Ray Dalio, Bubble Fears, & Nvidia’s Beat 11/20/25

0:00Bring in show music please. This is Squawk Pod and I'm CNBC producer Cameron Costa. On today's episode, NVIDIA has reported its third quarter results. The biggest report of earnings season, it did not disappoint. Blowing out expectations, calming investor jitters and boosting the broader markets. It's a very big deal for NVIDIA. It means big deal not just for NVIDIA, but also for the major averages. Then one big interview with Ray Dalio. Bubbles don't happen because of good estimates of what's in the future. It happens because of the need for cash. The famed hedge fund manager on AI hype and economic bubbles.

0:44We are in that bubble territory, but we don't have the pricking of the bubble yet. A markets lesson from the legendary financier, plus the red flags that he looks for in our economy. If you're looking at the private equity markets, they have a bunch of problems. It's Thursday, November 20th, and SquawkPod begins right now. Stand Becky by in 3, 2, 1, cue it please. Good morning, everybody. Welcome to SquawkBox right here on CNBC. We are live from the Nasdaq market site in Times Square. I'm Becky Quick along with Andrew Ross Sorkin. Joe is out today. The biggest report of earnings season, it did not disappoint.

1:26The AI bulls are winning. Christina Partinellos joins us right now at the table with more on NVIDIA's third quarter results. And maybe some more, the idea of the bubble can continue to go a lot bigger. Yeah, and that was one of the first issues that Jensen addressed on the call. But NVIDIA just showed exactly to your point how to calm those AI bubble fears. So much so that Cantor Fitzgerald is now saying AI is actually under-hyped. Yes, they wrote that in their note. The CFO on the call said the$500 billion border book, Jensen Wong, announced back in October at GTC will, quote, grow. Saudi Arabia and Anthropic both added a net new demand, so that's not included in that original$500 billion that was announced.

2:11Blackwell sales are, quote, off the charts and exceeding supply. the Q4 revenue guide of$65 billion be even high buy-side whisper numbers. CEO Jensen Wong really pushed back directly on spending concerns right now. He said from NVIDIA's vantage point, they see something very different. Agentec AI is growing rapidly. And then he went after depreciation fears, perhaps amplified by Michael Burry recently, by saying NVIDIA's older chips don't become obsolete. Chips they shipped six years ago are still running at full capacity, according to him. Even last generation chips are still selling strong in their third year.

2:47That line sent many of the NeoCloud players higher, Iron, CoreWeave, Nebius, you can see still up dramatically in pre-market trading. The$100 billion OpenAI investment, though, has raised some questions about whether NVIDIA is essentially funding its own chip sales, you know, that circular financing that's going on. Well, Jensen Wong defended that as well, saying they're not giving up equity in NVIDIA, they're getting equity in OpenAI. Here's his comments. Rather than giving up a share of our company, we get a share of their company. And we invested in them in one of the most consequential, once in a generation company.

3:24I was going through the 10Q last night, and it did show that NVIDIA hasn't committed a specific amount to OpenAI and that the deal isn't finalized. Perhaps it's just lawyer speak to, you know, dot all the I's and the T's. But for now, it's not a done deal. The stock, though, as you can see, trading much higher for now. NVIDIA has answered those skeptics to your point, Andrew. It is shocking when you look at the numbers. I was just digging through some of the stuff yesterday. We know it's big. We know it's growing fast. But there was a revenue chart that was in one of the stories I checked out.

3:56It was CNBC.com story. And if you look at their revenue on a quarterly basis, I think we've got the chart in the back, too. the growth that they saw in revenue just from quarter over quarter is more than they were seeing on a quarterly basis 10 quarters ago. Yeah, two years ago, two and a half years ago. But isn't it incredible, too, that the magnitude of the beat is expected to be$2 billion? So no matter what all of the estimates on either sell size or buy size, it was minimum$2 billion. And then, OK, for the guy, they beat by$3 billion, so even better. So that is, how can they keep managing that magnitude of beats?

4:29And that is always an underlying concern now in today's market. But it's incredible how far we've come and how much they've grown. So the thing that I want to go back to your open AI piece and just the equity piece. What I can understand about that is he doesn't consider it sort of vendor financing in that way simply because he's getting the equity for it. But that has to assume that the equity has an extraordinary amount of value. Exactly. Exactly. Exactly. And that$100 billion is, dare I say, an arbitrary number at this point because they don't actually plan on fully committing that full amount.

5:03Right. It's in tranches of$10 billion a pop. Yeah, according to the gigawatts that OpenAI produces. But he would defend it saying that all of this is real demand. He's not going to agree to vendor financing, much like Corweave's CEO has said that this is real demand. And he actually said that on his earnings call and saying that they've sold out of A100s, H100s. And even the spot price for many of these rental GPUs has continued to climb despite them being a little bit older, which fulfills the narrative that people still want the old stuff. So, OK, speak to this because we talked a lot about this earlier in the week.

5:34This just the depreciation schedule for these chips, because there's sort of a wide variance. Some people think that these things only really it's not they only work for three or four years. They can work for a longer time, but they only have great value for some for some people. they may have to be replaced every three or four years because people are going to want the highest end stuff. And then the question is, what kind of value can you get out of the stuff later? There's other hyperscalers that are now putting it on their books and saying they can depreciate over six or seven years. Right.

6:02And that's a huge distinction. So Core Reef has six years, but it didn't used to be six years. Nebius is four years. Microsoft CEO recently in a podcast spoke to that, saying that how he can now take older chips and layer them on with newer chips. and then you just, you know, you connect them with NVLink or whatever product you want to use. And that maybe speaks to using these chips for a little bit longer. On the earnings call, he said that there is still a useful life. The depreciation problem is not going away because you can either look at it as an accounting problem, how are you going to take it off your books over the X number of years, or an actual useful life problem.

6:37And the argument from the CEO of NVIDIA, Corweave, is that the useful life is very long. But it's the question of how are the accounting? No, no, no. I'm of the position, by the way, the useful life is not the issue. Accounting is the issue. Not even. I think the question is, unlike building railroads where you have tracks and you can continue to use the tracks. Yes, maybe every 10 or 20 or 30 years you have to replace the tracks. You might, because of just the way this business is moving, you might, especially if you're in the large language model business or something else, you might need to constantly be upgrading.

7:12It's the same reason that I'm getting a new phone every two or three years. The phone still works six or seven years later. I just don't want the phone six or seven years later because it doesn't work for me. It's not the latest and greatest. But it's not just the latest. But if you're in the business of needing the most high-end technology, which a lot of these guys may need. But then isn't that a great business model for NVIDIA? Fabulous for NVIDIA. But bad for everyone else? Bad for everybody else and unclear whether then the math will make sense for them. But you keep buying it. Everyone keeps buying the phone, so the math is still making sense for you and many others out there.

7:43As long as you have buyers. No, so I'm, look, it's fabulous for NVIDIA, but if you're meta, or take meta out of it, if you're OpenAI or Anthropic or a company that's not profitable yet, yes, then you actually will have a math problem. Then that's how NVIDIA is trying to go up the stack and get everyone locked in on every single angle, not just the chips. Now they're going after Anthropic. Why are they doing that? when Anthropoc right now may not be the most valuable customer because that's where they see the next leg of customers going forward. But you are a buyer still. So to your point, even years later, and perhaps maybe you won't be the best demographic in terms of salary and all that, but people are still buying the next greatest Nike dunks, your shoes, upgrading their car, right?

8:30There's still a market for it despite how expensive it could be. Let's talk a little bit about what this gain in the market this morning means. I can see your eyes. I'm not totally sold. No, I know. No, it's more that I think it could put a lot of the big hyperscalers into trouble. Of course. And then if the hyperscalers are in trouble, you don't have a buyer. That's different than being a little iPhone. And that's why the hyperscalers are moving closer and closer to pushing out their own chips. And Google is the best example with Gemini 3 on the TPUs. These are the individual chips that they created themselves.

9:03It's not trained on NVIDIA chips for the first time. And that's a big deal. Amazon, too. Yeah. The Trinium chips, there's some debate on the quality for all of that. But it's all about the patience of investors right now and the patience of companies. How much can they keep spending until they realize, OK, these chips are costing too much, especially on an annual cadence. If Jensen-Wang continues with putting out these chips every single year, to your point. Which then gets to the next piece, which we haven't talked about at all. So we're talking about this is great from video right at this moment.

9:33At some point, whether you think Amazon's going to get their act together and make those chips to be the quality you think they should be, or, by the way, OpenAI wants to design its own chips, there will be huge competition at some point. And so they may get great value on the training end, but not on the inference end, for example, or something like that. And then NVIDIA will say, well, they don't have the CUDA software, which is exactly what he said to the depreciation argument, that the CUDA software that they provide that has the most amount of developers right now, right now, is their winning recipe almost at this point, because you need it to revamp all of the chips, etc.

10:12I just want to talk very quickly about what this means for the markets overall. The stock is up by just under 5%, 4.9%. That may not sound like a lot, but when you are talking about a company with more than$4.5 trillion in market cap, 5 % gain for them, which is where the stock is right now, is about$225 billion. That's more than the equivalent of a Disney market cap at$186 billion. It's close to the market cap of Goldman Sachs at$235 billion. So a 5 % move in the market is a very big deal for NVIDIA. It means big deal not just for NVIDIA, but also for the major averages. It was a big deal when that stock was down by about 4 % in the first couple of days of the week leading into this.

10:53There were a lot of questions and concerns about whether they were going to be able to deliver. They did. The market is up substantially, and that is going to have implications throughout the rest of the major averages.

11:05President Trump wading into the debate over AI regulation and whether the U.S. should have a federal standard. Emily Wilkins joins us right now. She has more on this story. And Emily, this has been complicated. It's not when they were able to pass into law, but this does change things. Becky, I mean, this is it's been a complicated issue. It's been one that D.C. is trying to work on now for a couple of months. They failed this summer, but there is a huge momentum under an effort to try again. And look, the A.I. industry, they would be notching a major win if the White House and possibly even Congress are able to somehow block states from enacting an assortment of A.I.

11:41laws in favor of that single national standard. Of course, Trump himself came out on Truth Social calling for that national standard to go in place. And that prompted Mark Andreessen to tweet that a 50 state patchwork is a startup killer. He said that federal AI legislation is essential. There's no bigger issue for little tech, the builders who create the future for America. Now, that White House executive order that they're working on, it's not finalized yet, but a draft seen by CNBC would create a task force led by Attorney General Pam Bondi that would challenge some of these state AI laws under the grounds that they are regulating interstate commerce, and that's something the federal government has jurisdiction over.

12:21The draft executive order would also bar states from getting certain federal grants if they had AI laws that were not in compliance. Meanwhile, on the other side of Pennsylvania Avenue, Republicans in Congress, they're planning to use one of the upcoming must-pass bills, either on defense or funding the government, to limit how states regulate AI. They're looking at things like maybe allowing states to regulate police fraud, safety, but maybe stuff that's not AI development. However, any measure is going to require Democratic support, and that is going to be an uphill battle for them. Senator Brian Schatz tweeted that any sort of ban on state AI laws was going to be a poison pill, adding that it's not like we are doing regulation and preempting states, we are preempting states and doing nothing.

13:09We'll be watching to see what if any final policy does come out of either the White House or Congress, and then to see if there are any states that are going to try to be pushing back on their ability to regulate AI. Andrew? Okay. Cheese will be next. Coming up on SquawkPod, Bridgewater founder Ray Dalio. What makes a bubble? What pops it? And maybe most importantly, what does an investor do about it? Don't sell just because there's a bubble. OK, don't sell just because of bubble. But if you look at the correlations with the next 10 years returns, when you are in that territory, you get very low returns.

13:51AI hype, the gold advantage, private credit and market history. Hedge funder Ray Dalio is next.

14:03This is Squawk Pod with Becky Quick and Andrew Ross-Sorkin. Here's Andrew. Now to our newsmaker of the morning. He's here joining us to discuss the markets, the history of economic bubbles, artificial intelligence, so much more. Bridgewater founder Ray Dalio is with us. Good morning to you, sir. Good morning. So you've been providing lessons for us for a very long time about economic cycles and where we are and what's going on. The big question in the market right now, because we're looking at NVIDIA this morning, and I think a lot of people are waking up thinking, where's the market headed next, has been this question of an AI bubble.

14:36And I'm curious where you ultimately come down. Are you with Jensen Wang, who says he doesn't see it that way, or do you see something else? There's definitely a bubble in markets. Bubbles, what is a bubble, right? What is a bubble? What is a bubble? A bubble is that there's a lot of creation of wealth from various ways, such as you decide that you're going to sell$50 billion worth of stock and value it at a trillion dollars, or you have multiples like that, and then you create the wealth that way. And then the question with all this wealth relative to money is who needs the money. So it's a matter of who the buyers and sellers are.

15:27For example, if we had a wealth tax or if you had a tightening of monetary policy, then there has to be the selling of those assets in order to pay those things. So there's a mechanics of who owns it, is it over-owned, and so on. It's not the long-term duration of the earnings. So you think about it. Isn't it interesting that we have such a short term reaction? It's great that what the results are, but this is a it's valued as a long duration asset. So for 25 years, the next 25 years, it's very unknown. We don't know what's going to happen and bubbles don't go happen because of good estimates of what in the future.

16:11It happens because of the need for cash. Do you sell that asset? You have to sell that asset for cash for some reason. Well, that's the question, though. So your book, I thought, did such a good example. I mean, I thought, by the way, your book is fabulous. Thank you. But you take the 1929, and then you say, what made it go up and what made it go down? And that dynamic is the way bubbles work, right? And so if you take who has exposures, how much leverage is used, and so on, This is about 80 percent into a bubble that was 100 percent would have been 19, 29 and 2000. So how much leverage do you think is supporting all of this?

16:54Because we were talking to Patsonevelis in the last hour, and she was explaining this idea that Jensen Wong doesn't see a lot of the transactions that he's been doing. I mean, we've talked about these circular transactions where he's effectively taking an equity stake in some business and that business is then committing to buy his chips, if you will. Yeah. A form of vendor financing with sort of a equity overlay, maybe, is the way to think about it. Do you look at that and say that's a problem? It's not a problem. Should we think about vendor financing in a different way? That was something that got a lot of the fiber guys in trouble in the late 90s, for example.

17:30I think it's an issue, but I don't think it's the main issue. I think the real issue is who owns the stock. Is it in strong hands? Not just one stock. Isn't it amazing we're talking about one stock for the stock market bubble and we're talking about for the economy bubble. So you have such a small percentage of the economy, such a small percentage of the American population in terms of wealth and so on, so concentrated. and everybody in it and in a leveraged way. In various ways, it has leverage. What are strong hands? How would you define that? Weak hands would be the public. So retail investors are weak hands.

18:16Right. Strong hands are the owners of these companies? Right. In other words, strong hands is that they primarily invest their own money. You don't have public, right? The weekends is largely, let's say, a leveraged public and all united about that. That's that's one of the key ingredients of a bubble. So it's not just pricing because we have to find out. You know, your book did such a great job of looking at the wonderful companies in the 1920s and then enumerated how electricity came out in general electric and RCA and so on. And those companies went down a 90 percent, not because of the economy and so on.

19:01The bubble burst first because they needed cash. So what is you have to understand that wealth can't be spent in order to in order to get money. You have to sell wealth in order to get cash to get the money to buy things. And so when that happens, bubbles burst. So bubble bursting means, let's say, a tightening of monetary policy is classic. But also something like wealth taxes can happen. For example, think about wealth taxes forcing sales of assets that you have to break. By the way, there's a proposal for wealth tax in California right now. So state and national. Right. State and national. I'm just trying to describe the mechanics of a bubble.

19:49But you said we're at 80 percent. Yes, I have a bubble indicator that goes back to 1900, and it has a number of indicators. How much leveraging? Who has the leveraging? What is the amount of money in wealth relative to the amount of cash that needs to exist and so on? These indicators, there are a number of them, show that we're on that chart. If I was to show you the chart, it's about 80 percent of where it was in those two times. That doesn't mean that that's the end of the move, OK, because bubbles have to be pricked. Right. And so you can measure that there's vulnerability, a lot of vulnerability.

20:39There's a lot of vulnerability of who's holding what. Here's my question to you about that, which is if the bubble is at this 80 percent range on your index or metric, there are folks, we had Paul Tudor Jones come in here a couple weeks ago, and he said, look, I think we're in October of 1999. By the way, the market would still have 40 percent to go. Back in 1928, Charles Merrill, who founded Merrill Lynch, told everybody, get out of the market because he thought it was a bubble. And he was right and he was wrong insofar as from the beginning of 1928 to September of 1929, stock market went up 90 percent.

21:14And so. But I think that when you're looking at that, you're looking at it. Incorrectly, OK, what is a bubble is the definition, right? I want to reiterate, a lot can go up before the bubble burst. A bubble is an unsustained set of circumstances. It has unsustained amount of buying. It has an unsustained amount of valuation. It has, and then there's something that pricks the bubble. Is there a way to sustain this, though? You can't look at, you can't look at, don't sell just because there's a bubble. Okay. Okay? Don't sell just because of bubble. But if you look at the correlations with the next 10 years' returns, when you are in that territory, you get very low returns.

22:04Um, yeah. J.P. Morgan just did a report on this that showed that if you got in at over a 23 P.E. multiple, typically over a 10 year period, your your return is a delta between 2 percent and minus 2 percent a year over 10 years. Oh, yes. To your point. I just put out today a post on this and you'll see the charts and if you go on that. So I think that you have to say it's unsustainable. Then you have to go to the timing. What is it that pricks the bubble, right? Typically a tight-knit monetary policy. We're not going to have that now. Okay. Right? But you could have something. In other words, the need for cash.

22:49Right. The need for cash is always that which pricks the bubble. Because when you have wealth, you can't spend wealth. Right. You have to sell wealth in order to get the money to buy the things you need or pay the bills you have. That's the dump. So I think that the picture is pretty clear in that we are in that territory of a bubble. We are in that bubble territory, but we don't have the pricking of the bubble yet. So what does a Ray Dalio, what does the Dalio family do with its money at a point in this bubble of 80 percent? and you still think it can go higher, but now it becomes a game of timing, which is always very difficult.

23:31Of course it's a game of timing. Well, and elements of diversification, you know. I think you start off with what is a well-balanced, diversified portfolio, and we talk about certain things like gold as being part of that. There's a lot of ways to diversify. That's a whole other conversation of what you don't know. But I think that that mix, we could talk about gold, we could talk about these other assets, but I think that that kind of diversification is the... What about cash? I think that, yes, gold and cash have negatively correlated real returns. So when you take... I believe that we have a debt problem.

24:18OK, because we have a debt problem, we have a value of money problem. But but that's right. If you just said if I held cash, but cash can also be gold. I think we have to realize that money is what is money. There's fiat money and then there's, you know, gold is the second largest reserve currency. It's money. So when it diversifies the portfolio, that is, I think, an important element. It's negatively correlated. It does very well in such bubbles. So when you think of cash, I think of cash as being a mixture of those two things. But you don't love the U.S. dollar, is what you're saying, because you think we have to do that.

25:02I don't like debt. I think debt assets have a problem. They're overheld, and we're producing too much. The supply and demand is a problem. And so if I was to take my, you know, what is the tactical view? I would rather be short debt in a sense. Right. OK. And then that means that I'm what money do we have? I think a big question of our time is what money? Right. What is money? What is a storehold of wealth? That's a big question, too. There are these two big questions happening simultaneously. There is the AI question and then there's the debt question. We were talking about that a lot during this broadcast in the last couple of weeks about private credit and the like.