This is just a continuation of the circular financing that has plagued Nvidia for the past 5 years. If the last earnings call was any guidance, their plan to renegotiate their current usage deals with neoclouds is going to severely shoot themselves in the foot.
Now, if you believe the entire market hinges on NVDA, this will absolutely heighten our current correction.
Nvidia's investment in neoclouds was an alternative to building their own GPU cloud. Jensen "didn't want to compete with his customers". If Nvidia used their cash to build their own data centers and then rent them out, would it be still considered a "circular financing"?
What Nvidia doing isn't new for any industry. But they could have explained themselves using submarines PRs. On the other hand, it is clear to me the whole thing isn't just about "circular financing" but someone wants to bring down Nvidia.
> The cost to protect against default on Nvidia bonds has jumped to a fresh all-time high. It’s been moving up all week and jumped by the most since July this afternoon.
> Goldman Sachs Group Inc.’s panic gauge climbed near its maximum level as US stocks tumbled Friday, signaling that another substantial wave of selling will follow over the coming weeks. Goldman’s trading desk said the selloff spread across the so-called systemic investors, projecting that commodity trading advisers could offload another $65.8 billion in a flat market, $28.6 billion if stocks rise and as much as $80.6 billion if equities continue to decline.
(no paywall bypass for live page, so link and copy paste must do, mea culpa)
I mean, it's certainly an attractive offer. I bought some 10Y when it was touching 4.5 because I couldn't imagine it going any higher, and yet here we are.
I mean isn't circular financing how the economy works as a whole? Its just what level we tolerate before people think its risky. Economies are circular in nature.
This is why so many companies keep coming up with newer, even more “inventive” extensions of EBITDA: to draw attention away from the numbers which reflect customer acquisition costs as expenses.
> You wind up with truly crazy numbers like WeWork’s “Community Adjusted EBITDA"
That one must reach back to a 2019 aborted IPO probably indicates that we didn't "wind up" with this.
Anthropic went public with audited financials showing GAAP figures. While it did also have management numbers in its prospectus, as is expected, the market seemed to fixate on price to revenues. Again, entirely GAAP figures.
This is being repeated by people who seem to believe this. But the firings of the CEOs of McDonald's, Intel and HP for this exact misbehavior indicates what actually happens when this is uncovered (to anyone other than Elon Musk).
My neighbor told me a few months ago that Nvidia contacted him and offered a program where he could finance GPUs and then immediately lease their usage back through an intermediary.
It was slightly more complicated than that, but he claims NVidia was practically telling him it was free money. He is a natural salesman/ entrepreneur and thus prone to some exaggerating, but he's not what I'd call a liar.
Wait what? Which customers specifically are you referring to? All of the financing I have heard about has been for customers I definitely believe are using them: CoreWeave, Nscale, Crusoe, Firmus, Nebius, etc... Not sure how to evaluate if those companies "want" or "need" them, but I also can't imagine how they wouldn't, seeing their whole business model is selling access to these chips.
If these GPUs were not needed, then GPU rental price should be going down as all these neoclouds attempt to sell capacity, but GPU rental prices are going up.
Correct, however, NVIDIA has massive amounts of reserved wafer allocations at TSMC with no product/customer match. Those companies are also buying at their own pace, not at the pace NVIDIA needs them to. I would be REALLY surprised if NVIDIA pulls through this nonsense. They are profitable on paper, however, based on everything I've read, they are actually in a ton of trouble unless they can renegotiate everything. Like..."Bear Sterns" trouble.
EDIT: One small thing I wanted to mention, is that while I compared this to the downfall of Bear Sterns, NVIDIA is actually lower risk than the infamous investment banking company. There is a lot of room for things to go right. TSMC knows they'll have a ton of demand, regardless, so they could negotiate down allocations with NVIDIA. NVIDIA could also use their free cash flow and make more inventory, even without buyers, and sell it for smaller margins or a slight loss. They can even use those allocations for future gaming chips and other stuff. That second path will cost them big time. Hopefully, they can negotiate deals with TSMC.
Nvidia is not just profitable on paper, they are wildly profitable on paper. 70% + margins on 115 Billion a quarter revenue with even more projected for Q4.
I doubt Nvidia would have trouble finding a buyer for their chips in the medium term. Demand is incredibly high and most of the demand is actually originating from the big tech companies - SpaceX, Microsoft, Google, Oracle, Amazon, etc.
Since when did the misinformation on here get to be as bad as WallStreetBets
NVIDIA never raised the price of a 5090 Founders Edition. They are still $2k although you won't find one.
Resellers charge $6k (NOT $10k) and NVIDIA does not get more than they did before so there is no margin compression for NVIDIA if they charge $2k in the future.
Do not want or need? There is a 12-18 month backlog right now and the bosses of OpenAI just said yesterday they needed compute wherever they could find it.
Maybe everyone's lying to the point of committing felonies, but I'd need more proof than conspiracy theory posting. Could I buy a few racks of vera rubin right now?
GPUs are an extremely hot commodity right now and the bottleneck for a lot of high-demand production. It's very hard to imagine there being sales that are artificial, or companies buying overpriced units that they don't need, given the manifest extreme demand for them.
Isn’t the whole point of this that demand is being artificially exaggerated by NVDA subsidizing consumption? I.e. the extreme demand emanates from NVDA loaning everyone money and then turning around and saying “wow, you guys sure want these chips huh!??” Of course demand is extreme when you hold all of the resource and then juice demand!
Circular financing has a very specific meaning with respect to supplier and consumer and generalizing it for the whole economy is either naive or trollish.
Legitimate vendor financing is a thing and generally accepted as OK. For example a car company giving loans to credit-worthy folks to buy a car. Or leasing cars to folks via similar mechanisms. If the risk is managed properly it’s a nice profit center since now you get car revenue plus loan interest.
What Nvidia has been doing is closer to a car company finding a bunch of broke folks that will require a small miracle to pay back the loan, and financing cars for all them, then telling the market demand for cars if off the charts. If magically somehow these folks can generate cash to pay for it one gets away with it. If not, then you have a mini-2008 financial crisis unfolding on your balance sheet.
That second paragraph is basically how the aircraft leasing business works, and has worked for a very long time. Lend planes on credit to buyers intending to monetize and pay them back.
Boeing Capitals whole purpose is to provide financing to cash strapped airlines hoping to monetize the airframe.
I don't see a reason to be especially cynical about this business model because it has been proven to work in concept. I think the only thing left to answer is how profitable inference actually is, and how easy it is to use GPUs as collateral (it's very easy with a Boeing 777). We have 80 years of airline data and about 1-2 years of neocloud data.
Except only around 1% of Boeing sales are vendor financed. Leasing aircraft is popular by airlines but that balance sheet risk is carried by separate leasing companies. And there is a very strong market for second hand aircraft including those decades old. So it’s a totally different market setup than GPUs.
Well yes running an airline is a famously difficult business. But my point is that financing airlines has been fantastically profitable.
If GPUs turn out to be highly analogous to airplanes, (in the sense that they are monetizable capital goods offered on credit), then the neoclouds might suffer just like airlines over the long run while Nvidia and other creditors do handsomely.
> If GPUs turn out to be highly analogous to airplanes
That’s some hefty if you got there. I don’t know much about aircraft financing, but I know that if you default on your airline company I paid for, I have a fledt of very very expensive planes ready to make money next 20-30 years. I also know they will have demand since people need to fly and demand is linked to GDP. Also, did not Boeing amputate it’s financing portfolio years ago precisely because it is so dangerous, whereas Nvidia takes money from left pocket, gives to neocloud to receive it back and puts it in his right pocket?
If CoreWeave defaults you got bunch of GPU’s deprecating as we speak and being out of demand in 5-6 years. Also you have AI demand of which everyone say is off the roof, but the numbers on the board don’t seem to add up. Basically it is almost like throwing trillions of dollars to Wright -brothers after their first flight.
I think the fundamentals will be there one day, but it doesn’t look they are there now.
First, in the limited period of time we have had to observe inference economics, it's clear that these chips have an economically useful life that is longer than their accounting depreciation schedule. A100s sold in 2020 are still running 24 hours a day earning money. This is also true of airliners. United is still flying planes built in 1991. That has long long been depreciated and is still earning profit every day.
Second and far more important; Yes, an airliner lasts longer than a GPU, but the economic payback is also way way longer. A 737 can take generally 15 years to economically recoup the investment. Meanwhile they're saying a Blackwell is what, 9 months payback? And a Vera Rubin even faster.
> Legitimate vendor financing is a thing and generally accepted as OK
> What Nvidia has been doing is closer to a car company finding a bunch of broke folks that will require a small miracle to pay back the loan
Did you noticed Nvidia being no.1 on Wall Street or something like that? They was doing it at global economy scale Not some random individual wanting to get car worth maybe 3 x DGX Sparks.
And do you know what is worth ~5 Sparks ? Commercial bakery furnance with 60 years warranty for heating elements (German steel, German welding ;) and maybe 3 days of work of building it in bakery.
That computer things are way overvalued and they keep pumping it - just one company ordered 40% of global RAM production without even having builded computer room. And that company do not even plan to be profitable so from where they had money to pay for order of global 40% of RAM ?
I know someone that works at a high credit risk used car lot. Their motto is literally, "We sell to sell again" and that's kind of a joke. It's not uncommon for them to resell a vehicle that has been reposessed several times before it finally ends up somewhere permanent or is beyond economic repair.
I'm sure nvidia's plan is/was to take over the neoclouds that failed, and then sell turnkey solutions (pre-racked systems or whole datacenters) to other hopeful neoclouds.
Yes, money exchanges hands. Wow, we engage in a society.
The issue lies within the entire market being concentrated on 7 companies... and then 5... and now 3... and with these companies being so high up on stilts, their relative risk is heightened.
I thought the issue with NVIDIA is at times money was not changing hands, but they were getting investments in companies that needed GPU's at unsustainable markups? Or am I misunderstanding the controversy? If money was changing hands, this would not be an issue?
The market has a tendency to overreact in both directions. I'm not buying NVDA yet, but it's certainly on my watch list. I think it will be below $100 again, soon.
The failure of data center construction at any scale would stagger Nvidia and provide an opportunity for those hyperscalers to reduce their reliance on nvidia by siphoning those gpus off in whatever scale you want. It is in Nvidia's best interest to keep demand strong.
If you're arguing that a large KPR is a good thing, I would point to their CEO's exit as to why that's not the case.
The circular financing claims are unfounded. The vast majority of their products ship to hyperscalers that they're not funding/financing. A glance at the 2026 10-K confirms this.
Somehow nvidia had so much money laying around in 2021 that they managed to circularly finance their way to 1200% the revenue across five years and keep going without any relevant debt.
“Positive cash flow” and “circular financing” put together is playing with fire. It lets you take your own money, give it to others, then they give it right back to you and now magically you have more “revenue” that’s really just the cash you already had.
What’s even worse is when this is in the form of investments in overvalued startups. On the way up your cashflow comes back both as “revenue” and big balance sheet gains. But it works in the reverse directions too when things slow. Revenue dries up and your balance sheet has to write off the startup equity declines too, amplifying the pain during a slowdown.
The market is sensing that amplifying downturn phase is starting to unfold hence the market beginning to freak out.
> “Positive cash flow” and “circular financing” put together is playing with fire. It lets you take your own money, give it to others, then they give it right back to you and now magically you have more “revenue” that’s really just the cash you already had.
Even so, if NVDA has no debt to service. Where does the contagion come from?
> The market is sensing that amplifying downturn phase is starting to unfold hence the market beginning to freak out.
Actually most of what's happening currently in the market appears to be endogenous market dynamics, quite possibly due to people using tons of leveraged instruments with nvda.
Because the CFO is the only C suite on record opposing circular financing and calling it out by name. If anyone internal would give them deniability it's her.
Disagreeing on business strategy is one thing, accounting fraud is a completely different thing. After SOX, no CFO is going to sign off on accounting they don't agree with.
Whether a CFO agrees with the accounting is not the same as whether the accounting is fraudulent. The point is after SOX a CFO can't use the excuse of "the CEO made me do it." The CFO has to certify under penalty of prison that they personally reviewed the accounting and agree with it.
My thoughts exactly. She's there because Nvidia is about to have some very bad news while she's in charge. Including her making some very unpopular decisions (Layoffs, benefits cuts, etc)
Once those all land, she'll be fired and I'd not be shocked if Jensen is brought back. And, of course, none of the changes she's made will be reversed. This sort of play has happened a bunch of times.
For those thinking how could it be possible that CFO didn't know, well companies this large, public or not, will have in-house counsel they pay for, for advice (sure, in-house counsel will farm out to third parties). If the counsel tells the C-suite they shouldn't do a specific thing, the C-suite is more than welcome to ignore counsel. Counsel is giving its legal opinion and C-suite can override their counsel's advice.
A more fun way to look at it, search Better Call Saul, Kim Wexler talking to Kevin Wachtell giving him a mouthful.[0]
It sounds like a polite way of saying "right in the outcome, wrong about all of the details."
A good historical example of this is Billy Mitchell, saying in the 1920s that bombers made battleships obsolete, which sounds like what would happen by the end of WW2, but on closer inspection, he was betting very heavily on the wrong kind of aircraft development (and would have sabotaged the right kinds of aircraft!) that listening to him would have left you with a lot of dead pilots and a battleship still bearing down on you.
No one can predict the exact breaking point when people wake up from a bubble, and so "directionally correct" is the best one can do when pointing out the dogshit fundamentals of a company's offering.
I think we can be a little kinder than that to Zitron, he called out the circular financing in some detail but didn't perfectly nail the timing of the collapse or the political and infidelity angles. Still directionally correct on the big picture.
Aged poorly, that article. Paints a picture of Ed Zitron standing on the beach, saying, "Later on, the water will be higher," and the author looking at his watch 30 minutes later and proclaiming, "What a fool! The water is lower than it was when he made his prediction!"
The comment is good. Ed Zitron's loudest critics are the people who under no circumstances will accept that his central point is one of physics ineluctable, and that the AI trade is going to be drawn in.
More like quotes Ed on saying “No later than 30 minutes will the water be above the lifeguard stand” and then 30 minutes later its low tide.
Ed Zitron is like Freud where everybody who believes in him doesn’t actually listen to the exact words. Of course when you substitute the words its correct. The danluu article uses exact quotes.
Yet he was reporting about Nvidia’s chips ending up in China and the mysterious Chinese lady Jensen was often seen with back in March:
> Things get a lot weirder as Bloomberg digs in, including a woman called “Huang” that may or may not be both the CEO of Megaspeed and an associated company called “Shanghai Hexi,” which is also owned by the Yangtze River Delta project… who was also photographed sitting next to Jensen Huang at an event in Taipei in 2024.(Long article, Nvidua ar very bottom: https://www.wheresyoured.at/the-ai-industry-is-lying-to-you/)
Most of this stuff has been out in the open for a long time, if you just care to look for it. He is just reporting same stuff Bloomberg, Reuters, FT, WSJ.
Colette Kress the acting CEO of nVidia has been with the company since 2013 as Chief Financial Officer. It's difficult to believe the CFO was unaware of what could be years of blackmarket chip sales. It would ultimately be the financial department that has to manage that revenue on the books.
If the several investigations into this end up drawing her in, then nVidia might need another acting CEO in the not too distant future.
nVidia also has an issue where many of the C level people at the company have ties to a specific foriegn country that I will leave unnamed. It will be interesting to see who replaces her and if they bring in someone from the outside.
The stock continues to fall as well so who knows what will happen next.
Isn't everybody implicated? The stock-for-discounted-chips deals, huge numbers of dark GPUs held by shady companies (or just gone), it's all so shady that it boggles the mind.
IF the bubble is finally popping, just be careful guys - because no matter how cheap things look, they can always get a lot cheaper !
If what many people think is right, ie, that the whole business of AI is unprofitable and pretty much only exist for now because Nvidia & co is directly or indirectly financing their own demand, this whole house of card will collapse really quickly. As a reminder - before the 'boom' Nvidia was making a 3B profit, in a good year.
The whole business of AI can’t be unprofitable, open models like DeepSeek are clearly being served profitably and if it turns out GPU, HBM, etc. should be cheaper than they currently are, serving them will only become more profitable. Given that, the frontier closed models with a pretty big price premium must be very profitable to serve as well, even if you don’t take their word for it. Training can stop today and the existing models (open ones at least) are still very useful and aren’t going anywhere. A valuation pop may drag down some companies, but the .net pop didn’t mean the whole business of internet is unprofitable.
But unless someone suddenly pulls a moat of out of their butt, the layer of AI on top of that compute could very possibly end up being a minimally profitable business, like say VPS hosting.
Sure, you can make a few bucks doing that, but not what you want to try to build companies with trillion dollar valuations off of.
I believe inference as an industry will be as big or bigger than pre-AI AWS, just with current model capabilities, bubble popping or not. In fact if you add up this year’s AI revenue from all the big names, that should already be close to or surpass AWS’s 2024 revenue of $108B? Will it support multiple trillion dollar companies? No idea. But it’s not a few bucks either.
- nvidia stopped reporting revenue via singapore->china, and last year the revenue from china was >20% . they are now under strong scrutiny and likely can't sell to china nearly as much or at all
- in their earnings call they announced 2 major things: $150 billion of stock buy backs (very big vote of confidence /s), and tighter constraints on circular financing for neoclouds
- lost the leader who organized the china revenue and most of the current financial deals
- is now officially under investigation
- datacenters have more pushback and regulation is likely to increase once democrats go into office (not a guarantee), reducing demand which was already artificially increased (see below)
- was pressuring companies to buy gpus and warehouse them to pump their numbers, despite the inability of companies to actually build the datacenters in time
- and we don't know the full extent of their financial commitments or how much of it was loaned against their stock as equity
Yes this is very much a buy from me :thumbsup: Everyone put on your knife proof gloves and buy the dip
Over the last few days I actually started averaging down by buying in staggered increments. I still have 3 GTCs left with limits set in advance. I’m amused to see where this is going.
The problem is that the hit jobs keep coming and I have no idea what Jensen did (or didn't do), but clearly it's open season now. As much as I want to buy more, it clearly is not over yet, and it feels like gambling right now.
Nvidia appears to have dug itself a hole it’s not going to be able to get out of. The vendor financing games work until they don’t, then they backfire spectacularly.
They know if they show a quarterly decline, or even just a big growth slowdown, the AI valuation boom will implode and everyone will return to regularly scheduled programming. Valuations of all these AI gems, including Anthropic, would then see an 70-80+% haircut with standard valuations for growth companies.
So to avoid that they’ve been signing increasingly questionable funding deals and allegedly pushing folks to take delivery of chips they don’t need because data center construction is hitting massive snags. Hence too the reports of GPUs sitting in warehouses collecting dust. It appears this is now all coming to a head and the next 12 months will be very interesting.
Early moves such as all the “Jensen was actually kinda not a nice guy” articles hitting the press today suggest the strategy may be throw Jensen under the bus and use that to try and clean up this mess and reset the company. Get your popcorn ready. Some are already calling this “Jenron.”
I'm not sure, it seems like everyone and their mother has a decently capable TPU design at this point or is in the process of designing one. I think the moat isn't as big as it was for GPUs.
Nvidia’s got a time-limited (crumbling) moat and it needs to capitalize it to death before it goes away, replaced by competing stacks.
If the US datacenter industry can’t absorb their max capacity, its only option is to throw its momentum towards building clouds on their own stack, wherever they can. Asia.
Next step would’ve been boostrap their own AI lab on OSS models, and fork.
All this, without getting caught “entering their clients’ businesses”, American clouds and American ai labs
Trump already said the US is a country of "second chances" while talking about the topic. They care more about big numbers getting bigger than anything else
There is no "they" now. The admin is schizophrenic. Lutnick and Bessent are feuding, Rubio is a useful tool, and Trump is sundowning. It's anyone's guess where the chips will fall.
US taxpayers now own ten percent of Intel through current federal government’s actions. Now maybe taxpayers will have to own part of NVidia when the dust settles. Who knows what the Trump family and other insiders are trading but I bet they benefit from final outcome and each step of the way.
It'll be interesting to see what they find. My read as a guy who read about Enron but doesn't do much investing of my own is that the SEC is the lifeguard blowing the whistle asking you to turn your music down at the public pool, while the DOJ is the actual cop who slams your head into the hood of his squad car while yelling "STOP RESISTING!". If the lifeguard also notices you have a huge bag of drugs right next to the stereo, you're screwed because they'll pass that info along to the officer.
If they find the proverbial bag of drugs, well, mark this down as another instance of poor board leadership allowing the cult of the founder to take hold at Nvidia and the rest of us paying the price.
if you are hoping for an enron type of accounting scandal you are going to be massively disappointed. they are way too smart for that. what's probably under scrutiny are whether they followed accounting rules but even there ton of specialized lawyers that can stretch what the boundaries are, there's no real case here other than just a nice headline from what I understand.
Enron outright committed fraud by recording profits that didn't exist that wasn't very smart.
What NVIDIA is doing isn't illegal and they are far too wise to do something brazen like Enron did which seems to be what a lot of the misinformed users are hoping for
It’s like distributed Enron, the SPVs are held by the datacenter and hyperscaler people. So it’s arms length but that doesn’t make the whole thing any less susceptible to a big repricing. End of the day, higher interest rates, borrowing money, stretching depreciation, using SPVs, all of this is really reminiscent of Enron except instead of one company it’s the entire AI sector. ORCL being the most egregious example since they are also in the middle of an AOL/Time Warner style deal pledged with overvalued shares. High yield spreads are really taking off again and the bottom line is the market is out of money. You also have boomers hitting peak RMD, geopolitical stuff hitting energy and commodities. We have a long way to drop, the question is will it be fast or slow. But it’s healthy, prices were too high, people were speculating. Risk off now.
Now, if you believe the entire market hinges on NVDA, this will absolutely heighten our current correction.
What Nvidia doing isn't new for any industry. But they could have explained themselves using submarines PRs. On the other hand, it is clear to me the whole thing isn't just about "circular financing" but someone wants to bring down Nvidia.
> The cost to protect against default on Nvidia bonds has jumped to a fresh all-time high. It’s been moving up all week and jumped by the most since July this afternoon.
https://www.bloomberg.com/news/live-blog/2026-11-20/markets-...
> Goldman Sachs Group Inc.’s panic gauge climbed near its maximum level as US stocks tumbled Friday, signaling that another substantial wave of selling will follow over the coming weeks. Goldman’s trading desk said the selloff spread across the so-called systemic investors, projecting that commodity trading advisers could offload another $65.8 billion in a flat market, $28.6 billion if stocks rise and as much as $80.6 billion if equities continue to decline.
(no paywall bypass for live page, so link and copy paste must do, mea culpa)
We live in interesting times.
I'm not saying I believe this, but I think this is the real concern for many skeptics.
This then has the effect of artificially raising profits (you are providing a subsidy, but not counting it as an expense).
While I think eg taking equity is one thing, providing finance incentives is pretty standard in "traditional" industries
That one must reach back to a 2019 aborted IPO probably indicates that we didn't "wind up" with this.
Anthropic went public with audited financials showing GAAP figures. While it did also have management numbers in its prospectus, as is expected, the market seemed to fixate on price to revenues. Again, entirely GAAP figures.
This is standard vendor financing and it’s fine. The weird part with Nvidia is the taking of equity in its customers. (Also sleeping with customers.)
Perks of the job. CEOs don’t just get free snacks like the rest of us.
This is being repeated by people who seem to believe this. But the firings of the CEOs of McDonald's, Intel and HP for this exact misbehavior indicates what actually happens when this is uncovered (to anyone other than Elon Musk).
In 2024 I received an email to an address setup for a side hustle
Email was from a neo cloud with H100s for around 23% the cost of the usual suspects
Immediately wondered why a no-name neo cloud has such a fancy discount for a nothing side hustle with no traction
Emailed back hypothesizing channel stuffing. Reply was a simple
"You might be onto something ;)"
Since then been no question this all been a too big to fail push to make big deals and expect a bailout or the entire economy collapses
It was slightly more complicated than that, but he claims NVidia was practically telling him it was free money. He is a natural salesman/ entrepreneur and thus prone to some exaggerating, but he's not what I'd call a liar.
Wait what? Which customers specifically are you referring to? All of the financing I have heard about has been for customers I definitely believe are using them: CoreWeave, Nscale, Crusoe, Firmus, Nebius, etc... Not sure how to evaluate if those companies "want" or "need" them, but I also can't imagine how they wouldn't, seeing their whole business model is selling access to these chips.
EDIT: One small thing I wanted to mention, is that while I compared this to the downfall of Bear Sterns, NVIDIA is actually lower risk than the infamous investment banking company. There is a lot of room for things to go right. TSMC knows they'll have a ton of demand, regardless, so they could negotiate down allocations with NVIDIA. NVIDIA could also use their free cash flow and make more inventory, even without buyers, and sell it for smaller margins or a slight loss. They can even use those allocations for future gaming chips and other stuff. That second path will cost them big time. Hopefully, they can negotiate deals with TSMC.
I doubt Nvidia would have trouble finding a buyer for their chips in the medium term. Demand is incredibly high and most of the demand is actually originating from the big tech companies - SpaceX, Microsoft, Google, Oracle, Amazon, etc.
Customers are there I think...
NVIDIA never raised the price of a 5090 Founders Edition. They are still $2k although you won't find one.
Resellers charge $6k (NOT $10k) and NVIDIA does not get more than they did before so there is no margin compression for NVIDIA if they charge $2k in the future.
In other words, they are not $2k then.
Maybe everyone's lying to the point of committing felonies, but I'd need more proof than conspiracy theory posting. Could I buy a few racks of vera rubin right now?
> I'm not saying I believe this
Saying things you don't believe is called "lying". You are lying.
What Nvidia has been doing is closer to a car company finding a bunch of broke folks that will require a small miracle to pay back the loan, and financing cars for all them, then telling the market demand for cars if off the charts. If magically somehow these folks can generate cash to pay for it one gets away with it. If not, then you have a mini-2008 financial crisis unfolding on your balance sheet.
Boeing Capitals whole purpose is to provide financing to cash strapped airlines hoping to monetize the airframe.
I don't see a reason to be especially cynical about this business model because it has been proven to work in concept. I think the only thing left to answer is how profitable inference actually is, and how easy it is to use GPUs as collateral (it's very easy with a Boeing 777). We have 80 years of airline data and about 1-2 years of neocloud data.
If GPUs turn out to be highly analogous to airplanes, (in the sense that they are monetizable capital goods offered on credit), then the neoclouds might suffer just like airlines over the long run while Nvidia and other creditors do handsomely.
That’s some hefty if you got there. I don’t know much about aircraft financing, but I know that if you default on your airline company I paid for, I have a fledt of very very expensive planes ready to make money next 20-30 years. I also know they will have demand since people need to fly and demand is linked to GDP. Also, did not Boeing amputate it’s financing portfolio years ago precisely because it is so dangerous, whereas Nvidia takes money from left pocket, gives to neocloud to receive it back and puts it in his right pocket?
If CoreWeave defaults you got bunch of GPU’s deprecating as we speak and being out of demand in 5-6 years. Also you have AI demand of which everyone say is off the roof, but the numbers on the board don’t seem to add up. Basically it is almost like throwing trillions of dollars to Wright -brothers after their first flight.
I think the fundamentals will be there one day, but it doesn’t look they are there now.
First, in the limited period of time we have had to observe inference economics, it's clear that these chips have an economically useful life that is longer than their accounting depreciation schedule. A100s sold in 2020 are still running 24 hours a day earning money. This is also true of airliners. United is still flying planes built in 1991. That has long long been depreciated and is still earning profit every day.
Second and far more important; Yes, an airliner lasts longer than a GPU, but the economic payback is also way way longer. A 737 can take generally 15 years to economically recoup the investment. Meanwhile they're saying a Blackwell is what, 9 months payback? And a Vera Rubin even faster.
> What Nvidia has been doing is closer to a car company finding a bunch of broke folks that will require a small miracle to pay back the loan
Did you noticed Nvidia being no.1 on Wall Street or something like that? They was doing it at global economy scale Not some random individual wanting to get car worth maybe 3 x DGX Sparks.
And do you know what is worth ~5 Sparks ? Commercial bakery furnance with 60 years warranty for heating elements (German steel, German welding ;) and maybe 3 days of work of building it in bakery.
That computer things are way overvalued and they keep pumping it - just one company ordered 40% of global RAM production without even having builded computer room. And that company do not even plan to be profitable so from where they had money to pay for order of global 40% of RAM ?
I'm sure nvidia's plan is/was to take over the neoclouds that failed, and then sell turnkey solutions (pre-racked systems or whole datacenters) to other hopeful neoclouds.
The issue lies within the entire market being concentrated on 7 companies... and then 5... and now 3... and with these companies being so high up on stilts, their relative risk is heightened.
I am certainly not buying $NVDA right now
If you're arguing that a large KPR is a good thing, I would point to their CEO's exit as to why that's not the case.
The circular financing claims are unfounded. The vast majority of their products ship to hyperscalers that they're not funding/financing. A glance at the 2026 10-K confirms this.
“Positive cash flow” and “circular financing” put together is playing with fire. It lets you take your own money, give it to others, then they give it right back to you and now magically you have more “revenue” that’s really just the cash you already had.
What’s even worse is when this is in the form of investments in overvalued startups. On the way up your cashflow comes back both as “revenue” and big balance sheet gains. But it works in the reverse directions too when things slow. Revenue dries up and your balance sheet has to write off the startup equity declines too, amplifying the pain during a slowdown.
The market is sensing that amplifying downturn phase is starting to unfold hence the market beginning to freak out.
Even so, if NVDA has no debt to service. Where does the contagion come from?
Actually most of what's happening currently in the market appears to be endogenous market dynamics, quite possibly due to people using tons of leveraged instruments with nvda.
You mean uplifted and made Nvidia king and is by design?
Once those all land, she'll be fired and I'd not be shocked if Jensen is brought back. And, of course, none of the changes she's made will be reversed. This sort of play has happened a bunch of times.
because its due in a few weeks.
They last one was from August 26th so 3 months would bring them to near the end of November.
And pedantic side note, its called a 10-Q
A more fun way to look at it, search Better Call Saul, Kim Wexler talking to Kevin Wachtell giving him a mouthful.[0]
[0] https://www.youtube.com/watch?v=KjnNV2JLMX0
Good comment by yousif_123123 in that thread: https://news.ycombinator.com/item?id=49534777 (imho)
A good historical example of this is Billy Mitchell, saying in the 1920s that bombers made battleships obsolete, which sounds like what would happen by the end of WW2, but on closer inspection, he was betting very heavily on the wrong kind of aircraft development (and would have sabotaged the right kinds of aircraft!) that listening to him would have left you with a lot of dead pilots and a battleship still bearing down on you.
I agree, he was just early.
The comment is good. Ed Zitron's loudest critics are the people who under no circumstances will accept that his central point is one of physics ineluctable, and that the AI trade is going to be drawn in.
Ed Zitron is like Freud where everybody who believes in him doesn’t actually listen to the exact words. Of course when you substitute the words its correct. The danluu article uses exact quotes.
> Things get a lot weirder as Bloomberg digs in, including a woman called “Huang” that may or may not be both the CEO of Megaspeed and an associated company called “Shanghai Hexi,” which is also owned by the Yangtze River Delta project… who was also photographed sitting next to Jensen Huang at an event in Taipei in 2024.(Long article, Nvidua ar very bottom: https://www.wheresyoured.at/the-ai-industry-is-lying-to-you/)
Most of this stuff has been out in the open for a long time, if you just care to look for it. He is just reporting same stuff Bloomberg, Reuters, FT, WSJ.
If the several investigations into this end up drawing her in, then nVidia might need another acting CEO in the not too distant future.
nVidia also has an issue where many of the C level people at the company have ties to a specific foriegn country that I will leave unnamed. It will be interesting to see who replaces her and if they bring in someone from the outside.
The stock continues to fall as well so who knows what will happen next.
Genuinely unsure what you're hinting at here. Even Israel doesn't seem to check out (none born, studied or worked there).
Colette Kress, Interim Chief Executive Officer (CEO); Executive Vice President & Chief Financial Officer (CFO)
Nick Parker, Chief Commercial Officer (CCO) / EVP, Worldwide Field Operations
Debora Shoquist, Executive Vice President, Operations Tim Teter, Executive Vice President, General Counsel and Secretary
Chris A. Malachowsky, Founder & NVIDIA Fellow Alison Wagonfeld, Chief Marketing Officer (CMO)
Bruce Andrews, Chief External Affairs Officer
Scott Gawel, Vice President & Chief Accounting Officer (CAO)
Bill Dally, Chief Scientist & Senior Vice President of Research
I have no idea what they're getting at either in all seriousness though
If what many people think is right, ie, that the whole business of AI is unprofitable and pretty much only exist for now because Nvidia & co is directly or indirectly financing their own demand, this whole house of card will collapse really quickly. As a reminder - before the 'boom' Nvidia was making a 3B profit, in a good year.
Unless your competitor keeps training, then you'll be starved for customers. The idea that R&D can stop is absurd.
Offering compute is likely to remain profitable.
But unless someone suddenly pulls a moat of out of their butt, the layer of AI on top of that compute could very possibly end up being a minimally profitable business, like say VPS hosting.
Sure, you can make a few bucks doing that, but not what you want to try to build companies with trillion dollar valuations off of.
- nvidia stopped reporting revenue via singapore->china, and last year the revenue from china was >20% . they are now under strong scrutiny and likely can't sell to china nearly as much or at all
- in their earnings call they announced 2 major things: $150 billion of stock buy backs (very big vote of confidence /s), and tighter constraints on circular financing for neoclouds
- lost the leader who organized the china revenue and most of the current financial deals
- is now officially under investigation
- datacenters have more pushback and regulation is likely to increase once democrats go into office (not a guarantee), reducing demand which was already artificially increased (see below)
- was pressuring companies to buy gpus and warehouse them to pump their numbers, despite the inability of companies to actually build the datacenters in time
- and we don't know the full extent of their financial commitments or how much of it was loaned against their stock as equity
Yes this is very much a buy from me :thumbsup: Everyone put on your knife proof gloves and buy the dip
I like to call them gauntlets. :D
They know if they show a quarterly decline, or even just a big growth slowdown, the AI valuation boom will implode and everyone will return to regularly scheduled programming. Valuations of all these AI gems, including Anthropic, would then see an 70-80+% haircut with standard valuations for growth companies.
So to avoid that they’ve been signing increasingly questionable funding deals and allegedly pushing folks to take delivery of chips they don’t need because data center construction is hitting massive snags. Hence too the reports of GPUs sitting in warehouses collecting dust. It appears this is now all coming to a head and the next 12 months will be very interesting.
Early moves such as all the “Jensen was actually kinda not a nice guy” articles hitting the press today suggest the strategy may be throw Jensen under the bus and use that to try and clean up this mess and reset the company. Get your popcorn ready. Some are already calling this “Jenron.”
The only thing that matters is AI demand.
AMD is the clear winner in this market. They will own the graphics market and will create the best next gen AI chips from now on.
Nvidia’s got a time-limited (crumbling) moat and it needs to capitalize it to death before it goes away, replaced by competing stacks.
If the US datacenter industry can’t absorb their max capacity, its only option is to throw its momentum towards building clouds on their own stack, wherever they can. Asia.
Next step would’ve been boostrap their own AI lab on OSS models, and fork.
All this, without getting caught “entering their clients’ businesses”, American clouds and American ai labs
EOG stands for Enron Oil & Gas.
https://www.telegraph.co.uk/business/2025/11/25/we-are-not-e...
Nvidia has lost at least $2trln in market cap almost instantly with this going on.
This is the titanic to Enron’s little dinghy.
Folks are already calling it JENRON
https://bsky.app/profile/carlquintanilla.bsky.social/post/3n...
Maybe a gold bar, a Rolex or a 747 shows up too.
If they find the proverbial bag of drugs, well, mark this down as another instance of poor board leadership allowing the cult of the founder to take hold at Nvidia and the rest of us paying the price.
What NVIDIA is doing isn't illegal and they are far too wise to do something brazen like Enron did which seems to be what a lot of the misinformed users are hoping for