저자: OxTalks · 외국인
요약 보기
Enron 붕괴와 현재 AI 데이터센터 부채 구조(SPE/SPV, 오프밸런스시트)를 비교하며 AI 투자 버블과 하이퍼스케일러 이익 과대계상 위험을 경고. 마이클 버리의 감가상각 관련 주장(2026~2028년 1,760억 달러 과소계상, 오라클·메타 이익 과대 추정)을 인용. 버리가 펀드를 청산하고 11월 25일 발표 예고, 팔란티어·엔비디아 숏 포지션 언급.
다듬은 글
So in 2001, late 2001, Enron, uh if you guys remember, this goes back many years now. Enron totally collapsed. Uh and it resulted in about $80 billion in destruction of value. Shareholder equity, $70 billion got wiped out in the Enron collapse. Employee pensions 401k losses was between 1.2 2 and2.1 billion dollar had evaporated.
Creditor losses 63.4 billion in assets on paper and the creditors got about 20 cents on the dollar. It was 10 to 15 billion and the secondary collateral damage Arthur Anderson the huge accounting firm collapsed and there were also massive counterparty writedowns and class action suits. So what led to without getting too much in the weeds what led to the collapse of Enron?
It was the way that the debt okay or private credit was being structured. And what they were doing is they were using these special purpose entities basically accounting shenanigans. Enron created hundreds of offbalance sheet entities. The purpose was to hide the debt, inflate the earnings, and push losses out of sight.
Hold that thought. I'm going to get to what's happening right now with the funding of trillions, billions, and it's gonna be trillions eventually of dollars into the AI data center credits or debt obligations. So, uh, purpose of hide debt, inflate earnings, and push loss out of sight. Management treated the special purpose entities like a trash bin.
Bad assets went in, fake profits came out. to share with you what Michael Barry is saying uh in the last few days just before he closed his $155 million uh asset management firm about the hyperscalers in the AI and what they're doing with their allegedly doing with their earnings that's going to cause this whole thing to come down.
Like I said in the thumbnail title, this Enron 2.0 is underway. Okay, it's going to make like Enron 1.0 seem like a warm-up act. And of course, they were playing mark-to-market games. Again, I'll share this this what's happening right now. And Barry's warning. Enron booked multi-deade energy contracts as instant profits on day one based on fantasy valuations.
Revenue numbers became fiction supported by no cash flow. The last few days I've been sharing with you all these phantom um you know entities that are basically not or they have tremendous debt obligations but they're not showing any net profits or real cash flow. Of course Wall Street and all these you know darlings these AI stocks are going up on the expectation that the AI data centers will eventually produce massive cash flow.
This is once Wall Street expects continuous growth, you're trapped and and once one quarter is bad, the whole thing collapses. Okay, so that's uh what basically happened uh was going on with Enron. Now fast forward to uh late uh 2025 where we're here present time as every day we're seeing massive uh you know well massive a large amount of news stories every day about huge collaborations uh with these uh you know AI companies uh and tech companies and what they're doing to finance this uh this AI trade to build build out these data centers.
Case in point, out of Reuters just on November the 4th, um Meta agreed to a $27 billion financing deal with Blue Owl Capital. Blue Al Capital, another one of these non-depository, non-bank financial institutions, right, for its biggest data center project using a complex structure. And here we go, that keeps the debt off its own books.
So Meta will be able to keep that debt off its books. Look what look exactly what they were doing back in in the early 2000s with Enron. JP Morgan estimates AI lint companies account for 14% of its investment grade index surpassing US banks as a dominant sector. All right, think about that. 14% of JP Morgan's investment grade index is linked to AI companies.
you guys think this is not going to be massive as it uh collapses. Um so let me explain to you what the big tech firms are doing and I'll share with you two stories uh that I think are very important. So hang with me. Uh if you guys enjoy the content, please subscribe to the channel. If you're not subscribed, I appreciate everyone uh watching.
And to those subscribers that have been with me from the beginning, I appreciate you. To all the new subscribers, thanks for being here. So big tech firms are raising billions of dollars via private credit and structured vehicles. Okay? Rather than their traditional corporate debt lines. All right? And these deals are structured through you you you uh guessed it SPE special purpose entities or SPV special purpose vehicles.
Same thing. Joint ventures. Just talked about Meta's joint venture with Blue Al. Okay. So uh the tech company contributes the asset the data center compute hardware into the SPV private credit funds provide debt. Okay. And sometimes uh the they try to keep well they do it again it says to keep the debt off its balance sheets the parent company like Meta.
Okay. Accounting and regulatory sources confirm that offbalance sheet entities allows liabilities to be hidden from the main balance sheet and this kind of structure was historically abused. Here we go. Enron. Okay, that's according to Investopedia. So, let me share with you the two stories I have that bring this full circle and why it's important that everybody pay attention to this.
All right, you think Enron's losses were steep. Look at Enron was $80 billion. Right now, Meta alone, just their partnership or a joint venture, right, with Blue Owl is 2728 billion. All right. And it's going to keep going from here. So, out of the business insider just a couple days ago and um this got some coverage on mainstream media, but not too much.
Michael Barry the the the big short latest argument against hyperscalers like Meta and Oracle echoes one made by an investor who shorted Enron Barry who recently came out of social media hiatus says he's concerned about hyperscalers the mega cap tech firms like MetaOracle Microsoft that are shelling out big on AI infrastructure okay u those investment plans could result in painful losses for Those companies Bur suggested given that semiconductor chips like Nvidia's have a relatively short lifespan and don't forget they're booking on a much longer lifespan to generate massive returns on these investments to pay off these debts.
Massively ramping uh capital expenditure capex through purchase of Nvidia chips servers on a two to three uh year product cycle should not result in the extension of useful lives of compute equipment. Yet this is exactly what all the hyperscalers have done according to Bur. Let me read to you what a hyperscaler is.
Okay, I had to dig into it a little bit myself. So hyperscalers are the mega cloud. This is from chat GBT are the mega cloud operators that build and run massive industrial scale compute infrastructure. Think Microsoft, Azour, Google Cloud and Meta, OOracle and Alibaba. Think about these the massive size of these companies involved.
They are called hyperscalers because they can scale compute they storage and networking capacity globally on demand at it at insane volumes. So who are the hyperscalers? Number one, the developers. They build the biggest AIcentric data centers on the planet. Number two, the buyers. They are the largest global buyers of GPUs, networking gear, and power.
They sign multi-billion dollar orders with Nvidia, Broadcom, Cisco, etc. Okay, the circle of doom, right? Number three, they are the customers of private credit. Here we go. Hyperscalers often need fast capital to expand AI data center infrastructure and the and they basically go to the private credit funds to get the money and the private credit funds get the money from JP Morgan life insurance companies pension funds right and of course the hyperscalers also become the tenants they are usually they don't own the facility they lease it back remember I shared yesterday about coreweave it was going to be the tenant and Some of the creditors in that deal are questioning whether or not Coreweave will generate enough income, right?
Because they're losing now. They're losing their butts. Will they be able to generate enough income to service their debt obligations to pay their lease? Uh one of their obligations pay their lease in these huge AI data centers. Okay. So that's who that's what a hyperscaler is and and and what the the different let's say um they are the buyers they are the tenants they are the customers uh of the private credit.
So continuing with the article from Reuters um where Barry is saying he estimates that Meta, Google, Oracle, Microsoft and Amazon, five hyperscalers and some of the biggest names in the AI trade would understate this is his position. Uh but you know, do you think it's coincidental that he closed his fund down and he's announced now something is going to be coming out uh a big announcement on November 25th of this month just here in a few days.
Says that they would in those companies he's alleging would understate depreciation by $176 billion between 2026 this next year and 2028. This is going to come down folks. By 2028, Oracle, he says, will likely overstate its earnings by around 26% and Meta could overstate earnings by around 20%. He speculated without elaborating on the calculations.
But it gets worse and he says more details coming with an announcement on November 25th. And then in closing on this article, his warning sounds. It says similar to what Jim Chanos, the famed Enron shortseller, said about the AI trade. And Chanos says uh he was concerned about the billions of meggaap tech firms.
H the billions that mega cap tech firms are shelling out on AI hardware. Amazon, Meta, Microsoft, Alphabet, and Apple are on track to spend a collective $349 billion on capex this year. much of which is directed towards AI infrastructure. And Chanos, the Enron shortseller, said, "We're getting to the point now where within a year or two, some of these large companies are going to start having to make some very uncomfortable decisions as to when and how they will monetize AI and what the returns will ultimately be on this massive spending." He says, "Because the chips basically depreciate over two years, I will tell you to guarantee you those lease obligations." Okay.
Um are much longer than two-year terms on these massive AI data centers. So he says Wall Street analysts have been thinking well the article concludes Wall Street analysts have been thinking about the depreciation issues since at least last year. An analyst told business insider in November that the impact on earnings from how costs are calculated and how hardware depreciates is a potential time bomb at the heart of the AI trade.
Investors are taking stock of the market's monster gains. Many also recognize that AI monetization plans for many firms is unclear. It's unclear that let you yet you have JP Morgan allocating 14%, okay, of that one category into this industry. You've got massive funds flowing in from all the big big companies, life insurance companies.
So guys, this is um this is serious serious serious serious stuff. And then uh another article out of Reuters from yesterday uh Michael Barry the big short closing his hedge fund. All right. And he basically said that uh he's closing SkyOn asset management and he says my estimation of value in securities is not now and has not been for some time in sync with the markets now.
Yes, he is a short seller. Okay. Uh one one of the biggest names out there. So he gets a lot of flack for it. But can you honestly tell me that he's not on to something here? Come on. I mean, if you see how this is all laid out, they have no plans to really monetize this stuff. And if the let's say the um the useful life of this hardware uh is a couple of years, uh how are they going to repay the debt?
And when's it going to explode? 2026, 2027, 2028? My estimation of value, he says, has been for some time not in sync with the markets. It's a polite way of saying that this thing is gonna crash. Okay. So he closed down uh Skyion which managed 155 million in assets as of March. He says on to his quote was from X Bur says on to much better things November 25th.
That'll be interesting to see what it is that he uh that he uh comes up with. Okay. And again we shared I think on a video last week the news was out. Michael Bur took a short position against Palunteer and against Nvidia. All right. So you you don't think for a second he and his teams are not laying out and seeing uh you know what is happening here.
Um you're going to hear the term and one of uh the subscribers uh brought to my attention yesterday. You're going to hear the term uh uh data center credits come up. And so I pulled what that is real quickly. I'll read you the definition uh because it's confusing a little bit. Data center credit in this AI context simply refers to debt financing.
That's all it is. Uh it's not equity not ownership. Okay. There's no credit. It's loans, notes, and financing facilities made to data center operators or data or data center projects. Okay. It says there's no equity upside. You're underwriting the borrower's ability to service debt from data center cash flows.
And so it comes full circle. There is a huge assumption and huge bets being placed right now, billion and trillion dollar bets that the cash flows will be there to support all of this. I don't know like to get your thoughts and comments. We're a little bit long today, but I wanted to, you know, you know, delve into what really is a complex a bunch of overlapping concepts uh and models that appear to be a patterning exactly off of what happened with Enron.
So to say that no lessons are ever learned is an understatement. The $80 billion that was lost during Enron. Uh yeah, I don't know. Pick any number you want. What do you want the number to be? 80, you know, uh 500 billion, a trillion, five trillion, 10 trillion. What will the losses be? Okay. How many people will get wiped out?
Uh you know, of their equities, of their of their retirement portfolios. I don't know. I know that sounds doom and gloom, but this is actually happening. Thanks for watching. If you enjoy the content, leave me a like. Subscribe if you're not subscribed. uh thoughts and comments. I do uh enjoy uh receiving those and reading them, time permitting.
It's getting a little bit more difficult now to keep up with you guys, but I appreciate it. And with that being said, I will talk to all of you soon. Bye.
Comments 0