원문 보기
Katie Wood just posted one of the most optimistic videos I've ever seen. This is her final episode of “In the Know,” and I have to say, I’ve never seen Katie Wood more determined to grow. After watching it, I literally thought that it's entirely possible that GDP will grow by 6% per year, or maybe we'll see Elon Musk's growth levels—10 to 15% per year.
I want to break down some of the main components of her video and see where our opinions diverge, because , look, I can be optimistic right now. We have our little scale: bear, bull. Just so you know, I rate myself at 7.6 out of 10. That's really how I want to buy and the pace at which I want to build a 10-year portfolio, right?
And for me, I'm in the optimistic camp, but that doesn't mean I want to invest every last penny in stocks because I'm so confident right now. I was left with the thought that this was a pessimistic view . That's how optimistic Katie was. And some of it was even pessimistic for Treasury bonds, despite the optimism about the market.
So I'm trying to reconcile all of that and I'll do my best to do it. She compares it to the Industrial Revolution and says: " Look, GDP was 0.6% then, artificial intelligence is much better, and we will show a much better result, somewhere 5 times or more." So she thinks our current growth could easily double to 6%, and she thinks the IMF will greatly underestimate not only global growth but especially U.S.
growth. She also argues, as we've heard for a long time, that inflation could surprise downward, perhaps even turn negative. And, frankly , I don't disagree with her opinion that disinflation is near. There are two ways in which disinflation comes. The first is technology, to be honest . The second is the lack of printing money.
And there's probably a third option: a recession, when people get scared, or a depression, right? So there are three main forms of deflation that you get. Warsh, you know, he's kind of against printing money. That's why I think the next time there's a crash—we don't know when—it's going to be a kind of protracted crash.
This will be a little unpleasant because he is not going to print money. So this will help contain deflation. I really think there is technological deflation happening, and that is a good thing. And I hope we don't have a collapse, right? Of course, you don't want to invest now thinking that the markets or GDP will double, and go all out on the margin , and then we'll get into a recession.
Nobody wants to hear this. And that's not what I'm talking about. I just want to see what assumptions are actually needed to arrive at these GDP growth rates. So I'll move on to the key component of her video, which came right at the end. Which, I have to say, let Katie's team down a bit, because we started with GDP doubling, and 50 minutes later we heard, "Here's the rationale for why we think the return on investment in AI today is very good and much better than the lack of return we had during the railroad boom." That the railroad boom was largely built on hopes.
So, her argument that today's development of artificial intelligence is not based on hope is a reference to Elon Musk's deal with Anthropic for $50 billion per gigawatt, although the exact numbers are a deal for $40 billion, which is also not bad. She tried to remember it as she went. No problems. This is a $40 billion deal, which is $1.25 billion per month until May 2029 with a 90-day cancellation clause.
And that was pretty much it in terms of real data, well, we got a little bit of information about scaling and jobs. We'll sort that out in a moment, which disappointed me too , but it's okay. Let's try to understand this. Let's focus specifically on this deal with Anthropic. In Anthropic's deal with OpenAI, or, sorry, with Elon Musk, Elon Musk probably overestimated the power that xAI would need for Grok.
This put Elon in a very advantageous position because Anthropic is going to show the highest growth rates, which is quite logical, you know, it's completely logical. They really want to accelerate their growth right now, and they can't do it perfectly without emergency computing power as they prepare for an IPO.
So now they really have an incentive to pay any premium to get computing power right now, even if it's three to four times more than the terms under which industry agreements are usually signed. Well, even, you know, two and a half times, right? It's still more, you know? $40 billion per gigawatt, sometimes written as millions per megawatt, it's the same thing.
Well, at the same time, we're seeing new long-term deals being signed with companies like Nebius for closer to $15-16 billion per gigawatt. This shows that Elon had a real advantage, effectively selling water in the desert right before Anthropic's IPO. This is very useful for Elon to indicate future spending on artificial intelligence, but I mean computing power and Nvidia chip purchases, which, frankly, plays into Nvidia's hands.
This is a drain on funds , taking money out of the economy after the IPO to then spend it on Jensen's equipment. Who doesn't want to have Jensen equipment right now? I don't know if that's sustainable enough to justify doubling GDP. Because you're actually talking about laboratory-to- laboratory transmission, right?
XAI: " No, we have too much computing power here. Use it ." This is sufficient reason to argue that GDP can truly grow rapidly. Of course, there are other components here, but I think this was one of the weakest arguments we could have made. And to claim that the railways were built on the hopes and prayers and enthusiasm of Elon Musk.
I don't know, to me these things seem almost similar. Okay, having said that, let's move on to other facts, because another component that she mentions is, "Hey guys, you know what's probably going to happen here? We're probably going to see a hiring boom because AI is so useful." And Ramp conducted a study where companies using artificial intelligence are simply experiencing a hiring boom, and corporate profits are growing rapidly.
Okay, we need to dissect both of these statements because they both miss too many details. So, first of all, yes, if you look at the year- over-year EPS growth rates for these companies, just for tech companies or S&P 500 companies, that's actually an ex- tech metric. We have a pretty solid annual EPS growth rate , even excluding technology .
We are growing, by how much? The rate is now around 23%. Profits last year increased by almost 15%. This is phenomenal, and this is outside the technology industry, right? So does this prove Katie right? Well, that's great, right? And if we look at the S&P 500 as a whole, the latest quarter indicates that annual earnings growth could be in the region of 44%.
The problem is , you can't just look at the S&P 500 and say, "Hey, how much is the EPS growth of tech companies compared to without them?" We really need to look and say, "What is the real operating profit growth rate?" Now you might ask: Kevin, why do we care about operating profit? Isn't earnings per share enough?
I mean, the economy is on the rise. Why are you trying to downplay this? I'm not trying to downplay that. I'm just trying to be realistic about the level of GDP growth we can expect, because if I'm wrong, I should be more optimistic. But if Katie is wrong, I'm going all in on margin and increasing my leverage on stocks, because I just came out of that video thinking, " Let's go, baby." Industrial, er, freedom, industrial revolution 2.0.
Of course? Then, then I'll be in trouble if I get "George Damon" and the recession that's already on the doorstep will just eat my lunch, okay? I will feel very bad. I don't want to get " George Damon." I want to get "Kathy Wood". You know, I want to see big green growth. But I know, I think the truth is somewhere in the middle, and I love both of these people.
I think they are wonderful people. I'm just saying that we need to find a middle ground here. That's the thing. You need to look at the operating income of these companies because after operating income you get what is called capital gains, and that's what affects earnings per share. So, many companies have been influenced by SpaceX.
Of course, Google had it. I mean, we're talking about $90 billion in SpaceX exposure, and at its peak, it was over $100 billion. So, earnings per share increased significantly because of the increase in the cost of equity . But it's a vicious circle, because if the stock market falls, it becomes a negative factor for corporate balance sheets.
That's why I like to look at the growth of companies' operating income, because it gives you the phantom wealth effect . So let's call this the phantom wealth effect. If we just remove, literally just remove Google and Amazon from the S&P 500, we're just removing their growth rates. We don't take all their profits.
If we just remove these two companies, what would be the growth rate for the S&P 500? Well, we would literally go from that 48% that you saw to 28%. So, simply by removing the growth of Amazon and Google, the S&P 500's EPS growth rate would fall by almost half. This is important because it shows that these two companies, which have made huge profits, have helped to distort the statistics of the S&P 500 companies.
And so you can use another simple tool to check: look at revenue. Just look at the revenue of S&P 500 companies . The annual revenue growth of S&P 500 companies is not 50%. Not even 28%, right ? Because we have profits from stocks in all other companies too. Sure , margins may be up a bit , but revenue growth across the S&P 500 is only 11.9%.
It's still not bad. We're still in the heyday, baby. I sincerely believe that we are still in a heyday. But is it as powerful as Katie says? I'm a little more skeptical about this. I said I wanted to postpone this because there was something else I wanted to talk about—the phantom wealth effect. One of Katie's main arguments is that—she mentioned several times that low- income consumers are in trouble.
But she showed a chart that suggested, "Hey, you know, debt-to- equity ratios are at historic lows right now , and they're supporting people who are spending money and have stocks." And people don't have that much debt. You see, debt is falling because debt-to-equity ratios are very low. But wait a minute.
Low debt-to-equity ratios don't mean that people have suddenly paid off a bunch of debt. Let's say I have $100 in debt and my stock investments are $200 . In that case, my debt-to- equity ratio is about 0.5. Or, to put it another way, my equity to debt ratio is two, correct? I have $2 of equity for every dollar of debt.
But what happens if all of this is suddenly caused by a stock market surge? I still owe $100. But my stock is now worth $500 with the same $100 in debt. Yes, my debt ratio looks incredibly low. But what happens during the correction? What happens if suddenly Anthropic's IPO, well, doesn't work out, they break the contract with Elon Musk, and you suddenly see that orders for Nvidia chips have decreased significantly.
Well, then the stock market, which is rapidly falling, will start to bankrupt you with huge debts. So I think it's risky to use the equity-to-debt ratio, like a stock market-based graph of global wealth in relation to people's debt. In fact, I think a better tool would be to look at what's happening with people's marginal debt.
And if you look at what happened in July 2025, we reached over a trillion dollars in margin debt for the first time in a month. In June 2026, this amount increased to $1.5 trillion, right before the Leo fold brought it down to $1.4 trillion. This is just margin debt. I bet it will surpass that mark again in August.
We will receive this data in a few weeks, and I would not be surprised if this balance turns out to be even higher than in June. Isn't that right? So the debt is growing. It is simply supposedly being offset by the growth rate of the stock market. S&P equal weight shows incredible results. And I'm not here to play for relegation.
I'm just saying that I have... you know, do I believe that GDP will grow to 10%? I mean, man, if everyone needs to build 10 gigawatts of power, and we have to build it in space because we need so much AI, fine. But I actually think there's another red flag associated with artificial intelligence. You see, Meta had to lay off a bunch of people in two waves of layoffs .
But in May, the night before they were going to announce more layoffs—although there were some cuts—they effectively reversed the decision because their planned operational transformation at Mark Zuckerberg's estate in Hawaii didn't go as well as expected, because they realized, "Look, agent-based AI is cool , and it's useful for a lot of things, but it can't completely replace humans yet." So, on the one hand, the idea that agent-based AI is so capable that we can replace a bunch of people and therefore we need just as many new data centers is still a bit far from reality.
On the other hand, this is not a call for an economic recession with high unemployment. I'm just saying, "If you have huge companies with billion-dollar capitalizations and unlimited budgets, and they tell you, 'AI is good, but it's not yet at the level where we can really transform the business by laying off a lot more people than we planned,'" There were scenarios where Meta assumed it could lay off up to 60% of its staff because the AI was supposed to become so incredibly effective.
But he's just not there yet, is he ? I think the vast majority of us look at it and say, "Yeah, listen, this is really cool as an encyclopedic AI." We could, you know, finally work on our codebases because of memory constraints. We could do a lot of productive things. There will be many companies that will benefit from this." But has it reached a level where we can just fire everyone?
No. I mean, that would obviously create its own problems and unemployment, but to argue for a 6–10% GDP growth, you need to see profits skyrocketing everywhere, not just on the back of a market boom. And I'm not convinced we're seeing it yet. On the other hand, Katie argues that productivity is so high that people are actually starting to hire more workers, and she cites the Ramp study.
I was already sorting this out when she mentioned this study last month. She must not have watched my video , because when she referenced it last month, I said, “Kathy, no. This is a mistake. You can't claim that everyone is hiring through AI, because let's look at the facts of the Ramp study. I'll just give you the main conclusions, because I've already made a whole video about it .
So I'll just refresh your memory, okay? Ramp actually conducted a study among 21,000 firms. First, Ramp has a much higher percentage of startups. Therefore, there are people who are more willing to use a fintech bank than a traditional one. They are more likely to implement AI . Of course. Okay. The problem is that they noticed: companies that intensively implemented AI increased their total headcount by 10.2% 24 months after the start of active use.
But we completely ignore the fact that these could literally be AI companies. This does not mean that AI is driving hiring . Isn't that right ? So here we have some causality without correlation. And they also tell you that, well, hey, a lot of the hiring, by the way, right here, is actually for entry-level positions.
This was unexpected. 12% increase in staff for active users, but again, only for those who actively implement technology, that is, for companies focused on innovation. But they noted that they did not see any statistically significant staff reductions in any of the regions studied. So they didn't see any layoffs due to artificial intelligence.
That's good. But the big problem is that job growth was observed only in the IT sector. Again, there are more AI-focused companies out there. So this doesn't mean that AI is creating jobs everywhere. It's just that startups and businesses that actively use AI are hiring people. Well, of course, because they are probably new companies.
They literally need to hire engineers, support staff, and financiers because they may not have any at the start. In addition, this study ignored companies that went bankrupt. It was necessary to survive for 5 years during this period. So the idea that "AI doesn't kill jobs" ...Of course, but all the companies that didn't make it to the end of the 5- year study period were automatically excluded.
So all the firms that went bankrupt due to AI were simply removed from the results. But seriously? Such research cannot be cited. Well, you know, if someone came up to me and... I don't know. Maybe I'm being too categorical. But if someone said to me, "Kevin, companies selling AI products and services are hiring more people right now." I would answer: "What are you, really?" You're not even a company if you're not hiring anyone right now .
Yes, we also recently made a new hire. Moreover, we are expanding our team of programmers by about 10-20%, we are working on this, but we are also an AI company that has support in the real estate sector. But this video is not about my startup. This video is about Katie Wood's comments, so let's focus on them.
OK. Understanding some of her theses, I was surprised by what she claimed: one of the main drivers of GDP growth will be not only higher growth, thanks to some of the things she mentioned, but also lower inflationary expectations. In saying this, she suggested that the yield on 10- year Treasury bonds could continue to rise.
But if you advocate deflation and then say that 10-year bonds are going to rise, you have to have complete faith that this economy is incredibly thriving . Here's how this math would work. I printed this out to make it a little easier. OK? How 10-year Treasury bonds can stay high. Let's say Katie Wood believes that the yield on 10- year bonds could reach 7%.
How is that possible if our inflation expectations are potentially falling, right? Well, here's an example. If inflation expectations are now, say, 3 %—no, they're around 2.75% , but let's take 3% for example. Let's assume GDP growth of 5%, which is significantly higher than the approximately 2.3% we will likely end this year with in America.
So, that's twice as much. Let's say 5% GDP growth plus 2% inflation expectations. Here's how you can justify 7%. Many will say, "Hey , 10-year bonds don't really work that way ." And this is a subject for discussion. This doesn't work so clearly. Nobody, when buying a 10- year bond, says, "By the way, you get this much compensation for inflation and this much for growth." But just so you understand the logic of why this might work: people who buy bonds are saying, "Look, if I can make 7 % somewhere else because the S&P 500 is growing revenue at 11.9 %—not 44% on EPS, which is a misleading metric.
Then I need compensation for buying these bonds." In practice, this is how it works : people simply don't buy bonds until the yield is high enough. Then they buy these bonds. Good. Another thing she claims, which at this point is obviously speculative, to say the least, but I agree with her in the long run.
She believes oil prices could potentially fall to $30 a barrel as the UAE ramps up production. She says that oil production in the UAE has already increased by 78%. I have not been able to confirm these statistics. I only see growth of about 25%. I think she's comparing it to the daily minimum in some kind of "abyss", which is also acceptable.
In that case, that would be true. I couldn't confirm this. Yes, oil production in the UAE has increased, and we expect Saudi Arabia to increase it even more. You know, when will this war in Iran end, assuming it does n't last forever, which I, for the love of all that is holy, hope so. I agree with her.
Inflation may go down, oil prices too, all of this is possible. I completely agree with her. And that actually supports GDP. But do you know how much it will support GDP if there is a downturn in AI? This won't help. If you need a danger indicator, watch what happens with the Anthropic IPO. This is, so to speak, my indicator, if you will.
Of course? This indicator is singing quite loudly right now. I think, “Ah, yes, we can buy on the dip. We can buy on the dip. " Okay, cool." But if the indicator goes silent and Anthropic bends, the mine will become unstable. Run away. Right? This will be one of the early indicators. I hope they don't " bend their fins." She also acknowledged that gold peaked around the time Varsha was elected.
I have personally been making this argument for many, many months now. I thought Varsh was a signal that we would not print money, which again leaves us in a state of L- shaped recovery for a long time. Bad during a recession? It doesn't matter. I'm not here to campaign for money printing. I just want to say: if we find ourselves in a deep hole, then with Varsha we will be in it much longer.
That's all. The main conclusion. I do n't argue with her about the gold. And Katie Wood...she sounded moderately optimistic about a potential Bitcoin breakout, which would be great. You know, Bitcoin has been "bottoming out" for a long time, and I hope the entire stock market and risk assets will move along with it.
She sees Bitcoin as partly a risky asset and partly a hedge against inflation. I see it mostly as a risk asset. And I personally think that from now until my birthday, we will show growth, and the market will feel very good, because Anthropic is not the kind of indicator that will die on the day of its IPO.
If he dies, it will be sometime in the 27th or 28th year. And she may not die. She can actually go through this period of programmatic adjustment. So to fully understand this, you should just watch another video I made. So, if you search for “Meet Kevin Most Dangerous Chart in AI,” that will completely explain what an “ Anthropic canary” is.
Here's what the preview of this video looks like. I highly recommend you watch it. This is very, very good...Oh, it just started playing the video. That's it. I highly recommend you watch it. This is really very good material. OK. So, having said that, I like her enthusiasm. I truly believe that some of the key components of her predictions of such rapid GDP growth are not well -founded.
I think more could be done to confirm the " bullish" thesis. I understand there were a lot of graphs about the velocity of money, comparing that velocity to labor force participation rates, debt-to-GDP levels that have n't peaked yet, and so on. Well, okay. I looked at every chart. I'm just expressing my reaction to the key components.
I'm not going to argue about every single chart because I honestly believe that the others are not as important as the main points we've discussed here. So, with that in mind, watch this video. Check this out: " The Most Dangerous Graph in AI." And look at that spectacular frame we got. Oh my God. It's much bigger than I expected. .
Comments 0