Comment by tgsovlerkhgsel
3 months ago
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?
>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.
Poe's law applies, it's deadpan humor, finance Borat
IMHO it is very well executed, pushes the right buttons, and ultimately raises the question of financial realism (if the market acts like it's true is it true? how far is it from something that you can use to pay your taxes with? and so on)
6 replies →
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)
2 replies →
someone has to service the debt, and Meta has obligations to pay if the company cannot find any other tenant - if I understood correctly
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?
I think the point is that Meta really really wanted to avoid getting their credit rating downgraded, and ... went shopping for off-book financial vehicles.
On one hand they (supposedly) have enough "discounted cash flow" to do whatever, on the other hand I guess even they know it's a bubble and fallow years are coming.
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...
https://www.spglobal.com/ratings/en/regulatory/article/-/vie...
this is the original report
https://longbridge.com/en/news/265411465?channel=WHAB0001
> This arrangement comes at a steep price. The interest rate on these bonds is as high as 6.58%, significantly above the 5.5% yield of bonds from similar companies to Meta.
so it's definitely not completely junk, but the market priced in the gap (though, for me, it doesn't seem that big of a price difference!)
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.
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.
So your evidence -is- the bubble.