah, that's the best letter. You took the best letter. Of course I did. Well, I'm an early adapter, so, you know, first mover advantage is the key. He's also an author. I think you wrote the book without any agentic help. Correct. Are you going to put this book, by the way, into some kind of, would you want a large language model to to be able to read this book so that people can then ask the book questions? I will, because what this really is doing is is taking the word. So the King Center asked me to do this and it's taking the words of Dr. Martin Luther King and helping this new generation understand the relevancy, the importance of it. So it should be widely distributed. It should be part of what I call the American psyche, should be part of the world psyche about how do we lead forward so that we have more inclusive, productive, prosperous societies, which is what we're all after. What's your favorite Martin Luther King line? When you think about the way that they characterized the march on Washington for jobs and freedom. OK, everyone loses the jobs part. And to me, that's my favorite line, because it really is like, listen, we are focused on economic prosperity. And that's it. And that's if you think about what we do in a right now is creating a opportunity for massive prosperity for economies, for communities. So that's what I'm focused on. I want to get him in conversation with Walter Isaacson, also has a book out this week on what he calls the greatest sentence ever written. And because, by the way, Martin Luther King talks about that sentence as well. So you could get to anyway. OK, thank you so very much. Congratulations on the book. This is great to see you. Yeah, thanks.