If I understand the explanations on HN, the complaint is that Meta is taking on debt, which would normally affect its credit rating, so they're "hiding" the debt in a LLC without materially changing anything. Thus, alleging that Meta is "faking" a higher credit rating than it should have.
However, it looks like this construct might actually protect Meta against the main two risks that might make the datacenter be unprofitable (force majeure like a disaster destroying it, or a collapse of datacenter demand), i.e. keeping the good credit rating may be justified because the construct is actually very different, protecting Meta from risk, even though the article suggests that it's just a fig leaf?
The issue is the short, 4y renewal cycle, which allows Meta to attempt superficial arguments to avoid accounting consolidation for the variable interest entity that they...pretty much (a) control and (b) have skin in the game.
Now that am thinking about it (not accountant) does a lease show as debt or not?
It is a joke. This is a humor post on a comedy blog. This substack is not actually a bond rating agency.
I've seen at least two different commercials each focused entirely on the personal story of a relatable, folksy person living in a small town in a fly-over U.S. state, talking about how the town was declining and times were hard - then Meta built a new data center nearby and this person along with many others got jobs there and now things are great. They are very well-produced with cinematic shots of rustic small-town main streets, dusty pickup trucks in rural settings and local high school football games. Aside from the obvious brand-washing, it would be extra on-brand if it turns out Meta doesn't even own the data center but still tries to take credit for it.
Creating such bustling workplaces as https://maps.app.goo.gl/fc9AGtsVwiLA1vd88 https://maps.app.goo.gl/fHvTWK4rWqrsqsmr9 https://maps.app.goo.gl/RzggPfd3xbBQbdoo6 and https://maps.app.goo.gl/MBjun6ad4zJmmrRV7
These facilities will sometimes employ as many as 100 people - so a state that can attract three such data centres creates almost as many new jobs as an entire wal-mart store. Truly, a transformative number of jobs.
https://www.youtube.com/watch?v=xCVkA1xebrQ
It turns out the one in this ad is in Altoona, Iowa. The ad focuses on how it revitalized the community by providing jobs, kind of glossing over how that might be reflected in the massive facility's ~30 car parking lot.
And incidentally, that data center currently shows no open positions on Meta's career website, although third-party sites still have some dated listing for advanced IT positions that were probably filled by non-locals.
Ugh.
Lots of construction workers in the areas where they're putting up new buildings, though.
Also given the 24/7/365 nature of data center operation, the number of employees will be larger than the number of parking spots.
[0] https://corridorbusiness.com/data-centers-bringing-big-numbe...
[1] https://www.pa.gov/content/dam/copapwp-pagov/en/dli/document...
What? Me exaggerate? Never! :-) I did actually spend a minute looking for the commercial on YouTube but YT search sucks, so thanks for finding it. Setting aside the fact it's crassly manipulative corporate propaganda, as someone with a lot of film and video production experience, the production team that made it did very nice work. That's the only reason I actually ever even saw it. I normally skip through all commercials on the DVR, but I happened land on a couple of really nicely shot frames as I was skipping and rewound to see the spot. I thought, "Wow, blatant bullshit but great work!" :-)
"If you add two pounds of sugar to literally one ton of concrete it will ruin the concrete and make it unable to set properly which is good to know if you wanna resist something being built..."
> The bonds for the Hyperion data center priced with a coupon of almost 6.6%, roughly a percentage point higher than Meta’s outstanding corporate bonds and in line with the average junk bond. That’s a higher yield than investors would expect given that S&P rated the Hyperion bonds A+, safely within the investment-grade spectrum.
Apparently the bond market is pricing the guarantees made by Meta to this other entity as not quite as good as bonds that Meta issues itself, and Meta is willing to pay the higher interest rate. So, not entirely a free lunch?
I guess sometimes a company wants to issue junk bonds and its rating gets in the way.
I don't know enough about finance to tell for sure, but this seems backwards?
Wasn't that the root of the 2008 crash? The debt spiral was acceptable because people were making enough money in the present that regulators were powerless to advise against it. In a sane world people often go to jail for decades when doing this at pennies on the dollar.
On the other hand, Meta has great creditworthiness. And guarantees this vehicle. So... it's not the same.
This is debatable but subprime loans were mostly accurately rated. They were rated very low. That low rating was the ultimate precursor to the crash, because it means banks carrying those poorly rated vehicles needed to balance them with different highly rated vehicles to keep their own rating high enough to qualify carrying and lending other financial assets on their books. There were so many of these shitty loans that they had to repackage them to dilute their value/rating against their other highly rated assets, because there are limited number of highly rated assets any given bank could acquire at a moment.
That dilution was called a credit default swap, which is bundling under the guise of an insurance vehicle. This magnified the problem for two reasons: First these shitty assets can now be traded in large bulk and secondly any given bank can now carry more of them before further eroding their value. That proved catastrophic because this toxic debt could not be moved fast enough by anybody that held them. Its like hot potato or musical chairs, like Bitcoin. The only real difference between those credit default swaps and Bitcoin is only that everybody knows Bitcoin is intrinsically worthless and only exists as an instrument of speculation while many people actually thought these credit default swaps were real financial assets and that they were insured.
I mean, yeah, but at the same time, and?
The lesson learned from 2008 was that no one was going to do anything of consequence to degenerate gamblers who kneecapped a generation's economic prospects. Then, in 2024, we doubled down on that position.
The behavior will continue until an effective consequence is introduced.
(Clarification: I used a diabrowser.com feature to clarify the article, which uses ChatGPT underneath)
==============
Meta wants to build a huge AI data center campus in Louisiana. It costs about $28–29 billion. Instead of just borrowing the money itself and putting the debt on its own balance sheet, Meta uses a maze of LLCs and contracts to:
- Get $27.3 billion of debt raised by a special company called Beignet Investor LLC (80% owner of the project).
- Keep that debt off Meta’s official balance sheet, even though:
▫ Meta designs the campus,
▫ pays for overruns,
▫ pays the rent,
▫ guarantees the value at the end,
▫ and will basically be the only user.
In real life, this is basically Meta borrowing to build its own data center. On paper, it’s “someone else’s” debt.
Why is this off-balance-sheet?
The accounting rules say you only have to put an entity on your balance sheet if you “control” it and take on most of the risk/benefit.
Meta’s position is: “We don’t control this JV company, even though we do all the important things and take on all the risk.”
The rating agency in the piece is mocking this. They list all the ways Meta obviously controls and supports the project, then say: under current accounting rules, if Meta insists it doesn’t control it, we all politely pretend that’s true. So the $27B debt doesn’t show up on Meta’s balance sheet, even though economically it’s Meta’s problem.
The "fire sale prices" would be so delicious as to guarantee that the entity(-ies) involved stay solvent as long as meta stays solvent.
So that they will lend you the money...
It's not always required, depends on the amount and the business.
Nobody is trying to pull one over on a bank here. Pricing the risk of the loan is a bank’s whole business, they’re happy to loan to meta because meta is meta, and they’re a good candidate for a loan.
I'm asking how you would believe this vehicle would go broke, which is the usual reason to go to bankruptcy.
As an example to stimulate your imagination, Walmart has settled as recently as 2019 to resolve liability due to weak internal controls that allowed “third party affiliates” to bribe local officials and others in various ways.
https://news.ycombinator.com/item?id=45628186
But the one thing that doesn’t compute is the commitment. There is a long term obligation now incurred by meta to use this infrastructure. If it’s a capital lease I assume this is now a liability on their books (and disclosures)?
Fade-Dance had a fairly reasonable answer to it:
Maybe they don't want to securitize their core assets and introduce a new favored class of investor. Ex: If they are securitizing their AI data centers as part of the initial capital raise, those investors would be higher up the capital stack. They would get the datacenter in a theoretical bankruptcy before the bond/equity holders got their cut of the liquidation. Intel securitized their new fab builds with Brookfield and Apollo and, as a shareholder at the time, it didn't feel great. No idea what the precedent is regarding Meta by the way, just a thought. Maybe they think that the lenders are a bit "overzealous", and they want to push the risk of things like write down on GPU racks entirely onto external parties who are apparently all too happy to take the risk. I'm guessing it's a mix of both, combined with the fact that we're seeing some copy and paste thinking. This is proving to be a way to get fast access to the huge private credit market. I would assume there must be some very wide deal flow pipes cranking currently, so why not tap into them if the demand is there in the other end.
Meta wants to fund this project, but doesn't want the debt on own its books (because it would impact its vanity AA credit rating). Debt investors are happy to finance a special purpose vehicle guaranteed (in a non debt way) by Meta at a credit rating almost as good as Meta's (say, A). No one is confused this is Meta getting financing for their own project; they've just put it in a wrapper for vanity credit score reasons.
Levine wrote about it and his writing is better than ChatGPT, this snarky website, and obviously mine: https://www.bloomberg.com/opinion/newsletters/2025-10-29/put... .
So it's probably valuable to retain that credit rating.
The real issue here is how simple it is to game the rating agency in this way and how the market allows Meta to "launder" this activity through the ratings agency.
This is, in fact, a fairly close analogue to the housing crisis and the ratings laundering that was done with the CDOs[1]. The difference is, instead of drilling down to thousands of mortgages - each with different characteristics - you really just drill down to Meta ... which might not be too risky ...
[1] https://en.wikipedia.org/wiki/Collateralized_debt_obligation
…someone needs to shake the tree and see what falls out, like Peter Thiel did for SVB.
So if a company drops their AA rating it could force them out of a lot of funds and investment vehicles.
This complicated vehicle where the debt and assets are in another LLC isn’t actually tricking anyone in finance. If you’re reading about it from blogs then it’s already common knowledge. The structure isn’t actually a one way trick, it’s a set of tradeoffs and protections for the company. They probably could have achieved better terms going direct but with higher risk.
Surely the ratings agency people are "in finance"? Or are they in on the game, and sliding their way back to 2008, writing ratings for "deals structured by cows"?
The mistake throughout this comment section is to assume that the debt is functionally equivalent to Meta haven taken it on themselves, consequences and all. It's not.
Putting things in an LLC vehicle provides some protections. Both for large corporations and you and me as individuals. However if you put an asset in the LLC the lenders also know that those protections exist and will adjust terms accordingly. Meta has taken this into account, found some favorable terms, and pursued that direction.
The narrative that this is a secret loophole that lets them take on debt but also not take on debt is the substack authors doing their thing to make it more sensational than informative.
They will also have to pay a premium and give up more for debt to the LLC because the lenders know this.
The same is true for Meta.
The finance world isn’t blind. None of us hear are stumbling upon hidden knowledge that the lenders didn’t already have.
Meta is borrowing a whole lot of money and they're lying about it to investors.
https://archive.ph/2015.11.08-145615/http://www.wired.com/20...
Just desperate, stupid, or naive lenders trying to get solid returns (and convincing themselves there are no major risk).
Just like ‘08, frankly.
Have enough lawyers, and you can make almost anything legal and aboveboard, no matter how sketchy it actually is. Buyer beware!
Meta: (hiding debt behind its back) No. It's Jimmy's.
Investor: Now Meta, you know lying is wrong.
Meta: No it's not. All the kids do it so it's OK.
It's just a more advanced crypto fraud.
It’s easier to think of this as “project risk” as opposed to corporate risk overall.
This isn’t different than creating a subsidiary to embark on a new program, with its own debts and assets, collateralized by a parent company.
It’s effectively the same as what happens every time a major movie studio starts a new film project.
That’s how you can decide if this is disingenuous or not. If Meta is obligated to repay the loan and used to synthetic means to get it off the balance sheet that’s a problem.
If they have in fact successfully transferred risk to other parties then that’s what deals like this are for. It’s the whole reason the concept of limited liability exists.
I am fully willing to believe it’s the former. But that’s the test.
If those have been offloaded to the LLC, wouldn't that be a pretty key difference?
They aren't, but they're obligated to pay leases for it (they can't just build the datacenter and then walk away), which is kind of like having to repay the "loans".
If the Generally Accepted Accounting Principles don't require that to manifest on the balance sheet, then it sounds like the principles aren't very good ones.
Being conceptually similar isn’t the same thing as identical.
Basically, if we’re reading about it from substacks and Matt Levine’s newsletter then it’s already fully common knowledge in the finance world.
So ChatGPT put this sentence in list form and reordered it a bit. AGI is imminent!
I personally think it’s a great response and makes it clearer what’s happening
Times are changing quickly!
It is definitely not taboo to say you’re writing your own code.
Can you link to another one?
> (…) this is functionally Meta borrowing $27.30 billion for a campus no one else will touch, packaged in legal formality precise enough to satisfy the letter of consolidation rules and absurd enough to insult the spirit.
> The structure maintains a precarious technical separation that, under current interpretations of accounting guidance, allows Meta to keep roughly $27 billion of assets and debt off its own balance sheet while continuing to provide every meaningful form of economic support.
Downstream of the capex to build the data centre is, presumably, a sister capex to build a power station. At what stage do these come hand in hand? Or does this financing include provisions to pay the electricity bills for the next ten years which, in turn, gets used by the power company to finance the construction of a new power plant? The power company gets some kind of heads up?
If I finance the construction of a mile long dinner table due for late November 2026, presumably some of that had to trickle down into a local turkey farm, lest everyone go hungry?
Mostly things like this, yeah. The hyperscalers don't want to get into the power business.
Pedantically, that's one ninth of the Three Gorges Dam. One tenth would be the 0.3 Gorges Dam.
So? As usual, xkcd 1053 applies :) https://xkcd.com/1053/
> I should say that the big tech companies did not invent this technology to build AI data centers. This sort of thing — project finance, non-consolidated joint ventures, borrowing out of boxes — has a long history in a lot of capital-intensive industries.
Levine attributes a recent increase to private credit.
https://www.bloomberg.com/opinion/newsletters/2025-10-29/put...
The problem is that even standard practice, without nefarious intent, can cause massive financial collapse. If, say, the vast majority of economic growth were being focused into such vehicles, the lack of transparency could make people misanalyze the situation and result in bad valuations that collapse when it all becomes transparent.
RE throws off much income.
It's also difficult to transfer the RE to another entity without realizing gains.
ChatGPT isn’t really a solution because the source is both low quality and has questionable motives. Going to any of the other good articles on the subject that have been linked in this comment section is much better.
The target audience is people who want to be angry at Meta and think that they’re smarter than finance people.
> We assign a preliminary A+ rating to the notes, one notch below Meta’s issuer credit rating,
It’s hard to get away with that when the report is attributed to a company and person which don’t seem to exist, hosted on some randos substack. Wording like that works way better when it comes from a sender with an address ending with @bigbank.com
Of course, the latter parts of the post (Disclaimer and Limitation of Liability) do reveal pretty definitively that this is obviously not intended to be a serious report.
As for the content itself? The author tries really hard to turn a whole lot of nothing into something, and horribly misinterprets the GAAP in the process.
Things that are hard to read because you lack context is not the same as poor writing.
The actual subject matter has already been covered well by good writers like Matt Levine, WSJ, and others.
What on earth does your second sentence have to do with the quality of the writing? Try just a bit to separate your emotions from the text.
Not to mention that asking help to explain a text is extremely common. I can read English, but I have never read a US supreme court ruling. There are much better ways for me to understand those rulings to me as a non-lawyer.
The most publicly notable cases (on things like abortion, gerrymandering, gun control, etc.) aren’t so tied down in complex precedent or laws the average person is unfamiliar with.
Although, even some of those (like, for me, issues around Native American sovereignty or maritime law) are quite readable as well.
Having admitted to never having read a SCOTUS ruling, how can you then proclaim there are better ways for you to understand? How could you possibly make that assertion if you've never read a SCOTUS ruling?
A SCOTUS ruling is a primary source, and there's a pretty good universal rule that primary sources can be difficult to properly digest if you don't fully have the context of the source; for most people, reading a secondary source or a tertiary source will be a superior vehicle than the primary source for understanding. Although that said, some secondary and tertiary sources do end up being just utter garbage (a standard example is the university press release for any scientific paper--the actual merits of that paper is generally mangled to hell.)
1. Facebook creates a shell company.
2. The shell company borrows billions of dollars, and builds a data center.
3. Facebook leases the data center.
4. The fact that it is technically only a four-year lease with only one possible tenant can conveniently be ignored, as Facebook assumes essentially all possible risks. The shell company could only possibly lose money if Facebook itself goes under, so the lenders can treat the loan as just as reliable as Facebook itself.
5. Because Facebook technically only has a four-year lease, it can pretend it doesn't actually control the shell company: after all, it can always just decide not to renew the lease. The fact that is assumes essentially all possible risks can conveniently be ignored, so Facebook can treat it as a separate entity and doesn't have to treat the debt as its own.
So the lenders are happy because there's no real risk to them, and Facebook is happy because they can pretend a $27B loan doesn't exist. It's a win-win, except for the part where they are lying to their shareholders about not taking on a $27B loan.
And then going to the comments, excitedly no less, to find…this?
Jfc :’(
What's very interesting to me is what happens when Meta doesn't exercise those lease options. If there isn't some kind of penalty for declining the option, well...
As has been mentioned though if you purely want the info there are more succinct articles out there, e.g.: https://www.forbes.com/sites/petercohan/2025/11/25/metas-ai-...
https://bsky.app/profile/mailia.bsky.social/post/3lwys6d6r6s...