A la Enron but completely legal. It is not just a circle, it is more like a swirl. Wallstreet invested in Ai Farms to purchase NVDA chips and then NVDA invested in Ai Farms and not only that it is investing in its customer, Anthropic and its IPO.. They are recycling the cash that they harvest from stock markets till it becomes the famous singularity in funding like a financial hurricane.
Pretty much fanfic. The gap between premise and execution remains limiting. Nvidia has adapted to changing markets better than anyone else up to this date which is what got them to the top in the first place. What makes you think "this time it's different?"
The fallacy I see repeatedly is someone spots a legitimate issue that could hurt Nvidia and by the time they have stood up the challenge to pull that off, Nvidia has pivoted and addressed it. Or more recently, they just acquired it like Groq.
For your premise to hold, you're asserting Nvidia is blind to inference. And yet his public positioning suggests otherwise.
I'm not saying the bet is against Nvidia the company as in how it executes, or enters new markets.
the bet is against the high margins in inference - when other capable players have entered the market as models become commoditized and the inference serving chips as well. Cerebrus ai etc are already showing custom silicon can make a dent while being served cheaper & faster.
I don't think anyone expects nvidia to "sink". We're concerned that competitive pressures and possible reductions from sky-high demand could make it really hard for nv to justify or maintain their current valuation.
That is, revenues may be unusually cyclically high.
It's going to be hard for frontier labs to keep spending at this rate without starting to make money. And even if they do, lower cost alternatives will likely impact at least to -some- extent how much of that revenue nv gets.
Yes, but the first mover advantage really is a thing. As an example, Yahoo really had to mess up to lose its spot to Google, but they were up to the challenge.
There is the Law of Demand. As ai costs plummet, demand increases, just nobody knows the curve shape yet. But could be that cheap inference is how all the AI supply meets the demand
Demand functions (price as a function of quantity demanded) take all kinds of shapes. Veblen goods are the silly example of wrong-sloped demand [1]. The in-vogue example of sigmoid demand, however, is hot water–make hot water (or lighting, for that matter, as another comment today pointed out for LEDs) cheaper and there is a limit to the things we want hot water for. Halving the cost of hot water doesn't induce much new hot-water demand, it increases demand for other goods and services.
Wut? Enron was literally shutting down energy grids in one location to spike prices in another. Way way way more to Enron story than financing to begin with.
Enron was doing some bad things in energy markets, but they were profitable bad things and unrelated to the collapse. What brought Enron down was accounting problems, with many of the specific details having clear parallels here.
In particular, the "circular financing" agreements look very structurally similar to things like the Merrill Lynch barges. They're not exactly the same, but Nvidia's statements in the source article make me more rather than less concerned; does it just so happen to be the case that their investment targets all want to spend lots of money on Nvidia products, or does an Nvidia investment come with implicit and unaccounted guarantees that the target will spend lots of money on Nvidia products?
Enron comparisons folks are making is about why they failed, not why they made a lot of money. They ultimately failed for reasons scarily similar to what’s going on now across the industry with special purpose vehicles and off balance sheet liabilities.
The WSJ recently reported that there’s around 3 trillion in off balance sheet liabilities floating around in AI. It’s very unclear where the $3 trillion to pay those bills will come from.
Nobody is saying it’s “illegal” but it was news in the WSJ as the companies using creative accounting aren’t exactly going out of their way to make sure everyone knows that this $3 trillion in liabilities exists.
What is your point about electricity? Enron was cooking their books by controlling other companies they said they had no control of to improve their own financial health.
I think there questions for the commitments your talking about but they are not due today and I suspect when spread out over the life of the commitment are dwarfed by cash flow.
The mechanics are legal but the objective is the same which is buy your own output by manufacturing a demand. I am sure they can buy good lawyers to keep it completely legal.
Vendor financing[1] has a long and successful history. Here's a good WSJ article from 2001 about the practice and risks[2], written after the DotCom crash in March 2000.
There is nothing illicit or illegal in anyway about what NVidia is doing. It's reasonable business practice, and people on HN are simply ignorant to think otherwise.
NVidia is very aware of the risks it entails, but has the money to cover those risks.
A big problem in discussions about Nvidia is that people can't distinguish between:
1. The equity investments Nvidia has made in its customers.
2. The guarantees/backstops it has extended to some of its customers.
3. Vendor financing.
The vendor financing is the least interesting of the bunch. Nvidia has already disclosed that when it provides vendor financing, the average customer pays in less than 60 days. These are not long-term financing arrangements and virtually every big company sells on these type of terms (net-30, net-60, etc.).
The equity investments and guarantees are where there is room for legitimate debate.
> equity investments and guarantees are where there is room for legitimate debate
The guarantees dwarf the equity investments. If there is a shenanigan, it's going to be there.
A problem: the line between the guarantees and traditional vendor financing is blurry–one could argue use commitments are no different from repurchase commitments.
Enron was illegal but also a sham. The point is we have no evidence Nvidia's financing is a sham. It could be. And if it is, it's a huge problem. But the shammiest parts of what Enron did do not apply to Nvidia, which makes the comparison a bit like saying OP is Hannibal Lecter but legal while glossing over the fact that OP never murdered anyone but once drank red wine.
Their implication ("A la Enron", "they can buy good lawyers to keep it completely legal") that there is something illicit or wrong in what NVidia is doing.
The OP clearly is implying that it should be illegal for some reason. This is wrong - not only is it nothing like Enron (!?) but it's a great way for both NVidia and the companies building on them to build what they want.
Not a lot of people were clamoring for copilot in their outlook. But NVIDIA are giving money to labs so that they will buy their hardware. Manufacturing demand for their hardware, which is used for more AI, which eager PMs stuff into every crevice they can find and call demand.
ChatGPT was the fastest growing software of all time not that long ago. Citing a bad AI product as evidence for lack of demand is saying the failure of the Homer car was due to the fact that people don't like automobiles.
You could have said this about Google's Search product...
How much would the average normie pay if Google suddenly charged? Sure the user-base would drop.
And what? Did Google suddenly become a bad business?
I find it funny how people have these strange and hypocritical viewpoints when it comes to OpenAI and Anthropic when the hyperscalars like Google, Amazon etc, followed these exact same kinds of playbooks for years and years.
I think you, and many LLM fans, have a grossly distorted view of per unit SaaS costs vs per unit current LLM costs.
Google Searches, once the infra setup was finished, were ridiculously cheap. Not the case for OpenAI & co.
Inflation adjusted, launching most of Google's core services (Search, Gmail, Maps, etc) cost less than launching ChatGPT 5. Running them at similar scales is also much cheaper per user.
There is a point at which it's just too much.
Oh, Google and Amazon never had these kinds of humongous losses and they reached profitability much sooner.
> ChatGPT was the fastest growing software of all time not that long ago.
Google+ was the fastest-growing social network of all time not that long ago.
When you're one of a pair of VC darlings that have effectively-infinite money because of -in part- handwavy promises to cure cancer and eliminate 90% of payroll everywhere, or if you're an established company that has total control over very widely used consumer products, you can do all sorts of things to manufacture amazing growth numbers.
In the case of those VC darlings, we're seeing their shift towards providing their products that are most expensive to create exclusively to B2B customers and also the shift towards justifying the elimination of most of their R&D expenditure. Every company performs belt-tightening in advance of their IPO, [0] and those two are no exception.
[0] ...which is when their finances will be scrutinized by the public and regulators...
> Investing in your customer, what is that if it is not buying demand
It can be manufacturing demand that wouldn't otherwise exist. It can be facilitating demand to come online sooner and smoother. You can't tell which it is by only looking at the transaction; you need to know how many dollars are going into the ecosystem as a whole for purchases of goods and services. Until Anthropic's S-1 lands tomorrow-ish, we won't have that publicly.
Well, Nvidia started out as a manufacturer first & foremost.
Maybe I'm in the middle of the road :)
Seems like economy, scale, velocity, and even "critical mass" are related in some way, but not the same at all.
I would say it's quite possible that economies of scale can go from positive to negative without much warning.
I think it's most sustainable financially when the underlying "economy" is what drives the resulting scale-up, which usually does occur in phases or stages where each successful milestone informs the next campaign more realistically than you can get any other way.
As market demand grows beyond baseline sustainability it becomes less costly to serve each additional customer this way.
The opposite effect could occur if meeting lofty scaling goals requires an ever increasing cost of customer acquisition beyond the point of unmet initial pending demand.
When the scaling process itself is what drives the activity without being limited by the actual buying power of the ultimate consumers at any one point, things can really get ahead of themselves. Accounting practices can be so diverse that the only way to be sure whether scaling ahead of the curve was actually "economical" is after liquidation ends up occurring.
Unfortunately, liquidation of one kind or another is more likely when the scale is based on hyperbolic dreams rather than more reasonably optimistic estimates. But who's to say which is which, and the continuum between them is blurry enough without any highly interested parties trying to muddy the waters even further. Who even knows if they've given it as much thought as it deserves, or if more clear-headed thinking could be the primary factor given what there is to work with :\
> mechanics are legal but the objective is the same which is buy your own output by manufacturing a demand
Eh, I think it's an open question whether OpenAI and Anthropic would be buying GPUs like they are with or without Nvidia's financing. Financing customers' purchases isn't proof per se of demand creation versus demand inducement. Anyone who claims they've seen a certain fact in these financings is deluded or lying.
The Ai companies are not really profitable now so they need cash incoming from the skies to grow even more and they are still losing money. Ai farms and much of Ai software world are money sucking machines with assumed profit in future. Their P/E is assumed to be positive in future but it isn't now. The hardware suppliers are making the real money now from the AI hardware pipeline and NVDA is pretty much the center of hardware pipeline sucking most of the cash. In my opinion NVDA knows that the profits from AI have to flow from Ai software to keep the music playing and that it has a lot of competition incoming and so it is funding its customers and buying its own output to circulate the cash towards itself to make as much money as possible now. It has no other choice in reality.
Their assets are depreciating faster than they can pay it off and that means that the profits are pretty much a temporary illusion in my opinion that is because the so called ai chips have short lifespans and don't age gracefully. What happens when you buy a car and say use it for delivery for income for example and then the engine dies. You have to buy a new car and pay for the old one as well.
> The utilization rates of GPUs for AI workloads in a datacenter run by cloud service providers (CSP) is between 60% and 70%. With such utilization rates, a GPU will typically survive between one and two years, three years at the most, according to a quote allegedly made by a principal generative AI architect from Alphabet and reported by @techfund, a long-term tech investor with good sources.
This is just extremely unbelievable to me. I am certainly not operating at a level the hyperscalers are and have much more limited direct experience. But I do actually put various GPUs inside datacenters (and much harsher locations) and have operated them at balls-to-the-wall 100% utilization for over a decade now.
You get the typical bathtub curve of failures. Unless the hyperscalers are operating these things even more overclocked and beyond thermal specification limits than early GPU crypto miners used to do, I simply cannot believe that the average hardware life is less than the useful life of the whole chip generation itself.
I have plenty of decade old GPUs that operate today just fine. Both consumer and datacenter form factors. The failures tend to be board-level like capacitors and such, so if you are operating at a massive scale partnering with someone who can fix those relatively cheaply is not all that difficult either.
It could be that these H200 and above class sort of stuff is engineered extremely fragile, but I seriously doubt it. The prevailing "common knowledge" pre-AI for GPUs were that they'd burn out in a year or two of heavy use, and that was simply untrue. I saved a ton of money buying batches of used units because everyone was terrified of this - and had no more early failures than I did buying brand new after basic refurb of re-pasting and putting a new fan on them.
I've been pitched data-center deals. They depreciate on an 18- to 24-month schedule, well under the Tom's Hardware terms. The ones who went online a year or two ago aren't losing their chips like ducklings through a storm gate; if anything, their resale value has remained remarkably stable because compute production is the bottleneck.
You've given a source (a great one, btw) for depreciation but not revenue. If you can name a company, I can look if I have a public source that confirms what I know. But broadly speaking, no, unit economics in the AI economy is weirdly sound, though I suspect it's because every non-AI CEO is blowing out their budgets on frivolous spending.
Within the whole industry (beyond Nvidia) this is starting to look a lot more and more like Enron. It does appear “legal” but the special purposes vehicles and off balance sheet commitments are having the same net effect. Convincing everyone things are fine while hoping they don’t notice the massive liabilities building up that are kept out of reporting.
If you haven’t read “Smartest guys in the room” it’s important reading now as it’s scary similar to what’s going on now across AI. Nobody has alleged anything illegal but the net effect on building a house of cards in the AI bubble can be the same.
And that gap in where people are watching (AI company press releases or the creative accounting going on) explains why those watching this are saying “oh no, we’ve seen this movie before” when others are blinding all rah rah about the AI bubble going on forever.
Enron's mistake was doing it too early. Nvidia investing in AI labs to use their own chips falls under anti trust, but I don't think the latest administration is interested in investigating it.
Which is a fundamental difference. When Apple extends me credit to buy an iPhone, that isn't circular financing in a problematic way. I was buying the phone anyway, the financing just made it easier.
It's still questionable. Similar to car manufacturers or IKEA, who all have their own banks, Apple has an incentive to hand out credit to people who should not get it, increasing the likelihood of people getting underwater with their debt.
Generally, the fact that most non-Boomer people simply don't have the means to even save up for basic consumer goods like cars, furniture or a phone but have to go into debt instead is scary. Our entire economy has become a house of cards.
Oh hell yeah. But a lot of folks are treating the existence of customer financing as damning per se. The scale is daunting. But the scale of the entire AI enterprise is massive.
> But a lot of folks are treating the existence of customer financing as damning per se.
Yeah because it's gotten completely predatory. That is what people are getting ever more pissed off about - advertising, social media and gamification (have you seen the ads for Tiktok, Whatnot, Wish and whatever else goes with "live shopping" recently?) leads people to go way deeper into credit than they can afford.
I've also seen zero evidence that Nvidia is extending this credit to related parties to put in sham orders
Have you bothered to look at the finances of Nvidia's AI clients?
None of them are making any money. They're borrowing money they don't have in oder to buy from Nvidia. And now some of this money is coming from Nvidia itself.
In a round about way, Nvidia is buying it's own product.
It's pretty clear that this sort of thing can't continue indefinitely --- just like any Ponzi scheme.
All the "frontier" AI vendors are borrowing money to invest in AI (and buy from Nvidia). None of these Nvidia customers are actually making money from it.
Anthropic and OpenAI are two cash burning machines that Nvidia has invested billions into --- so they can continue buying from Nvidia.
Bottom line: A lot (if not most) of Nvidia's cash flow is borrowed money --- and some of it is borrowed from Nvidia itself.
Some of it isn't even "cash flow". It's contract futures being counted as cash flow --- a la Enron accounting.
All of what you said is totally wrong and deluded.
The amount of money Nvidia has put into the ecosystem is much, much less than money coming into the ecosystem from actual customers who are willing to pay for the products!
The idea that somehow Nvidia is financing the entire AI industry is laughable. The numbers do not add up at all if you look at the numbers of people paying for Google cloud GPU compute, AWS GPUs, Azure GPUs, Nebius, Coreweave, etc, not even including companies like Fireworks, BaseTen, Together AI, etc .
The reality is this, enterprise companies are spending HUGE amounts of their money on AI products because they are gaining value from them. This money (which doesn't originate from Nvidia) is flowing into the ecosystem. The money being spent by enterprises combined is far, far more than Nvidia puts in.
The idea that somehow Nvidia is financing the entire AI industry is laughable.
Yes, it is --- and this is not something I said. This is an absurd extrapolation done by you.
But there is no denying that Nvidia is investing billions in it's own customers (aka "lending") and others up and down the AI infrastructure stack.
And all of it has one objective --- to create and enhance what is being marked as "sales" for Nvidia. In a round about way, Nvidia is buying at least some of it's own product.
Where are those 99$ coming from? Who is actually paying this money? Because it seems like so far everybody is losing money with no reversal of this trend in sight
If this isn't a house of cards, AI companies' customers. The companies and individauls ponying up for a Claude subscription or compute through OpenRouter.
Customer financing isn't inherently fucked. It's just highly suspect at the scale Nvidia's doing it. There was another thread where I noted that Nvidia's investments are literally monetarily significant, to the point that I expect them to start being directly referenced in the Fed's beige book [1].
Money flowing from the rest of the economy means money no longer being spent on something else. Assuming for the moment it's not all a big circular inflationary scheme, what exactly is being cut back on to fund this infinite spend on AI inference?
Any one actor has finite cash flow (plus their cash-on-hand buffer). Directing cashflow towards inference necessarily directs it away from something else. Either these companies are spending less on something else, or they are returning less profit.
And this is the actual problem. I can't see framing it as a ratio as anything other than an attempt to mislead. What do you predict will happen when the market deflates? Whether or not a segment will behave like a line of dominoes and catastrophically implode is the question.
As I understand it they are risking that even if the major AI labs fail all the compute capacity that's been built out will remain in demand at sufficiently high prices.
> Where are those 99$ coming from? Who is actually paying this money?
Companies who pay 99$ to make >99$ in return. I am not saying it works in all cases but that's the idea when a company spends money.
> Because it seems like so far everybody is losing money with no reversal of this trend in sight
I am not sure what you are seeing: Anthropic (as one of only two major companies that do just AI) is starting to return profits, while demand for AI is accelerating and, clearly, compute is maxed out. And I mean: On the entire planet. They are turning profits despite everything being in full buildout mode.
Well, it is. Thermodynamics. Earth is pretty much a closed system, except for the sun.
So for example if you invest $100 in a farm and get a return of 10%, where does that come from? The nutrients in the soil, the effort expended by the workers, and the power of the sun to turn seeds into food. All value comes either from finite resources in the ground (nuclear, oil, ...), from solar power, or from human effort (work, innovations, etc.)
So can you trace back Nvidia's incredible 10000% return on investment to any of these sources? Which ones?
But it seems this "economic value" is often created by depleting actual physical resources (e.g. people's health, clean air and water), usually resources belonging someone other than the person capturing the "value".
Only when more money is created than production increases.
If money grows faster than goods for consumption, then buyers bid up price resulting in inflation.
It's not that simple. Of course there is inflation but in the long run in history, real money has increased massively. We have give opportunity of education to almost everyone, abolished slavery, given food and basic medicine to almost everyone. 100 years back having all of these sounded like a perfect world.
You, when their circular financing scheme fails and you're the one left holding the bag as your government says "they're too important to let them fail".
This month I spent $50 on OpenRouter credits. I find it very useful for coding assistance (not coding per se). It's $50 that I never spent before on this kind of service. That's what we call growth - it quite literally came out of nowhere.
This month I reduced my spend on cloud AI and started using local models as primary - it's maybe 2 or 4 times slower for some things but it still is getting the job done just fine. I talked to a friend of mine who's been doing the same running smaller Qwen models locally for coding.
Your $50 isn't enough to sustain trillions of dollars in CapEx investments. This investment cycle - at this scale - only makes sense if within the next 5 years AI replaces something crazy like 5-10% of jobs, if not more. Not changes, replaces completely.
I merely pointed out, that there is real growth behind this - 50$ from my wallet went to this, that did not go towards it at any point in the past. Your argument that my 50$ is not enough is duly noted, but also intellectually lazy.
It's the marketing they were doing all the time that AI will replace developers and other jobs, then the companies started this bullshit firing thousands of people without really thinking through. It's more sane to advertise that AI will augment us and I guess if they followed this path a more sustainable growth is expected to happen!
The current valuations are only inflated until they're not. Especially if the big American labs can convince the US government to work out a joint "AI non-proliferation" agreement with China which will allow both countries to essentially carve up and techno-colonize the rest of the world without any other competition except between themselves.
We're already seeing signs of this strategy from Open AI and Anthropic warning about the dangers of AI and the need for safety regulations. None of that stuff matters if China is not also on board with it.
> The current valuations are only inflated until they're not
You could say that about any historically inflated valuation all the way back to the tulip mania. Either the expected profit materializes or it doesn't.
> We're already seeing signs of this strategy from Open AI and Anthropic warning about the dangers of AI and the need for safety regulations
I would read this as a desire to pause training to be able to present a profit in anticipation of the IPO. The major AI labs mad scramble to IPO is if anything a sign that they aren't at all confident in the valuation. If they were they would be no hurry to cash out.
I first read that "every $1T brings back $100", thinking, wow, is it THAT bad?
No, it's very unlikely at this point every $1 Nvidia invests brings back $100. Stating something like that is almost a red flag that things are overheating.
But even that isn't really true since Nvidia P/E was unreasonably high even 4 years ago. So how is this not lying to investors (a thing that is illegal)?
As long as it really is true and continues to hold true for all the big players involved then things should be OK.
Though it does imply that there's got to be a long-term source of returns which is 100x richer than Nvidia is now, and that source is willing & able to give up 100x what Nvidia can afford to spend now, or where else is the return actually going to come from?
OTOH if the leverage turns out to be unsustainable, or never was quite as extreme as estimated, then adjustments will have to occur, whether that amounts to carefully planned compensation for any shortcomings, or more abrupt developments which relieve undue pressure.
+20% is a sizeable return, but I don't think anyone would look at me right if I made such an investment and then said "I earned $100 for every $1 invested". This is a mathematical statement with concrete numbers, you can't use it as a figure of speech.
They also said that theur hardware scales faster than Moore's law, when in reality it is way slower than Moore's law. Which I guess at least is a good thing for hyperscalars because it helps with their decision to write off their GPUs over a much longer period.
But still, it seems that billionaires are lying left and right. It isn't just Elon.
they're talking about book value, which is all the AAR garbage VC's throw around with their infinite TAMs and all the other clutches of bullshit capitalists need to put in their nests to make themselves feel like they arn't just very complex parasites.
So, in one aspect, like the Matrix, they're not lying: they litterally see that on the paper they're reading from.
But like in the matrix, they just need to be unplugged from the bullshit machine to recognize that this money doesn't exist in the real world.
Scientists, particularly the ones who get ragged on, you know, social scientists, are now well aware that no matter how well any given social goal works in the lab; no matter how much pscyhology they evaluate, depend on, etc, what matters is real world implementation.
Thier sheets of cash raining in from some future valuation simply do not exist in a vacuum, but they want to pretend it does.
So we're litterally watching capitalists look at their piece of meat inside the matrix, and telling us: they do not care that it doesn't actually exists, because on the Capitalism ledger, it feels like real meat.
It is a bit circular but it also a growing real market. NVIDIA is accelerating the market and also making money. It is a smart strategy on their part.
We are not at the top of this trend yet so I do not see this as over investment.
I remember the criticism Microsoft got for investing in Facebook/Meta that gave them a whooping $15B valuation and getting 1.7% in 2007. Not all deals will turn out that prescient but a few will and make up for any duds. This is a real market.
I think it’s safe to venture into science fiction futures at this point. We have AI and government mass surveillance technology. We’re living in science fiction futures.
sure, as long things go up, smart. when things no longer go up, they will take the whole economy down with them akin to the banking crisis in 2008 or the dotcom bubble before that (which both had far less circular incestuous activity going on) and will shamelessly act surprised after the fact.
as long as things go up people conveniently ignore the lessons from the past.
we are long overdue for a correction and given the fear mongering oligopoly shenanigans we must be very close.
I don't know about $100 back ... but their profit margin is $0.90 on the gpu that is bought with the $1 they lend / swap for equity + they get ongoing interest / equity stake.
So if they just had spare 100 billion they could make 10 trillion? Sounds reasonable and plausible right? Sadly market manipulation is not being prosecuted.
They are definitely in NVIDIA’s own rooms! It’s like, they invest $1 in AI companies, that pay back NVIDIA $100 for GPUs. If you draw the lines you will see something that looks a bit like an ellipse. Or maybe a boomerang path. Or like, a 2d sphere. A bit frustrating we don’t have a simpler word for such a shape
Embedding the image would have ruined the comedic timing and made it less funny. Comedy is mostly about subverting expectations. If your eyes are drawn to the punchline before you read the set up it doesn't work.
I hope that this is temporally and spatially far away enough. In order to fully appreciate the joke, one should have read https://openai.com/index/navier-stokes-solution/ (I didn't beforehand).
I spent a few hours trying to run all the numbers myself from first principles (e.g. sum of all American salaries is 11.7T, plus international), to see what I could come up with.
What I came up with was Nvidia is maybe slightly overpriced currently, but a bad stock to buy or hold onto because the risk of it shrinking is significantly higher than the risk of it ever doubling again (and it trades at a tech/growth multiple). The case is far worse for Anthropic/OpenAI who, at current pricing, capture such a tiny slice of the pie it's hard to see how they will stay in business long term.
It does neither, as money is simply a representation of wealth. Increasing the supply of money without increasing the wealth will mean you'll have more money representing the same wealth. This is called inflation.
But the wealth, or value if you will, remains unaffected.
Money isn't wealth. It's just a representation. It used to be wealth back in the days before paper money, when a coin actually had the value it said it had.
Anything that puts pressures on Nvidia's margins is going to have an outsized effect imo - margins declining from 75% to ~50% would undo all of its projected 2027 growth of 70%.
And what about the purported "AI frontier slowdown" - if frontier labs stop pushing better and better models, the main way to grow the pie is more users and that will eventually stop, too. Then providers would probably stop buying new hardware hand-over-fist and move to a slower depreciation-based replacement.
Because it isn't a simple loan, and the value of the equipment itself as collateral is highly doubtful anyways.
Nvidia is investing in their customers by buying their stock. Nvidia is directly buying compute from their customers. Nvidia is making "if you can't find a customer, we'll take all of your capacity" deals. Nvidia is making a ton of fake huge-number deals which are supposed to be realized over time but most likely never will.
They are doing everything they can to make that customer look like a healthy and valuable business which everyone should invest in and loan money to - which in turn flows directly back to Nvidia to buy GPUs and boost their revenue. They are essentially creating sockpuppets to keep external money flowing in so they can keep the money printer running.
If I'm reading that correctly, then 59% of them claim to plan to use AI, 11% of them claim to use AI and 0 of the AI buyers (absolute number) claim that AI pays for its costs.
Is it really that? AI is still heavily subsidized.
Yes, it seems very possible to me that Nvidia makes a lot of money. They are literally the largest company in the world.
Who pays? Only all the companies, desperately raising as much capital as possible, issuing new stock to do it, going for unprecedented investment rounds, burning cash reserves to fund the largest data centre rollout in history.
Who builds the things that go into said datacenters? The thing that has also raised in price dramatically in recent years?
It is always about scale. At scale today's Nvidia does anything is so big that 100x is not anymore reasonable. Or then it is not significant...
From sub million to hundreds of millions I could theoretically see. Even tens of millions to billions. And even then I would be extremely skeptical.
But with Nvidia we are talking somewhere in the scale at tens or hundreds of billions. And that is actually very large sums of money. Even if it often does not look like it.
In a very narrow sense it may be true that Nvidia expects to make some multiple of the money invested (unlikely to be 100x but whatever). But you can't talk about that without talking about the wider economy. If the wider economy, and especially the pension funds putting money into the scheme, cannot sustain this, it'll eventually collapse and then everyone including Nvidia will lose lots and lots of money.
The only metric by which they are the largest company in the world is valuation. There are thousands of companies more important to your daily existence than Jensen's leather jacket factory.
They are also talking out of their asses, as they are massively overselling their capacity, and everyone is on a 5 to 10 year backlog.
I find that the people who are most concerned about circular financing in the case of AI, are those who fearmonger about deflation and talk about the importance of monetary velocity for the economy.
NVidia just has the money to overleverage themselves by buying companies in AI industry (coincidentally their own customers).
If they put all of their money in the chip business they'd have major problem once someone else makes good enough chips in volume. And Chinese are obviously going to do that very soon.
But if NVidia owns comapnies who buy chips it's winning no matter who they buy the chips from.
With what NVidia is doing, it's making the ride better for themselves and provide a soft cushion for when it ends.
The only way for NVidia to lose is for AI to fail utterly. Which is pretty impossible.
A la Enron but completely legal. It is not just a circle, it is more like a swirl. Wallstreet invested in Ai Farms to purchase NVDA chips and then NVDA invested in Ai Farms and not only that it is investing in its customer, Anthropic and its IPO.. They are recycling the cash that they harvest from stock markets till it becomes the famous singularity in funding like a financial hurricane.
80-90% of AI costs are related to inference. Once zAI served their new model entirely on Chinese chips for free then the nVidia jig is up.
who is going to pay back the money nVidia invests in AI labs, AI datacenter companies if the models are being served dirt cheap.
the Chinese are not the only competitor - Amazon with their Trainium, Google with their TPUs etc.
Nvidia might have a moat on training but on serving it's gonna be a blood bath.
But for now they're capturing 80% of all the AI spend so they gonna keep making money.
Pretty much fanfic. The gap between premise and execution remains limiting. Nvidia has adapted to changing markets better than anyone else up to this date which is what got them to the top in the first place. What makes you think "this time it's different?"
The fallacy I see repeatedly is someone spots a legitimate issue that could hurt Nvidia and by the time they have stood up the challenge to pull that off, Nvidia has pivoted and addressed it. Or more recently, they just acquired it like Groq.
For your premise to hold, you're asserting Nvidia is blind to inference. And yet his public positioning suggests otherwise.
https://qz.com/nvidia-gtc-2026-jensen-huang-keynote-takeaway...
What makes you think they've hit some sort of metaphorical iceberg and all they'll do before they sink is rearrange deck chairs?
I'm not saying the bet is against Nvidia the company as in how it executes, or enters new markets.
the bet is against the high margins in inference - when other capable players have entered the market as models become commoditized and the inference serving chips as well. Cerebrus ai etc are already showing custom silicon can make a dent while being served cheaper & faster.
I don't think anyone expects nvidia to "sink". We're concerned that competitive pressures and possible reductions from sky-high demand could make it really hard for nv to justify or maintain their current valuation.
That is, revenues may be unusually cyclically high.
It's going to be hard for frontier labs to keep spending at this rate without starting to make money. And even if they do, lower cost alternatives will likely impact at least to -some- extent how much of that revenue nv gets.
Have you ever checked out the beta on Nvidia stock? Not to mention its surprisingly sane P/E ratio?
Compare and contrast with Tesla.
I know the P/E isn't completely nuts.
It still prices in revenue growth.
It is possible that nv maintains an absolutely dominant market position and total amount spent on GPUs continues to rise.
It's also possible that either of these things doesn't happen.
Yes, but the first mover advantage really is a thing. As an example, Yahoo really had to mess up to lose its spot to Google, but they were up to the challenge.
That doesn't really have anything to do with what I said.
> But for now they're capturing 80% of all the AI spend so they gonna keep making money.
And this is why the stock remains high. This hasn't even started to make a dent in Nvidia's bottom line yet.
These comparisons to Enron are absurd when Nvidia is generating this much cash flow.
Let's see the other alternatives come online and start taking away sales.
There is the Law of Demand. As ai costs plummet, demand increases, just nobody knows the curve shape yet. But could be that cheap inference is how all the AI supply meets the demand
> There is the Law of Demand
What is this?
Demand functions (price as a function of quantity demanded) take all kinds of shapes. Veblen goods are the silly example of wrong-sloped demand [1]. The in-vogue example of sigmoid demand, however, is hot water–make hot water (or lighting, for that matter, as another comment today pointed out for LEDs) cheaper and there is a limit to the things we want hot water for. Halving the cost of hot water doesn't induce much new hot-water demand, it increases demand for other goods and services.
[1] https://en.wikipedia.org/wiki/Veblen_good
I don't think that follows when free plans to get people interested are still a thing. Most people aren't paying anything at all.
Wut? Enron was literally shutting down energy grids in one location to spike prices in another. Way way way more to Enron story than financing to begin with.
Not everything is an Enron
Enron was doing some bad things in energy markets, but they were profitable bad things and unrelated to the collapse. What brought Enron down was accounting problems, with many of the specific details having clear parallels here.
In particular, the "circular financing" agreements look very structurally similar to things like the Merrill Lynch barges. They're not exactly the same, but Nvidia's statements in the source article make me more rather than less concerned; does it just so happen to be the case that their investment targets all want to spend lots of money on Nvidia products, or does an Nvidia investment come with implicit and unaccounted guarantees that the target will spend lots of money on Nvidia products?
> Enron was literally shutting down energy grids in one location to spike prices in another.
Kind of reminds me of the current lack of supply of GPUs and ram to consumers.
How? Two very different things but because we don’t like AI we make bad comparisons?
Enron comparisons folks are making is about why they failed, not why they made a lot of money. They ultimately failed for reasons scarily similar to what’s going on now across the industry with special purpose vehicles and off balance sheet liabilities.
Just wrong. It’s a superficial comparison at best. Enron was cooking books using entities they controlled but making it look like they did not.
You could make an argument that maybe this is similar to oracle (I think?) in the com bubble financing networking gear to customers.
I still think that misses the mark since the customers using hardware are have demand for computer by their customers.
And people were really buying electricity too.
The WSJ recently reported that there’s around 3 trillion in off balance sheet liabilities floating around in AI. It’s very unclear where the $3 trillion to pay those bills will come from.
Nobody is saying it’s “illegal” but it was news in the WSJ as the companies using creative accounting aren’t exactly going out of their way to make sure everyone knows that this $3 trillion in liabilities exists.
What is your point about electricity? Enron was cooking their books by controlling other companies they said they had no control of to improve their own financial health.
I think there questions for the commitments your talking about but they are not due today and I suspect when spread out over the life of the commitment are dwarfed by cash flow.
Boots. You could cook boots but they still won't be edible.
who is "[harvesting cash] from stock markets"? Show me where consumers are being bilked.
NVDA hasn't sold any stock to the public since 1999. If they want to trade shovels for ownership stake in the mines - good for them.
They have issued billions corporate bonds, but buyers like Goldman Sachs and J.P. Morgan have armies of analysis and lawyers.
> Enron but completely legal
Nvidia's financing is disclosed. Enron lied about its schemes.
I've also seen zero evidence that Nvidia is extending this credit to related parties to put in sham orders–that was part of Enron's shtick, too.
The mechanics are legal but the objective is the same which is buy your own output by manufacturing a demand. I am sure they can buy good lawyers to keep it completely legal.
Vendor financing[1] has a long and successful history. Here's a good WSJ article from 2001 about the practice and risks[2], written after the DotCom crash in March 2000.
There is nothing illicit or illegal in anyway about what NVidia is doing. It's reasonable business practice, and people on HN are simply ignorant to think otherwise.
NVidia is very aware of the risks it entails, but has the money to cover those risks.
[1] https://en.wikipedia.org/wiki/Vendor_finance
[2] https://archive.is/mOIfg
A big problem in discussions about Nvidia is that people can't distinguish between:
1. The equity investments Nvidia has made in its customers.
2. The guarantees/backstops it has extended to some of its customers.
3. Vendor financing.
The vendor financing is the least interesting of the bunch. Nvidia has already disclosed that when it provides vendor financing, the average customer pays in less than 60 days. These are not long-term financing arrangements and virtually every big company sells on these type of terms (net-30, net-60, etc.).
The equity investments and guarantees are where there is room for legitimate debate.
> equity investments and guarantees are where there is room for legitimate debate
The guarantees dwarf the equity investments. If there is a shenanigan, it's going to be there.
A problem: the line between the guarantees and traditional vendor financing is blurry–one could argue use commitments are no different from repurchase commitments.
> There is nothing illicit or illegal
Which is exactly the point of the person you're responding to. What part of “but completely legal” isn't clear enough?
Enron was illegal but also a sham. The point is we have no evidence Nvidia's financing is a sham. It could be. And if it is, it's a huge problem. But the shammiest parts of what Enron did do not apply to Nvidia, which makes the comparison a bit like saying OP is Hannibal Lecter but legal while glossing over the fact that OP never murdered anyone but once drank red wine.
(Also note that illegal != illicit.)
Their implication ("A la Enron", "they can buy good lawyers to keep it completely legal") that there is something illicit or wrong in what NVidia is doing.
The OP clearly is implying that it should be illegal for some reason. This is wrong - not only is it nothing like Enron (!?) but it's a great way for both NVidia and the companies building on them to build what they want.
Are you claiming NVIDIA manufactured the demand for AI?
Not a lot of people were clamoring for copilot in their outlook. But NVIDIA are giving money to labs so that they will buy their hardware. Manufacturing demand for their hardware, which is used for more AI, which eager PMs stuff into every crevice they can find and call demand.
ChatGPT was the fastest growing software of all time not that long ago. Citing a bad AI product as evidence for lack of demand is saying the failure of the Homer car was due to the fact that people don't like automobiles.
ChatGPT was the fastest growing gratis software of all time.
I wish we still lived in an age where things were sold to cover their costs.
My money's on their product having 10% of its current user base if they charged 50 cents per month.
Which doesn't matter if they're still not profitable.
You could have said this about Google's Search product...
How much would the average normie pay if Google suddenly charged? Sure the user-base would drop.
And what? Did Google suddenly become a bad business?
I find it funny how people have these strange and hypocritical viewpoints when it comes to OpenAI and Anthropic when the hyperscalars like Google, Amazon etc, followed these exact same kinds of playbooks for years and years.
Probably not a normie but I pay for Kagi search because it’s a lot like pre-seo google. So far haven’t tried any AI since OpenCYC.
The labs have raised far, far more money than any of the hyperscalers ever did, with no sign of a route to monopoly profits.
Oh, besides my other comment.
Yes, I'm against all predatory economic behavior. Selling user data, price dumping, platforms/walled gardens/monopolies/oligopolies/cartels/etc.
I think you, and many LLM fans, have a grossly distorted view of per unit SaaS costs vs per unit current LLM costs.
Google Searches, once the infra setup was finished, were ridiculously cheap. Not the case for OpenAI & co.
Inflation adjusted, launching most of Google's core services (Search, Gmail, Maps, etc) cost less than launching ChatGPT 5. Running them at similar scales is also much cheaper per user.
There is a point at which it's just too much.
Oh, Google and Amazon never had these kinds of humongous losses and they reached profitability much sooner.
> ChatGPT was the fastest growing software of all time not that long ago.
Google+ was the fastest-growing social network of all time not that long ago.
When you're one of a pair of VC darlings that have effectively-infinite money because of -in part- handwavy promises to cure cancer and eliminate 90% of payroll everywhere, or if you're an established company that has total control over very widely used consumer products, you can do all sorts of things to manufacture amazing growth numbers.
In the case of those VC darlings, we're seeing their shift towards providing their products that are most expensive to create exclusively to B2B customers and also the shift towards justifying the elimination of most of their R&D expenditure. Every company performs belt-tightening in advance of their IPO, [0] and those two are no exception.
[0] ...which is when their finances will be scrutinized by the public and regulators...
ChatGPT is in my tool belt now, and I barely touch Google anymore.
The world is fundamentally and momentously being rewired.
Don't need to go far. Sora by the same company was the fastest incineration of dollars in software of all time.
Yup. Investing in your customer, what is that if it is not buying demand. They get the cash to buy your output.
> Investing in your customer, what is that if it is not buying demand
It can be manufacturing demand that wouldn't otherwise exist. It can be facilitating demand to come online sooner and smoother. You can't tell which it is by only looking at the transaction; you need to know how many dollars are going into the ecosystem as a whole for purchases of goods and services. Until Anthropic's S-1 lands tomorrow-ish, we won't have that publicly.
I'd argue that in this case, Nvidia is accelerating a flywheel to try and reach critical velocity rather than manufacturing demand.
> to try and reach critical velocity
Do you mean economies of scale?
> I'd argue that in this case, Nvidia is accelerating a flywheel to try and reach critical velocity rather than manufacturing demand.
I'd argue that in this case, if the flywheel never reaches critical velocity, they are manufacturing demand.
Well, Nvidia started out as a manufacturer first & foremost.
Maybe I'm in the middle of the road :)
Seems like economy, scale, velocity, and even "critical mass" are related in some way, but not the same at all.
I would say it's quite possible that economies of scale can go from positive to negative without much warning.
I think it's most sustainable financially when the underlying "economy" is what drives the resulting scale-up, which usually does occur in phases or stages where each successful milestone informs the next campaign more realistically than you can get any other way.
As market demand grows beyond baseline sustainability it becomes less costly to serve each additional customer this way.
The opposite effect could occur if meeting lofty scaling goals requires an ever increasing cost of customer acquisition beyond the point of unmet initial pending demand.
When the scaling process itself is what drives the activity without being limited by the actual buying power of the ultimate consumers at any one point, things can really get ahead of themselves. Accounting practices can be so diverse that the only way to be sure whether scaling ahead of the curve was actually "economical" is after liquidation ends up occurring.
Unfortunately, liquidation of one kind or another is more likely when the scale is based on hyperbolic dreams rather than more reasonably optimistic estimates. But who's to say which is which, and the continuum between them is blurry enough without any highly interested parties trying to muddy the waters even further. Who even knows if they've given it as much thought as it deserves, or if more clear-headed thinking could be the primary factor given what there is to work with :\
Hence the designation "Hyper-Scalers".
> mechanics are legal but the objective is the same which is buy your own output by manufacturing a demand
Eh, I think it's an open question whether OpenAI and Anthropic would be buying GPUs like they are with or without Nvidia's financing. Financing customers' purchases isn't proof per se of demand creation versus demand inducement. Anyone who claims they've seen a certain fact in these financings is deluded or lying.
The Ai companies are not really profitable now so they need cash incoming from the skies to grow even more and they are still losing money. Ai farms and much of Ai software world are money sucking machines with assumed profit in future. Their P/E is assumed to be positive in future but it isn't now. The hardware suppliers are making the real money now from the AI hardware pipeline and NVDA is pretty much the center of hardware pipeline sucking most of the cash. In my opinion NVDA knows that the profits from AI have to flow from Ai software to keep the music playing and that it has a lot of competition incoming and so it is funding its customers and buying its own output to circulate the cash towards itself to make as much money as possible now. It has no other choice in reality.
> Ai companies are not really profitable now
You need to be more specific, because there are absolutely sections of the AI economy that are clearly and presently profitable.
> Ai farms and much of Ai software world are money sucking machines
If AI farms refers to datacenters, plenty of existing ones are currently profitable.
Their assets are depreciating faster than they can pay it off and that means that the profits are pretty much a temporary illusion in my opinion that is because the so called ai chips have short lifespans and don't age gracefully. What happens when you buy a car and say use it for delivery for income for example and then the engine dies. You have to buy a new car and pay for the old one as well.
> Their assets are depreciating faster than they can pay it off
Whose assets? Where are you getting this from to be able to state it with this level of certainty?
I said in my opinion..
But here a reference; https://www.tomshardware.com/pc-components/gpus/datacenter-g...
> The utilization rates of GPUs for AI workloads in a datacenter run by cloud service providers (CSP) is between 60% and 70%. With such utilization rates, a GPU will typically survive between one and two years, three years at the most, according to a quote allegedly made by a principal generative AI architect from Alphabet and reported by @techfund, a long-term tech investor with good sources.
This is just extremely unbelievable to me. I am certainly not operating at a level the hyperscalers are and have much more limited direct experience. But I do actually put various GPUs inside datacenters (and much harsher locations) and have operated them at balls-to-the-wall 100% utilization for over a decade now.
You get the typical bathtub curve of failures. Unless the hyperscalers are operating these things even more overclocked and beyond thermal specification limits than early GPU crypto miners used to do, I simply cannot believe that the average hardware life is less than the useful life of the whole chip generation itself.
I have plenty of decade old GPUs that operate today just fine. Both consumer and datacenter form factors. The failures tend to be board-level like capacitors and such, so if you are operating at a massive scale partnering with someone who can fix those relatively cheaply is not all that difficult either.
It could be that these H200 and above class sort of stuff is engineered extremely fragile, but I seriously doubt it. The prevailing "common knowledge" pre-AI for GPUs were that they'd burn out in a year or two of heavy use, and that was simply untrue. I saved a ton of money buying batches of used units because everyone was terrified of this - and had no more early failures than I did buying brand new after basic refurb of re-pasting and putting a new fan on them.
> I said in my opinion
I've been pitched data-center deals. They depreciate on an 18- to 24-month schedule, well under the Tom's Hardware terms. The ones who went online a year or two ago aren't losing their chips like ducklings through a storm gate; if anything, their resale value has remained remarkably stable because compute production is the bottleneck.
You've given a source (a great one, btw) for depreciation but not revenue. If you can name a company, I can look if I have a public source that confirms what I know. But broadly speaking, no, unit economics in the AI economy is weirdly sound, though I suspect it's because every non-AI CEO is blowing out their budgets on frivolous spending.
Within the whole industry (beyond Nvidia) this is starting to look a lot more and more like Enron. It does appear “legal” but the special purposes vehicles and off balance sheet commitments are having the same net effect. Convincing everyone things are fine while hoping they don’t notice the massive liabilities building up that are kept out of reporting.
If you haven’t read “Smartest guys in the room” it’s important reading now as it’s scary similar to what’s going on now across AI. Nobody has alleged anything illegal but the net effect on building a house of cards in the AI bubble can be the same.
And that gap in where people are watching (AI company press releases or the creative accounting going on) explains why those watching this are saying “oh no, we’ve seen this movie before” when others are blinding all rah rah about the AI bubble going on forever.
Enron's mistake was doing it too early. Nvidia investing in AI labs to use their own chips falls under anti trust, but I don't think the latest administration is interested in investigating it.
Right, that's why it's legal. We all get that
> that's why it's legal
Which is a fundamental difference. When Apple extends me credit to buy an iPhone, that isn't circular financing in a problematic way. I was buying the phone anyway, the financing just made it easier.
It's still questionable. Similar to car manufacturers or IKEA, who all have their own banks, Apple has an incentive to hand out credit to people who should not get it, increasing the likelihood of people getting underwater with their debt.
Generally, the fact that most non-Boomer people simply don't have the means to even save up for basic consumer goods like cars, furniture or a phone but have to go into debt instead is scary. Our entire economy has become a house of cards.
> still questionable
Oh hell yeah. But a lot of folks are treating the existence of customer financing as damning per se. The scale is daunting. But the scale of the entire AI enterprise is massive.
> But a lot of folks are treating the existence of customer financing as damning per se.
Yeah because it's gotten completely predatory. That is what people are getting ever more pissed off about - advertising, social media and gamification (have you seen the ads for Tiktok, Whatnot, Wish and whatever else goes with "live shopping" recently?) leads people to go way deeper into credit than they can afford.
But the scale of the entire AI enterprise is massive.
Yes --- and the scale of the money being set on fire is epic. And if/when the burning comes to a screeching halt, the resulting crash wil be likewise.
I've also seen zero evidence that Nvidia is extending this credit to related parties to put in sham orders
Have you bothered to look at the finances of Nvidia's AI clients?
None of them are making any money. They're borrowing money they don't have in oder to buy from Nvidia. And now some of this money is coming from Nvidia itself.
In a round about way, Nvidia is buying it's own product.
It's pretty clear that this sort of thing can't continue indefinitely --- just like any Ponzi scheme.
> None of them are making any money
Who are you thinking of? Because yes, I have, and they're not in line with the YouTube influencer consensus.
Who are you thinking of?
All the "frontier" AI vendors are borrowing money to invest in AI (and buy from Nvidia). None of these Nvidia customers are actually making money from it.
Anthropic and OpenAI are two cash burning machines that Nvidia has invested billions into --- so they can continue buying from Nvidia.
Bottom line: A lot (if not most) of Nvidia's cash flow is borrowed money --- and some of it is borrowed from Nvidia itself.
Some of it isn't even "cash flow". It's contract futures being counted as cash flow --- a la Enron accounting.
All of what you said is totally wrong and deluded.
The amount of money Nvidia has put into the ecosystem is much, much less than money coming into the ecosystem from actual customers who are willing to pay for the products!
The idea that somehow Nvidia is financing the entire AI industry is laughable. The numbers do not add up at all if you look at the numbers of people paying for Google cloud GPU compute, AWS GPUs, Azure GPUs, Nebius, Coreweave, etc, not even including companies like Fireworks, BaseTen, Together AI, etc .
The reality is this, enterprise companies are spending HUGE amounts of their money on AI products because they are gaining value from them. This money (which doesn't originate from Nvidia) is flowing into the ecosystem. The money being spent by enterprises combined is far, far more than Nvidia puts in.
The idea that somehow Nvidia is financing the entire AI industry is laughable.
Yes, it is --- and this is not something I said. This is an absurd extrapolation done by you.
But there is no denying that Nvidia is investing billions in it's own customers (aka "lending") and others up and down the AI infrastructure stack.
And all of it has one objective --- to create and enhance what is being marked as "sales" for Nvidia. In a round about way, Nvidia is buying at least some of it's own product.
https://www.cnbc.com/2026/05/09/nvidia-embraces-ai-investor-...
Where are those 99$ coming from? Who is actually paying this money? Because it seems like so far everybody is losing money with no reversal of this trend in sight
> Where are those 99$ coming from?
If this isn't a house of cards, AI companies' customers. The companies and individauls ponying up for a Claude subscription or compute through OpenRouter.
Customer financing isn't inherently fucked. It's just highly suspect at the scale Nvidia's doing it. There was another thread where I noted that Nvidia's investments are literally monetarily significant, to the point that I expect them to start being directly referenced in the Fed's beige book [1].
[1] https://news.ycombinator.com/item?id=49673871
Demand for AI inference is infinite. It's coming from the rest of the economy.
Money flowing from the rest of the economy means money no longer being spent on something else. Assuming for the moment it's not all a big circular inflationary scheme, what exactly is being cut back on to fund this infinite spend on AI inference?
Economic activity is a flow, not a stock. There might well be a bubble, but you can't prove it like that.
> Economic activity is a flow, not a stock
Any one actor has finite cash flow (plus their cash-on-hand buffer). Directing cashflow towards inference necessarily directs it away from something else. Either these companies are spending less on something else, or they are returning less profit.
sure elon
And this is the actual problem. I can't see framing it as a ratio as anything other than an attempt to mislead. What do you predict will happen when the market deflates? Whether or not a segment will behave like a line of dominoes and catastrophically implode is the question.
As I understand it they are risking that even if the major AI labs fail all the compute capacity that's been built out will remain in demand at sufficiently high prices.
> Where are those 99$ coming from? Who is actually paying this money?
Companies who pay 99$ to make >99$ in return. I am not saying it works in all cases but that's the idea when a company spends money.
> Because it seems like so far everybody is losing money with no reversal of this trend in sight
I am not sure what you are seeing: Anthropic (as one of only two major companies that do just AI) is starting to return profits, while demand for AI is accelerating and, clearly, compute is maxed out. And I mean: On the entire planet. They are turning profits despite everything being in full buildout mode.
Is Anthropic turning a profit? I thought that was only when considering discounted compute from spaceX
You do realise it’s not a fixed pie right?
Well, it is. Thermodynamics. Earth is pretty much a closed system, except for the sun.
So for example if you invest $100 in a farm and get a return of 10%, where does that come from? The nutrients in the soil, the effort expended by the workers, and the power of the sun to turn seeds into food. All value comes either from finite resources in the ground (nuclear, oil, ...), from solar power, or from human effort (work, innovations, etc.)
So can you trace back Nvidia's incredible 10000% return on investment to any of these sources? Which ones?
Economic value isnt a physical resource being depleted....
But it seems this "economic value" is often created by depleting actual physical resources (e.g. people's health, clean air and water), usually resources belonging someone other than the person capturing the "value".
So many don't want to admit things like that.
"Economic value" is great stuff and kind of all-encompassing.
But a lot of it is mainly moving money around after wealth has already been created, sometimes generations earlier.
To this day the lion's share of "creating wealth" is traceable to natural resources.
Money is created all the time. It is not a zero-sum game.
When money is created, the total amount of money in circulation increases, decreasing the value of each unit of money overall. It’s called inflation.
Only when more money is created than production increases. If money grows faster than goods for consumption, then buyers bid up price resulting in inflation.
No it’s not fixed like that. Total prosperity of world has increased inflation adjusted over 100 years. Why?
Surprised basic stuff is now being questioned.
It's not that simple. Of course there is inflation but in the long run in history, real money has increased massively. We have give opportunity of education to almost everyone, abolished slavery, given food and basic medicine to almost everyone. 100 years back having all of these sounded like a perfect world.
If it’s a fixed pie why do we have more collectively now than 200 years ago?
Value also comes from technology which you seem to suspiciously remove
>Who is actually paying this money?
You, when their circular financing scheme fails and you're the one left holding the bag as your government says "they're too important to let them fail".
Also you if your retirement funds are heavily invested in AI, directly or indirectly through big caps who are.
This month I spent $50 on OpenRouter credits. I find it very useful for coding assistance (not coding per se). It's $50 that I never spent before on this kind of service. That's what we call growth - it quite literally came out of nowhere.
This month I reduced my spend on cloud AI and started using local models as primary - it's maybe 2 or 4 times slower for some things but it still is getting the job done just fine. I talked to a friend of mine who's been doing the same running smaller Qwen models locally for coding.
Your $50 isn't enough to sustain trillions of dollars in CapEx investments. This investment cycle - at this scale - only makes sense if within the next 5 years AI replaces something crazy like 5-10% of jobs, if not more. Not changes, replaces completely.
And I'd argue the timer started in 2023.
I merely pointed out, that there is real growth behind this - 50$ from my wallet went to this, that did not go towards it at any point in the past. Your argument that my 50$ is not enough is duly noted, but also intellectually lazy.
It's the marketing they were doing all the time that AI will replace developers and other jobs, then the companies started this bullshit firing thousands of people without really thinking through. It's more sane to advertise that AI will augment us and I guess if they followed this path a more sustainable growth is expected to happen!
I've spent several 1000 dollars on AI products this month, roughly 3000 dollars.
Trust me, there are many other people like me in the world and the enterprises are spending even more.
There is more money flowing into the overall AI ecosystem (by far) than the money Nvidia puts in.
The idea that Nvidia is artificially creating the whole demand is laughable and doesn't add up.
The AI numbers only work if it replaces white collar work.
Power tools for knowledge workers, which is what we are getting, isn't enough to save it.
Inflated valuations, mostly.
The current valuations are only inflated until they're not. Especially if the big American labs can convince the US government to work out a joint "AI non-proliferation" agreement with China which will allow both countries to essentially carve up and techno-colonize the rest of the world without any other competition except between themselves.
We're already seeing signs of this strategy from Open AI and Anthropic warning about the dangers of AI and the need for safety regulations. None of that stuff matters if China is not also on board with it.
> The current valuations are only inflated until they're not
You could say that about any historically inflated valuation all the way back to the tulip mania. Either the expected profit materializes or it doesn't.
> We're already seeing signs of this strategy from Open AI and Anthropic warning about the dangers of AI and the need for safety regulations
I would read this as a desire to pause training to be able to present a profit in anticipation of the IPO. The major AI labs mad scramble to IPO is if anything a sign that they aren't at all confident in the valuation. If they were they would be no hurry to cash out.
I first read that "every $1T brings back $100", thinking, wow, is it THAT bad?
No, it's very unlikely at this point every $1 Nvidia invests brings back $100. Stating something like that is almost a red flag that things are overheating.
Every $1 they invest brings back $100 in their valuation maybe.
But even that isn't really true since Nvidia P/E was unreasonably high even 4 years ago. So how is this not lying to investors (a thing that is illegal)?
Because a PE is a fact not an interpretation.
It's interesting how people who stare only at P/E make consistently bad investment decisions. Perhaps P/E is not such an important indicator overall?
I agree. But that still doesn't make it a lie or illegal.
>every $1 it invests brings back $100
That's great leverage.
As long as it really is true and continues to hold true for all the big players involved then things should be OK.
Though it does imply that there's got to be a long-term source of returns which is 100x richer than Nvidia is now, and that source is willing & able to give up 100x what Nvidia can afford to spend now, or where else is the return actually going to come from?
OTOH if the leverage turns out to be unsustainable, or never was quite as extreme as estimated, then adjustments will have to occur, whether that amounts to carefully planned compensation for any shortcomings, or more abrupt developments which relieve undue pressure.
Reminds me of the kind of stuff Bernie Madoff said before he got busted
Its definitely consistent over the past several years.
> The figure was rhetorical, not a disclosed 100-times investment return.
seems no one is reading the article..
It's a very stupid thing to say though, it's like Trump's promise to reduce drug prices by 1000%. I guess that was also just a figure of speech.
It's not nearly that stupid. If their investments are, on-average, bringing sizable returns, it's a perfectly cromulent figure of speech.
+20% is a sizeable return, but I don't think anyone would look at me right if I made such an investment and then said "I earned $100 for every $1 invested". This is a mathematical statement with concrete numbers, you can't use it as a figure of speech.
They also said that theur hardware scales faster than Moore's law, when in reality it is way slower than Moore's law. Which I guess at least is a good thing for hyperscalars because it helps with their decision to write off their GPUs over a much longer period.
But still, it seems that billionaires are lying left and right. It isn't just Elon.
they're talking about book value, which is all the AAR garbage VC's throw around with their infinite TAMs and all the other clutches of bullshit capitalists need to put in their nests to make themselves feel like they arn't just very complex parasites.
So, in one aspect, like the Matrix, they're not lying: they litterally see that on the paper they're reading from.
But like in the matrix, they just need to be unplugged from the bullshit machine to recognize that this money doesn't exist in the real world.
Scientists, particularly the ones who get ragged on, you know, social scientists, are now well aware that no matter how well any given social goal works in the lab; no matter how much pscyhology they evaluate, depend on, etc, what matters is real world implementation.
Thier sheets of cash raining in from some future valuation simply do not exist in a vacuum, but they want to pretend it does.
So we're litterally watching capitalists look at their piece of meat inside the matrix, and telling us: they do not care that it doesn't actually exists, because on the Capitalism ledger, it feels like real meat.
It is a bit circular but it also a growing real market. NVIDIA is accelerating the market and also making money. It is a smart strategy on their part.
We are not at the top of this trend yet so I do not see this as over investment.
I remember the criticism Microsoft got for investing in Facebook/Meta that gave them a whooping $15B valuation and getting 1.7% in 2007. Not all deals will turn out that prescient but a few will and make up for any duds. This is a real market.
> It is a smart strategy on their part.
No, it's a high risk gamble.
If the market grows enough they will win the bet, but if the market doesn't or we get a recession that dries up capital they will be holding the bag.
> If the market grows enough they will win the bet
So for this to become true there are some number of jobs that that pay $N salary are replaced completely by LLMs that do the job for $N-0.01?
Is that what you mean by "market grows enough"
> No, it's a high risk gamble.
Even in a recession I think the shift towards AI would just accelerate since it is usually cheaper than humans.
The real risk I feel if what if AI is too successful and there is a lack of human demand because of dropping wages/employment?
(Then I wonder if sentient robot demand will make up for it? Although I realize that veers into science fiction futures.)
I think it’s safe to venture into science fiction futures at this point. We have AI and government mass surveillance technology. We’re living in science fiction futures.
What do you think the difference is. If you are making a quarter trillion in profit per year (where NVIDIA is tracking), you want to invest.
What else should they be buying? US bonds, housing?
Tell us how you know we are not at the top of this trend yet please
Because the benchmarks are stilling going up fairly rapidly. There are rumors of RSI. We are not at the top definitely.
And then the following trend will be humanoids and similar and they are barely getting started.
sure, as long things go up, smart. when things no longer go up, they will take the whole economy down with them akin to the banking crisis in 2008 or the dotcom bubble before that (which both had far less circular incestuous activity going on) and will shamelessly act surprised after the fact.
as long as things go up people conveniently ignore the lessons from the past.
we are long overdue for a correction and given the fear mongering oligopoly shenanigans we must be very close.
I don't know about $100 back ... but their profit margin is $0.90 on the gpu that is bought with the $1 they lend / swap for equity + they get ongoing interest / equity stake.
Its a mighty fine deal for Nvidia
So if they just had spare 100 billion they could make 10 trillion? Sounds reasonable and plausible right? Sadly market manipulation is not being prosecuted.
After the downfall - it will be prosecutable. Of course before that we (simpletons) will have to pay for those companies to avoid “economic collapse”.
are those 100$ in the room with us right now
They are definitely in NVIDIA’s own rooms! It’s like, they invest $1 in AI companies, that pay back NVIDIA $100 for GPUs. If you draw the lines you will see something that looks a bit like an ellipse. Or maybe a boomerang path. Or like, a 2d sphere. A bit frustrating we don’t have a simpler word for such a shape
I asked Astra to explain this and it thought for a really long time and then came back with this:
https://www.marginalia.nu/junk/financing.png
That’s hilarious, I wish HN rendered pictures just for that one!
Embedding the image would have ruined the comedic timing and made it less funny. Comedy is mostly about subverting expectations. If your eyes are drawn to the punchline before you read the set up it doesn't work.
I hope that this is temporally and spatially far away enough. In order to fully appreciate the joke, one should have read https://openai.com/index/navier-stokes-solution/ (I didn't beforehand).
This reminds me of the “it’s not a pyramid scheme. It’s a reverse funnel system”
I spent a few hours trying to run all the numbers myself from first principles (e.g. sum of all American salaries is 11.7T, plus international), to see what I could come up with.
What I came up with was Nvidia is maybe slightly overpriced currently, but a bad stock to buy or hold onto because the risk of it shrinking is significantly higher than the risk of it ever doubling again (and it trades at a tech/growth multiple). The case is far worse for Anthropic/OpenAI who, at current pricing, capture such a tiny slice of the pie it's hard to see how they will stay in business long term.
Those $99, where do they come from?
Shush there! Don't ask any difficult questions!
https://www.youtube.com/watch?v=tO5sxLapAts
The Fed is printing them. Just like they print all other money.
From the article:
> I put in one, and a hundred comes back.”
> The figure was rhetorical, not a disclosed 100-times investment return.
There's a lot more words in there, but it doesn't seem to say anything else.
Printing money doesn’t create value. It dilutes value.
It does neither, as money is simply a representation of wealth. Increasing the supply of money without increasing the wealth will mean you'll have more money representing the same wealth. This is called inflation.
But the wealth, or value if you will, remains unaffected.
Money isn't wealth. It's just a representation. It used to be wealth back in the days before paper money, when a coin actually had the value it said it had.
The future we want. Not the wasteful F-35 'Fat Amy' lemon future noone except the fighterjet mafia wants..
Anything that puts pressures on Nvidia's margins is going to have an outsized effect imo - margins declining from 75% to ~50% would undo all of its projected 2027 growth of 70%.
And what about the purported "AI frontier slowdown" - if frontier labs stop pushing better and better models, the main way to grow the pie is more users and that will eventually stop, too. Then providers would probably stop buying new hardware hand-over-fist and move to a slower depreciation-based replacement.
nvidia invests $1 in AI company , datacenters with GPUs , AI company / datacenters buy $x worth of GPUs from NVIDIA , in a circle
Isn't vendor financing a common business practice to finance the cost of equipment? What makes it nefarious in this instance?
Because it isn't a simple loan, and the value of the equipment itself as collateral is highly doubtful anyways.
Nvidia is investing in their customers by buying their stock. Nvidia is directly buying compute from their customers. Nvidia is making "if you can't find a customer, we'll take all of your capacity" deals. Nvidia is making a ton of fake huge-number deals which are supposed to be realized over time but most likely never will.
They are doing everything they can to make that customer look like a healthy and valuable business which everyone should invest in and loan money to - which in turn flows directly back to Nvidia to buy GPUs and boost their revenue. They are essentially creating sockpuppets to keep external money flowing in so they can keep the money printer running.
Woah, better than Madoff! How can i give him money?
AI isn't showing up in S&P 500 company earnings as a line item. proof: https://aiearnings.dev
given the stakes, we'd better hope it starts showing up soon.
If I'm reading that correctly, then 59% of them claim to plan to use AI, 11% of them claim to use AI and 0 of the AI buyers (absolute number) claim that AI pays for its costs.
Is it really that? AI is still heavily subsidized.
I always here how Nvidia's investment "scheme" is like previous company X that was destroyed using said scheme.
But its quite obvious that while there are some similarities its simply not the same. Most analysis, so it seems, are quite surface.
Are there any deeper finical analysis on the investments of Nvidia and how they are financed?
Isn't most of the financing just another way of expressing discounts and payment plans for very expensive hardware?
They sell shovels in the gold rush and invest on the miners as well. The gold is the tokens . As long as the tokens are consumed,they are fine .
> so why does the stock keep falling
When I zoom out to 5 years, it does not look like "keeps falling".
Zoom back in
Like a grade school student claiming they peed for 10 minutes straight yesterday
Amazed by these comments?
Yes, it seems very possible to me that Nvidia makes a lot of money. They are literally the largest company in the world.
Who pays? Only all the companies, desperately raising as much capital as possible, issuing new stock to do it, going for unprecedented investment rounds, burning cash reserves to fund the largest data centre rollout in history.
Who builds the things that go into said datacenters? The thing that has also raised in price dramatically in recent years?
It is always about scale. At scale today's Nvidia does anything is so big that 100x is not anymore reasonable. Or then it is not significant...
From sub million to hundreds of millions I could theoretically see. Even tens of millions to billions. And even then I would be extremely skeptical.
But with Nvidia we are talking somewhere in the scale at tens or hundreds of billions. And that is actually very large sums of money. Even if it often does not look like it.
In a very narrow sense it may be true that Nvidia expects to make some multiple of the money invested (unlikely to be 100x but whatever). But you can't talk about that without talking about the wider economy. If the wider economy, and especially the pension funds putting money into the scheme, cannot sustain this, it'll eventually collapse and then everyone including Nvidia will lose lots and lots of money.
The only metric by which they are the largest company in the world is valuation. There are thousands of companies more important to your daily existence than Jensen's leather jacket factory.
They are also talking out of their asses, as they are massively overselling their capacity, and everyone is on a 5 to 10 year backlog.
I find that the people who are most concerned about circular financing in the case of AI, are those who fearmonger about deflation and talk about the importance of monetary velocity for the economy.
Isn't this a contradiction?
Ken Lay would like a word
10,000% growth! No way that could be wild claim! How dare we doubt them.
"Every $1 invested brings back $100"
--- Bernie Madoff
You see, it's not a circle, it's a pyramid! Just buy in right now and you'll get unprecedented returns! Nothing could possibly go wrong!
My name is Charles Ponzi and I approve of this message.
100$ for every 1$ invested would definitely be branded as a ponzi scheme if it was done by a regular Joe
Cisco has entered the chat.
Reads like satire. "I get 100$ for every 1$ I put in" is exactly the kind of phrase that triggers ponzi alarm bells on anybody reading this x)
Where does the $99 come from you ask?
Your salaries if AI works (and takes your jobs), and your taxes if it doesn’t (bubble bursts, your taxes bail them out Becasue it’s too big to fail).
NVidia just has the money to overleverage themselves by buying companies in AI industry (coincidentally their own customers).
If they put all of their money in the chip business they'd have major problem once someone else makes good enough chips in volume. And Chinese are obviously going to do that very soon.
But if NVidia owns comapnies who buy chips it's winning no matter who they buy the chips from.
With what NVidia is doing, it's making the ride better for themselves and provide a soft cushion for when it ends.
The only way for NVidia to lose is for AI to fail utterly. Which is pretty impossible.
Every $1 it invests brings back $100
Mega eyeroll
God heavens. It's worse than 2000/2001!
Gotta start selling my portfolio
Nvidia is full of shit. Leatherman thinks Astra = AGI.
So for every trillion that was invested, 100 trillion were made. Got it.