원문 보기
Private credit and the debt market are beginning to plummet into the abyss. In this video, we'll look at what this means for you and how bad things really are. Look at this. It's a big arrow pointing down. Spoiler: nothing good. But first, let 's start with what JP Morgan just warned about . Then we'll talk about the situation in the AI bond market, and after that—what it means for us.
So, let's start with the fact that we've all heard about private credit. Many companies financed by private credit started going bankrupt late last year, which created a lot of fear: "Oh my, we're just at the beginning of that black swan event where we're going to have a massive wave of bankruptcies that the Federal Reserve will eventually have to bail out." Well, so far, although many companies have closed down since the end of last year, quite a few of them are, um, potential scammers, as claimed.
Concerns about private credit have not gone away. They were just trying to hide them. So JP Morgan asked the question: " Hey, this problem with private credit— we keep being told that the stress level is between 1% and 3%." But there are so many companies that are living on private credit, that is, private firms that have accumulated debt and are now selling for less than their value, or have agreements to pay in kind, or, so to speak, " carry on and pretend " style loans, that the actual default rate is probably much higher.
So JP Morgan decided to find out: " What is the likely actual default rate, taking into account all the so- called shadow defaults?" And here we come to the crisis of shadow defaults in private lending. The name doesn't sound very familiar, but that's exactly what it is . This graph here is just amazing and critically important because it shows us that what we're being told about the risks of private credit is a big story about one company going bankrupt, and then people say, " Don't worry, the rest of the ship is fine, there 's just a few squeaky wheels;" " You know, we replaced them, we patched them up, they're fine." As you can see right here, we only write off about one to three percent of transactions along this black line; it's not a big problem.
But when you factor in troubled payment-in-kind (PIK) deals— companies that are paying with their own shares instead of cash because they have no money left and are on the verge of bankruptcy while trying to force rates down—they are becoming more and more common. You see this red sector, these PIK deals, they're starting to grow.
There are also extensions of repayment terms, they are also increasing; Previously, they occupied about two percent, and now they occupy about six percent, meaning this small blue sector has almost tripled. Next, we see restructured capital, that is, the actual “throwing away” of the company’s internal owners in exchange for assistance in obtaining financial support .
And if we add to this companies whose debt is valued below 80%, that is, sold at a 20% discount, then suddenly it turns out that the level of shadow defaults in our market reaches about 16 %. Note that the private lending market is about $1.3 trillion. So when we look at private credit, we realize, "Oh my, there's a real debt problem here." Hopefully, interest rates will not increase.
Oh wait, we all know that 10-year bonds are at their highest in 24 years, and it looks like they recently peaked, which gives a little respite. This gives some room for maneuver, but the trend here is quite obvious. Look at this. If we simply draw a line through the total percentage of shadow defaults, we see that their number has actually doubled.
And so many people ask: what does this mean? Is it just private credit or is the broader credit market also having problems? This is where you can turn your attention to public capital and public debt markets. So, how to do it? Well , there are two ways. First, you can simply look at a stock like CoreWeave, for example.
Note that Nvidia has invested in CoreWeave. They invested $2 billion at a price of around $87, and yet...well, they invested more money earlier at a lower price. So Nvidia is probably still in the black from its shares in CoreWeave, but if you step back and look at the situation, the stock dynamics have been pretty bad.
After a high of $187, the stock is trading at just $77, and it is in a steep downtrend, barely staying afloat. However, this does not necessarily mean that the company is bad. There are many great companies that just have n't had their prime yet, so to speak. So I like to look at what's actually happening with their corporate bonds.
This can sometimes be an early signal of risk in the field of artificial intelligence. So, we have two things. There is the risk of private lending debt, and there is the risk of artificial intelligence debt, which we will talk about now. And , of course, there is the national debt. And when everyone has so much debt, this trio of debts makes me strive to have as little debt as possible.
Personally, I don't have any debts. I got rid of all my debts. I am very afraid of debt in this economy. People ask, "Kevin, are n't you worried about housing?" No, I think now is a great time to buy a home with no or minimal debt . We are debt-free. No bank debts, right? Now, when we look at stocks, people say, " But Kevin, aren't you worried about the stock market bubble?" I answer: there are no debts.
I don't care. I'll just buy great companies. And the government—well, I can't do anything about it . I tried in California. Nothing came of it. Politics is something else. But hey, I tried. I gave myself to it completely, baby. Ha. OK. So what do we have here? Coreweave bond prices. Oh, that's not good.
So, Coreweave is currently trading at an effective yield of around 15% on some of its debt, effectively putting it in the " junk" bond category. In fact, if you look at the spreads of some Triple C ratings, for Coreweave they are now over 1000, which, simply put, means: this is bad. They are actually trading at " junk" levels, and you can see that they have recently collapsed even further.
Part of the reason for this collapse, in my opinion , is not necessarily the increase in default risk. I think there are just more releases on the market. You see, I don't think Coreweave is any riskier now than it was back then. In my opinion, she was just as lousy there as she is here. So, what has changed?
We just issued significantly more debt. And when you issue more debt, you're like sucking money out of other places. People were asking, “Why would I buy CoreWeave bonds when I could take new SpaceX or Broadcom bonds that would give me access to Anthropic or OpenAI ?” “Why would I buy CoreWeave bonds where 62% is in one customer— Microsoft, right?” So there 's a certain amount of risk involved, which essentially means, " Hey, if people look at CoreWeave, they're asking themselves, do I really want to take a risk with this company, or is it better to choose another one?" In fact, if you look at CoreWeave's balance sheet, things are bad, man.
They have about $5.5 billion in cash plus $2.5 billion in accounts receivable . They also just raised about $3 billion , which is not bad. This brings their cash holdings to about $10 billion. But here's the problem, friends. These people have accounts worth $18 billion. So let's make this very simple.
If you have 10 and you have to pay 18, you are short of money. Of course? But that's been true ever since we first analyzed CoreWeave's last earnings report, which was at least a month ago. Oh, here. When we first opened this, it was September 17th. But I'm pretty sure we also looked at this right after the company reported earnings in August and saw the same thing.
This file just happens to be dated September 17th, and I added these two little notes in red to point out that this company's bonds have been selling off for the past 8 months. And we're talking about short-term, medium-term, and long-term bonds. They are all sold at a discount. People would rather exit CoreWeave at a loss, potentially due to the risks they see in this company, than stay.
And this is happening again at the same time that shadow defaults are doubling. So you have a doubling of shadow defaults. You have problems in the AI debt market . At the same time, you continue to issue new debt. Now this gets a little more complicated, and I'll explain it, but it goes along with other nonsense I see in our environment right now.
Let me talk about some more nonsense, and then we'll move on to the stock issue, because that's a little more boring. And here's another example of something absurd. There's a company, a private equity firm, a partner with Sequoia, and apparently they're running their own fund now or something. They decided to buy Bumble because they thought, "Oh, well, AI is going to make human communication more important." So we're going to buy this loss-making company at a 38 % premium, I think, at $6.08 per share.
And Bumble is literally losing revenue. They have fallen by 21.5% over the last 2 years. They have fallen by 15% over the last year. So, the losses are accelerating . Also, if you actually look at the company's earnings reports, and I'm starting to doubt that these venture capitalists are doing this, when I look at these reports, I think, "Wait a second." What are we doing here?
"Oh, now Bumble is going to bring together best friends." What? Yes, they are trying to make best friends forever. This will become their new advantage. Hey, maybe this will work, right? But this seems like an adventure with a completely new business model. See? Generation Z really prefers to hang out with their best friends.
We believe Bumble is a natural bridge from meeting to chatting and then to dating. We are already seeing signs of this in our BFF initiative. We are starting to see strong growth in both the number of active groups and the average number of active members per group. This is very exciting. Great, especially among Generation Z women.
Good. But then we go from "this is really exciting" to " hey, will this make money?" And when we are asked, will it make money? In other words, hey, what is your tolerance for a decline in average revenue per user? You know, like when are we finally going to see some money here, people? We love money.
And here they say, "Well, with the average revenue per user, you know, with the revenue going down, um, we're working on things that will grow the business." " In other words, we don't really want to talk about it right now." We're not ready to talk about it because we're barely breathing here and still trying to figure it out.
Okay, to me, this is just a small sign that some of these people are reckless. Sorry, Ravi Gupta, maybe you're a good person. Okay, sorry for bumping into you and mentioning Sequoia, because you say you're a partner there. But I think that's madness. Over the past 6 months, we've looked at hundreds of companies, asking which ones could become more valuable with the development of AI.
Yes. So, you're looking for a business that you think could become more valuable with the development of AI, but over the past 2 years, while AI has been developing, their revenue has been falling precipitously. Got it. Bumble stood out. As intelligence becomes more accessible, we believe that connecting with the right person and building meaningful relationships will become even more relevant.
And our goal is to apply cutting-edge technology to what really matters. We buy long-standing businesses and transform them with technology. Oh, my mistake. All Bumble needed was more technology. They simply lacked the technology. You know, this is something that really pisses me off. And I do n't say this as an advertisement.
I'm just trying, I think it's instructive, and that's my point. So one of the reasons we're launching what we call the Actions and Research Terminal is because there's so much artificial intellectual garbage out there. In my opinion, you really need people who can provide real insights into what the hell is going on there.
We're actually going to hire people in 2027 for this. I think quite a lot. I really believe that we will create a cool research team , and all this research will be available in our, uh, terminal . So if you like to research and read and learn and you like chips, like semiconductor or memory analysis or the Federal Reserve or whatever, hey, send us an email.
Write to [email protected]. Give us a little pitch, a resume, a short video , whatever, you know? But, uh, no AI junk. If you send me AI junk, that's the first alarm, son, and I have a junk detector. This is garbage. Uh, but anyway, again, the point I'm making is that people believe that technology fixes bad business.
And I think people forget that technology doesn't improve bad businesses. Technology enhances great business. Technology and AI make a great business even better. Adding AI to a bad business leaves you with a bad business. It's like having a bad query or, you know, a bad strategy. If you rely on this , it wo n't get better just because the model has improved.
Okay, enough about that. This angered me and makes me think that we are still in the early "drunken " stage of the cycle, where, despite the emergence of credit problems, fears of shadow debt are growing and there is an increasing amount of such material from sellers of discounts. I don't know if George Noble is really playing for the downside right now, but he's making these exact line-by-line comparisons to Exodus Communications , which essentially collapsed in 2001.
And so they say line after line: “Oh, they’re selling critical infrastructure. Sells critical infrastructure. Extremely high capital expenditure, extremely high capital expenditure. They use significant amounts of debt and asset-backed financing. Active use of debt and mezzanine financing. Blah-blah -blah.
Although, to be fair, this is a slightly different situation, because CoreWeave debt is backed by GPUs and server racks. And frankly, in my opinion, if a company like CoreWeave went bankrupt, someone would be drooling to buy their assets at a discount. These GPUs, these server racks, even if they're H100 or older chipsets— even if they're older chipsets, as the models get compressed, they ironically become more valuable.
So, the more we play with open scales, the more valuable old things become. This is the strangest paradox. And let me tell you, I watched "V for Vendetta," you know? I discovered the possibility of using the word "paradox". But I also find it quite strange that they ask the man in the mask who he is.
Oh, no, sorry, that's the wrong line. It's also amazing that you can see things like , well, take my plane, for example. Why did I buy a plane for $12.9 million ? He flew it himself, spun it around , flew it back and forth , and did stupid things . I flew 800 hours on this plane in 3 years. And then sold it for $69,000 more than he paid for it.
Then of course there were the costs, like fuel or something, and maintenance, right? We took very good care of our baby. I mean, who doesn't take care of their baby ? The point is, it's weird. Usually, such things should not increase in price. This is because there is a huge shortage of aircraft. But the same thing is happening in the chip market, only the situation there is even more extreme.
So I almost wonder if the falling cost of airplanes is an early sign that production is catching up with demand. Maybe when airplane prices fall, it's a sign that production is finally starting to catch up. I do n't know, okay? This may be a bit far-fetched , but the point is that on red days we see a lot of comparisons, especially from short sellers, who say, "This is like 2001." And I agree that there is a surplus.
I agree that CoreWeave has no money. But I think the residual assets are strong enough at this point to argue that even if such a company has higher debt costs, which eventually leads to bankruptcy, the residual assets are valuable. They are actually more valuable at CoreWeave than at any random private lending company.
So I would be more worried about the private lending issues than the issuance issues that we see in the AI space. God, that was hard to say. OK. Let's dig into the issue a bit, because someone left me a comment. They say, "Oh, but Kevin, if people are buying bonds instead of stocks, shouldn't yields fall and prices rise, right?" Like, why don't yields fall if people suddenly switch to buying bonds instead of stocks?
This is a very valid question. So, here's how it works. The key to understanding is the word " suggestion." Let's say there are $50 million worth of bonds on the market right now, and they're all sold out , right? Someone owns them all. If there is no new issue, and someone comes with $100 and there is no issue, they can say, "Okay." I can either buy one of those discounted bonds or just buy the stock on the open market, right?
I'm intentionally simplifying this to get the point across . Now suppose that additional $100 bonds are issued. Well, now let's say this insurance company says, "Oh, I can get new bonds during this fundraiser." Okay, cool. I'm an insurance company, or whatever. I'm going to go and buy these $100 bonds.
"Well, since they're buying the newly issued bonds, what happens to the yield? Nothing. Got it? The yield only changes when there's an existing supply. So when that 50 million, that initial 50 million, if suddenly everyone comes in and buys it, you're going to get 20% more buyers. And these things start selling at a premium.
So, the yield is going to go down. But we're not actually necessarily buying more of these things in this example. What we're doing is, if someone comes out and says, 'Hey, we should issue new bonds for $100.' "And someone says, 'Okay, I'll buy these bonds for $100.'" "It's like a new market. It's a brand new product.
It actually has very little impact on reducing those profitability. It certainly doesn't have any impact on reducing those profitability. If anything, it increases profitability because now they're not necessarily being bought in the secondary market, and people are going to buy new cars. Of course, at some point in the future, they're going to be used cars too.
You know what I mean? So I mentioned that because I think there's a greater risk in those private equity companies that JP Morgan is talking about when we see the shadow debt doubling. That's bad. The failure of those companies does n't necessarily affect you or me yet. They do affect us. Well, I guess technically it would be you or me.
The companies that go bankrupt, they're an entity. Yeah. The failure of those companies doesn't necessarily affect you or me. The failure of those companies leads to layoffs. And because some of these businesses don't necessarily have residual value, zombie companies that go bankrupt just die and decompose.
There are no chips to resell. Those layoffs become permanent. That's when the labor market suffers , and that's probably why the Fed says, "Oh, yeah, this private lending problem—it's a problem." "If Coreweave goes bankrupt, shareholders will suffer, of course, but someone will pick up those chips, and bondholders will also suffer to some extent, although they will benefit from the residual value of those chips.
But really, for the entire AI market to collapse, there would have to be enough bankrupt companies, say, like Coreweave, and enough of those companies would have to fail to suffocate the demand for AI hardware. At least for now, until OpenAI and Anthropic go bankrupt, and they may never go bankrupt, we are definitely spending more on these products than ever before.
If they don't go bankrupt, then we'll just say, ' Great.'" I would happily buy your H100s after bankruptcy. It's literally like getting a bargain on chip equipment. So to look at the AI debt market and say, "Oh my god, Coreweave is doing badly." This is not a sign that the entire AI market is heading for disaster.
This is my opinion. I think this is a reasonable opinion. And if you want to be part of Reinvest Terminal, remember, we are discontinuing lifetime access. So go to reinvestone.com. If , it will be yours forever. Which is pretty cool. This means that if you purchase a subscription in 2027 or later, you will pay a regular fee.
Therefore, I advise you to join in 2026 . Check it out at reinvestone.com. , private equity .
Comments 0