Nvidia's $750B in Deals Reignite Circular AI Fears
bloomberg.com72 points by petethomas 5 hours ago
72 points by petethomas 5 hours ago
Alternative source https://finance.yahoo.com/technology/ai/articles/nvidia-plan...
> The concern is familiar: NVIDIA money funds customers who then buy NVIDIA chips.
Nvidia had a free cash flow last quarter of $48.5 billion, a 36% quarter-over-quarter increase. $750B does not seem that huge in comparison. (Yes, the gravy train could theoretically stop anytime, but that still seems like a large but localized unwind rather than a wider economic crash.)
If you were NVDA and had that much cash on hand and wanted to grow your business, where would you put it?
Circular is a dumb way to describe it IMO, because it's not like both parties end up in the same place.
Nvidia is making trades for people to buy their GPUs.
Sometimes companies are trading stock for GPUs, sometimes money, other times something else.
In summary, Nvidia is selling GPUs.
Right. The risk isn't accounting fraud, its the equity-to-debt loop that relies on all these companies making "enough money to pay it back someday."
Nvidia invests, that equity check gets used to secure 10x it in debt with the GPUs as collateral, and then they buy the chips.
Nvidia gets paid, so they don't hold the debt liability. But, if AI revenue doesn't cover those debt payments before the GPUs depreciate, the loop starts to unravel, and fast. CoreWeave, Oracle, all the "neoclouds" etc. will blow up, and there could potentially be a ton of PE debt that is now under-collateralized due to depreciation, causing a pretty big haircut to basically all of private credit.
why AI revenue will not cover?
Chinese models pushes prices down and quality up, that makes GPU-based automation more affordable, while covering more and more cases to automate.
You can debate that llm producers will go bankrupt, some of them at least for sure.
How do you lose in this market if you do gpu?
> How do you lose in this market if you do gpu?
You don't. Nvidia gets paid either way. They were never the ones in danger (outside of the buildout going bust and having a massive surplus of cheap, used GPUs flood the market).
> You can debate that llm producers will go bankrupt, some of them at least for sure.
And that's the risk that will cascade down and kill off a bunch of companies and cause a debt crisis. If (for example), OpenAI goes to Oracle and says "I promise I'll pay you, at some point in the future, $1T to build my datacenters" and then Oracle funds that build out with debt, and then OpenAI goes bust, or just doesn't make enough money or can't raise enough cash to start making payments on their IOU, Oracle now also can't pay their debt and will eventually go bust, and now the private credit market takes a huge haircut, potentially bankrupting entire funds (like what happened in '08).
But if there is one thing trivial to see then it is the connection between loaned money and stock prices AS A WHOLE.
https://www.sciencedirect.com/science/article/abs/pii/S01651...
It's stronger for momentum stocks, but it's not like something like a gold mine escapes from it either.
Yeah, the "circular" language is obviously intended to imply unsustainability, as a system without external inputs must eventually run down. But this system is intended to have external inputs, revenue from customers that buy the services of the data centers. So the fundamental issue is just whether there will be enough such demand to justify the scale of the build-out.
These deals give Nvidia more exposure to that, in both directions. Certainly Nvidia shareholders should be cognizant of this. But nothing structurally problematic is occurring here.
Its only a problem if you think about it, just don't think about it and no problem!
They're selling GPUs in exchange for scrip which may or may not be able to pay Nvidia's operating expenses depending on whether AI has a profitable business model. This isn't hard to understand.
No, Nvidia is getting paid. Nvidia puts down a fraction of equity cash, and the recipients are taking that to PE to finance debt, using the GPUs as collateral. So Nvidia pays $1B, receiving company uses it to secure $10B in debt and buys $10B worth of GPUs.
Nvidia gets real cash, pays TSMC, etc.
The people in real trouble are companies like CoreWeave, Oracle, etc. that took an IOU from OpenAI (for example) to start a buildout, entirely debt financed. It works out so long as demand keeps going up, but the moment the music stops and that debt comes due and there's no revenue to pay it, game over.
Nvidia's concern isn't not actually getting paid, it's being faced with a glut of cheap, depreciated GPUs flooding the market impacting their future revenue. They'll live.
But OpenAI, not being able to pay CoreWeave, for example, that IOU, and then private credit coming for the debt payments from CoreWeave, is what would start the chain reaction. We may actually get to live to see Oracle fall.
Looks like you discovered an infinite money glitch! As long as you’re selling things at a profit, all you need to do is take those profits and give them to your customers to buy more things, repeat the loop a few times and you can become a billionaire food vlogger just like Jensen
> all you need to do is take those profits and give them to your customers to buy more things
Those customers aren't buying Nvidia chips with Nvidia's money. They're using Nvidia's equity check to finance debt, and then using debt to buy the GPUs. Nvidia invests $1B in someone like CoreWeave, CoreWeave then takes that check, goes to PE a borrows $10B w/ the GPUs as collateral. Nvidia basically paid $1B to get $10B in sales, and now CoreWeave is saddled with debt based on an IOU from the AI labs.
Nvidia is insulated from the debt exposure, but the companies doing the datacenter build outs are the ones in real trouble if the house of cards comes falling down.
[flagged]
This comment struck me as odd, and I went and read some of your other comments. And then I see this comment, on a completely separate thread:
> Now I have the full picture. You're right to push back, and that's on me. The load-bearing seams of language are the smoking gun I should have been aware of.
Pardon my curiosity, but whats going on? Is this a bit, or a bot, or simply how you tend to write?
Hahah sorry, I just really like the joke. This is me being satirical with the claude-isms. I take it you don't use Claude Code eh?
EDIT: By joke I mean this is I guess meme culture shaping as meta references within the workforce using these tools, I find it relevant in certain context on HN as irony and a way to nudge at the topic
Apparently "load-bearing seam" is something Claude (or one of the frontier US LLMs) really likes to say.
OPEN AI is living true to their name and foundation principles. Non profit and NO PROFIT, lol.
If anything this means that the insane amounts spent on AI are just mostly virtual speculative stock deals and in reality the amount spent in AI is kinda normal.
At what point do I start taking money out of my VTI holdings and parking it in cash - there is no way the market keeps going up.
First, don't park it in actual cash or you'll lose value to inflation which is currently running high. At a minimum put it in treasuries.
Second, trying to time the market is almost always a suboptimal strategy. The question is when will you likely need the money? If you won't need it for 10 years or more, keep it in index funds. Otherwise, treasuries.
Personally I tend to assume anyone talking about VTI knows about the risk-free rate.
Others have given practical replies, so here's a philosophical one: Sometimes it's just not practical to make much money from being right.
I want to acknowledge and empathize how much it sucks, while also putting it out there so that nobody suffers blaming themselves for something that might not be achievable.
Disclosure: I've been waiting-and-seeing too long myself, and I should probaby stop trying to time/strategize.
Historically even if you invest into index at the worst possible time (prior to a crash) and keep holding you still outperform inflation long term. Timing the market is impossible. Just keep an emergency fund in a money market or savings account and hold the rest.
Just need a larger emergency fund to mitigate the risk, especially if you work in tech and you feel the crash would heavily impact your labor earnings (including possibly extended unemployment)
There is no law that prices must revert to a mean.
The market can keep going up in dollar terms while losing real value if we enter a phase of high inflation.
Park it in BRK?
BRK has been roughly flat since the beginning of 2025; you might be better off in bonds or money markets (depending on your beliefs about near-term inflation).
BRK is 40% cash at the moment. I think "roughly flat" is evidence they're performing the strategy the OP wanted.
BRK has stated that they'll buy back in when prices are reasonable again, so it's an automatic "sell-high buy-low" strategy.
buffet may be holding his breath for some time
markets...irrational... longer than you stay solvent, etc.
Greg Abel has mostly replaced Buffet. Neither really care about the market as a whole. They're willing to buy any reasonably priced security with a promising future regardless of where the rest of the market is at. It's just that there are usually more of these available when the market is down.
Not that I have any skills in stock-picking whatsoever, but couldn't the recent lukewarm performance not also be an argument for BRK?
I mean their cash pile is also invested in money markets (so you get that), and the rest of the portfolio consists of quality companies where their (combined) valuation didn't explode in the last 1,5 years. So it's an opportunity to invest into something that might not be overheated.
> there is no way the market keeps going up.
There is no way the US’s leaders let the prices of publicly traded securities go down or even stagnate relative to the US dollar. These publicly traded securities make up a significant portion of the US leaders’ and most active voters’ assets, plus almost all state and local US governments depend on the securities’ price growth to meet their deferred compensation obligations.
The alternative to risk in US securities isn’t the USD, it’s a stake in other stable countries with resources.
already did mate, after Trumps first pump and dump with Iran
what's your risk tolerance?
can someone explain why this is actually bad?
nvidia spends X amount to invest in data centres or investments on the agreement that the counterparty spends Y amount back, the net delta is the actual amount of value being transferred aka Nvidia sells chips as usual despite the high numbers of X and Y?
The frontier labs do not have enough chips to meet demand, and AI demand is ferocious and climbing, so I'm not sure what the story is here
The problem is that when a vendor finances their customers, they can create the illusion of 'real' demand for their product, when most of the the end-users are only actually using something because it's cheap. When the vendor runs low on cash and starts requiring payment, the customer may not be able to afford it, taking both vendor and customer down, and leaving the end-users who have a real need, and were willing to pay sustainable prices without any options.
that makes sense to me in a conceptual sense
however inference is very profitable and plummeting in cost for a given point on the intelligence curve, and nvidia gpus can serve different models so they are protected post-buildout
You are describing the justification that NVDA is using to explain their behavior; they see it as something of a 'bridge-loan' until the LLM business model reaches steady-state. The problem is that this explanation has been used for many bubbles, where companies mis-categorize ongoing costs as one-time expenses.
> however inference is very profitable
Is there any actual evidence of that?
I imagine nvidia is securing these loans on the hardware being bought?
So if the company defaults they can take the GPUs and servers etc and sell those.
They are just adding zeros to already obnoxious numbers that make no sense. The endgame is on.
if demand for "safe" tokens explodes, this is a no brainier
if open source throws a wrench into the frontier revenue growth, then its gonna be biggest bubble explosion
It is far better for society when companies like Nvidia spend their money rather horde it like Apple.
Nobody (including the dragon) benefits from sitting on piles of gold.
Apple is a $5t company that has $45b cash-on-hand.
Your post is "cute", but 3 or 4 months of operating cash isn't a great example of "sitting on piles of gold".
DRAM will get to the moon before we get ...
I guess it gets bigger with each passing day. Tens of billions goes to Hundreds of billions.
It has to get bigger. As soon as it starts shrinking, the next round of debt will no longer be able to cover the prior round of commitments. What is happening in AI is essentially a gigantic version of what is happening in consumer auto loans, they just keep refinancing for more and more money. Eventually there will be no one willing to lend them more, and then they'll go to the government to bail them out.
Yeah, I'm pretty sure we've already passed some sort of fiscal singularity where economic and political interests are so intertwined that there can only be growth. If there ever isn't growth, then the legislators (who are elected via corporate sponsorship) will pull whatever levers necessary to make sure that there will always be asset growth.
It's some sort of tragic positive feedback loop that isn't going to stop until the whole thing comes crashing down for everyone and we're paying $37,000 for a loaf of bread.
I'm not sure a government can bail out these irrational companies and have it not be political suicide.
Those companies are already roundly hated by the general voting public, so I agree.
The public generally doesn't like banks and investment firms either, but they had bailouts in the past.
There's one problem with the idea of a bailout. What comes after?
Say, for the sake of argument, that the US government bails out OpenAI. Are they now suddenly not a money-burning operation anymore that will need another bail out in a period better measured in weeks?
The thing is just very unsustainable.
It's come to a point (or has it passed it) that these numbers are completely meaningless. One hundred billion here, $750B there, $1.2T over a year or so, toss in $300B for a few hyperscalers there. There's no imaginable scenario where these are actually backed up with real profit to where the investments make sense. Just passing the same hundred dollar bill among everyone and all booking it as revenue. I can't wait until it pops.
Displacing a vast amount of labor is what makes the numbers make sense. We don't know if they'll succeed or not, but it's obviously what they're chasing.
In addition to actual lost jobs, replacing a skilled white collar worker with a fungible operator of AI lowers the salary for that role significantly.
The labour needs to be eliminated — essentially nobody in the economy paying them for an equivalent job — for enough money to be freed up for it to work. But then these people who are either unemployed or job sharing some remaining work don’t have the money to buy the products of all the AI companies’ customers. At the necessary scale it will trigger economic demand collapse.
The whole idea is impossible.
Have they thought about the expenses in running continental Death Pits?
OK, dumb attempt at funny over, but certainly someone is thinking about instability costs? Even if everyone is super cool with literal Death Pits, they don't run for free. And not everyone will be cool with watching their entire family die, which will mean substantial costs in security - and money spent on security, that's just setting money on fire, that money doesn't work any more.
I know they've batted around the ideas of "compliance collars" and suchlike for the guys running the Death Pits, but I haven't seen anything that wouldn't be ultimately defeated by a typical zoo chimpanzee, let alone a psychopathic Delta Force guy with more advanced degrees than your entire family. He will not be pleased with your "compliance collar" thing.
And the brain control chips they've been trying to get working . . well, they're not ready yet. You'll just make the Delta Force guy even crazier .
[flagged]