The AI Ecosystem

0:00Hello, and welcome to the personal wealth coach with Jake McClure. This week I'm going to go into an area that a lot of people have been talking about, but from a lot of ignorance. So what's going on in the AI world, the AI ecosystem? I don't mean the large language model. I don't mean when you get online and you type in or you put in your phone and you ask get a question. I mean, the companies that are spending huge amounts of money on this stuff. We hear about that regularly. Who are they? Who's spending the money? How much of it is borrowed? How does it flow? We hear about Nvidia. We hear about open AI. Most often when people think of AI, they think chat GPT. So what is meta, the owner of Instagram and Facebook and what's happened? All of that. What do they have to do with this stuff? What is that? Spending money, are they in competition with chat GPT? And this is a lot of what I've been hearing lately.

1:01I hear a lot of people asking me strange questions with these assumptions built into them that are not factual. Like Nvidia's borrowing tons of money to be competitive against Microsoft. So hopefully I can clean up some of this confusion and talk about what's actually happening here and compare it across the world. Who is making AI? How are they making it? And I want to kind of compare it to some things that we're used to and we understand, but we didn't when they were part of the new big thing. So what I want to kind of parallel here is the advent of the computer or the advent of the internet and how people spent money and where and what was needed. So let's start at the beginning. But first we've got to play some bumper music. When we get back we'll be talking about the AI ecosystem. We're going to follow the money. We're going to look at what is it talking about when we talk about a stack.

2:03And how is it that all these companies expect to make all the money back that they are obligating to AI on the other side of the bumper music? We'll be right back. Once more onto the breached dear friends. Else close the wall up with our English dead. Welcome back ladies and gentlemen boys and girls to another exciting episode of the personal wealth coach with Jake McClure, the AI ecosystem. Alright, so let's talk through it. Let's talk about all the industries that are being touched here that we hear about independently in the news. We've got power and construction. Utilities, data centers, cooling, fiber. We've got lines being laid across the country to increase our bandwidth. We've got data centers showing up, but not in my backyard. We've got utility companies expanding at a rate that we haven't seen since the 1930s.

3:09That's all physical stuff. That's all stuff we can see actual dirt being moved and bricks being stacked. That's visual, but it's, it falls into the background. So who's doing it? The next layer is what goes into those data centers, the semiconductors and the computational hardware. Those are two different groups. Semiconductor inputs are memory, both the hard and soft memory. The stuff that is stored indefinitely on a hard drive and the stuff that's holding its values and calculation in the RAM. You have computation, which is what's kind of holding that data and moving it back and forth to figure out what the end result will be. Those are all, we're still talking hardware. We're still talking about data centers and actual physical stuff being made. We're not getting into the software area where what is it that we're actually making?

4:09What are we actually selling? It's all stuff that kind of floats out there and we're not really sure about. So this is still the hard stuff. The first three areas of this stack, if you will, are hard. The next layers are cloud. What is this? Well, this is Azure from Microsoft, AWS from Amazon, the Google Cloud. You're all pretty aware of that. This is where we store our photos and Apple does this and where we store our documents for backup purposes and malware, fighting and ransomware. We can get it back and all that good stuff. Well, that's kind of amorphous. It's floating somewhere between a lot of that hardware we just talked about. Then we have foundation models where large language models are being created to do something. So all the way up to this point, I could have been talking about our technological infrastructure going back 15 years.

5:10This point forward, we're talking about strictly AI. The foundation models are being made by companies. They're not in the utility business, so the data center business, so the cooling and fiber business. They're not making memory chips. This isn't about a processor. And they're certainly not a cloud platform. So this is open AI and chat GPT and anthropic and cloud. It's Gemini and Lama and XAI and GROC. These are foundation models that are built around the concept of how do we find the best answers or come up with the best results based on the input from our user. And then you have how it shows up in your life, the distribution of this stuff, whether that's in a search engine or an ad that pops up when you're on Amazon, commerce software to help with logistics, reorganizing your maintenance and reorganizing how you measure what you

6:13have to sell. All of this stuff comes into actual rubber meeting the road in a business that has nothing to do with AI. The business itself might be selling shoes, so how can AI benefit them? And simple approaches are like if you maintain inventory control in your software, you don't have to spend days and days every year just going through and cataloging everything you have. For small businesses, that's what you do. Small businesses have been in competition with the big businesses that have had these big software stacks and they know what they have in their inventory without having to go through and count everything 14 times and have people looking over each other's shoulder and triple it, triple counting it again. So distribution of the technology is the last end of that chain. So who's doing it? Why are they doing it? So let's kind of step back, let's follow the money. The money obviously gets spent first in the physical structures when we're talking about

7:17a supply chain. If you're buying shorts from the Amazon, where did it start? Somewhere they made the thread that went into the shorts. That thread could be made with cotton or with plastic. You probably don't want to wear wool shorts. There's different things you can use to make the shorts with, but they're physical. So the beginning investment on making shorts starts with where you get the base product. You step that up, then you have to manufacture it, then you have to distribute it, then it is actually used. When it comes to AI, the first step is always physical. You have to have a large language model to begin with, but even that had to start with something physical. I'll go back to Google's release of the algorithm that is AI. A few years back, I'm going to be able to say a few years for a while now, Google's research

8:20department came out and said, we're going to make this available to everybody, but we did a weird thing. We took our normal algorithm what we used in the back end when you're doing a search, and we cranked up the compute power on it, and we cranked up the utility power. We had to provide a huge amount of electricity to this thing, but then suddenly it was answering our questions as if it was thinking, thus the rise of the concept of artificial intelligence. There's a limited artificial intelligence in a calculator. It can generally come up with answers to mathematical questions far faster than most humans. That's artificial intelligence, but it's a very limited version of it. You can't ask it what the weather is going to be tomorrow. You can't ask it when your favorite show is playing. Those are things that it cannot answer. It can answer in numbers. The next layer here is this Google search result that wound up actually reading the internet

9:20and providing the answer to the question. This was released to everybody. Open source. People started using it. The mass behind this is something that we've been using at a firm for a very, very long time decades, because it's based on economics models. Google took it in rocketry and handed off how to optimize a search result. I can spend another episode just talking about that. Okay. When I say it has to start with hardware, they couldn't crank up the power without some power to crank up. They had to have the hardware in place for the large language model to function. It was a rudimentary large language model without much in the way of training. It was just sipping data from everywhere. The results were less correct but more popular. Let me touch on that for a second. The way our search results occur, the way our correct answers occur across all of these large language models,

10:22is based on popularity in an area where it hasn't been trained. What does that mean? The most popular result. If you're thinking to the old school Google days when everybody would Google it, just as a side note, that verb Google it is starting to disappear. It's being used less and less, almost not at all, mostly by the older generations because that's now people are getting their information anymore. They might go to Google to ask it an AI question. It's not Googling anymore. It's fascinating to see this transition because that was a big thing when it came out. Now we're behind the scenes without a lot of input or fanfare for that matter moving away from it. Okay. You go to Google, you ask it a question and it has optimized the search result to say other people have asked this question before. And these are the places they clicked on in my answer to them. And most people didn't click any further beyond this link. So that's probably the right answer.

11:22It's based on popularity, not on correctness. And I hear a lot of people go, oh, that's scary, that's dangerous. Well, that's how we run our government. Winston Churchill said this very, very well. Democracy is the absolute worst form of government on the planet, except for all the others. We are essentially treating data on the internet or anywhere else. The same way we do politicians. We want the most popular result. Now over time, as we chain the models and we tell it no, one plus one is in 12, no matter how popular that answer is, we start getting better answers. But on untrained models, you can get a 12 as the answer because it's popular. People have asked about how do we get hallucinations in AI. That's your answer right there. If you ask it a strange enough question and it gives you a confluence of popular resulted answers combined into one thing, you get a weird result because it's popular.

12:24So that's the easiest method to say what's going on there. Okay. So moving forward, we've got all of these data centers being built. Who's doing it? Who's getting paid when they do it and who's doing the pay? Well, the companies that are investing in their cloud platforms like Microsoft and Amazon and alphabet are putting in data centers. Sometimes they are hiring that out. The little independence that are coming in and saying we'll provide data for you. But a lot of times it's their own stuff. So as a, for instance, meta is renting cloud storage and compute and utility usage and so on from Azure and AWS and Google. So are they in competition with each other? Well, first we have to figure out what they're doing. So let's kind of step back a second. What's going on out here? There are four big companies that are called the hyper scalars here.

13:25That's kind of an internal industry word in the economics world. Hyper scalar. What does that mean? They're taking their cash flow and turning it around and focusing it on internal investment. That's what capital expenditure is. Capital, as in capitalists, it's the same root word as capable. They're investing in capability for themselves. It might be infrastructure, more power. It might be more compute. It might be more ability to know who their customers are and what they want. All of that takes money. So who's doing it and how much? There's four big ones. The really big spenders right now are Microsoft, Alphabet, Meta and Amazon. They've each committed to a massive amount of expenditures this year, 2026. And to some extent have already done those expenditures. So Amazon said that in their last reported plan, that they were going to spend $220 billion on internal investment this year, 2026.

14:31Alphabet said between $195 and $205 billion. Meta between $130 and $145 billion. Microsoft is spending a lot, but they haven't really said how much of their free cash flow they're using. But they had 127.5 billion operating cash flow through the quarter three, that 127.5 billion cash flow. That's money coming in from their operations just through the third quarter of their fiscal year. That's not even the whole year. Wow. And Videa is not doing a whole bunch of capital expenditure. And Videa is in this list only kind of vaguely. It's not one of the four big capital expenditure folks. They're not. It's not a hyperscaler. Most of its cash is just coming in and sitting in its coffers. Massive, massive loads of cash coming in. They've purchased a large language model producer called Hugging Face recently. Because they just have a lot of cash and they don't know what to do with it.

15:32So where's the cash coming from? Where's all that money coming from? We're talking about cash flow. We're talking about use of this huge hundreds and hundreds and hundreds of billions of dollars. Put it all together. It's trillions of dollars being spent. Where's it coming from? Well, if you recall back to the Tax Cuts and Jobs Act, way back, back in ancient history like six years ago, there was a law that said, hey, we'll lower, it's more than six years ago, but not that far back. Well, we're going to lower corporate tax rates. Why are we going to do that? Well, because all these big companies are sitting on mounds of cash overseas and they don't want to bring it back for investment purposes because the tax rates too high. If they bring it back, they're going to pay way too much money in taxes. They're never going to do it. So we lowered the corporate tax rate and that revenue increased. That's something called the laffer point on a revenue curve when you lower taxes and increase revenue. It didn't happen when we lowered taxes for everybody else. Revenue dropped everywhere else.

16:32For corporations when we lowered taxes, revenue went up. That means the tax rate was too high. Good, good use of it. They brought their money back here. Where did that money come from to begin with? Well, they're free cash flow. They're imminently profitable companies, hugely profitable companies. Money comes flowing in. What do they do with it? Apple was sitting on huge amounts, Microsoft sitting on huge amounts, all the big companies sitting on these massive piles of money. Well, they brought a lot of it back to the United States after the taxes dropped. Well, then what are they going to do with it? Most of these companies don't pay dividends or if they do very tiny dividends, why don't they pay dividends? Because it costs tax money when you do that. People, a lot of people don't want to pay taxes every year on a stock that they own. They want those companies to reinvest it, but they're kind of out of loss for what to do with some money. And if you doubt that, think of the company Mehta. It used to be Facebook. That was the name of the company. Now it's Mehta. Why did they change it to Mehta? Well, because the plan was to go fully virtual.

17:35All of their platform would be through goggles and such. You would go through that and you'll be in this virtual world and conduct all your business online. They didn't do a huge amount of market testing on that. What they found out is that most people actually don't like to spend their whole day in heavy goggles. That yes, it's fun to play games that way sometimes. And maybe even watch a few videos, but otherwise you just feel gross all day. It didn't work, but they changed the name of their company around it. Why? What is this about? Because they had so much profit coming in from their advertising revenue. Huge amounts. So they needed to do something with the money. They decided to go here. They bought Oculus. They concentrated on VR and AR. Their smart glasses are out now. Because they've got this big expense that they put into it. This technology that they built around it. What do they do with it? How are they going to make a profit on it? Well, the shareholders aren't stomping their feet and screaming about the misuse of cash. Because the profits from the rest of the company have far more than made up for it.

18:40And then along comes AI. And everybody changed their look. Everybody looked at what they're doing and said, OK, what's going on? So Microsoft is spending 57% of its cash generation this year. 63% at Alphabet. 61% at Meta. Amazon. If you look at trailing earnings, they're at 102%. So they've used up all their cash flow. Does that mean they're going into debt? No. What do you mean no? When you're 102%. That means that you're spending more than you're making. Yeah, but they have huge cash reserves. That's what we're talking about. Is there debt or financing happening? Sure. Amazon's using some debt and leases while funding heavily on its operation. What is the debt in leasing? Well, if they're leasing computers and computer parts, or leasing places for the data centers, or they're borrowing money to create the data center, rather than spend it all up front out of cash flow. So it's a long term debt for short term returns.

19:40That's pretty normal in business when you're building a factory. Alphabet. They're financing activity. What does that mean? They're loaning not borrowing. They are turning around and saying, Hey, you want to buy a data center? You want to build a data center? We'll loan you the money. Meta on the other side has about almost $84 billion in long term debt. There's still massively positive on their net worth. We call that a book value in the stock market terms, but they've massively positive there. But they've got some debt. It makes sense that they would have debt. They're trying to extend the amount of money they have on hand for whatever they need to do. Microsoft has had debt for a long time. Not massive. Small amounts of debt. They usually do it as kind of a cash flow thing. Same sort of thing. If we can finance this for a longer term, it costs us less money up front and we can recoup that money year by year.

20:41An Nvidia is not going into debt. So they use cash and strategic investments to move up the stack. There's this rumor going around that Nvidia is somehow leveraged to its neck. It's borrowing everywhere. I don't know where that rumor came from. I keep hearing it. But I don't see the source for it. There's no data that says that. Nvidia is just making a lot of money because all of those other companies need to buy from Nvidia. They are the center point here. When talking about liquidity and debt, both can exist at the same time. For instance, Amazon, Amazon in total, long-term debt, $133 billion, but cash and marketable securities. Not talking about the value of the rest of the business, which has a value in itself, but just cash, short-term cash, and the marketable securities that they hold for that short-term. What is marketable security? Short-term US Treasuries. So it's cash equivalent.

21:43They have 123 billion sitting in cash versus 133 billion of long-term debt. If you had your entire mortgage balance or almost your entire mortgage balance sitting in cash, while you had a mortgage, most people would not say that's unhealthy. They would say, you've got enough to pay off your mortgage almost just in cash. Well, that's where Amazon is. Microsoft has nearly double the cash that it has long-term debt. Meta has significantly more cash, 90 billion in cash versus about 84 billion in long-term debt. Why am I bringing this up? Because I'm hearing this being compared to the .com bubble when all that speculation was going on. There's some speculation going on today. Don't get me wrong. Don't hear me saying, or acting like I've got rose-colored glasses, there's a lot of money being spent. Where it's coming from would have just sat there, though, if it wasn't being spent on this. So I see it as a definite positive.

22:43All right, so that debt sitting out there at the same time that cash flow is sitting out there, just recognize when you hear long-term debt and short-term cash flow, we're talking about very different things. Okay, so Meta said they would spend 137 billion or so, 130 to 145 billion. They've already spent 50 billion as a second quarter. Alpha bet said it was going to spend somewhere around 200 billion, somewhere between 195 and 205 billion. They've already spent 81 billion. Amazon has spent 118, almost 119 of its 220 billion dollar guide. So this isn't maybe they'll spend the money. They're doing it. And a lot of the chip manufacturers, they're saying that semiconductor inputs, the TSMC and Samsung and micron and SK high-nix and ASMR, these companies that are making chips, they're the same companies that have been making chips. This is not new.

23:44They're out making these chips, and they've recently in South Korea had a massive crash in their stock market because everybody said, well, maybe this is all going to dry up tomorrow and no money is going to come. Well, the numbers that I just told you about money being spent, that means that they've already spent the money, and it hasn't shown up on the books that these semiconductors yet, because they've rather reported the end of the quarter. But what they are reporting is that their earnings are way, way up, and their projected earnings are way, way up, but the stock market is still nervous about this. I have no problem with that nervousness, that allows us to buy things cheaply. Compute hardware, Nvidia, AMD, Broadcom. These are companies that are making processors, either video processors or optimizable processors that are good for AI thinking. And again, they got hit relatively recently because people said, oh, this is too much too fast, something's wrong. The money's being spent. The debt's not rising. These are really important factors.

24:45I am aware of potential danger in this entire industry. But what we have is a set of six sort of separate areas of the ecosystem that function separately from each other, and we're specializing in those areas in the United States, which is what we would expect to see. Instead of everybody trying to do everything themselves, we're specializing. So Nvidia is doing the hardware on the processing side, or AMD or Broadcom. The RAM and ROM chips are being made, and I know it's weird to think about a hard drive being a chip, but that's really what it's come down to these days. It's not a block anymore. So these things are being made out there. Next layer, where are the major AI players participating in this stack? So when we look at the United States, this is, I'm going to start here in the domestic market. We've got some players. I'm going to run across the big players people hear about.

25:45Nvidia, Microsoft, Alphabet, Amazon, Meta, OpenAI, and Thropic. Where are they looking? Well, Nvidia is the primary processor creator for the American AI ecosystem. Everybody else is sort of doing it as a background. Microsoft's making some processors, but they're not trying to compete with Nvidia. Alphabet's doing it. Amazon's doing it. Meta's doing it. It's weird to think that they're creating hardware, Meta. They are on a smaller scale. The semiconductor side of it, the chips, memory kind of stuff. Nvidia's a partner on that. They are investing in other chip manufacturers. But essentially, Microsoft, Alphabet, Amazon, and Meta are out of that business. An OpenAI and Anthropic aren't in any hardware. They're not going to get into that area. Okay. So Nvidia, what does Nvidia do? They're a hardware manufacturer. They do use the cloud. How do they use the cloud?

26:46Well, in all their design, they're using Microsoft and Alphabet and Amazon to help with their cloud compute side of things for all the designing that they're doing. Microsoft's not doing it for them. They're Alphabet and Amazon aren't in there saying, here's your new design. They're providing the platform. Microsoft provides Windows. And if you're doing it in the cloud, you're getting their compute power and everything else. You don't even have to have it for AI. Just storing stuff. So Nvidia is buying from them. And Microsoft and Alphabet and Amazon are all primarily in the cloud business. Amazon makes more money from its server rentals than providing goods for sale. Just think about that for a second because most of us think of the Amazon as the boxes and bags that show up at our door. And there's a lot of them. So if there's a lot of them, the server side is even more. They're a cloud company,

27:47just like Microsoft and Alphabet. Their platform is out there and the three of them are in direct competition on the cloud platform. Open AI and Anthropic, when you go to chat GPT or to cloud, you're actually going to Microsoft Alphabet and Amazon to get to chat GPT or cloud. So Open AI and Anthropic are paying Amazon, Alphabet and Microsoft for their cloud computing. Microsoft, Alphabet and Amazon are buying the large language model usage from Open AI and Anthropic. Does that sound incestuous? Well, no. Not when you consider a technology infrastructure. If Microsoft is creating an operating system and they're buying computers from Dell and Dell has Microsoft Windows on the computers that are running inside the Dell Corporation, is that a problem? No, it's not a problem. It's like if you buy a car and it has tires on it,

28:47did the car manufacturer make the tires? No? Cool. Well, the delivery vehicle that brings the tires was probably not made by the tire company. So this isn't some kind of weird, nasty relationship. This is how an economy works. You are building that stuff and I am building this stuff. There's another comparison and the earlier days of the internet, Dell and Compact were building computers and the whole idea was get these computers out there because the internet's here and everybody needs to be on the internet. So Dell and Compact are in direct competition with each other. But both of them were buying printers from Hewlett Packard until Compact and Hewlett Packard became the same company. And then Dell said, whoa, we're going to have to figure out how to make printers because I don't want to buy from my competition. Right now we don't have a lot of the competition being forced to buy from each other. That's not happening here. And that's one of the big dangers that I keep hearing people talk about.

29:48It's so weird and strange that Microsoft is buying Nvidia and Nvidia is buying Microsoft. Well, Nvidia needs an operating system and they need to run on the cloud and Microsoft needs processors to make their cloud work and Nvidia needs to use the cloud to design better processors. So this is just how an ecosystem is supposed to function. None of this is to say there's no danger in this market and none of it is to say go out and buy these companies. What I'm trying to establish is some baseline understanding that this isn't necessarily a bad idea. Now here's the next layer, the developer platform and the foundation models. So Nvidia's building foundation models, LLMs, so is alphabet and so is meta. Even though open AI and Anthropic are the primaries there, Microsoft is not. Microsoft's not trying to build large language models

30:50neither is Amazon. So what are they doing instead? Well, they're investing in other companies that are doing it for them. So Microsoft owns a big chunk of open AI and alphabet owns part of Anthropic. So you can see why they would want to focus more on those areas. But Microsoft also owns part of Anthropic. When you go on to co-pilot to use the platform, the developer platform itself, the enterprise software, or the consumer app at Microsoft co-pilot at alphabet with, you go through the list, these companies are using the foundation models from other companies, the developer platform, the software that you're using to get to those models is made by them. Co-pilot is Microsoft. Well, what are you getting out of Microsoft? Well, you're getting Claude and you're getting chat GPT. You may not even know which one you're getting in an e-given time because Microsoft is trying to find both the most effective and the least expensive method to get your answer. So they may switch back and forth between these two models.

31:52Is that a bad situation? No, it makes perfect sense. If you're selling cars and you have two different providers for tires, that's perfectly OK. If you own part of the tire companies, is that wrong? No, it's all good so far. So the consumer apps are coming out. And these are important. Nvidia is not going to have a consumer app. You're not going to go to Nvidia.com to ask it a question that AI is going to answer for you. That's not what Nvidia does. They just bought hugging face, which is a large language model creator and an open source version of it. And you'll probably go through them or they'll sell that to other platforms to get to it. The end result is probably not going to be that you log into chat GPT. Or you log into Claude. It's more likely that Claude is going to show up in the other places that you're doing business or asking questions. And that chat GPT will as well. At some point, meta wants to be the one that's providing that. So meta is developing as is alphabet

32:54their own large language models. Some of that is coming from open AI and entropic. But most of it is their own internal efforts to make their foundation models. Meta and alphabet are in advertising. Alphabet's got Claude as well. But their big revenue stream is advertising. Meta's in advertising. They've got 2 1-2 billion customers every month, unique customers. That's a massive chunk of the population of the planet. And they would really like to increase your time on their sites. So their AI's development is around, how do we keep them on here with their eyeballs attached to our screens so we can advertise to them for longer? That's their goal. Alphabet's saying, how do I get this continual stream of advertising revenue to me for showing our customers the best advertisement, the most likely thing that they will use to buy? Amazon, to some extent, is doing that as well. But they're using other models, other large language models to do it. They're renting those models or using those models

33:56from other people, other companies. So alphabet is the one that's trying to do the most here. They've got a cloud structure. They've got foundation models. They've got a developer platform and enterprise software. They've got apps going straight to the consumer and they're doing advertising where Amazon has most of that stuff, but they're also using their AI on logistics. Their ability to same day or next day deliver to your door is an AI thing. This is why FedEx and UPS are having trouble keeping up. If Amazon decides to shift over and say, hey, we're going to do shipping now. If you want to send stuff to people, here's how you do it. Just wait for your Amazon delivery and we'll pick it up. And if they increase, that's a potential thing out there that AI might kill UPS, it might kill FedEx and Amazon might take over. That's not a, that's not a for sure anything. But the writing is starting to be on the wall. Okay, so across the board,

34:56they're in different areas doing different things, but they're all in the same ecosystem. So who else is out there? Well, there are some smaller US players. By smaller, they're still really big. Two of these are sort of marrying together. GROC and XAI. GROC is building chips and hardware. They're building processors. GROC, why are they doing that? What is this about? Tesla's doing it too. And these companies are sort of mixed together, X and SpaceX and XAI and GROC and Tesla. Tesla's using GROC. They're building their own hardware because they need to be able to put that hardware in things that aren't standard computers, cars, spaceships, infrastructure, where they're actually putting the compute in the place or in the thing that's moving around to different area. Then you have hugging face. Well, they're a little bit into cloud computing,

35:57they're building models, they've got platforms. There's a couple of others I'm going to give you the names really quick, because they're not really worth spending a lot of time on when the big names are being named. Cochir, Miss Trial, AI, ALEF Alpha, Black Forest Labs. These are European folks as well. So ALEF Alpha, Miss Trial and Black Forest, they're European. They're not being left out of the game. Miss Trial's trying to kind of hold that full stack as well. They're trying to do all the things, but everybody else is partnering up. And now the Chinese players, because we hear about them. We hear about them a lot. While the Chinese open-weight models, and they're cheaper and they're using fill in the blank, be scared a little bit when you say it, and you can get, oh, this is going to be bad somehow. Where the Chinese differ from the European and United States models for business is that most of their companies are trying to do all of it. They're not going the partnership route,

36:59they're not ordering the chips from one company and the processors from another company and the cloud computing from another. They're trying to do it all. So Huawei. Huawei is trying to be the competitor for Nvidia. They're also trying to be the competitor with AWS and Google and AlphaBet and Microsoft for Cloud. They're trying to compete with Anthropic and OpenAI on Foundation models. Their developer platform is in competition with most of the major tech groups. This is true for Alibaba, for Baidu, Tencent, Bite Dance. Those companies, all of these companies are in, they're trying to create the technology stack from one end to the other themselves, which is why they're a little behind because they're not specializing. And if you think about this, if you're hiring data scientists to make your AI better, but you also have to hire hardware specialists to make better processors and you have to, so you get this massive ecosystem all in one company

37:59and that's the way most of the Chinese companies are building AI. They're saying we're gonna put it all in here. We're not gonna outsource any of it. So this is a massive competing experiment between the United States and Europe and China. China's saying each of our companies is kinda trying to build the whole stack because it's cheaper to do here. We already have the hardware. We already have the software. We're just gonna combine it all and do it. And it's making them have a slower output, but less expensive. So we'll see how this goes long-term. This is a big question and nobody knows the answer yet. What is the outcome of AI? Well, nobody knows that. There's a lot of potentials. There's wonderful, amazing things and horrible, dystopian things. And we're probably gonna get a mix of all of it. When we say what is the benefit of all of this to your standard business, how can AI help a small mom and pop shop do anything better?

39:03And I mentioned this at the beginning of the program. Inventory control is one of the largest time sinks for small businesses when they're actually selling physical products. Inventory control. Wasn't this something that we've had under control for a long time? I mean, Drayton McLean's company McLean developed this concept of real-time transport. You've got a barcode, you scan it, you're selling the toothpaste at the aisle and the computers are already ordering the next toothpaste tube to be delivered to stock the shelf with. You don't have to have a warehouse, you don't have to have a big inventory. This is the concept that drove big business into high profitability over the last 20 years. Now when supply chains got disrupted, we realized that just in time delivery is a little hard, but we still have that model built into the system. Where we can, that's the least expensive approach that hasn't benefited the mom and pops at all. Because getting all of that technology wired up to a logistical framework,

40:05how do you even do that? I mean, you've got a point of sale software that uses a credit card purchasing app. You can put that in there. And that sort of helps you control your inventory. The next layer is that AI is going to help them tremendously because it will find methods of integrating logistics and real-time purchases and smaller stores. And that has an immediate time saver. Well, I've got clients that absolutely dread inventory because they've just got to stop everything they're doing and get out of all their boxes and all the room. And you're doing all these countings. And usually you have employees doing it that don't want to be doing it. It's a very manual task. That's an immediate benefit. And that's part of the reason why even though we have higher unemployment for computer science folks, it's not as high as a lot of other sectors. Even though AI is replacing them at high, high speed,

41:06and if you talk to any major tech company, and the standard is how many and what large percentage of your programmers did you already lay off? Not did you lay off, but how much? And if in the middle of all of that, we still don't have massive unemployment in the computer programming world. It's higher than it was, but it's not massive. And you look at all these big layoffs. Where are they going? Where they're going immediately to the mall and past stores and putting together a cheap platform built on AI or built with AI, but hopefully with a lot of extra expert guidance to the AI to build the software for these smaller companies. So the tremendous productivity that we're seeing based on all of this is clear. What I am not seeing is huge borrowing. I'm not seeing leverage in this space. I'm not seeing people borrowing extra to invest more so that they can magnify their returns. Some of that's taking place out in the hedge fund world.

42:08But the massive investment isn't from institutional investors. The massive investment is internal on these big hyperscalers. They're taking cash that they didn't have a clue what to do with. That's just a concept that's mind-boggling to begin with. I mean, the whole concept of what do we do with this cash goes back 15 years. Prior to that tax cut, they've got all this cash. They can't bring it home because they'll pay taxes. So what do they do with it? Well, maybe we'll buy back some of our stock. We don't want to pay dividends. Let's invest in virtual reality. Oh, no, that didn't fan out. Nobody's upset because we're still making a lot of money. That concept of what to do with the cash so that it actually builds infrastructure and productivity into the future. We've got the cash flow happening now. It's moving into infrastructure. It's moving into utilities and cloud compute and a whole bunch of things that are imminently useful for the rest of us. The fact that our power grid is being expanded

43:10in the middle of this because of this money being spent. When over the last decade, that's been one of the largest infrastructure complaints across our citizenship. Where's the power? If the power grid goes out, where are we going to get it? What do we do? And those people are using power and we're using power and we've moved more and more to the model of the city runs the power plant. That's not how the grid was manufactured. The original grid was made by four profit companies trying to reach more customers. But over time, it just became all, it's for the greater good. We're going to have to use easements on property. So we might as well centralize these companies into government ownership, state run companies running the utility or municipal run companies. This new expansion is for profit again. That's likely to bring the price of power down drastically. It's really hard to believe that today. When power prices are going up and the energy prices

44:11are going up for all kinds of geopolitical reasons, we've got wars going on, we've got refineries blowing up everywhere. What's that mean? Well, it means higher energy prices, right? At the same time that we have a higher energy demand. But I'm looking ahead and saying, we're finding other alternatives for energy. We're finding more methods of creating the energy, not just alternatives, but hey, we'll put the natural gas right and plant right here next to the natural gas refining. Well, the power right here. So this expansion is phenomenal. And some of the best money that can be spent in an economy is in the ability to expand what's happening. Our roads need help. Our bridges need help. But a lot of what we're doing is happening in the virtual space. And I don't mean what goggles on. I mean, you've got your cloud storage and you've got invoices moving around. And a lot of that is virtual. The expansion that we're seeing right now, the fact that it's seeing the profitability that it's seeing at this moment, that's the unusual part.

45:15Usually when we have this big and infrastructure spend, it makes all the balance sheets negative for a while. Spending all that money. Yeah, it's a big expansion. But man, that's expensive. Well, we're doing it out of free cash flow for the vast majority of this. Okay. So I'm not trying to say that there's no bubble in AI. There's some stuff that's overpriced. There's some prices prices that are below where they should be. What I am saying is that the danger around these companies doing business with each other is the same danger in any manufacturing complex. It doesn't matter if it's Dell and Compact. It doesn't matter if it's AT&T buying energy products from general electric. Is that where they in competition with each other? Well, they both have wires, but they're not really in competition. I hope kind of laying this out. What is Amazon trying to do with AI? Well, they're doing something totally different than Microsoft.

46:16What's Meta gonna do? Well, Meta's just released its muse AI. And this is one of the things that they've been working toward. They want you to stay on their platforms and get advertised to. Google wants the same thing. So if there's companies that are in competition with each other, Alphabet's in competition with Meta, Microsoft's in competition with Amazon and to some extent Alphabet. Where's Apple in this? Well, Apple is coming to the game a little bit slow. They're looking around and saying, we don't need to spend the most expensive money up front. The bleeding edge is always the most expensive. Let's look around and see what models are working best and let's move forward. Is that the best decision? It's a decision. We don't know what the best decision is yet. We'll find out later. We know that Microsoft is making money hand over a fist on its AI investment already. Its income is increasing hugely. Does that mean it's making good decisions? Time will tell. What I can say is that a lot of the rumors about the super leverage

47:18buying of each other and the speculative investing and so on, they're bunk. They're not true. Open AI and Anthropic names that we hear all the time. And when someone hears AI, they think chat GPT almost always the amount of money being spent by open AI this year is somewhere around $30 billion. Some of that's with borrowed money. They made $23. So $30 billion, that's peanuts compared to just one of these hyper scalers. That's like 10% of what Amazon's spending. Alphabet spending. That's tiny. But when we hear open AI borrowing money, people tend to get scared. They're not profitable yet. Why are they borrowing money? This is a more standard startup. More standard startups don't start profitably. The fact that most of the money being spent on this stuff is coming from profits. Is the big good news that I'm trying to say here. Does that mean it's all going to work?

48:18Does it mean that it's not going to be downturns? No. But it means that we are nowhere near as shaky as we were in 1999 during the dot com lead up. What was different then? Most of the dot coms. The vast majority of them. We're in the same boat as open AI is today. They're trying to spend. They're trying to borrow. They're trying to get to profitability. Facebook was not profitable when it had its IPO. And I realized this is mildly after the dot com bubble. We're talking 2007 and 2006 and eight and come forward from there. But they weren't profitable when they IPO'd. Amazon lost money for years. The fact that this expansion is coming from the coffers of bigger companies kind of rhymes more with some of the early industrial revolution where it wasn't speculative investments. It was people changing their factory models. And that's what I'm seeing across Microsoft, Alphabet, Amazon and Meta. They're changing their factory models.

49:19What they build and what they're focused on. The profitability is what's leading them there. I think I have given you at least a mildly clearer picture of this crazy convoluted mess of what's going on out in the big, big wide world. I'm not sure I removed enough confusion. But enough so that you can see the dangerous web that people are talking about here is a normal new industrial expansion. A lot of this is hardware that's not going to go away if the company that produced it does. The data centers will be there even if the company that made it goes away. Now, there's a lot of people would say, oh, that's not good. Data centers are bad. Yes, but they don't want their Google pictures to go away. And they don't want to go back to getting paper mail for all of their bills. So how is that being stored by Citibank or by Chase or any of the rest? Well, they're using Amazon or they're using Microsoft or they're using Alphabet.

50:22So those data centers are what makes it so easy to pay your credit card bill or or not get paper in the mail about all the other things that are going on. We have been complaining for generations about our lack of industrial capability. This is the new industry. It's not the only new industry. There's a lot of other things to be made, but we're making processors and chips and vehicles and all of that's being wrapped into this. We're also making the utilities the power and to some extent, improving the roads to get from one spot to another. And that's my wrap up for this week. There's a lot going on in the world. I'm probably going to need to touch on again on Ukraine and Iran and talk about what's happening there and the outlook for oil and so on. That'll be another episode. This one I think has been building and needing to be out there. I hope you enjoyed it as much as I did putting it together. Until next week, this has been the personal wealth coach and Jake McClure signing off.

51:36The personal wealth coach is an SEC registered investment advisory firm. Just because we're registered with the SEC doesn't mean that the SEC endorses us in any way. They don't do that. They're the regulators. And just because we're registered to give investment advice doesn't mean that that's what's going to happen on the podcast. In fact, we won't be giving any investment advice on the podcast that needs to be given to people that we know thoroughly. And there's a number of listeners to this podcast who I've never met in person, though I wish I could. And I'm going to be quoting a whole bunch of information from different sources all over the planet. We're getting that information from sources. We deem to be reliable, but we're not warranting, guaranteeing or any kind of team. How accurate that data is. And finally, our contact information. If you need to get ahold of us for whatever reason, our phone number is 2549471111 or 1-800-9147526.

52:42You can also go to our webpage, tpwc.com or thepersonal wealthcoach.com.