Hi, this is Steve Eisman and this is another episode of the Real Eisman Playbook. The biggest debates going on right now are clearly involving AI, how profitable it's going to be, how sustainable it's going to be. The debate also changes almost on a week-to-week basis. It's really quite extraordinary. And so today, I'm going to interview two tech analysts. Dan Ives who has left Wedbush and has gone on to create his own investment bank and Gil Luria of Davidson. What I like about these two is they cover a broad swath of the tech sector. You know, most tech analysts cover chips or they cover tech equipment or they cover software, but these two cover pretty much everything. So, I think they're going to have a lot to say about the breadth and length of debate. And I'll be back at the end to talk about lessons learned. But before we start, if you like what we're doing on our interviews and our weekly rap, the best way to support the Real Eisman Playbook is to subscribe as free subscribers on YouTube and on Substack.
Hi, this is Steve Eisman and welcome to another episode of the Real Eisman Playbook. So there's so much going on in tech literally every single week. You know, I talk about it on the rap. I've never seen a group where the pace of change is so big that every single piece of news is not like incremental. It's almost thesis changing. So today we have as two guests. First recurring guest Dan Ives.
Great to be here as always.
Dan is doing a new gig which we won't talk about. And Gil is at Davidson.
And they cover what one of the reasons why I have you guys on is you know most people just cover semiconductors or they cover software but you guys cover a very broad swath and what's happening impacts so much. So let me say a couple of things and I'll give it to you guys. If we were here a year ago and we were talking about AI, I know Dan would have been incredibly positive and he would have talked about how Nvidia, you know, the revenue growth is huge and the hyperscalers are growing very rapidly and you'd be hardpressed to find a negative story, a negative thesis. It's a year later you're not so hardpressed to find a negative thesis. So let me hand it off first to you Gil. Why don't you just summarize, take a couple of minutes, give us from a high level what are the terms of debate and where do you stand in the terms of debate?
Absolutely. So there's two really big debates happening in technology. One of them is are we going to get a return on all this investment? Are these data centers going to create the return on investment for the companies building them and the capital that's being deployed that will justify the extreme expenditure that we've had.
And everybody has to factually admit it's extreme.
It's extreme. It's unprecedented.
So that's one really big debate that's being had and back and forth and our views are a little nuanced on that. Let's frame the other side of the debate which is how is it going to impact all the other companies especially software companies. How are software companies going to do in a world of AI and again the.
The so-called SaaS apocalypse.
Yes. Well, and the SaaS apocalypse was this perspective that they're all doomed and there's nothing to look here like they're all dead in five years. There's no software. And now we have a more nuanced discussion. We could talk about where we stand on that. But those are the two big debates. So let's bring it together for a company like Microsoft, right? Because Microsoft gets the raw end of both of those deals. It's, oh, you're building so many data centers, you're not getting a return on investment. And since AI is not worthwhile, AI is bad, that means you're wasting capital at the same time. At the same time, it's oh, you're a software company and AI is so good that it's going to destroy your business, right? So, they get the raw end of both deals and we argue that, well, hold on a second. I could walk you through why I think we are getting a good return on investment and the return on investment will improve from here. So that probably makes sense for them to invest and then I could make an argument that five years from now I'm still going to get up in the morning, turn on my computer and get on Outlook, okay, and use Teams, okay? And then PowerPoint and Excel and Word. And by the way, there will be agents using my Excel and Word and Outlook and Teams, but I'm also going to be there. And guess who is going to stand in front of the model when that happens? Microsoft. And so those two debates are what's going on right now. And again, Microsoft's getting the raw end of the deal. And that's what makes this interesting right now.
I mean, such a phenomenal summary.
So you have nothing to say.
So look, what I would say is that you're in year three of an 8 to 10 year buildout of the AI revolution. I mean, I view it as it's kind of being like building out the Vegas strip 1955. So, inherently in that there's going to be questions about when does capex ultimately lead to modernization. Does Anthropic eat everyone else's lunch valuations? Is this a dot-com 2000 moment or is this truly a fourth industrial revolution? I believe, you know, obviously the latter. So I think you're going to go through what I'll call like these gut check moments three to four times a year. But I just take a step back and be like in our recent Asia trip demand to supply is 15 to 1 for chips. So I'm just someone that I don't get caught up sometimes in narratives. If you got caught up in narratives a year ago New York City cab drivers bearish on Alphabet, AI is going to crush search. DOJ is going to break it up. I just think right now we're in a narrative shift where if memory is skyrocketing or if it's come down significantly since the SK deal right away it's like it causes definitely these sort of white knuckle moments but in my view like this is going to change society in a good way. I believe more jobs are going to be created from AI than taken away. And for the first time in 30 years, the US is ahead of China when it comes to tech. And for so much of my life, I land from some far off place, land in New York airport, there's some fist fight to Dunkin Donuts, and then I go back to I just came away from, you know, a fab where they're working 18 hours a day in terms in Taiwan.
Speaking to the view, the disparity that you saw maybe in Asia versus here. I think that's narrowed significantly and I think now it's the US's game to lose.
So, let me press you both. Okay, I'll press you on three counterarguments. Tell me what you guys think. Number one, it's not just that they're spending a lot of money. It's that you have companies that haven't raised capital basically since inception. I mean, Google went public. I can't remember the year, but it's early 2000s. They never raised any capital since then. Microsoft never raised any capital. Meta never raised any capital. All of a sudden, because of the incredible amount of money that's being spent, this is now a very capital intensive business, which all other things being equal is a negative.
I think that's a fair statement.
Okay. Number two, from an outsider's perspective, it feels like there aren't any moats in this business. Every week, somebody's got a press release on some new AI LLM that's the new hot toy, and everybody's switching from this one to that one to the other one. Google had a moat around search that was insurmountable. You know, maybe they'll get 30% of this business. I don't, but they're not getting 100%. So, feels like there aren't a lot of moats in this business, which is a negative. And then third is pricing. You know, last week there was this news about this new AI, Chinese AI model, Kimi. Kimi, I love that name. Kimi K3. How they came up with the name Kimi K3 was Kimi K2, I guess. But why Kimi? So, the price that they charge for tokens is like a fifth of what the other LLMs are charging. So, feels like I could make an argument. Again, I'm an outsider. This is not my area of expertise. I got a business with no moats. The everybody's spending a ton of money. Somebody, all these Chinese companies are coming in at a much lower price. That spells to me price war. Okay. That's my argument. You tell me what you think.
So I'll say and then you agree or disagree. Well, first of all, the moat, my view is like the models are going to get cheaper and cheaper over time. They will get more and more commoditized. I think the value continues to be in the data and the install bases. So I think what all these companies doing on hyperscale, whether it's what Meta is doing, whether it's what Oracle is doing, whether it's what Microsoft is doing, the data, like the hearts and lungs of this are all going to be the data centers and compute because every company, every individual as they go down the AI path, you're basically going to have the choice, you're going to be able to put on one or two hands in terms of who you go with. So right now the moat maybe doesn't seem as obvious but they're basically building out their own ecosystems where you're either going to go Microsoft, you're going to go Alphabet, you're going to go Oracle. There's going to be minimal choices and companies are going to have to go down that path and the enterprises and consumers are ultimately going to have to pay the piper. I mean they're going to have to pay these companies. So today it doesn't seem like there is a moat but the reality is they are actually step by step building their moat in front of us. So whether it's physical AI, whether it's autonomous, whatever it may be in the future, it's no different than today. It's like what are your choices when it comes to, you know, content? Netflix was first. They built it. They spent a ton of money. At first investors didn't recognize and now where do you go? Netflix basically owns content that speaks to their opportunity and their install base. That's like, so that's like my own way of kind of viewing it in terms of going back to Vegas strip.
See, but let me press you for a second. It's one thing to say that there's only going to be a few hyperscalers and so you're going to use Oracle or you're going to use Microsoft's data center or Amazon's data center. My point is I'm taking this from the position and I agree with you. Sure. That's a ton of money though. You know, once those businesses get going, they'll be great businesses. I'm thinking of this from the point of view of Anthropic and OpenAI, the creators of the models. And if I was the head of Anthropic or OpenAI, that announcement that came out last week from Kimi K3, I'd be petrified because I'm charging five to seven times more than this model. And supposedly this model is just as good as my model. So what am I going to do?
You referred to AI as one business. It's not. Okay. We're talking about a whole value chain that's being created. There's the companies that make the stuff that makes chips, primarily ASML and TSMC, but a whole other slew of companies. There's the companies that make the chips, Nvidia, AMD, Micron, etc. There's the companies that buy those chips to provide compute, primarily the three big hyperscalers, Microsoft, Amazon, and Google. And then there's the model companies. There's a lot of value being created throughout. And I'll point you to one important data point to show that which is the cumulative run rate of OpenAI and Anthropic right now is clearly above $75 billion in revenue. Okay, so that's to get combined actually we're probably over a hundred billion of revenue by the time you include Gemini's revenue and maybe a little bit Meta and xAI we're above a hundred billion dollars of revenue from what was zero a couple of years ago. Okay, I'll call that value and I'm focusing on that number for a very specific reason which is that is people and companies willing to spend money for AI. So that is real economic activity. So maybe we've put a trillion dollars into the ground so far, but that's already a hundred billion dollars that consumers and companies are willing to pay. That's not a great return yet, but that was zero two years ago. And now it's 100. And we keep building more and more. And those first three parts, the companies that make the equipment, the companies that make the chips, the companies that provide the compute, they add just as much value, if not more, if the model is open source. So yeah, models open source as a threat to open.
Just to find for the viewers because not everybody.
And this is a very and actually I think Gil this is like an extremely important point that goes in terms of like the open source relative to like cheap.
Just define open source and who is using open source so everybody's on the same page.
Yeah. So Anthropic and OpenAI's model, you can really only use it through Anthropic and OpenAI because it's closed. They control all the parameters. They don't tell you what those parameters are. They control all the code. They don't tell you what's in the code. There's two ways to make that more open. One is to share what the weights are, the parameters of the model, and the other is to share what the code is. How does this model work? If you share both of those, it's an open-source model because you can now take this model, take it offline and use it without connecting.
Change it as you wish, right?
Without being connected to the model company, you can own the model, use it, which makes it far less expensive. That's how a lot of the technology stack works right now. By the way, open source software is a lot of the technology world. Linux software is how much of our operating systems works. That's free and open source. Apache, open telemetry, a lot of our technology stack is built on open source. It's a big part of the picture as it will be with AI models. They will be a big part of the picture in the future. Companies will use OpenAI's Anthropic most advanced model for their most important mission-critical tasks. But they'll use open-source models for everything else either from a data center or even on premise or sometimes it'll just be on our device. There'll be a smaller model that's on our device that runs on our own GPU in our own memory that we use to do really simple AI tasks like summarizing emails and drafting emails, things like that. You don't need an Anthropic or OpenAI model. You can just use a small model. And so that is part of the future. Now what's been confusing so far is that the only companies that have been willing to do that are Chinese. American companies have avoided so far having open-source model and the reason is that you can charge a lot more for closed source model.
And Meta tried, if you think like with Llama, like tried, didn't really go that well, right? So I think that was.
So Meta tried an open source model.
You essentially tried it and the problem is that in the open source world there's a view that Anthropic and OpenAI they're so far ahead, it's like what do you, it's like trying to chase Usain Bolt.
But we will end up with American open source models and we're already seeing that happen because Nvidia for instance who is a key player in this ecosystem is saying hey if none of the labs will build open source models we'll just do it because we know open source models use just as much compute as closed source models and we don't want anybody to use the Chinese models so if you want we'll make a model we'll call it Neotron and it's free and so everybody could use that because by the way you still need a GPU you still need memory. You still need to deploy it in a server whether in a data center on premise and therefore it's in our best interest NVIDIA for you to have that. Microsoft now coming on board with that. Palantir's coming on board with that and saying, 'Hey, look, beware of Anthropic.' There's a lot of tricky parts to working with them.
What's his name? The head of Palantir.
Alex Karp was on CNBC and I listened to that interview and I have to say I didn't understand what he was talking about. Full confession. So maybe.
Dan and I are big fans so we can help translate.
Please, please translate into plain English. He was so exercised about, I figured this sounds important but I didn't know what he's talking about. What was he talking about?
I mean look he is just like he is what makes him so unique. Not just in terms of like what he built at Palantir but it's like his view of the world. He's almost a philosopher historian to some extent. So he's able he views things through a certain prism that has ultimately sculpted Palantir. But at the reality is he is a core believer. The models are almost like you know you don't want to be closed into the models. The value is ultimately going to be in the data side. I'm saying from a Palantir perspective because the reason that's so important is he's saying it could be an Anthropic model, a Google model. It doesn't matter.
And he's threatening us that if you allow Anthropic to see your business, if you put things directly in Anthropic's model, you put your data into Anthropic's model, they actually not only do they have their data, they know how your business operates, right? And if they decide to compete with you, they can compete with you. So that's one very important thing he said. Oh yeah. The other thing he's alluding to which I think is even more important is if you build your model, if you build your business on top of a model from either Anthropic or OpenAI and something happens to that model, you're screwed, right? And that happened just a couple of weeks ago when the government told Anthropic to rein in Claude and they didn't. You're done. If you were a business that built your business directly on top of a Claude model, you're out of business.
But the ramifications for that are I mean that was kind of the first wakeup call because but it just goes back to like you're going to have many models. The view of Palantir and many other companies is you could be model agnostic. It's about the data, the ontology to some extent the technology that you're building around it.
Let's move on for a little bit. Let's talk about Google.
Hi, Steve Eisman here. Hiring help shouldn't be a headache or a drain on your budget. Upwork makes it easy to hire specialized freelancers quickly so you can get the expertise you need now without weeks of recruiting or a full-time hire. Upwork is a one-stop platform to find, hire, and pay expert freelancers across web and software development, data and analytics, marketing, business operations, and more. Upwork helps grow your business by giving you fast access to specialized talent across 125 categories. So you can fill skill gaps, launch projects faster, and scale support up or down without committing to full-time headcount. You can browse profiles, review past work, and get help scoping the role so you can hire with confidence and get started quickly. With Business Plus, you can access the top 1% of talent on Upwork. And with AI powered shortlisting, you'll get matched to the right freelancer in under 6 hours. No endless searching required. Thousands of growing businesses already trust Upwork to hire flexible, high-quality freelance talent for everything from one-off projects to ongoing support. It's free to sign up and posting a job is easy. Visit upwork.com right now and post your job for free. That is upwork.com to connect with top talent ready to help your business grow. That's upwork.com.
Because Google's a great company. There's no question about it. Prior to AI, they had search. They controlled 90% of search. I mean, it was basically a monopoly. They hadn't raised any capital from inception. Today, they're in the hunt for AI and they just raised 85 billion in equity capital, which was kind of shocking to everybody.
Yeah. What are your thoughts on Google in this world?
