The First Crack in the AI Bubble Just Appeared

So Meta, the company behind Facebook and Instagram, is reportedly thinking about laying off over 20% of its current workforce. While the company didn't confirm the rumors, it also didn't deny them either. More importantly, these theoretical layoffs do not appear to be trying to replace the current workforce with AI robots. Instead, they bear all the hallmarks of conserving cash, which raises a number of questions, not just about what Meta's up to, but also the conditions in debt and the debt market. It is the one dog which so far hasn't barked in the private credit bus. We've heard about fraud and bad business in the auto sector or e-commerce, really anything that goes along with consumers these days, but the AI expansion, call it a bubble because it's already soaked up a massive amount of monetary resources. And whenever that happens during the upswing in any credit cycle and it starts to run up against systemic limits, you know what comes next. To keep the good times rolling, Wall Street gets creative. And this time, no different, literally no different than the last time, meaning 2008. We have seen the return of the SPV, off-balance sheet vehicles whose sole function is to increase borrowing capacity through complexity. And it works great until it doesn't because debt that is held on balance sheet behaves differently than what's owed off balance sheet, especially during periods of stress like right now. And we're only just beginning to see more serious strain. So even the BIS is out with a note today and we'll go over this, warning about the potential for repeating the same problems as last time. So Meta is allegedly thinking about making drastic cuts to its operations. Is this a message to the stock market saying we're going all in trying to win the AI race? Or is this coming from the bond market that up until recently, couldn't wait to shovel unlimited cash and money into an AI bubble, now thinking maybe this is not a bubble, but instead a credit cycle bust. That's what we need to think about here. As I said in the intro, no confirmation from the company itself, but every single media outlet is reporting the rumors. It started with a Reuters report, which said Meta is planning, that's the word they use in their headline, not just thinking about, but already planning to lay off 20% or even more of its current workforce. And this comes from three sources who Reuters said are familiar with the plan. Two of those anonymous sources claim top executives at Meta have already told senior company managers to begin drawing up plans about how to go about cutting back. The only comment from the company itself came from a spokesperson who only said, quote, this is speculative reporting about theoretical approaches, not exactly a denial here. So on the one hand, they're making this all about AI. However, it should be noted, we have yet another example of macroeconomics impacting microeconomics at the corporate level. Just like we talked about a couple of weeks ago, Jack Dorsey's block, which said it was gonna lay off 40% of its staff or Amazon, which reported last month or a couple of months ago, it was getting rid of another 16,000 of its own workers, its own corporate workers. What's really behind these moves is lack of business and recovery. All these firms, and not just in tech, they staffed up big time in 2021 and 2022, expecting the economy to match the consumer price rate. They fell for the idea it was a red hot recovery because that's what all their economists said it was gonna be. Another lesson in never listen to economists where it comes to the economy. What happened instead was price illusion, initially hiding the impoverishment of the vast majority of people on planet earth. Prices zoomed ahead because of the supply shock while incomes lag too far behind and then never caught up. That's the more important part. That's not the recipe for sustaining booming growth, it's how you get forgot how to grow. And that's what all these tech titans and real economy businesses have been trying to come to grips with over the past couple of years. This is exactly where the flat beverage environment is coming from. There was no recovery. After waiting years for one to finally show up, companies finally stopped listening to the economists and started downsizing or rightsizing as they call it. They're payrolls and Meta is no different. Its head count peaked in 2022 as Zuckerberg's firm was one of the first to shift into forgot how to grow mode, calling 2023 the year of efficiency after starting some job cuts back then. But notice how Meta hasn't been adding to staff since, like Amazon and all the rest, basically not hiring for years. There's your no hiring economy. And keep in mind as we go through this, the flat beverage world is behind a whole bunch of stuff, not just weakness in jobs or overall economy, but also the credit markets. The reason we're seeing this, the reason we're seeing the credit market bust, the reason we're seeing liquidity problems in the credit market, all goes back to flat beverage and negative payroll reports. The more we get of them, the more trouble we're gonna get. Therefore, Urdal University, we're having a webinar March 26th. It's a Thursday, 6 p.m. Eastern time. We're gonna go over where things stand in the credit market bust, as well as the various stages, what they mean, what they suggest about going forward. So you're gonna wanna be there. Again, Thursday, March 26th, 6 p.m. Eastern time. There's a link in the description. You are not gonna wanna miss this because of everything that's going on right now. And so here's where AI really enters the picture. You've got an economy, a recovery that isn't booming, that isn't growing, it isn't even going forward. Suddenly AI creates all these opportunities that didn't exist before. So a lot of businesses said the real economy sucks, there is no recovery. AI, at least there's a future there. At least there could be a future there. So more and more of these technology firms leaned into the tech. And as it got increasingly expensive, initially that wasn't an issue because Wall Street was still in on the whole recovery idea because it was being pushed by the Fed, as well as the stock market. So AI hyperscalers needed a trillion to make it happen? No problem. The financial services industry was only too happy for the volume. Risk? Nah. What's the risk? These are brand name firms building a completely new technology at size humanity has never experienced before, whose future is totally uncertain as it follows the same general pattern as every other tech bubble humans have blown throughout history. What risk? And up until last year, no one cared about all that. It was all positive. Banks, shadow banks, inside sources, outside sources, private credit, indirect lending, the money was flowing in from everywhere. But even then at the height of the credit bubble mania, there were some constraints. These weren't easy to see and they sure weren't being felt by anyone on the outside, but it didn't matter. SIVs and SPVs were increasingly brought back and used to finance the AI boom, which had to get creative all over again because of the amount of money that was being involved. Now you may remember the term SPV or special purpose vehicle from the 2008 crisis. These were heavily used in the mortgage bubble and for the same general purpose, to creatively take on more debt than otherwise would have been possible using traditional means. And I can already feel you shaking your head. And I agree with you. After all, if you can't borrow enough money in your own name, such as you have to borrow using someone else's, isn't it already pretty dodgy? But the answer you get back from Wall Street is that SPVs create more space to borrow because you're not just borrowing in your own name, you're bringing in other investors to borrow with you. Not one-to-one, all different kinds in the structure, all different kinds of obligations and incentives and return characteristics. It's spreading everything out over a wider base, which means borrowing with partners and all their resources and expertise and all that goes along with it. And this is absolutely true. When done correctly, that's exactly what structured finance accomplishes. And it has many legit uses and has been used legitimately for decades without problems. It's when things go too far, when the bubble itch begins to take over the thought processes, when the allure of size and scale overwhelm common sense and sound purpose. Greed combines with rationalizations. And that's when a good idea gets bastardized and it becomes something bad. So the main problem is in very broad terms, what I just said, complexity. When everything is just fine, it works great because you have all these different minds, all these different pocket books coming together, all these different resources, creating a wider base that allows for more lending to take place than otherwise would have. That's the whole point here, other than creating repo of a collateral. But either way, you've got more investors coming together, but when times flip, when we go from the bubble going up to potentially the bubble going bust, that complexity can become information asymmetry among other problems, which leads to cascading issues that become bigger and bigger and bigger. I mean, after all, different investors from all over the financial landscape with very different ideas about what they want to do. Yes, they can come together and make off balance sheet vehicles viable, but when things aren't fine, when financial stress really begins to mount, in theory, the wider base investors should help bolster its resistance. But what we found out the last time, we found out the hard way the last time is it had the opposite effect. Complexity, it worked against structured finance and therefore the entire system in which structured finance had been deeply embedded. So let's talk about this approach in the context of 2020's private credit and specifically Meta. Now, at the end of last October, which was only four and a half months ago, though it seems like a lifetime ago with everything that's gone on since then, Bloomberg reported on one massive deal that Mark Zuckerberg's company had already worked out. This is the October 31st. Quote, just this month, Meta platforms has secured about 60 billion in capital to build data centers, part of its spending to get ahead in the artificial intelligence race. Now, half of that won't show up on the social media giant's balance sheet as debt. Meta is among firms popularizing a way for debt to sit completely off balance sheet, allowing enormous sums to be raised while limiting impact on its financial health. Morgan Stanley structured a $30 billion deal, the largest private credit capital transaction on record where the debt would sit in a special purpose vehicle or SPV tied to Blue Owl Capital. There's that name again. That made it easier for Meta to raise another 30 billion this week back in October, the usual way in the corporate bond market. Off balance sheet debt through an SPV or a joint venture tied to assets like chips or real estate is becoming the go-to for AI data center deals that bankers say. Now, the SPV, the special purpose vehicle, is the borrower for the 30 billion, not Meta, at least not directly. The vehicle gets equity investment from Wall Street plus Meta puts in some of its funds up front, usually a minority stake. But then Meta will commit to leasing whatever the project ends up creating, either a data center or the purchase of massively expensive equipment. And in the example we just went over, it's data center build-out. Now, the special purpose vehicle can then raise the full amount needed because it has all the right names as owners and it has a commitment from a top tier tech titan to buy its products or services. Bond dealers can then go to every corner of the debt market and sell individual bonds and bond types to different types of investors who are looking for high returns with little risk as we've talked about before, little assumed risk at the time the deal gets done. Now, private credit providers take some of the debt, private investors and institutions, some others, and yes, regulated banks absolutely do supply some of the funding on top of everyone else's. And just like that, Meta effectively raises $30 billion to build a data center or buy brand new chips and servers with debt that it didn't really borrow and doesn't legally owe. Its name is only on a lease to go along with a relatively small minority equity position. So when the bubble is still being blown, nobody cares about the details, nor did it take a whole lot of time thinking over any of the downside cases. Bubble behavior is overwhelming, which is exactly what we've been seeing and saying ever since Tricolor pulled back the veil on everything. So what would a downside case be? Well, the credit cycle itself would be a big one. What happens if lending tightens up so much that it becomes more difficult to put these complex AI funding arrangements together? If the money isn't coming in from all over Wall Street like it did before, it might mean the AI capacity doesn't ever get to where it needs to be to make this stuff all pay off. So what seemed like a risk-free investment only six months ago starts to sound like there might be a touch more risk than everyone thought. Now the BIS among others has started to sound the alarm over this off balance sheet craze, which has already been going on for quite some time. Like I said, the tactic itself is not really the problem here, securitization, structured finance, all that stuff has been used in many good cases for a very long time. It's when you get into these bubble behavior periods, these bubble periods where behavior becomes altered, underwriting suffers, common sense gets thrown out the window. And as Jamie Dimon said, people start doing dumb things. Only in this case, they're doing dumb things off balance sheet at a scale that we didn't even see in 2008. Not overall, but in these individual deals, which has the potentials the BIS is saying to create more trouble than just for special purpose vehicles, individual special purpose vehicles. It's always the scale, underestimating and therefore mispricing risk. Therefore the complex structure becomes much less robust under pressure. So let's go here to the note from the BIS, which comes summarized by Bloomberg. Quote, alongside traditional bonds, hyperscalers have turned to off balance sheet arrangements to finance infrastructure expansions, often in partnership with private credit firms. That's according to the BIS officials in a note on Monday. These structures strengthen links between hyperscalers and non-bank investors. Those are the shadow banks. The comments add to warnings about the debt frenzy to support the artificial intelligence revolution. With so many lenders lining up to throw cash at assets formally, the risk is a bubble could eventually leave credit players facing substantial pain. We've already seen the bubble. Now we're working on the bust. Back to the BIS. These arrangements amount to shadow borrowing. Why not? It's shadow banking. Obligations that are economically akin to debt, but largely reside outside corporate balance sheet. Banks support such off balance sheet vehicles with funding lines, potentially creating new shock transmission channels, the note said. That could leave lenders potentially exposed to refinancing pressures at the vehicle level, pro cyclical shifts in private credit appetite, or the activation of guarantees. All the kinds of stuff that we saw almost 20 years ago. So when you have one borrower, Meta in our example, there's not a whole lot of complication to work through to figure out how the borrower will react under duress. They're gonna pay back every loan because not doing so would be catastrophic unless something catastrophic happens to Meta, then we've got bigger problems to worry about. And Meta has the ability to do it, even if it would cost Zuckerberg a ton to do so. But when the borrower is an SPV with all kinds of equity holders who are at risk of bigger proportional losses, the downside case gets to be much less predictable. Could everyone just walk away? Well, maybe not, but some people could. Could the owners try to pay off at discounts to preserve most of their stakes? Lost potential is no longer as simple as saying, hey, Zuckerberg, you owe us 30 billion. And that higher degree of uncertainty creates feedbacks, which ripple throughout the rest of the credit market in the broader system, which is what the BIS was getting at. So what happens if a big bank is now stuck having made a loan to one of these things and is unsure what it might get out of it? That bank is gonna be making fewer loans to other SPVs just as a start. Banks and credit providers begin to go into protection mode. That's the pro cyclicality. You get what the BIS cautioned is a new shock transmission channel that leads to pro cyclical shifts in private credit appetite or as they say, the activation of guarantees, which just means a major problem in one SPV becomes a major headache for everybody and a general tightening of money flow and JP Morgan potentially collateral. So we've already seen that, just not an off balance sheet debt tied to AI and hyperscaling, not yet. But this is where we come back to meta and its layoffs. Not just why, but why now? And I asked the same question with Block, Jack Dorsey's employee chopping firm from a couple of weeks ago. Why now? With Block, the company wasn't getting rid of workers and replacing them with robots. They intend to maybe down the road if business ever picks up for them. But is there financial pressure? Is the bond market and Wall Street saying in private credit, I need to see more commitment. I need to see more capacity to pay back. Meta is a clear cut case of conserving cashflow with its layoffs. But for what purpose? To pay for AI capacity out of retained earnings? If so, wouldn't that indicate a credit market tightening up to the point the company is concerned Wall Street may not be there for the full amount like it was before? Notice how all the roads here go into increasingly dark territory. And this appears to be another credit cycle signal which points to perhaps some serious reconsiderations going on in that cycle. And therefore more defensive action on the part of meta like Amazon, trying to create more internal cash in case the debt market isn't willing to come up with it moving forward. Let alone what might happen if something happens with the risk profile for these various SPVs which is another downside case all its own. As I said, the AI part of this is the one dog in the private credit bus which so far hasn't barked. Now what meta might be doing and keep in mind not confirmed yet, but it's very likely given all the reports. But what meta might be doing is another layer behind everything. Another signpost that suggests credit cycle turning and it's something we need to keep our eye on the AI, SPVs, off balance sheet funding and knock on effects from it given the scale as well as the true risk not the risk that were sold to all the investors the true risk that we increasingly see are involved all across the formerly bubbly space. So SPVs, all the stuff we just went over another layer of indications you wanna keep an eye on outside of strictly private credit. And along those lines, Deutsche Bank and Wells Fargo there's a couple of big signals there and I went over what they say in the video link below. As always, thank you very much for joining me. Join me March 26th for our webinar there's a link in description for that. Huge thanks for Eurodollarversity members and subscribers and until next time, take care.