The Pomp Podcast
00:26:01 · Anthony Pompliano
[00:26:01] S3: Mhmm.
00:26:02 · Jordi Visser
[00:26:02] S2: Or was this a message that this is our hope? Now you said, are they using this for my... I mean, I've run a business. I I know what micro... Like, you can go through their compensation. So let's assume it's $400,000 a person when you add in health care at Microsoft. That's $24,000,000,000. Not 24,000,000,000 for one year. That's every year going forward unless they go hire more people. That number, if they all took it, would take them back to 2,020 levels. Their revenue has grown... Will have grown 70% if they just meet numbers into next year. So revenue per employee's gone through the roof. So this has a lot of issues to the government, and this is why it's positive Bitcoin. This is why I wrote a substack about this whole thing, comparing what Microsoft was doing to what was happening in France, where the common thread is gonna be affordability. Because on the one side, what do employees making 400,000 have to pay in taxes? Yeah. 50%, 50 plus, depending on where they live. What does Microsoft pay on corporate taxes? There's an arbitrage here that's becoming a bigger story to the government because if Microsoft doesn't have to pay the same amount, you're you're losing receipts Mhmm. From this if people don't get jobs or if they end up taking less of a job, whatever the case is. I think this is becoming a labor capital story that is much, much bigger than people realize because we're just entering the productivity boom. That's the other message that Microsoft is saying is they believe they can do this buyout at five to six... Or six to seven times the last one. What is that saying about their expectations about what AI is gonna enable their business to be like in the future?
00:27:44 · Anthony Pompliano, Jordi Visser
[00:27:44] S3: Listen. I can can talk about this from a personal perspective, because we just went through this with our own public company. We have done two things, a massive cost cutting exercise, compute cost down substantially, Compute cost literally... I... I've talked about publicly. $550,000 in March 2026 [00:28:04] S2: Mhmm.
00:28:04 · Anthony Pompliano
[00:28:04] S3: Per month. Yep. Down now 50%, give or take. My hope is... Hope that we're actually gonna be able to get completely off the frontier models, run everything on our own software and models and hardware. If we do that 97% cost savings on a token for token processing, massive cost savings. Going to all of our key vendors, finance, accounting, legal, etcetera. Hey. We we gotta do something here. I know you guys are using the technology. So are we. Let... Let's reorient here. Okay? Mhmm. We mutually separated with some of our executives. Cost savings there as well. So you take basically the, you know, the most well paid people at the company, and you're like, hey. Can we do this for cheaper? Look at that. What we publicly said is we think we're gonna save $10,000,000 between cash equity, etcetera, in this whole exercise. Okay. But that's actually, to me, what people will pay attention to is because it's like a headline. Mhmm. The more important thing is we have a public company. There's 10 people there. I'm not claiming we're Facebook. I'm not claiming we're whatever. But if you go and you look at the product, this week, there's five days in the workweek. We launched five separate products or features, one every single day. And these are features that aren't like, hey, let me just put a prompt in Claude, you know, or something, and you just get it. You gotta be very thoughtful about it. You're dealing with people's sensitive financial information, etcetera. But I explain to people, think about what that means. If we're launching something every single day, and we only have 10 total people, including the nontechnical people at the company, When you extrapolate that out to a Meta, an Amazon, you know, right now, actually, SpaceX. If you look at the updates to Grokbot
00:29:48 · Jordi Visser, Anthony Pompliano
[00:29:48] S2: Mhmm. [00:29:48] S3: They're releasing something like 50 to a 100 new features a week. It's just you're cutting cost while that productivity is just exploding. I don't know what this looks like in two, three years, but, like, you start getting pretty excited about just how productive any of these companies can be. Right?
00:30:07 · Jordi Visser
[00:30:07] S2: Yeah. And and so let's bring it back to Anthropic, because I think the most important news item that came through this is that Anthropic announced in their s one that they're dealing with basically 250 of the Fortune 500 companies, so 250 of the S and P 500. But in terms of their total expenses, which again are dominated by compensation, of the Anthropic portion is point less than 1%. I think people have to fully grasp that if this company has this kind of momentum, and they can give you insight into the adoption right now of AI, and also for everyone that thinks that they're gonna be able to go build data centers and do all the stuff, which I'm hearing more and more on other podcasts. I'm like, where are they getting the memory from to go build their own data center? I mean, guys, come on. Like, if the banks wanna go build their own data centers, that's all well and good, but if you can't get the memory, you can't do it. You can buy as many GPUs as you want, you can go through this, but if you can't get the memory, where is this gonna go? So I just think for, as you go through the Anthropic thing and you look at it, where people talk about a bubble, they're talking about a bubble for the reasons that you said, oh, look, the AI trade is 20% off its highs. Some of the names are 40% off their highs, whatever the case is, even though they're up a 150% year to date, that's... That doesn't matter. Spreads are widening. This is going on. Triple c's are up at 16%. None of this is going away. This is the one thing I will say is as much as I believe yields are probably peak now at the long end, do I think we're going down to 3% in ten year yields anytime soon? No. And the main reason is because the capital needs are going to be massive, and that means we're gonna have debt, and the government has a lot of debt. And so the issuance alone will keep it in. Do I think that Bessent knows this and wants to control the back end? And this is one of the reasons why if you were waiting for yield curve control, if you were waiting for QE, we're not getting QE the way that we used to in the past. It's coming in a different form, and that form to me is starting to happen because the labor versus capital debate means a lot of voters, because you know how many votes Microsoft gets?