Isn't the point that they intentionally glossed over ("We did not model what would happen if data center demand collapses and Meta cannot secure a new tenant. This scenario was excluded for methodological convenience.") a pretty important one?
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?
My understanding is that the residual value guarantee only covers "most modeled cases" and this case (which might be the only one where the datacenter could not be meaningfully utilized and the guarantee would be relevant) was the not-modelled one...
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.
How is this different from a 5-10 year lease? You can sell the car to someone (or the construct) and he continue to service the debt or you can service the debt.
Now that am thinking about it (not accountant) does a lease show as debt or not?
>Isn't the point that they intentionally glossed over ("We did not model what would happen if data center demand collapses and Meta cannot secure a new tenant. This scenario was excluded for methodological convenience.") a pretty important one?
It is a joke. This is a humor post on a comedy blog. This substack is not actually a bond rating agency.
I sent it to my friend that works in corporate accounting and she thought the post was hilarious, so I guess the intended audience is pretty narrow? ¯\_(ツ)_/¯
That aside, I don’t think that this post by a made up bond rating agency called the Flexible Standards Group that uses phrases like “unbothered by reality” “downgrade when the shit hits the fan” is particularly difficult to parse as being humor (or at the very least not an actual bond rating)
One of the reasons these things are getting financed is the belief that the purchased hardware wont depreciate significantly. Which is bizarre when you think about the hardware production latter.
For these datacenters to keep value without the computational demand, you also assume new hardware prices & capability will stagnate.
I still dont see the "total addressable market" being large enough to satisfy even one monopoly.
>you also assume new hardware prices & capability will stagnate
Well shall I break the bad news to you.. for the past few years this is starting to look more and more true, for server hardware at least (CPUs, DRAM, SSDs). Sure, you can get more (cores, GB of DRAM, TB of NAND) now in a same package, but also at a higher cost.
It would be deeply ironic if this data center (or similar ones using creative accounting), are among those featured in the TV commercials Meta has been running in expensive national prime time slots in recent weeks.
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.
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.
Yep. Massena, NY totally got hundreds of jobs out of the shitty Bitcoin miners that came to town like Coinmint and whoever the hell "North Country Colocation Services" is. We were saved! Everybody is rich and healthy and happy!
... wait, no, we're not. We're still an absolute shithole at the top of New York, now with a bunch of sea cans sitting in front of abandoned industry we lost decades prior, humming away doing nothing for any of us.
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.
30 sounded low to me, but looking at the sprawling Altoona facility in Google Maps https://maps.app.goo.gl/KGLEpJRFiwVKYob89 satellite photos show 52 parking spaces in use across 11 buildings.
Lots of construction workers in the areas where they're putting up new buildings, though.
That data center started operation over a decade ago. There's no reason to assume it should have any vacancies currently, nor is there any reason to assume its positions were largely filled by non-locals. And even if a few external folks relocated in some cases, isn't that economically beneficial to the area regardless?
Also given the 24/7/365 nature of data center operation, the number of employees will be larger than the number of parking spots.
Meta says over 400 people work on-site at the Altoona facility [0], but most of those are clearly working for a variety of smaller contractors given that the initial tax terms anticipated a few dozen direct employees [edit: wrong state] and no datacenter companies show up in the county's 50 largest employers [1].
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.
If the article is correct and they are 144A then they will also be a little less liquid. But yeah, I have to imagine everyone involved knows what’s up. Just happens to work for everyone (for now).
The Forbes article says that "to be an operating lease [...] Meta must have the obligation to absorb the venture’s losses or the right to receive its benefits."
I don't know enough about finance to tell for sure, but this seems backwards?
> This treatment is considered acceptable because the people who decide what is acceptable have accepted it.
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.
The 2008 crash was in part caused by inaccurately rating synthetic bundles of subprime mortgage debt as extremely low risk (e.g. AAA). Subprime borrowers had a much higher risk of defaulting than a AAA rating implied.
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.
Right, this is more like wash trading, in that companies like Meta are trying to syntehtically make it look like there's more assets involved in AI than there really are.
Isn’t part of its creditworthiness how much debt it’s carrying? And if it’s shifting that off of its balance sheet, then it appears in better shape than it actually is.
Until they don't. Lest we forget that Facebook's new handle was borne out of a pivot that's sunk almost a hundred billion dollars while being having been largely sidelined/abandoned. (I know that they're still doing a good deal of R&D, which is good, and a worthy investment, but 1) Carmack left, and 2) We apparently don't judge corporations on whether or not they're contributing to society, but only on whether they're in the red or black.)
> 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.
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.
It’s buried in the article but this about a debt vehicle created to finance a “2.064 GW hyperscale data center campus”. That’s approximately equivalent to a One-Third-Gorges Dam (one tenth of the Three Gorges Dam.)
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?
> 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?
Mostly things like this, yeah. The hyperscalers don't want to get into the power business.
They also need artificial demand to keep the valuations of their data centers high, so generating multiple business interests creating wash trading benefits their interests.
Relevant excerpts to understand what's at play here:
> (…) 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.
This is HN, you're supposed to say "is anyone surprised" and sprinkle in the word Nothingburger and other naval gazing platitudes I'm sure I can think of
"Nothing is created. Nothing is contributed. It’s a loop. Borrow money, earn interest, and use the interest to claim you provided equity. The kind of circle only finance can call a straight line."
Meta's accounting games are entirely reminiscent of Enron, who famously named their off-balance sheet debt-hiding special purpose vehicles after Star Wars "Jedi 1, Jedi 2," Jurassic Park, "Raptors 1 through 7," and the crooked CFO's kids "LJM" etc.
AI companies are running the same frauds as multiple, but I think cryptocurrency/FTX is more apt. They're creating artificial demand by trading contract with themselves and using those assets to make it look like they've got more revenues & assets of value.
It definitely has the Voltaire/Onion like snark and cynicism with biting accuracy that really gets me going. We need more well informed rants disguised in heavy sarcasm
This article is poorly written. It’s so desperate to be clever and edgy that it’s hard to get the facts out of it.
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.
While I’ve seen a plenty of silly reports from big bank analysts, they usually have the advantage of not coming across like complete idiots when saying things like this
> 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.
How does it not have anything to do with the quality of the writing? The writing is supposed to convey some facts, but it's too busy pushing narratives and layering on snark that it fails to convey real facts. Even in this comment section the people who applaud the article don't really understand what's happening because they soaked up so much of the narrative-pushing from the article.
this is the future of human-written articles - they will obligatory be written like this as 99% of article comments on HN these days is “oh, this is AI written.” :)
It is not the reader's fault if the article is unreadable in the first place.
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.
Many SCOTUS opinions, especially the major ones, are very readable! The justices and clerks are excellent writers.
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.
> 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.
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?
> 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.)
The difficulty understanding this piece comes from lack of knowledge about finance and ratings, not from an inability to read. The blog assumes a large amount of financial knowledge which is not common among the HN audience.
It seems fairly understandable even without financial knowledge?
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.
Monroe LA is the former headquarters of Lumen, they realized that their corporate headquarters was a white elephant and donated it to the local university I think. However that means there is available power capacity from the local power company and of course, fair amounts of fiber optic cable nearby.
This isn't really about fooling lenders, they already underwrite this as Meta risk i feel. What is wild is how much economic substance you can legally shove into these non‑consolidated boxes, while GAAP still pretends form wins over who actually eats the downside
If this becomes the default template for AI capex, headline leverage stats for most of big tech are going to drift further and further away from their real balance sheets
> 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.
You would guess right, and I have even heard that this sort thing has been standard practice for a long time, without nefarious intent.
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.
I asked ChatGPT to make this more readable since it's a mix of satire and actual information:
(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.
I have skimmed through the article and if I get the details through all the humor, satire and sarcasm even remotely correct, the major assets are actually the duality of payment obligations and residual value guarantees, both from meta. One could include cost overrun protection at the construction time too.
The "fire sale prices" would be so delicious as to guarantee that the entity(-ies) involved stay solvent as long as meta stays solvent.
I thought the whole point of LLC was to limit liability so you wouldn't be liable for debt beyond your paid up capital? Why would you ever sign a personal guarantee?
Why would the CEO have the personal liability here and not the board? Does Sundar Pichai have to personally guarantee loans for Google? That would be weird since the CEO could be fired.
“If you are meta” in this case means “if you have a billion dollars already, and a credit rating that you don’t want to destroy.
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.
Meta has $44.4 billion in cash-on-hand as of September 2025.
I'm correcting you because people can't really fathom just how wealthy these companies are. They could buy startups for billions of dollars with literally the cash they have in their bank accounts, let alone debt or stocks.
Do you think any CEOs of gigantic corporations are personally liable for any loans made by the companies they work for? I would be incredibly, incredibly surprised to hear if that's the case.
Then why don't they do it? It's the easiest money they can ever make. Even I can litigate the hell out of it if I get $27B, take the money and close the LLC.
Meta would have to not renew the lease and somehow nullify the residual value guarantee. This would leave the LLC with no revenue at all. If the RVG works there should be no chance of bankruptcy.
Corporate bankruptcy happens for a lot of reasons other than being "broke". Chapter 11 is a court-supervised way of restructuring your debt. This has a lot of utility in many situations other than not being able to pay.
Meta may have lots of assets, but the LLC may not. The ability to have one wholly owned LLC go bankrupt by itself is one of the main reasons shell corporations exist.
I think it’s naive to focus on “what is meta getting” from Beignet.
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.
A lot of comments praising this summary, but I'll criticize it: it's still too verbose, and misses the point.
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.
I think "vanity" is the wrong term because their existing credit rating, which they attempt to preserve, impacts all other borrowing (and possibly other agreements and finance vehicles, etc.) that they undertake.
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 ...
Agreed. I know very little about financing but I’d bet if their rating fell that would trigger some debt repayment clause and the house of financial cards might wobble or fall.
…someone needs to shake the tree and see what falls out, like Peter Thiel did for SVB.
There are a lot of places where the credit ratings are hardcoded (to borrow a term) into funds. There are pension funds and other vehicles that might be bound to only invest in AA rated companies.
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"?
They're not being "tricked" in the way these sensational substack posts would like us to think. You're not being given some secret knowledge that the ratings agencies don't have.
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.
Is anyone actually limited to only AA+? The usual meaningful separation is investment grade (AAA, AA, A, and maybe BBB) vs junk (everything else). (I don't think Meta would meaningfully lose access to credit if they were downgraded to A.)
Instinctively I try and simplify things. It this was a person with an excellent credit score, it’s as if the person is taking on extra debt to start to create something they need, but trying to hide it.
But that simplification isn’t the whole story. If that person took on debt as part of an LLC they started, not their personal bank account, then they have certain protections in the event of default of the LLC.
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.
Ehh, tell me the credit ratings assigned by rating agencies to mortgage backed securities circa 2005-2007. Its an ecosystem with misaligned incentives, and some cohort of investor will be left holding the bag. Big Tech, investment banks, and ratings agencies will get off with no consequences when this Jenga-esq capital apparatus eventually collapses.
I don't see what's Jenga-esque about this capital structure. You've got some AA- bonds issued directly by Meta having to do with their core business, and some A+ bonds issued by different entities to fund their riskier and more speculative datacenter construction. If anything, wouldn't it be harder to track the risk if both these bonds were stuffed into the same bucket?
If it's related to AI, it's more like wash trading. The entire business interest in AI is making things look like there's a lot of investment when it's really just a small circle jerk of business interests.
Usually subsidiaries’ debt is not also debt on the parent company, especially when said parent is publicly traded and subject to accounting/disclosure rules.
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.
>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 is a publicly traded company. If they have an agreement wherein they may have to pay substantial amounts to another company or on behalf of another company, that's a liability and they need to disclose that. Whether it's on their balance sheet doesn't really matter, this is what analysts do, back out these types of financial arrangements from the footnotes and publish thumbs up or thumbs down. People like to say "do your own research" but people should not, in general, do their own research.
This is hilarious because I was at the Louisiana public utility commission meeting where the argument was basically it’s Meta borrowing the money so they’re good for it.
Is Meta actually obligated to repay the loans or not?
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.
To me it reads like the article intentionally pretends two major risks (force majeure and datacenter demand collapse) don't exist, by quipping that "rating agencies historically treat as theoretical inconveniences rather than recurring features of the physical world" for the former and explicitly saying they ignored it for "methodological convenience" for the latter.
If those have been offloaded to the LLC, wouldn't that be a pretty key difference?
I don't think Meta has a debt relationship with the loans involved here; that's the point. It does have strong contractual obligations to the wrapper business, though.
Even if they aren't obligated to repay, they have to in practice because it'll impact their ability to get loans in the future. If the shell company declares bankruptcy and gets the loans off Meta's books no one will ever loan money to Meta again.
>Is Meta actually obligated to repay the loans or not?
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".
I'm not an accountant, but "contractually obligated to pay" sounds like a debt to me.
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.
I can't for the life of me figure out who would fund this, other than Saudi oil money or Russian petro-oligarchs, both so they can whitewash or launder their cash. This just makes no sense to me otherwise.
I get what you are saying but this is one of the largest conpanies on the planet asking for what is little more than change for them… it may make just a little sense :)
"None of this is unusual except for the part where Meta designs, builds, guarantees, operates, funds the overruns, pays the rent, and does not consolidate it."
So ChatGPT put this sentence in list form and reordered it a bit. AGI is imminent!
Seriously. I thought about doing the same because I couldn't make heads or tails of the article, and then assumed it would just all be downvotes... glad to see it wasn't.
Serious questions: won't banks and ratings agencies simply treat this as Meta's debt since it it effectively Meta's debt? What changes if this was on their "official balance sheet"? How does playing with the wording actually help Meta overall?
The crux of this article is that they won't treat Meta's debt as debt, because Meta intentionally structured this debt to circumvent the agencies' definition of "debt." Should they change their definition of "debt?" Maybe, but what incentive do they have to do that, is any formal definition bulletproof to circumvention, etc.
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 in the 2008 crash, the ratings agencies were disincentivized to accurately rate these vehicles because they were superficially masked and paid by the companies asking them for ratings.
It seems to me that the lengthiness and opacity of the report is part of the joke, and therefore running it through ChatGPT kind of misses the point. (The "FSG analyst" would have intentionally spread a layer of BS on top of everything to make it a lot of extra work to understand that the debt should actually be on Meta's books. Of course it's satirical so it calls out its own absurdity instead of actually burying it.)
Setting up a separate company for a major construction project is so common that it would be surprising to hear someone didn’t do it. The structure of the OP makes it hard to understand if there’s some more specific aspect he’s objecting to.
Meta (which is short for the metaverse btw) occasionally remembers the metaverse existing, too, whenever there's a small break to be had from the AI stuff.
HN commenters flustered, baffled by the words on the screen: “Why would he say it like that? The phrasing is so foreign, it’s like the author wants me to laugh at it. The only way to understand this is to ask a chat bot what I should think the point is”
I appreciate this serious response to a joke about hn users not identifying jokes in a thread where hn posters are treating a comedy blog like an actual bond rating service
> The Outlook is Superficially Stable, defined here as “By outward appearances stable unless, you know, things happen. Then we’ll downgrade after the shit hits the fan.”
Had the same issue. This is what Claude 4.5 Opus gave me:
This is actually a satirical piece, written as a fake credit rating report. It's mocking how Meta (Facebook's parent company) is using a complex financial structure to keep $27 billion in debt off its official books.
Here's what's actually happening in plain English:
The basic setup:
Meta is building a massive $28 billion data center in Louisiana
Instead of borrowing the money directly (which would show up as debt on Meta's financial statements), they created a convoluted structure with multiple shell companies and a partner called Blue Owl Capital
The trick:
On paper, Blue Owl's company "Beignet" owns 80% of the project and borrows the $27 billion
But Meta still guarantees all the rent payments, covers cost overruns, and promises to pay if anything goes wrong
So economically, it's Meta's debt—they're on the hook for everything—but legally it's structured so it doesn't appear on Meta's balance sheet
Why the author thinks it's absurd:
Meta controls the project, uses the buildings, guarantees all payments, and bears all the real risk
The only reason it's "not Meta's debt" is a technicality in accounting rules
The author is essentially saying: "This is obviously Meta borrowing $27 billion with extra steps, and everyone pretends otherwise because the rules allow it"
The whole piece is written as if a rating agency is giving this deal an A+ grade while openly admitting the structure is ridiculous—it's a critique of how financial engineering and accounting rules let big companies hide their true debt levels.
Isn't the point that they intentionally glossed over ("We did not model what would happen if data center demand collapses and Meta cannot secure a new tenant. This scenario was excluded for methodological convenience.") a pretty important one?
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?
someone has to service the debt, and Meta has obligations to pay if the company cannot find any other tenant - if I understood correctly
My understanding is that the residual value guarantee only covers "most modeled cases" and this case (which might be the only one where the datacenter could not be meaningfully utilized and the guarantee would be relevant) was the not-modelled one...
All scenarios are modelled. It's not complicated.
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.
How is this different from a 5-10 year lease? You can sell the car to someone (or the construct) and he continue to service the debt or you can service the debt.
Now that am thinking about it (not accountant) does a lease show as debt or not?
>Isn't the point that they intentionally glossed over ("We did not model what would happen if data center demand collapses and Meta cannot secure a new tenant. This scenario was excluded for methodological convenience.") a pretty important one?
It is a joke. This is a humor post on a comedy blog. This substack is not actually a bond rating agency.
It's not good humor though.
This blog is in the "no man's land" of satire v. serious. Doesn't pick a lane and people get confused, but it's not funny, "bit the onion" confused.
I sent it to my friend that works in corporate accounting and she thought the post was hilarious, so I guess the intended audience is pretty narrow? ¯\_(ツ)_/¯
That aside, I don’t think that this post by a made up bond rating agency called the Flexible Standards Group that uses phrases like “unbothered by reality” “downgrade when the shit hits the fan” is particularly difficult to parse as being humor (or at the very least not an actual bond rating)
Look at all of the critical parsing here on HN of nonsensical premises, e.g., "we skipped model scenarios of default".
It also has an eerie tinge of AI generated satire wrapping a real article.
Can't have nice things anymore...
One of the reasons these things are getting financed is the belief that the purchased hardware wont depreciate significantly. Which is bizarre when you think about the hardware production latter.
For these datacenters to keep value without the computational demand, you also assume new hardware prices & capability will stagnate.
I still dont see the "total addressable market" being large enough to satisfy even one monopoly.
>you also assume new hardware prices & capability will stagnate
Well shall I break the bad news to you.. for the past few years this is starting to look more and more true, for server hardware at least (CPUs, DRAM, SSDs). Sure, you can get more (cores, GB of DRAM, TB of NAND) now in a same package, but also at a higher cost.
It would be deeply ironic if this data center (or similar ones using creative accounting), are among those featured in the TV commercials Meta has been running in expensive national prime time slots in recent weeks.
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.
> Meta built a new data center nearby and this person along with many others got jobs there and now things are great.
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.
Yep. Massena, NY totally got hundreds of jobs out of the shitty Bitcoin miners that came to town like Coinmint and whoever the hell "North Country Colocation Services" is. We were saved! Everybody is rich and healthy and happy!
... wait, no, we're not. We're still an absolute shithole at the top of New York, now with a bunch of sea cans sitting in front of abandoned industry we lost decades prior, humming away doing nothing for any of us.
I was sure you were exaggerating. But no!
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.
30 sounded low to me, but looking at the sprawling Altoona facility in Google Maps https://maps.app.goo.gl/KGLEpJRFiwVKYob89 satellite photos show 52 parking spaces in use across 11 buildings.
Lots of construction workers in the areas where they're putting up new buildings, though.
That data center started operation over a decade ago. There's no reason to assume it should have any vacancies currently, nor is there any reason to assume its positions were largely filled by non-locals. And even if a few external folks relocated in some cases, isn't that economically beneficial to the area regardless?
Also given the 24/7/365 nature of data center operation, the number of employees will be larger than the number of parking spots.
Meta says over 400 people work on-site at the Altoona facility [0], but most of those are clearly working for a variety of smaller contractors given that the initial tax terms anticipated a few dozen direct employees [edit: wrong state] and no datacenter companies show up in the county's 50 largest employers [1].
[0] https://corridorbusiness.com/data-centers-bringing-big-numbe...
[1] https://www.pa.gov/content/dam/copapwp-pagov/en/dli/document...
The datacenter is in Altoona, Iowa. Your second link is about the county containing Altoona, Pennsylvania. Ditto for your population size number.
Thanks, updated to reflect the correction.
> I was sure you were exaggerating. But no!
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!" :-)
I'm chuckling at the comments for the video:
"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..."
A quote from The Information via Matt Levine:
> 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.
If the article is correct and they are 144A then they will also be a little less liquid. But yeah, I have to imagine everyone involved knows what’s up. Just happens to work for everyone (for now).
There are better articles explaining this: https://www.forbes.com/sites/petercohan/2025/11/25/metas-ai-... and https://www.wsj.com/tech/meta-ai-data-center-finances-d3a6b4...
The Forbes article says that "to be an operating lease [...] Meta must have the obligation to absorb the venture’s losses or the right to receive its benefits."
I don't know enough about finance to tell for sure, but this seems backwards?
it seems to mean that Meta would have to "rent" the whole operation (with all the potential upside and downside)
> This treatment is considered acceptable because the people who decide what is acceptable have accepted it.
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.
The 2008 crash was in part caused by inaccurately rating synthetic bundles of subprime mortgage debt as extremely low risk (e.g. AAA). Subprime borrowers had a much higher risk of defaulting than a AAA rating implied.
On the other hand, Meta has great creditworthiness. And guarantees this vehicle. So... it's not the same.
That's not accurate.
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.
The other parts of the 2008 crisis are even more dissimilar to this scenario than the MBS ratings.
Right, this is more like wash trading, in that companies like Meta are trying to syntehtically make it look like there's more assets involved in AI than there really are.
Isn’t part of its creditworthiness how much debt it’s carrying? And if it’s shifting that off of its balance sheet, then it appears in better shape than it actually is.
Until they don't. Lest we forget that Facebook's new handle was borne out of a pivot that's sunk almost a hundred billion dollars while being having been largely sidelined/abandoned. (I know that they're still doing a good deal of R&D, which is good, and a worthy investment, but 1) Carmack left, and 2) We apparently don't judge corporations on whether or not they're contributing to society, but only on whether they're in the red or black.)
Enron had great creditworthiness, too. They are, famously, a very rich and powerful company today.
Except they're taking on a huge amount of debt, enough that it would lower their credit rating, which is why they're trying to offload it ...
> 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.
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.
[flagged]
Yep any day now!
It’s buried in the article but this about a debt vehicle created to finance a “2.064 GW hyperscale data center campus”. That’s approximately equivalent to a One-Third-Gorges Dam (one tenth of the Three Gorges Dam.)
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?
> 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?
Mostly things like this, yeah. The hyperscalers don't want to get into the power business.
They also need artificial demand to keep the valuations of their data centers high, so generating multiple business interests creating wash trading benefits their interests.
Of course not. We're eating roast beast. (I'm saying that the entire endeavor is a fairy tale that we're misguidedly bringing into live-action.)
> One-Third-Gorges Dam (one tenth of the Three Gorges Dam.)
Pedantically, that's one ninth of the Three Gorges Dam. One tenth would be the 0.3 Gorges Dam.
Relevant excerpts to understand what's at play here:
> (…) 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.
This is hardly a secret. Matt Levine blogged about it: https://www.bloomberg.com/opinion/newsletters/2025-10-29/put...
Does it need to be a secret to be noteworthy, especially if it’s apparently working despite not being a secret anymore?
I meant to say it's not new information. The blog post I linked is from a month ago. It is also more accessible for casual reading.
> I meant to say it's not new information
So? As usual, xkcd 1053 applies :) https://xkcd.com/1053/
This is HN, you're supposed to say "is anyone surprised" and sprinkle in the word Nothingburger and other naval gazing platitudes I'm sure I can think of
And? At least find the previous HN discussions if you are gonna say this is old news.
Unfortunately I didn't find a mention of any mathematical geometry in the article.
"Nothing is created. Nothing is contributed. It’s a loop. Borrow money, earn interest, and use the interest to claim you provided equity. The kind of circle only finance can call a straight line."
Levine wrote about this here: https://www.bloomberg.com/opinion/newsletters/2025-10-29/put... .
This have been covered by FT a while ago: https://archive.ph/zs7ul ( https://www.ft.com/content/d0344253-b0a2-4c6d-8b97-520243678... )
LOL "The entity is named “Beignet,” presumably because “Off-Balance-Sheet Leverage Vehicle No. 5” tested poorly with focus groups."
Meta's accounting games are entirely reminiscent of Enron, who famously named their off-balance sheet debt-hiding special purpose vehicles after Star Wars "Jedi 1, Jedi 2," Jurassic Park, "Raptors 1 through 7," and the crooked CFO's kids "LJM" etc.
AI companies are running the same frauds as multiple, but I think cryptocurrency/FTX is more apt. They're creating artificial demand by trading contract with themselves and using those assets to make it look like they've got more revenues & assets of value.
I assume you have no proof of this, correct?
https://media.licdn.com/dms/image/v2/D5622AQGWzJ28VHg98w/fee...
"A Love Letter to Enron" would sound really bad in the deposition.
It’s a fitting name for Louisiana at least. But this place is next Monroe which is…nowhere near New Orleans.
We get it you want it to be named Boudin
Delta would be a good name. It’s fairly near to the delta in reality.
This entire thing was a masterpiece I love it.
It definitely has the Voltaire/Onion like snark and cynicism with biting accuracy that really gets me going. We need more well informed rants disguised in heavy sarcasm
Folks in the comments here begging ChatGPT to teach them how to read
This article is poorly written. It’s so desperate to be clever and edgy that it’s hard to get the facts out of it.
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.
It's well written for its target audience, people who are used to reading financial analyses.
While I’ve seen a plenty of silly reports from big bank analysts, they usually have the advantage of not coming across like complete idiots when saying things like this
> 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.
Well, I guess it’s my turn to look stupid. The author actually has some pretty serious credentials https://stohl.substack.com/p/credentials-such-as-they-are
Hard disagree. I read a lot of well-written financial analyses and this isn’t it at all.
The target audience is people who want to be angry at Meta and think that they’re smarter than finance people.
It's actually written quite well, you just have to understand the underlying financial documents and methodology.
Things that are hard to read because you lack context is not the same as poor writing.
No it’s not. It’s sarcastic, snarky, sneery content that appeals to a certain group.
The actual subject matter has already been covered well by good writers like Matt Levine, WSJ, and others.
> No it’s not. It’s sarcastic, snarky, sneery content that appeals to a certain group.
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.
How does it not have anything to do with the quality of the writing? The writing is supposed to convey some facts, but it's too busy pushing narratives and layering on snark that it fails to convey real facts. Even in this comment section the people who applaud the article don't really understand what's happening because they soaked up so much of the narrative-pushing from the article.
this is the future of human-written articles - they will obligatory be written like this as 99% of article comments on HN these days is “oh, this is AI written.” :)
Don't say "I'm critical of AI", say "I have questionable motives"!
It is not the reader's fault if the article is unreadable in the first place.
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.
Many SCOTUS opinions, especially the major ones, are very readable! The justices and clerks are excellent writers.
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.
> 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.
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?
SCOTUS ruling: 213 page PDF.
News article: 500 words that provide everything I need to know.
Unless I am actually very interested in the ruling, this seems an easy choice. Because I just wouldn't open that PDF file at all.
> 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.)
The difficulty understanding this piece comes from lack of knowledge about finance and ratings, not from an inability to read. The blog assumes a large amount of financial knowledge which is not common among the HN audience.
It seems fairly understandable even without financial knowledge?
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.
This whole blog reeks of WSB, pretty sure the target audience is not people with a large amount of financial knowledge.
It is so hopelessly depressing. I was wrapt reading it from start to finish and thoroughly enjoyed it as few recent articles at length have been.
And then going to the comments, excitedly no less, to find…this?
Jfc :’(
Monroe LA is the former headquarters of Lumen, they realized that their corporate headquarters was a white elephant and donated it to the local university I think. However that means there is available power capacity from the local power company and of course, fair amounts of fiber optic cable nearby.
This isn't really about fooling lenders, they already underwrite this as Meta risk i feel. What is wild is how much economic substance you can legally shove into these non‑consolidated boxes, while GAAP still pretends form wins over who actually eats the downside
If this becomes the default template for AI capex, headline leverage stats for most of big tech are going to drift further and further away from their real balance sheets
I’m guessing Meta isn’t the only one doing this
Right.
> 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...
You would guess right, and I have even heard that this sort thing has been standard practice for a long time, without nefarious intent.
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.
Yeah, it's very dumb to own real estate like this directly for a c-corp
Is it dumb for tax reasons?
Yes. 21% CIT + 15-20% LTCG for the owners.
RE throws off much income.
It's also difficult to transfer the RE to another entity without realizing gains.
If someone feels this is a sham transaction, can't they buy Credit Default Swaps [0] betting that it will default?
[0] https://en.wikipedia.org/wiki/Credit_default_swap
I asked ChatGPT to make this more readable since it's a mix of satire and actual information:
(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.
If the llc declares bankruptcy does meta have to pay the bank for it - or can they buy the assets at fire sale prices?
I have skimmed through the article and if I get the details through all the humor, satire and sarcasm even remotely correct, the major assets are actually the duality of payment obligations and residual value guarantees, both from meta. One could include cost overrun protection at the construction time too.
The "fire sale prices" would be so delicious as to guarantee that the entity(-ies) involved stay solvent as long as meta stays solvent.
My personal experience with LLC loans and banks is that the bank is using the assets as collateral and me as a backstop.
I thought the whole point of LLC was to limit liability so you wouldn't be liable for debt beyond your paid up capital? Why would you ever sign a personal guarantee?
>Why would you ever sign a personal guarantee?
So that they will lend you the money...
It's not always required, depends on the amount and the business.
The liability I’m shielded from is not debt I specifically requested. I’m shielded from unknown events.
Why would the CEO have the personal liability here and not the board? Does Sundar Pichai have to personally guarantee loans for Google? That would be weird since the CEO could be fired.
No, but this person’s LLC is not capitalized quite as well as Google, and the bank is adjusting the loan to account for that fact.
banks are not stupid… you can’t just open LLC, borrow billion bucks, spend it and then be like “oops, LLC mates, not liable”
You can if you are Meta and are willing to litigate the hell out of it.
“If you are meta” in this case means “if you have a billion dollars already, and a credit rating that you don’t want to destroy.
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.
> if you have a billion dollars already
Meta has $44.4 billion in cash-on-hand as of September 2025.
I'm correcting you because people can't really fathom just how wealthy these companies are. They could buy startups for billions of dollars with literally the cash they have in their bank accounts, let alone debt or stocks.
Do you think any CEOs of gigantic corporations are personally liable for any loans made by the companies they work for? I would be incredibly, incredibly surprised to hear if that's the case.
Then why don't they do it? It's the easiest money they can ever make. Even I can litigate the hell out of it if I get $27B, take the money and close the LLC.
More like if you are Meta and are viewed as a lucrative business opportunity by the bank.
"Me" in this case being a stand-in for the principal owner, which could be a corporation, individual, or group of individuals
Meta doesn't actually owe the bank anything in this setup. That would be Blackrock and the other private creditors.
Mechanistically, how would the LLC achieve bankruptcy?
Meta would have to not renew the lease and somehow nullify the residual value guarantee. This would leave the LLC with no revenue at all. If the RVG works there should be no chance of bankruptcy.
you just... file for bankruptcy like any other person or corporation?
Yeah but when you come to bankruptcy court with significantly more assets than debt, they aren't going to let you sell the business for pennies.
I'm asking how you would believe this vehicle would go broke, which is the usual reason to go to bankruptcy.
Corporate bankruptcy happens for a lot of reasons other than being "broke". Chapter 11 is a court-supervised way of restructuring your debt. This has a lot of utility in many situations other than not being able to pay.
It'll go before a judge and creditors would be able to object, so if it's just a ploy to get rid of debt you can be certain it'd be contested.
Meta may have lots of assets, but the LLC may not. The ability to have one wholly owned LLC go bankrupt by itself is one of the main reasons shell corporations exist.
I think it’s naive to focus on “what is meta getting” from Beignet.
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.
A lot of comments praising this summary, but I'll criticize it: it's still too verbose, and misses the point.
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… ‘vanity’ ratings… what’s the point of them then.
I think "vanity" is the wrong term because their existing credit rating, which they attempt to preserve, impacts all other borrowing (and possibly other agreements and finance vehicles, etc.) that they undertake.
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
Agreed. I know very little about financing but I’d bet if their rating fell that would trigger some debt repayment clause and the house of financial cards might wobble or fall.
…someone needs to shake the tree and see what falls out, like Peter Thiel did for SVB.
There are a lot of places where the credit ratings are hardcoded (to borrow a term) into funds. There are pension funds and other vehicles that might be bound to only invest in AA rated companies.
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.
> isn’t actually tricking anyone in finance.
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"?
They're not being "tricked" in the way these sensational substack posts would like us to think. You're not being given some secret knowledge that the ratings agencies don't have.
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.
Is anyone actually limited to only AA+? The usual meaningful separation is investment grade (AAA, AA, A, and maybe BBB) vs junk (everything else). (I don't think Meta would meaningfully lose access to credit if they were downgraded to A.)
Instinctively I try and simplify things. It this was a person with an excellent credit score, it’s as if the person is taking on extra debt to start to create something they need, but trying to hide it.
But that simplification isn’t the whole story. If that person took on debt as part of an LLC they started, not their personal bank account, then they have certain protections in the event of default of the LLC.
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.
Ah ok, now that makes sense. Thank you.
Funds and investment vehicles subtly lowering their standards while the downstream investors remain clueless is how we got 2008, no?
Still too verbose. Here's a TL;DR.
Meta is borrowing a whole lot of money and they're lying about it to investors.
No one is lying or deceived here.
Ehh, tell me the credit ratings assigned by rating agencies to mortgage backed securities circa 2005-2007. Its an ecosystem with misaligned incentives, and some cohort of investor will be left holding the bag. Big Tech, investment banks, and ratings agencies will get off with no consequences when this Jenga-esq capital apparatus eventually collapses.
I don't see what's Jenga-esque about this capital structure. You've got some AA- bonds issued directly by Meta having to do with their core business, and some A+ bonds issued by different entities to fund their riskier and more speculative datacenter construction. If anything, wouldn't it be harder to track the risk if both these bonds were stuffed into the same bucket?
A nice article on the underlying systemic causes of the crash:
https://archive.ph/2015.11.08-145615/http://www.wired.com/20...
Except for a few instances, there wasn’t any lying or outright fraud in that situation, just like there probably isn’t here either.
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!
Investor: Is this your debt, Meta?
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.
If it's related to AI, it's more like wash trading. The entire business interest in AI is making things look like there's a lot of investment when it's really just a small circle jerk of business interests.
It's just a more advanced crypto fraud.
It’s not necessarily lying, but it’s certainly deceptive.
Not even deceptive. This is relatively normal business practice.
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.
Usually subsidiaries’ debt is not also debt on the parent company, especially when said parent is publicly traded and subject to accounting/disclosure rules.
I asked almost this same question a few weeks ago here:
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.
>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 is a publicly traded company. If they have an agreement wherein they may have to pay substantial amounts to another company or on behalf of another company, that's a liability and they need to disclose that. Whether it's on their balance sheet doesn't really matter, this is what analysts do, back out these types of financial arrangements from the footnotes and publish thumbs up or thumbs down. People like to say "do your own research" but people should not, in general, do their own research.
This is hilarious because I was at the Louisiana public utility commission meeting where the argument was basically it’s Meta borrowing the money so they’re good for it.
Please do say more about this!
What do you want to know? There’s a YouTube of the session.
Thank you, I actually understand what this is all about now.
Is Meta actually obligated to repay the loans or not?
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.
To me it reads like the article intentionally pretends two major risks (force majeure and datacenter demand collapse) don't exist, by quipping that "rating agencies historically treat as theoretical inconveniences rather than recurring features of the physical world" for the former and explicitly saying they ignored it for "methodological convenience" for the latter.
If those have been offloaded to the LLC, wouldn't that be a pretty key difference?
I don't think Meta has a debt relationship with the loans involved here; that's the point. It does have strong contractual obligations to the wrapper business, though.
Even if they aren't obligated to repay, they have to in practice because it'll impact their ability to get loans in the future. If the shell company declares bankruptcy and gets the loans off Meta's books no one will ever loan money to Meta again.
They would still be able to get loans, but the terms would be much worse.
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.
Eh, debt investors have short memories. They buy 100 year bonds from Argentina, for fuck's sake. It might limit Meta's ability to do this SPV trick.
>Is Meta actually obligated to repay the loans or not?
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".
I'm not an accountant, but "contractually obligated to pay" sounds like a debt to me.
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.
It’s not though. Signing a lease isn’t debt. If you rent an apartment you haven’t taken out a loan from your landlord.
Being conceptually similar isn’t the same thing as identical.
I don't get why Beignet doesn't also hire Meta and pay it to build the DC.
They can get a better interest rate by using a specialty data center lender.
Very useful, ty
Is coreweave a similar situation?
I can't for the life of me figure out who would fund this, other than Saudi oil money or Russian petro-oligarchs, both so they can whitewash or launder their cash. This just makes no sense to me otherwise.
I get what you are saying but this is one of the largest conpanies on the planet asking for what is little more than change for them… it may make just a little sense :)
wow, what a great summary.
reading is really hard. I'm so happy ChatGPT exists
"None of this is unusual except for the part where Meta designs, builds, guarantees, operates, funds the overruns, pays the rent, and does not consolidate it."
So ChatGPT put this sentence in list form and reordered it a bit. AGI is imminent!
This might be the first time an explicit ChatGPT response survived being the top comment
I personally think it’s a great response and makes it clearer what’s happening
Times are changing quickly!
One year ago was taboo to say you were using LLM to help you code, today is the other way around...
> today is the other way around...
It is definitely not taboo to say you’re writing your own code.
could get you fired in more and more places though… :)
If you only see the world through crazy headlines, this probably seems true.
not just headlines, former colleagues and friends
where?
This isn’t to code. It’s tk summarize - something LLMs are usually good at, since they’re essentially lossy text/knowledge compression at their root.
Yeah... no it's not.
This is testable:
Can you link to another one?
Seriously. I thought about doing the same because I couldn't make heads or tails of the article, and then assumed it would just all be downvotes... glad to see it wasn't.
[dead]
Serious questions: won't banks and ratings agencies simply treat this as Meta's debt since it it effectively Meta's debt? What changes if this was on their "official balance sheet"? How does playing with the wording actually help Meta overall?
The crux of this article is that they won't treat Meta's debt as debt, because Meta intentionally structured this debt to circumvent the agencies' definition of "debt." Should they change their definition of "debt?" Maybe, but what incentive do they have to do that, is any formal definition bulletproof to circumvention, etc.
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 in the 2008 crash, the ratings agencies were disincentivized to accurately rate these vehicles because they were superficially masked and paid by the companies asking them for ratings.
It seems to me that the lengthiness and opacity of the report is part of the joke, and therefore running it through ChatGPT kind of misses the point. (The "FSG analyst" would have intentionally spread a layer of BS on top of everything to make it a lot of extra work to understand that the debt should actually be on Meta's books. Of course it's satirical so it calls out its own absurdity instead of actually burying it.)
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-...
Soo they just borrowed money from themselves to pay for their data center? Nice.
No, the capital comes from someone else. IIUC, private credit.
Can anyone comment about how common this (apparently legal) practice is?
Setting up a separate company for a major construction project is so common that it would be surprising to hear someone didn’t do it. The structure of the OP makes it hard to understand if there’s some more specific aspect he’s objecting to.
Film industry, too, as I understand it. Pretty much every production is set up with it's own LLC.
A couple thousand of them in the US https://carta.com/data/spv-spotlight-q3-2024/
MBAs gotta have something to do.
About as common as breathing.
remember that time Facebook spent $10s of billions on the metaverse?
They continue to spend $4B/quarter on this as of 2025Q3 financials.
Meta (which is short for the metaverse btw) occasionally remembers the metaverse existing, too, whenever there's a small break to be had from the AI stuff.
https://bsky.app/profile/mailia.bsky.social/post/3lwys6d6r6s...
What's your point with this comment? How can we ever hope for another Bell Labs if we decry companies taking risks on things no one even asked for?
how could we ever deserve another juicero or quibi, right?
Do you really think the kinds of technology that are adjacent to or enablers for vr are on the same level as juicero?
HN commenters flustered, baffled by the words on the screen: “Why would he say it like that? The phrasing is so foreign, it’s like the author wants me to laugh at it. The only way to understand this is to ask a chat bot what I should think the point is”
This is the lamest place to do ragebait. Facebook is probably more fun.
I appreciate this serious response to a joke about hn users not identifying jokes in a thread where hn posters are treating a comedy blog like an actual bond rating service
even a really strong shot of cafe bustelo failed to make this an interesting read.
> The Outlook is Superficially Stable, defined here as “By outward appearances stable unless, you know, things happen. Then we’ll downgrade after the shit hits the fan.”
[dead]
[flagged]
Had the same issue. This is what Claude 4.5 Opus gave me:
This is actually a satirical piece, written as a fake credit rating report. It's mocking how Meta (Facebook's parent company) is using a complex financial structure to keep $27 billion in debt off its official books. Here's what's actually happening in plain English: The basic setup:
Meta is building a massive $28 billion data center in Louisiana Instead of borrowing the money directly (which would show up as debt on Meta's financial statements), they created a convoluted structure with multiple shell companies and a partner called Blue Owl Capital
The trick:
On paper, Blue Owl's company "Beignet" owns 80% of the project and borrows the $27 billion But Meta still guarantees all the rent payments, covers cost overruns, and promises to pay if anything goes wrong So economically, it's Meta's debt—they're on the hook for everything—but legally it's structured so it doesn't appear on Meta's balance sheet
Why the author thinks it's absurd:
Meta controls the project, uses the buildings, guarantees all payments, and bears all the real risk The only reason it's "not Meta's debt" is a technicality in accounting rules The author is essentially saying: "This is obviously Meta borrowing $27 billion with extra steps, and everyone pretends otherwise because the rules allow it"
The whole piece is written as if a rating agency is giving this deal an A+ grade while openly admitting the structure is ridiculous—it's a critique of how financial engineering and accounting rules let big companies hide their true debt levels.
The Matt Levine article on this financing is more readable: https://news.bloomberglaw.com/mergers-and-acquisitions/matt-...
Practicing a skill when it becomes challenging is exactly how you improve the skill. Giving up on it is how you stagnate.
It's probably unreadable because you have to understand the underlying financial discussion.