Not for gaming they don't have monitor outputs and most of them aren't even in a standard PCI-E format. Some of them might be useful as a home AI accelerator with adapters and jerry rigged cooling.
Depends, you buy a used server off of eBay or 3d print your own fan shroud. Waiting for prices to drop some and then I’m looking to build my own rack in the basement, so I can use open weight models and not directly support horrible data center practices.
Some of them, sure. You can invest in a server mb, and put a rack in your shed or basement. If someone was selling B100 like that, I'd happily buy a few for 15k. I somehow doubt its going to happen though.
I am not sure how inference constrained we are, btw. As an enterprise customer, I have zero problems with 429s and capacity from google or aws. All large volume inference jobs - so customer journey bots that help our company make money - run on dirt cheap models kept on short leashes in custom harnesses. Message delivery costs us 10x more than inference (with meta price hikes of October). All long-horizon jobs such as coding are probably somewhat inference constrained, but only to the point that sometimes the devs and corporate users with their own agents wait a bit for something to be completed, but nobody is noticing it. Sure, one notices a difference between /fast mode and normal while building software, but nobody really has a reason to care much, /fast modes are a pure luxury thing outside of incident remediation. I don't see what would Anthropic actually do if they had 10x the compute or 100x the compute they are planning for. Surely, some new business models, trolling mathematicians at a larger scale maybe?
Surely this is Nvidia deploying excess capital by acquiring equity in high growth companies at significant discounts (factor in that a major portion of the investment comes back to them against their high margin products). This isn't a new business concept and is generally accepted as smart business? Not some breaking news conspiracy?
> (factor in that a major portion of the investment comes back to them against their high margin products)
One problem is that there isn't one number that indicates what percentage of Nvidia's equity investments come back to it in the form of chip purchases. In some cases, it appears it's 100%; in others, it has been suggested that every $1 invested produces >$3 in revenue. $1 invested for $1 in revenue versus $1 invested for $3 in revenue are really different outcomes.
There are also the questions of the revenue and revenue quality of the companies Nvidia has invested in, how much debt they've taken on, what sort of backstops/guarantees Nvidia has given them, etc.
> One problem is that there isn't one number that indicates what percentage of Nvidia's equity investments come back to it in the form of chip purchases.
And why couldn't a single number be calculated for this? They know their portfolio companies, their investment amounts, and the portfolio company purchases of their products.
> There are also the questions of the revenue and revenue quality of the companies Nvidia has invested in, how much debt they've taken on, what sort of backstops/guarantees Nvidia has given them, etc.
These are generally good things to look at for all corporate investments.
The main takeaway is that Nvidia investing in their own customers isn't inherently a bad thing. Investing in a poorly performing company is a bad thing.
> And why couldn't a single number be calculated for this? They know their portfolio companies, their investment amounts, and the portfolio company purchases of their products.
They probably could calculate it but they're not required to and, if we're being honest, they probably don't have an incentive to.
> These are generally good things to look at for all corporate investments.
They are, but the issue with Nvidia is that the web of customer relationships, equity investments, financing, backstops, etc. is very hard to untangle. Not every counterparty is publicly-traded, and even when they are, not every aspect of their business that would go into this sort of due diligence is necessarily required to be reported.
> The main takeaway is that Nvidia investing in their own customers isn't inherently a bad thing. Investing in a poorly performing company is a bad thing.
The question is whether or not Nvidia is even capable of adequately performing due diligence on all the customers it has invested in, financed or backstopped, or modeled the macro effects of its investments.
Because not every deal reported seems to have that math.
Some of the deals look really good on paper, and others look like $1 in equity investment produces $1 in revenue.
You also need to factor in whatever non-equity financing Nvidia extends to these buyers, how much debt they've raised from lenders and what backstops Nvidia has committed to.
Investing a bunch of money in a company that is heavily indebted, and committing yourself to buying significant amounts of any unsold compute might not be so smart, especially if that company has only one or two big customers.
So this is the real problem with Nvidia: it has all these deals, some of which might be totally sensible and others that might be WTF, and it's really hard for investors to figure out what the "portfolio" looks like overall.
They have been reinvesting all their inflated profits into more of these deals, at increasingly uneconomic valuations. That already makes it Ponzi-like.
It doesn't make recurring revenue, either. Not without fresh cash infusions into all the startups.
Largely means P/E is a useless metric here, investors don't get to see their short term net profits, and even if they do see it, those don't extrapolate to future years/decades.
We've got a lot of investors out there not thinking about this with a level head, they just see %%% growth and a "low" P/E.
Even if NVidia crashed and there ends up being an oversupply of GPUs or TSMC fab slots, this doesn't change that OpenAi and Google still need memory for their own hardware accelerators, so the price of boards wouldn't go down by much
Part of me wonders (maybe naively) if with falling cost of AI inference and improvement of models means there is an oversupply of inference infrastructure. since we may reach a point where no one needs more and everything you could want can be done by the machine and adding more compute for most users is not any kind of benefit (like it was before LLMs there was an obvious ceiling where the majority of people did not need more compute).
I know Jevons paradox is a thing and I could see it going the other way as well but the first possibility seems to be ignored and I am not sure we can. Eventually demand for training might dry up as well if are models become largely very good and only a bit of fine tuning is required for most applications.
The author is a developer working on software that scales to 1,024 cores and greater -- which is within spitting distance now thanks to 256-core AMD EPYC processors. If you count hyperthreads as a "virtual core" then a dual-socket system gets you there!
That blog made me realise that almost all software can't utilise 1,024 cores, not even close! Microsoft SQL Server, as a random example, tops out at 64 cores for any single query.
Similarly, people couldn't utilise a machine that big, not with pre-AI software, except for certain embarrassingly parallel codes such as simulations and 3D graphics rendering.
Now... NOW... there's no upper limit. None. One person(!) can cheerfully utilise millions of cores, billions even, and there's no end in sight.
If you think one conversation thread in "ChatGPT" is all we need, you're not seeing the big picture of what's possible here.
Those math proofs OpenAI produced were created with tens of thousands of agents running in parallel, orchestrated by thousands more agents, and overseen by hundreds of agents, all triggered by someone typing as little as a sentence or two to start that avalanche of compute!
With that kind of AI power at your fingertips, huge, unthinkable projects become a fun exercise for the weekend:
- Translate the entirety of Wikipedia to every human language.
- Rewrite the entire Linux kernel to Rust in one go
I am arguing that their is an quantity intercept (The demand for compute if it cost nothing would be a finite number and may go down rather then up with advancements in cheaper inference and accuracy).
Sure you can do all these things you mentioned but eventually even if it was free they would be done and doing some of these things your self would be annoying and a waist of time. Once some has formalized a proof why would some do it again and why would some one rewrite a program again, it will still take time and human effort over just using it. If you just spend 1 minute with each agent of your say 10000
I really don't think most people want to be running tens of thousands of agents and would not find a need for that many agents. if you it takes you just 1 minute of looking at output of your agents that would take you 1 year and 52 days. I think you have to admit there is a limit. Though I could find busy work for my agents, you could do the same with computing before hand such searching for primes, there is a limited thing that people want out of their machines is my point and people who would run such things would be a minority if they exist at all.
I don't see how it helps to think so narrowly about the computation needed solely for training when we are clearly headed to a future where tons of realtime applications (think robots) need to do insane amounts of processing ASAP while doing time-sensitive tasks.
IMO its copium to think that chips which have 10x more demand than supply are suddenly going to be easy to get as "AI is eating the world" like software did over the past decades.
You are definitely right, mrgordon - this isn't an infinite scaling, "one extra copy cost essentially nothing," software issue. Video cards are physical. You can't just sell more copies with almost zero marginal cost like software. Each video card is a physical good that is subjected to all those physical laws.
I was not trying to deny the corporeality of GPU's (or cpu and memory), only that there might be an oversupply given the rate of buy up of these things and the amount of start ups who have brought these things and will fail and liquidate or liquidate hardware as they pivot or go into just offering inference.
Secondly that the frontier model providers will not necessarily want to buy these up as the aforementioned lower demand for training (or costly and memory intensive training) and cost of inference goes down (and possibly requests as the models get better or at least plateau).
Further TPU's and if model weights do eventually stabilize then ASIC's might make more sense.
And lastly as models improve and make less mistakes the requirement trying things in parallel traditional compute may also go down. I think their is a quantity intercept (The demand for compute if it cost nothing would be a finite number and may go down rather then up with advancements in cheaper inference and accuracy) for compute both AI inference and traditional compute.
I think there is a question of how much of that demand is real. If GPUs are sitting unused in boxes and/or if NVDA has been artificially juicing demand either through basically buying them from themselves or selling them to China through an intermediary then the 10x demand comes into question.
That leaves the real buyers who are also starting to run low on capital.
Finally, demand doesn't need to go to zero for there to be a price correction. Just dipping from insane demand to elevated levels would lead to a significant price drop.
real time visual data processing for these applications seems to me to not be that expensive and the thinking part will be done as a swarm. There is some real quantity intercept for the demand of compute that does change over time but one can imagine it going down with better inference not to mention ASIC's.
I think there are two sides that want a particular outcome one want gpu's to get cheaper (like I admit I do), others want there investment to work out (both might want to huff copium at some point, who is huffing here is not easy to tell).
Same. I replaced the pads and paste on my now ancient RX 580 in the hopes of surviving to the next time prices are affordable.
For anyone wanting to do so - you can! You have to make sure you nail the pad thickness though. That can be tough because it can have changed between revisions of the same card. Thankfully, AI is pretty good at combing through past data to figure out what you need. Also it voids your warranty.
GPU scarcity is a really poor proxy for the health of the AI market. I don’t know if I should be surprised that someone so bullish about it doesn’t understand this.
Did you read the OP? Because Nvidia gave money to people who aren't you, so those people could give the money back to Nvidia for GPUs. In other words they gave away all of their GPUs in exchange for stakes in startups, because (1) they could not sell that quantity of GPUs at market prices ordinarily and (2) they were betting that VC hype and the perception of high demand generated from doing so would be worth more than selling GPUs normally.
But presumably those gpus are being actually used by those companies? Even if their prices are artificially inflated? Or are these buyers sitting on piles of unused gpus to make demand look good?
I mean even Satya Nadella famously said that Microsoft has plenty of GPUs just nowhere to put them. There are definitely plenty of unused GPUs out there. Jensen is also a big fan of scarcity (said in multiple interviews) so obviously he’d want to keep it up at all costs.
A different question might be, will these companies grow over the new few years?
If I give you 10 Million GPU in exchange for 25% of the company and the company goes bust or has negative growth (and thus is forced to sell assets), effectively I just put 10M GPUs on the market with a huge discount.
At a grand enough scale, this undermines the GPU marker and one could argue that Nvidia has been doing this on a massive scale.
> Or are these buyers sitting on piles of unused gpus?
Yes, they are literally doing exactly that. Noting that they aren't truly "buyers" so much as "gift recipients" (in exchange for ownership stake, but the value of their remaining ownership also goes up if these gifts generate sufficient hype for investors to invest more). GPU production is outstripping datacenter capacity.
> But presumably those gpus are being actually used by those companies?
But, even if they were all being used, this wouldn't mean much. The central point is: suppose Nvidia has produced 10 million[1] datacenter-class GPUs during the LLM bubble. Nobody can afford to, or is willing to, buy 10 million datacenter-class GPUs at their list price because the economics don't work out. But if they gift 10 million GPUs for stock, and frame it like they sold the GPUs (a framing that is effective -- although this has been an open secret, this is being reported like news by the WSJ, and to the majority of their readers, it is news), it makes Nvidia GPUs appear to be extremely desirable. Even though, logically, an item that you have to give away because it's not economically worth its market price is the exact opposite. The fact that you can use goods that you've effectively been given for free does not increase the actual economic value of those goods.
It's pertinent to note this scheme works only as long as the hype lasts. If Nvidia gives away millions of GPUs in exchange for stake in LLM startups, and the LLM startups go bust, Nvidia is then tens to hundreds of billions of dollars in the hole.
[1] (hypothetical number, it's vaguely around this order of magnitude)
Why would they sit on piles of unused GPUs when they could rent them out or use them? Even if they were free that is still not what a business would do with them.
I wonder if the economics of the two industries is similar enough to be useful here. Car demand itself seems fairly stable and repossession is a way of recovering value in cases of default. Credit scores of individuals purchasing cars being tied to salary/wages vs. startups without much history. I just wonder if the analogy isn't fraught with issues.
There's plenty. Electricity (and energy) is absurdly cheap in the modern world. 1GJ is roughly 10$. An H100 running 24/7 takes 20GJ/year, so that's 200$, which is orders of magnitude less than its retail price.
If energy wasn't cheap, you wouldn't see people driving large vanity SUVs around in the least economic way possible.
Just FYI, there are around 360 mil SUV's on the road today around the world. Let's say 1/3 are "large", so about 120 mil. Thus, out of population of 8.3 bil, that's 1.4% or apx 3% of households. :-)
There's plenty of electricity to run every GPU in the world, just maybe not enough electricity to run every GPU in the world and have poor people afford AC.
Right? This fever dream that there are boxes upon boxes of unused cards (some even claiming on this thread that it is 50% of GPUs sold) is utterly delusional. Some people want AI to go away so badly that they temporarily lose the ability to reason.
You're giving them a lot of leniency with that phrasing. Personally I get the impression many never do think in the first place, just imitate it by regurgitating what they've heard from others.
I try and give a lot of leniency because generally these people aren't stupid or meaning to be unpleasant, they are just scared. Which means sympathy, kindness and understanding is the likely cure. You can't reason someone out of a position they didn't reason themselves into.
With the recent craziness, perhaps not all of it is as far from the truth as we might think? There could be quite a gap between units "sold" and units actually made...
FWIW we absolutely know for sure that both AWS and Microsoft have racks upon racks of GPUs that are unused, if the cards are installed at all.
(AWS could have $20B of unused GPUs at this point — it's widely reported that there was at least $5B in August, with up to $3B more arriving each month)
So 50%? Who knows. Seems a particularly large figure. But if AWS and Microsoft can't plug them all in, why assume anyone else can? What if it's as much as 20%?
The clear indication is that cold/dark GPUs delivered to major customers, GPUs that Nvidia has sold but still has on its shelves (buy and hold), and GPUs that Nvidia has sold, leased back from customers and may or may not have plugged into their own cloud, together represent a large enough proportion of sales to have meaningfully pushed up the prices of GPUs and caused scarcity that is somewhat artificial.
If you take the scarcity of GPUs to be a solid proxy for assessing compute demand — or even for the sales of GPUs to be a solid proxy — then you are making a dangerous bet, IMO.
Exactly whether this is just misallocation or an absolute scandal of channel stuffing remains to be seen.
When a bubble pops, the pop typically takes years to happen. It's more of a slow deflation in most cases. Afterwards, you can easily spot the peak that would never be reached again but, during a "pop", it's not clear if you're past that peak or if you're just experiencing a momentary blip on the path to new highs.
Nvidia is likely not going to surge back to $240 soon. It might go lower actually. Skeletons are still falling out of closets. Huang may have been the tech founder's tech founder, but he kept some pretty nasty company in recent years and, as we're now finding out, got both himself and his company into shady territory. As it turns out, ethics are not just for aesthetics.
Is Nvidia slowing down for a year or two bad for AI as a technology? Probably not. Nvidia could find a leader that will return them to top form, or another company could take advantage of the situation to catch up and surge past. A lot of AI researchers have been calling for a pause or, at the very least, a slow-down. This might be just what was needed.
However, given Nvidia's utterly unique position as both hardware supplier and bank for the data centre build-out, what will the impact be on the rest of the AI sector? The market is going to have to price in a weaker and slower build-out as well as a loss of trust in Nvidia. That's clear downwards pressure. Is it a blip, or are we now past the peak?
It’s not the only indication that we are past the peak, I think. OpenAI’s failure to raise what they wanted, Anthropic’s likely junk rating, banks struggling to offload AI debt at fairly deep discounts, Oracle’s obvious problems.
As you say, if Nvidia is chastened enough to slow its banking endeavours and that lets air out of the bubble in a progressive rather than dramatic way, that’s probably better for everyone (except OpenAI, who are probably hosed in that scenario, as they have huge commitments but are struggling to raise what they need; Anthropic might last long enough to reassess)
I'll admit it, I am not very well educated in the ways of finance but I am a bit baffled how Anthropic can be valued at (reportedly) upwards of 1 trillion but also have a "junk" rating?
A lot of people want to buy their stock because of their fame, but the company’s underlying financials are junk (way too many commitments, way too little revenue)
The problem is they don’t have a unique enough product to sell at so high margin. If they were the only AI good enough to use, then I’d be ok. But given that cheap Deepseek is always catching up their stock is useless
Junk rating from a relatively high likelihood that they can’t pay the people they owe money to. If you buy their shares they are much more likely to go to zero than a more stable company in the near term. So you don’t want to buy anything Anthropic with money that you may need for anything.
High valuation because the public believe that if they do figure it out they could be worth hundreds of trillions. So a lot of people may want to buy it as a speculative bet of ungodly returns
Previous bubble villain Michael Milken and the 80s leveraged buyout bubble made it a name of common parlance because that bubble was built on "junk" bonds.
The old traders of that time who were against what Milken was doing called them "junk" and the name has stuck ever since.
It just means "risky". If you trade bonds it has no association with "bad". It is an inside bond market term, I think that is why it sticks too. It is a fun term if you trade bonds but terrible for communicating to the average person.
Junk means something specific here, it's not just a vague term, is it?
If you think it's terrible for communicating to the average person (and I am very much the average person) you could explain it yourself in detail.
So I will try, and I will accept constructive criticism:
You are right to imply that it is a sort of consensus rating threshold maintained by the ratings agencies that means their (Anthropic's) debts are not considered investment-grade — that they are speculative, high risk, that they are considered more likely not to be able to repay. It is ultimately the ratings' companies (and therefore some human) decisions as to whether they are above or below this threshold.
But for Anthropic it has concrete impacts, doesn't it? It means they will pay much more for their debt, they may not be able to negotiate long repayment terms, they may need to pay higher underwriters fees, they may be limited by the terms of any individual loan (how much future debt they are allowed to take on, whether they need that lender's approval to restructure their business in even small ways), and most importantly, some institutional investors (pension funds, mutuals, insurance companies, ETFs etc.) will be barred from investing (which drives why their debt will cost more).
All of this means it isn't just everyone thinks they are riskier. It means they will be operating in a more constrained framework.
Frankly I think it is baffling that Anthropic are still headed to an IPO when this is what it could look like on the other side.
Bond ratings are (supposed to be; see 2008 for failure case) based on actual, hard financial metrics. Stock valuations are that, plus fortune telling, plus feels, plus Jim Cramer hitting a big red button.
'junk rating' always means bonds, and it's the share price/market cap that's that high.
Very loosely a 'high' (price over earnings) share price is about growth and exciting prospects; high or well rated bond is about stability. (Not to suggest they move linearly in opposite directions though!)
They are different things, the valuation applies to their equity value (reported) but the junk rating is for its debt and how creditors asses the quality of the debt. If a company is burning tens of billions of dollars and is only "operationally" profitable, you would kind of have some doubts about their ability to repay. Those doubts can stand at odds with how the equity market sees the long-term potential of the company.
Would it not be extremely foolhardy for a new CEO to start slow rolling things considering the dominant position Nvidia currently finds itself in? Why would they do that? You say yourself it would clearly be a managed decline in that case. Why would an incoming CEO want that? Don't forget their pay would likely be performance related and probably not tied to decreasing the stock price
I’m pretty naive about this kind of thing, so bear with me, but if your stock is in free fall and you need a new CEO, isn’t there merit in a stabilizing force? No one is ever compensated just for, at least in the near term, stopping the bleeding?
Well the stock is high because shareholders have been predicting massive future growth. If a CEO is going to stabilize the price, that means they have to give investors confidence in the same or at at least similar growth trajectory. Which was aggressive.
Imagine you're about to get on a rocket ship, betting that this famous maverick rocket pilot can take you to Mars in 2 days. He already made it to the moon in 12 hours, record time. And he invented the ship, too. Then the pilot gets caught having an affair with the stewardess, gets fired before the trip, and they bring in a normal airline pilot who's never been on this ship before, never even flown a rocket ship, and who starts talking about good old fashioned normal speeds, and making it to Mars in a few years' time or so. Or, you know what, maybe just orbiting the Earth a few times instead.
You might be tempted to pass on the trip, no? And you wouldn't be the only one. Suddenly this rocket trip which had lots of passengers is completely empty, with no one to fund the trip. This so-called stabilizing force has killed the trip.
Nice analogy but there’s just one problem: it assumes arriving at mars is the end goal. But that’s not how companies seem to operate. They’re about indefinite growth. Which clearly isn’t possible and ends up killing a lot of companies as they eat themselves from the inside just to keep greedy shareholders happy.
There used to be a time when people would invest over the course of decades and enjoy the dividends they received. But these days it’s all just about the short term rewards.
Tragedy of the stock market: infinity can hit a company even if both management and shareholders are fully committed to modest long term stability (apparently this is absolutely not what happened at nvidia).
A company can get into the sights of "infinity believers" even if they try not to. Infinityists will honestly believe that the stock is a very good buy and whenever a "stability lover" has to sell, they will outbid the next stabilityist. Price goes up, it becomes more visible to infinityists and so on. Eventually infinityist will outnumber stabilityist and management will enter "or die trying" mode on promises they never made. Or be replaced.
If they discover tricky shenanigans, they might well announce them publicly to cover their own personal position. At the start, they're not going to have too much invested.
Yes, you publicly announce everything now so you can throw it all at the feet of the ousted leader. If not, enough time will pass where blaming the ousted leader looks like excuses for the current one and just looks bad to the public. Look at the current US administration as an example of how the constant blaming of the previous administration played to the public opinion.
Right. One of the weird side effects of the Jensen Huang report hierarchy is that instead of a handful of direct reports that have stories they need to CYA on, there are over sixty of them! This could make the congressional testimony aspect pretty intense.
That’s a good point. A new CEO never wants a slow unwinding, they go for quick pain to set the baseline of their bonus assessment into something workable.
But public perception has huge repercussions in business. If screwing Chinese models contributes to bad public perception, business will realign and oust CEOs who do.
Sure you can, if you are your own one-man business. Otherwise it's exploitation—you're stealing from your workers. This is not remotely controversial. (or profit sharing etc, which is so rare as to mostly not exist)
Stealing from your workers - have you ever been in business in your life? A workman is worthy of his hire - which means you pay your employees what they are worth and keep the excess. If they want to capture all of it they can start their own business.
How do you figure? There are certainly places where we manage to create value, but this has little to do with capitalism and more to do with hauling shit out of the earth that we didn't have before.
> When a bubble pops, the pop typically takes years to happen. It's more of a slow deflation in most cases.
That’s just a bear market. Bubbles pop, otherwise we wouldn’t call them bubbles. If you call every downturn a bubble it ceases to be meaningful, and we’d have to come up with a better term for actual bubbles.
Even if you know the timing of the collapse of a cataclysmic bubble what do you do?
All of the investable assets will collapse at the same time. Inflation will evaporate cash. Bonds will worth nothing (if the us gov bothers to repay them).
It is also so lucrative to participate in a bubble that if you focus too much on the problems you will miss out.
I think part of a bubble is knowing the problems but participating anyway. Over time then the problems start to feel normal and while things are good you rationalize why maybe the problems aren't such a big deal after all.
We did but people simply keep pretending it’s not happening.
Between the circular financing and behaviour some analysts call “channel-stuffing”, it does seem that with Jensen Huang gone, the music is perhaps about to stop.
No one seriously denies the actual activity was happening, they just dispute its implications, and for good reason. Vendors frequently extend financing with the hope of making more sales, the most obvious example being auto loans and BNPL
Why is Nvidia financing neoclouds to buy Nvidia chips to rent to AI customers? At least it was my understanding that run their own DCs for GeForce Now for example
I didn't know that, I thought it was just financing deals as I never looked closely into it. It looks like the ownership is lowish in both though (from what I can fine)
You might as well ask why they don't just keep all their GPUs and compete directly with Anthropic and OpenAI to make frontier models. The answer is the same, Nvidia's core competency is hardware and although they dabble up the stack they would rather stay focused on the GPU business. It also provides a degree of separation where they can bet on many different neoclouds and AI businesses, and some may fail but there will be survivors and winners.
Yeah obviously, but I just think of it in layers. Selling a product vs renting is a different business but closely related. Microsoft made their entire Cloud and O356 business by renting products they used to sell. Then to compete directly with Anthropic and OpenAI, it would need DCs to start. It can rent its own hardware back from cloud providers like Anthropic does or build DCs and rent to others and its own use.
It's basically the entire Anthropic IPO model too right? Start with the model, then get into productivity, design, and slowly other businesses and companies with your AI. I agree with another post here that Nvidia didn't seem particularly AGI pilled, but they are certainly DC pilled.
I just am unable to understand whether it's a bad thing or not, you could frame it as being circular, but you could also see it as Nvidia simply giving hardware in exchange for startup equity, is that really fundamentally problematic, is it the scale at which it was done, or is it a valid but risky tactic?
The scale is the largest factor, I think. If vendor financing is happening in a mature multi-vendor market on a small enough scale as to not constrain supply, it’s a less unhealthy situation than when there is only one vendor.
There’s a ton of circular financing in AI but I don’t see this as part of that. Circular financing is more like companies passing IOUs around. This is more like VC funding which is pretty reasonable imo.
It wasn't just startup equity. For example, they are backstopping agreements between customers and to-be-built data centers, so the companies building the data centers can get better loan terms from the bank.
All told there have over $500 Billion in equity deals and "future commitments" like above since July and even more after that.
Even for a $4 Trillion company, that's still quite a lot.
IMO Jensen's affairs might have been leaked because of him being out of control and making too many deals like this. The board might have wanted to fire him, but not say the real reason.
Does the math check out, though? For reference, Nvidia added $17B of non-marketable securities in 2025, and its total revenue was $215B. So, assuming all these equity investments went to GPU purchases, they only amount to 8% of its annual revenue.
Since these firms (neoclouds and AI startups) will get consolidated if Nvidia puts too much money into them, I assume the majority of their funds are coming from elsewhere.
According to Gemini, Nvidia added $43 billion in direct equity stakes,$36 billion in AI compute partnerships, and $100+ billion in loan lease guarantees, which works out to $63 billion this year and $20 billion for the next 3 years, and $10 billion for the last 6 years.
So about a third, a sixth, and a twelth of their annual revenue respectively.
I think the larger concern is that both the revenue and agreements depend on Nvidia maintaining sales, market share, and capacity TSMC/etc... If they run into trouble on any of those fronts, their revenue and assets could take a hit from multiple directions.
> Corporate development-led AI equity investments alone exceeded $40 billion in just the first four months of 2026
The accusation in the headline is that this activity contributed to the general shortage of GPUs which would, if true, increase their pricing power and profit margins even on chips sold to other customers. So I don't think it's a direct comparison to their equity deals.
Is this the only reason Nvidia made money? No, obviously. But is it adding fuel to the fire? If there is a bubble will this make the fallout worse? Seems likely.
More importantly this suggests a second plausible reason the board might have wanted to wrest control of the company away from him. (The other hypothetical one being that he was doing illegal stuff.)
Just like we knew for at least a year already about the sales through Singapore. A lot of people called it for what it is at the time already (see for example: https://www.ft.com/content/6b1ce190-24d4-4f39-a518-ef9dae5fa...).
And yet, things unrolled very slowly, then suddenly.
My theory is that once new CEO is found, especially if he comes from the outside, he will be highly incentivized to denounce past sins and disclose as much as possible to start with a clean sheet. I would bet that this is one of the main reasons the stock is tanking. House of cards is about to collapse, and that may be the trigger.
> My theory is that once new CEO is found, especially if he comes from the outside, he will be highly incentivized to denounce past sins and disclose as much as possible to start with a clean sheet.
Very much this. If you read the article it mentions that the current acting CEO was against investing into startups and disagreed with Jensen Huang.
What separates a company trying to “pump sales” through “circular financing” from a company with bucket loads of cash investing in startups who use their products?
How about companies who heavily discount their product to startups to support new uses for their products because it could create new demand?
I’d say “is Nvidia the only investor in this company?” If not then clearly other investors feel like the startup is an investment.
It’s a hell of an allegation to say it’s a pump and dump scheme since those same behaviors can just be smart business practice.
Why should there be a separation? If the startups using their product wouldn't have survived on their own or found other investors injecting as much money, that's a huge sign of circular economy.
At best, financing like that means any honest competitor that doesn't get on the honey train is at massive market disadvantage.
Kinda similar to when niche gets flooded by VC money. Now anyone with better product will get out-competed by people that can sell at cost or below it because they can ride the VC honey train till the competitors are extinct and they have captured the market
I mean this is their second forray into making a bubble. See bitcoin GPU price craze.
And yes Jensen lied to shareholders that GPUs aren't that tied to Bitcoin while they were selling trucks of GPUs to interested individuals.
As a result they never seriously pursued frontier research for open models - which could have easily absorbed any excess compute they planned for.
On the other side their customers are rapidly becoming their competitors and will happily eat up spare capacity at TSMC if nvidia can’t fulfill its obligations.
To be fair their Nemotron models are pretty good. The Parakeet speech-to-text model is also pretty good. That being said these pale in comparison to the frontier labs. They should've invested more into it, but I understand that they don't want to piss off their biggest competitors.
A friend of mine works on Nemotron. He stated quite plainly that they intentionally don't push the boundaries for the reason you state. It's similar to when companies create reference designs that show off the features of their real product, like Google and their Pixel phones
What would investing in the models give them ? They already are selling chips at massive markup.
Investing in tooling so more people get on their stuff is direct return, but having yet another AI model is of little value aside from being a test bed for their AI tooling
I wouldn't disagree with that. Astra was the first model that is actually really smart, doesn't constantly hallucinate, writes high quality code etc.
If it isn't AGI, it's at least something that shows that "AGI" is not a useful label. You wouldn't argue that we don't have Artificial General Flying because planes can't flap their wings.
A plane or a drone can do any aerial movement a bird can (in a shorter, less energy efficient manner). Can Astra do a software dev job alone without supervision? Let alone any other job.
Sorry, ramshanke - there really is a lot of promise to this tech. Will some of the current big payers collapse int he process? Definitely could happen. But there's no way around it: there's a lot of value in this new tech, and it's here to stay.
People don't have to be neutral. Neutrality between good and bad is mediocre. If you disagree with them make an argument, don't just lament the end of enlightened centrism.
Isn't that straightforwardly explained by typical hoarding behavior, not " afraid to let their competitors to make inference"? During the pandemic people were hoarding TP that they couldn't possibly use within a reasonable amount of time, but it'd be crazy to accuse people of doing that because they were afraid of their neighbors having clean asses.
>If you cannot wipe efficiently, you better make sure nobody else can too so that no-one can call you out on that.
But so far as we can tell, American proprietary models are more compute efficient than open models? They're also practically cheaper, if you factor subscriptions into account. Does that mean we should be decrying all the GPUs used for GLM, when they can be put to better use for claude or chatgpt models?
What scares me is that anybody could believe this garbage. It costs me $1 to make, I sell it for $5. Good deal! In fact, such a fabulous deal, I'm going to do this: Make 2 for $2, sell one for $5, then loan these other dudes the $3 I made in profit plus go $2 in the whole for a total of $5 so that they can give me the $5 back.
The more often you do it, and the greater the margin becomes due to squeeze, the dumber it is.
The first time around you sold a GPU for $5 and took payment as 5 startup shares at $1 apiece. Now you go back for round 2 and the startup says their shares are trading at $5. Will you still loan them the money and exchange 1 GPU for one share at $5?
You get less and less of their company each time and you've already peaked because future funding rounds can't keep being valued higher.
The trick only works once but everyone thinks it's a perpetual motion machine.
Many HNers didn’t know and are now speaking in hindsight. (They love doing that to appear smart) Otherwise they would have shown the receipts about it years ago.
Nvidia did a great job at engineering a central bank-style AI funding bubble for as long as possible. Unfortunately, Cisco did the exact same thing.
Before, there was denial about the circular financing bubble, now look at them believing that a crash is coming.
What happens to neoclouds when the bubble pops. nvidia stops circular financing, and all of a sudden they can’t pay their debt. This leads to the DCs getting bought cheaply by hyperscalers and you have a monopolistic AI market with google/fb/amazon and likes are running the show again.
I feel there's just enough value behind the crazy proposition that there won't really ever be a classic .com bubble pop. That's the thing: there actually is a lot of value in this tech. May some of the current big players crash and burn? Sure. But the tech is undeniably powerful - things will, and are, changing.
Sorry deaux, but I can't quite understand your response - you're using a "I disagree with you" format, but it sounds like you agree with me! Can you clarify a bit?
It's not about value, but return. Everyone likes to mention circular financing, which in of itself isn't an issue. If I have an amazing idea and NVDA wants to loan me money that's fine. My idea works out and we all end up billionaires.
The problem comes in if my idea doesn't end up generating a return. NVDA has not only sold me my GPUs, but it's also leveraged itself to my return. Repeat this across many businesses, none of which are generating returns yet. This is why when/if this unwinds it's going to be pretty catastrophic. As Charlie Munger said, "There are only three ways a smart person can go broke: liquor, ladies, and leverage".
40 year old gamer here having a flashback of all anti- competition practices Nvidia sustained for decades. Who here remembers GameWorks and PhysX effectively sabotaging multi-vendor optimization in titles like Witcher 3 and Watch Dogs? Expensive Gsync monitors? Trying to force OEMs to be GeForce partners in exchange for full GPU support? Forcing EVGA exit from the market? And then completely shafting gamers when crypto and AI markets opened up?
I always thought that Nvidia's presence in the Khronos Group was a farce. On one hand they were vendor-locking as much as possible at the expense of the industry, and on the other they were "contributing" to the open languages and protocols.
Dirty secret of "open" source: it's a strategic collaboration between big vendors. It's very rarely someone's hobby project. Almost all Linux changes come from hardware vendors.
>We should celebrate their fall, maybe there will be room for some real competition.
How are you going to get more competition when the entry cost is so high? The only competition will be the Chinese willing to dump endless cash into this but they are not doing it out of the goodness of their hearts either and it will still take years. If NVIDIA starts to falter then in the short term things will get even more painful for consumers as AMD becomes the new crummy company people love to hate.
What you need is for the interfaces people use to not be vendor-locked. At which point Nvidia doesn't have to "fail" in the sense of going out of business, but rather just have the ecosystem switch to APIs that are supported by not just them but also AMD, Intel, Apple, Qualcomm, etc.
At which point customers are buying for hardware price/performance or performance/watt rather than because they care which vendor they're getting it from.
And then you'd still have the biggest companies making the highest performance hardware, but you could also get something from the RISC-V people which had e.g. half or a quarter the performance of the enterprise GPUs but the same amount of VRAM, for less money.
This is the sort of self defeating mentality that creates evil monopolies. I used to read gamers berate AMD for their "driver problems" when their drivers simply couldn't emulate Nvidia's intentional tricks to force game developers to build for NV cards. And then they would moan when card prices skyrocketed. Don't know why you think AMD is "crummy" their greatest sin is that they didn't play these anti-competetive games and were screwed by other companies.
People seemed to be programmed to be sycophants or at least willing victims to rich and powerful people and companies, even as they royally screw them and empty their pockets. I never understood it. As the song goes: we'd make great pets.
The barriers to entry are not capital costs they are compatibility moats. Wake up and smell the coffee.
I also had the displeasure to work with nvidia. After years of pushing gaming cards on us for compute, the years I spent working with nvidia was full of conversations where they would list silly excuse after excuse as to why we should use the expensive enterprise-grade cards instead. All the while, we begged them to unlock driver features that were available to Quadro cards but not GeForce cards, features as simple as hardware acceleration pass-through on a remote desktop connection. It took them a full-on epidemic, where remote work became super common, to allow some of those features to GeForce cards.
So it's really on brand that they tried creating a fake scarcity to keep their prices up. I don't envy the situation Jensen is in but I won't be sorry for him.
Thanks for sharing the video. What's funny is that YouTube tagged a 10 year old video as AI-generated. Google is another company that's fallen hard from grace.
Don't Be Evil was never about not doing any evil. It was about the understanding that Google engineers were aware of their company's increasing power and influence, understood how their actions could cause great harm to society as a result, and were concerned enough to put their foot down and force management to codify the recognition in the corporate ethos.
Indeed it was marketing fluff, like any big corporation has for their newly employees about "the mission", with compulsory training, that everyone forgets after a couple of months.
That is correct, but in old google, sometimes employee pushback would get shitty and evil plans/products/policies rolled back. Not all of them, but enough of them. I've seen it first-hand.
If you are old enough to have seen Google under their founders (bunch of CS PhD program geeks) vs modern Google (which is almost entirely run by McKinsey MBA types), there is a ton of difference. Ask any older guy who still works there. The same thing happened to Boeing.
Well - the situation with his wife or being unable to control his Jenwong or whatever one would call it, is still a private matter (the allegation of the spy situation is a bit more complex, but the other affair he had going, or at the least one, losing track here of his personal love life), is at the end of the day not really something that is or should be in the spotline for others, even though it has already generated tons of meme videos. People are very happily putting memes about the ex CEO right now - he is probably one of the top #3 meme people right now. The other day I saw where he approached two female podcasters; it suddenly had such a creepy vibe, given his weakness for females ... (though that was not a meme as such, just some seemingly accidental Jensen sneakily creeping in situation ... can't stop him zooming in on the ladies, apparently)
But customers having to pay more due to NVIDIA milking them - now that's something that really would require mandatory jail time for all NVIDIA folks involved here, IF true and verified in court (right now we can only speculate, even if we have extremely convincing proof; the courts would have to come to such a conclusion at the end of the day; and for this, the court systems have to work. I have severe doubts the courts work in the USA right now. It seems a billionaires-controlled court. Why is the whole Epstein network not investigated? Why is only Ghislaine in jail? A single person master-minded +5000 sex slaves over some 20 years or so? How is that even logistically possible?).
> But customers having to pay more due to NVIDIA milking them - now that's something that really would require mandatory jail time for all NVIDIA
It’s the task of every capitalist business in existence to “milk” their customers for the maximum amount of money they can for any given product. Everyone in business would be in jail by your metric.
Well that sounds hyperbolic considering “capitalist business” runs the gamut between “trillion dollar company” and “guy down the street doing this in his spare time”
Not defending NVIDIA by any stretch, but I think Intel is the only one that allows ‘enterprise’ features on their consumer cards, like SR-IOV.
Such a complete failure of the market that NVIDIA hasn’t had any real competition. All of the vendors abandoned PC gamers though. Though the price choke point is memory, not cores.
Which makes me curious, if AMD did another HBM-based card like the Vega series, would that RAM be cheaper now than GDDR?
many have asked why AMD doesn't make its GPUs cheaper.
lisa su and jensen huang not competing with each other on price and being related is as much of a coincidence as wuhan having a virology institute experimenting on COVID-19 and being the center of the COVID-19 outbreak
I don't bring it up at all, ever, except here. Post histories are open. It's a jokey analogy. People are looking for exotic ways scarcity are made and it's usually just an agreement between two people made in a 15 minute phone call.
It’s been confirmed by agencies around the world including the rather conservative German BND [0].
It’s not a conspiracy theory - it’s fact that COVID-19 is of Chinese lab leak origin. The only open question is whether that happened out of malice or stupidity.
No. A bunch of agencies have started favoring the lab leak hypothesis with low confidence after recently reassessing the evidence. The BND has apparently favored it from the very beginning. Some others still back zoonotic spillover.
In any case, I'd wager no non-Chinese intelligence agency has the evidence to ascertain COVID-19 origin one way or the other.
Both the initial confidence in natural origin and now this new certainty in declaring the lab leak hypothesis a fact seem nonsensical to me.
> CIA now joins DOE and FBI in assessing the more likely cause of the first human COVID-19 infection was a research-related incident.
> CIA has low confidence in this judgment
> Given CIA's low confidence, credible new intelligence reporting or open-source information could prompt CIA to change its assessment.
> Four agencies and the National Intelligence Council (NIC) continue to assess that natural contact with an animal infected with SARS-CoV-2—or a closely related progenitor virus—was the most likely cause of the first human COVID-19 infection.
It's interesting that the BND has higher confidence in its assessment but not sufficient to support your statement that the lab leak hypothesis has "been confirmed by agencies around the world"
NVIDIA has been terrible to countless groups over the years(even more than you listed). They survived by being really good. You know who else have been jerks? Every other leading successful tech company (NVIDIA is nothing compared to Microsoft's misdeeds for example). Microsoft still going after 51 years. And now a new generation having never remembered the browser wars have fully embraced Microsoft products like VSCode. Your trip down memory lane reminded me about the numerous 'gamer boycotts' that all failed. There used to be a meme about how people signing a pledge boycotting a title would then be seen 'online' on Steam playing said game. At the end of the day most consumers didn't care. They wanted the best products and NVIDIA more often than not delivered. A company that is super polite to all their customers and partners is a company that will still be dropped the second their products stop delivering the value that customers expect.
Google used to be not jerks, Valve still are. HP - often considered the company that started Silicon Valley - were somewhat famously not jerks for a long time, but that ended decades ago when Hewlett and Packard left.
Valve aren't jerks, but they are stealthily inflating game prices by 30%. That massive fee is being paid by the consumer, even if the consumer doesn't realize it.
> Valve aren't jerks, but they are stealthily inflating game prices by 30%.
Nah.
From what I've heard from folks who used to do bricks and mortar distribution, that 30% is _way_ better than what you'd get cut out in that system. [0] From the perspective of the person publishing a video game, you also get waaaaaaaaaaay more for your cut with Valve than you'd get in a B&M distribution. _So_ much more. The same is true for the purchaser and player of video games.
[0] If for no other reason [1] than the fact Valve will sell your game forever, unless some government or powerful payment processor very strongly demands that it not be sold.
That's very ignorant perspective on it. 30% is very much standard cut in near any kind of online gaming stores and Steam's on average is lower than that.
Yes, they could drop the price for indies but they do have progressive tiering, at top you pay 20% to them and by revenue split it's much closer to taking 20% cut on average than 30%. But yes, indies are being a bit screwed there.
... if not for the fact every single cheaper platform will still get you massively less sales just because they can't compete on feature, let alone popularity of Steam.
Also the 30% funded ability to play games at all on my Linux box so I'm fine with paying the extra.
In the 1970s, installing the $500 power fail interrupt feature on the HP 2114 minicomputer was done by clipping a jumper.
Companies price based on perceived value and customer willingness to pay, not on cost to produce.
HP weren't alone. About the same time, modifying one of IBM's line printers to the more expensive, faster model was done by changing out a couple of gears.
As another almost 40 year old gamer, I find myself wondering why nobody else ever brings up 3dfx when discussing Nvidia. I've been salty with them since the buyout and I've remained an ATI/AMD user all this time because of it.
That is suggesting 3DFx does not allow support Glide for other graphics card. and Apple Lightning Made for iPhone programme is as bad Gsync Monitor certification. When they were the one to ensure quality matters across the whole spectrum. Just the same as Apple. And the reality it is just business.
Have they been pushy? Hell yes, I guess everyone have a different line on this. But on the scale of business partnership and leaving money for its ecosystem, Nvidia is far from the worst. Just look at Apple, or in other industry, Walmart. ( Note: Steve Jobs used to be pretty good at that, the same as Jensen. It was actually Tim Cook who pushed this to previously unseen level )
>And then completely shafting gamers when crypto and AI markets opened up?
I believe Jensen was the one who continue to push capacity in Geforce for consumer market when they could have done 100% AI. And there just isn't enough capacity for everyone. And had AI and Crypto not been that popular the R&D cost of amortised over GPU would have been far higher on leading edge node with Big GPU design, often pushed to Reticle limit. Not only have I recounted exactly zero appreciation on any technical forums, consumer simply want their GPU to be faster, cheaper and widely available every single time regardless of its cost, when it was Nvidia who is pushing the whole GPU industry forward.
> And there just isn't enough capacity for everyone.
You are in a comment thread related to an article about Huang intentionally engineering that shortage by investing into AI startups. Why do you need to defend a billionaire?
So the CEO 'laid into' the CFO but sounds like she didn't push back all that hard. Seems like something leaked by Kress to make her seem clean, but just makes her seem dirty and weak.
If this is truly a problem, it's her job to say no to the CEO. They need to clean house at the top and hire someone like Johny Srouji to right the ship.
522 comments
[ 29.8 ms ] story [ 256 ms ] threadI am not sure how inference constrained we are, btw. As an enterprise customer, I have zero problems with 429s and capacity from google or aws. All large volume inference jobs - so customer journey bots that help our company make money - run on dirt cheap models kept on short leashes in custom harnesses. Message delivery costs us 10x more than inference (with meta price hikes of October). All long-horizon jobs such as coding are probably somewhat inference constrained, but only to the point that sometimes the devs and corporate users with their own agents wait a bit for something to be completed, but nobody is noticing it. Sure, one notices a difference between /fast mode and normal while building software, but nobody really has a reason to care much, /fast modes are a pure luxury thing outside of incident remediation. I don't see what would Anthropic actually do if they had 10x the compute or 100x the compute they are planning for. Surely, some new business models, trolling mathematicians at a larger scale maybe?
Reducing circular financing agreements seems like ... a good idea?
> in response to which Huang allegedly lay into Kress, calling her a "deer in the headlights,"
One problem is that there isn't one number that indicates what percentage of Nvidia's equity investments come back to it in the form of chip purchases. In some cases, it appears it's 100%; in others, it has been suggested that every $1 invested produces >$3 in revenue. $1 invested for $1 in revenue versus $1 invested for $3 in revenue are really different outcomes.
There are also the questions of the revenue and revenue quality of the companies Nvidia has invested in, how much debt they've taken on, what sort of backstops/guarantees Nvidia has given them, etc.
And why couldn't a single number be calculated for this? They know their portfolio companies, their investment amounts, and the portfolio company purchases of their products.
> There are also the questions of the revenue and revenue quality of the companies Nvidia has invested in, how much debt they've taken on, what sort of backstops/guarantees Nvidia has given them, etc.
These are generally good things to look at for all corporate investments.
The main takeaway is that Nvidia investing in their own customers isn't inherently a bad thing. Investing in a poorly performing company is a bad thing.
They probably could calculate it but they're not required to and, if we're being honest, they probably don't have an incentive to.
> These are generally good things to look at for all corporate investments.
They are, but the issue with Nvidia is that the web of customer relationships, equity investments, financing, backstops, etc. is very hard to untangle. Not every counterparty is publicly-traded, and even when they are, not every aspect of their business that would go into this sort of due diligence is necessarily required to be reported.
> The main takeaway is that Nvidia investing in their own customers isn't inherently a bad thing. Investing in a poorly performing company is a bad thing.
The question is whether or not Nvidia is even capable of adequately performing due diligence on all the customers it has invested in, financed or backstopped, or modeled the macro effects of its investments.
What seems to be the problem?
Some of the deals look really good on paper, and others look like $1 in equity investment produces $1 in revenue.
You also need to factor in whatever non-equity financing Nvidia extends to these buyers, how much debt they've raised from lenders and what backstops Nvidia has committed to.
Investing a bunch of money in a company that is heavily indebted, and committing yourself to buying significant amounts of any unsold compute might not be so smart, especially if that company has only one or two big customers.
So this is the real problem with Nvidia: it has all these deals, some of which might be totally sensible and others that might be WTF, and it's really hard for investors to figure out what the "portfolio" looks like overall.
It doesn't make recurring revenue, either. Not without fresh cash infusions into all the startups.
Largely means P/E is a useless metric here, investors don't get to see their short term net profits, and even if they do see it, those don't extrapolate to future years/decades.
We've got a lot of investors out there not thinking about this with a level head, they just see %%% growth and a "low" P/E.
I know Jevons paradox is a thing and I could see it going the other way as well but the first possibility seems to be ignored and I am not sure we can. Eventually demand for training might dry up as well if are models become largely very good and only a bit of fine tuning is required for most applications.
The author is a developer working on software that scales to 1,024 cores and greater -- which is within spitting distance now thanks to 256-core AMD EPYC processors. If you count hyperthreads as a "virtual core" then a dual-socket system gets you there!
That blog made me realise that almost all software can't utilise 1,024 cores, not even close! Microsoft SQL Server, as a random example, tops out at 64 cores for any single query.
Similarly, people couldn't utilise a machine that big, not with pre-AI software, except for certain embarrassingly parallel codes such as simulations and 3D graphics rendering.
Now... NOW... there's no upper limit. None. One person(!) can cheerfully utilise millions of cores, billions even, and there's no end in sight.
If you think one conversation thread in "ChatGPT" is all we need, you're not seeing the big picture of what's possible here.
Those math proofs OpenAI produced were created with tens of thousands of agents running in parallel, orchestrated by thousands more agents, and overseen by hundreds of agents, all triggered by someone typing as little as a sentence or two to start that avalanche of compute!
With that kind of AI power at your fingertips, huge, unthinkable projects become a fun exercise for the weekend:
- Translate the entirety of Wikipedia to every human language.
- Rewrite the entire Linux kernel to Rust in one go
- Rewrite all open source C code to Rust.
- Formalise every maths paper ever written.
Etc, etc, ... without end, without bound.
Sure you can do all these things you mentioned but eventually even if it was free they would be done and doing some of these things your self would be annoying and a waist of time. Once some has formalized a proof why would some do it again and why would some one rewrite a program again, it will still take time and human effort over just using it. If you just spend 1 minute with each agent of your say 10000
I really don't think most people want to be running tens of thousands of agents and would not find a need for that many agents. if you it takes you just 1 minute of looking at output of your agents that would take you 1 year and 52 days. I think you have to admit there is a limit. Though I could find busy work for my agents, you could do the same with computing before hand such searching for primes, there is a limited thing that people want out of their machines is my point and people who would run such things would be a minority if they exist at all.
IMO its copium to think that chips which have 10x more demand than supply are suddenly going to be easy to get as "AI is eating the world" like software did over the past decades.
Secondly that the frontier model providers will not necessarily want to buy these up as the aforementioned lower demand for training (or costly and memory intensive training) and cost of inference goes down (and possibly requests as the models get better or at least plateau).
Further TPU's and if model weights do eventually stabilize then ASIC's might make more sense.
And lastly as models improve and make less mistakes the requirement trying things in parallel traditional compute may also go down. I think their is a quantity intercept (The demand for compute if it cost nothing would be a finite number and may go down rather then up with advancements in cheaper inference and accuracy) for compute both AI inference and traditional compute.
That leaves the real buyers who are also starting to run low on capital.
Finally, demand doesn't need to go to zero for there to be a price correction. Just dipping from insane demand to elevated levels would lead to a significant price drop.
I think there are two sides that want a particular outcome one want gpu's to get cheaper (like I admit I do), others want there investment to work out (both might want to huff copium at some point, who is huffing here is not easy to tell).
Similar to the fact that supply of consumer RAM is currently low.
That's nowhere in the realm of what I was talking about.
For anyone wanting to do so - you can! You have to make sure you nail the pad thickness though. That can be tough because it can have changed between revisions of the same card. Thankfully, AI is pretty good at combing through past data to figure out what you need. Also it voids your warranty.
https://www.theguardian.com/technology/2025/dec/28/nvidia-in...
A different question might be, will these companies grow over the new few years?
If I give you 10 Million GPU in exchange for 25% of the company and the company goes bust or has negative growth (and thus is forced to sell assets), effectively I just put 10M GPUs on the market with a huge discount.
At a grand enough scale, this undermines the GPU marker and one could argue that Nvidia has been doing this on a massive scale.
Yes, they are literally doing exactly that. Noting that they aren't truly "buyers" so much as "gift recipients" (in exchange for ownership stake, but the value of their remaining ownership also goes up if these gifts generate sufficient hype for investors to invest more). GPU production is outstripping datacenter capacity.
> But presumably those gpus are being actually used by those companies?
But, even if they were all being used, this wouldn't mean much. The central point is: suppose Nvidia has produced 10 million[1] datacenter-class GPUs during the LLM bubble. Nobody can afford to, or is willing to, buy 10 million datacenter-class GPUs at their list price because the economics don't work out. But if they gift 10 million GPUs for stock, and frame it like they sold the GPUs (a framing that is effective -- although this has been an open secret, this is being reported like news by the WSJ, and to the majority of their readers, it is news), it makes Nvidia GPUs appear to be extremely desirable. Even though, logically, an item that you have to give away because it's not economically worth its market price is the exact opposite. The fact that you can use goods that you've effectively been given for free does not increase the actual economic value of those goods.
It's pertinent to note this scheme works only as long as the hype lasts. If Nvidia gives away millions of GPUs in exchange for stake in LLM startups, and the LLM startups go bust, Nvidia is then tens to hundreds of billions of dollars in the hole.
[1] (hypothetical number, it's vaguely around this order of magnitude)
Nothing new, nobody pays a car cash anymore
there is plenty of electricity in china
If energy wasn't cheap, you wouldn't see people driving large vanity SUVs around in the least economic way possible.
It also requires major infrastructure to deliver that energy, which takes money and time to build, as do the datacenters.
Around 0.05-0.15% of the world population own such large SUVs.
While your datacenter is not built, guess where the AI chips are? That's correct. In boxes.
Are you sure about that? Might just as well be in Excel sheets only...
[1]https://www.techspot.com/news/114315-dark-gpus-amazon-show-l...
(AWS could have $20B of unused GPUs at this point — it's widely reported that there was at least $5B in August, with up to $3B more arriving each month)
https://www.businessinsider.com/amazon-nvidia-ai-gpu-data-ce...
We know that Satya Nadella has been open about how one of his big problems (as a CEO) is that he has bought GPUs he cannot plug in anywhere.
https://www.tomshardware.com/tech-industry/artificial-intell...
So 50%? Who knows. Seems a particularly large figure. But if AWS and Microsoft can't plug them all in, why assume anyone else can? What if it's as much as 20%?
The clear indication is that cold/dark GPUs delivered to major customers, GPUs that Nvidia has sold but still has on its shelves (buy and hold), and GPUs that Nvidia has sold, leased back from customers and may or may not have plugged into their own cloud, together represent a large enough proportion of sales to have meaningfully pushed up the prices of GPUs and caused scarcity that is somewhat artificial.
If you take the scarcity of GPUs to be a solid proxy for assessing compute demand — or even for the sales of GPUs to be a solid proxy — then you are making a dangerous bet, IMO.
Exactly whether this is just misallocation or an absolute scandal of channel stuffing remains to be seen.
Step 2: still have money even though the financial bubble popped
Step 3: just buy them normally, at market prices
Nvidia is likely not going to surge back to $240 soon. It might go lower actually. Skeletons are still falling out of closets. Huang may have been the tech founder's tech founder, but he kept some pretty nasty company in recent years and, as we're now finding out, got both himself and his company into shady territory. As it turns out, ethics are not just for aesthetics.
Is Nvidia slowing down for a year or two bad for AI as a technology? Probably not. Nvidia could find a leader that will return them to top form, or another company could take advantage of the situation to catch up and surge past. A lot of AI researchers have been calling for a pause or, at the very least, a slow-down. This might be just what was needed.
However, given Nvidia's utterly unique position as both hardware supplier and bank for the data centre build-out, what will the impact be on the rest of the AI sector? The market is going to have to price in a weaker and slower build-out as well as a loss of trust in Nvidia. That's clear downwards pressure. Is it a blip, or are we now past the peak?
As you say, if Nvidia is chastened enough to slow its banking endeavours and that lets air out of the bubble in a progressive rather than dramatic way, that’s probably better for everyone (except OpenAI, who are probably hosed in that scenario, as they have huge commitments but are struggling to raise what they need; Anthropic might last long enough to reassess)
Junk rating from a relatively high likelihood that they can’t pay the people they owe money to. If you buy their shares they are much more likely to go to zero than a more stable company in the near term. So you don’t want to buy anything Anthropic with money that you may need for anything.
High valuation because the public believe that if they do figure it out they could be worth hundreds of trillions. So a lot of people may want to buy it as a speculative bet of ungodly returns
Previous bubble villain Michael Milken and the 80s leveraged buyout bubble made it a name of common parlance because that bubble was built on "junk" bonds.
The old traders of that time who were against what Milken was doing called them "junk" and the name has stuck ever since.
It just means "risky". If you trade bonds it has no association with "bad". It is an inside bond market term, I think that is why it sticks too. It is a fun term if you trade bonds but terrible for communicating to the average person.
If you think it's terrible for communicating to the average person (and I am very much the average person) you could explain it yourself in detail.
So I will try, and I will accept constructive criticism:
You are right to imply that it is a sort of consensus rating threshold maintained by the ratings agencies that means their (Anthropic's) debts are not considered investment-grade — that they are speculative, high risk, that they are considered more likely not to be able to repay. It is ultimately the ratings' companies (and therefore some human) decisions as to whether they are above or below this threshold.
But for Anthropic it has concrete impacts, doesn't it? It means they will pay much more for their debt, they may not be able to negotiate long repayment terms, they may need to pay higher underwriters fees, they may be limited by the terms of any individual loan (how much future debt they are allowed to take on, whether they need that lender's approval to restructure their business in even small ways), and most importantly, some institutional investors (pension funds, mutuals, insurance companies, ETFs etc.) will be barred from investing (which drives why their debt will cost more).
All of this means it isn't just everyone thinks they are riskier. It means they will be operating in a more constrained framework.
Frankly I think it is baffling that Anthropic are still headed to an IPO when this is what it could look like on the other side.
Very loosely a 'high' (price over earnings) share price is about growth and exciting prospects; high or well rated bond is about stability. (Not to suggest they move linearly in opposite directions though!)
Imagine you're about to get on a rocket ship, betting that this famous maverick rocket pilot can take you to Mars in 2 days. He already made it to the moon in 12 hours, record time. And he invented the ship, too. Then the pilot gets caught having an affair with the stewardess, gets fired before the trip, and they bring in a normal airline pilot who's never been on this ship before, never even flown a rocket ship, and who starts talking about good old fashioned normal speeds, and making it to Mars in a few years' time or so. Or, you know what, maybe just orbiting the Earth a few times instead.
You might be tempted to pass on the trip, no? And you wouldn't be the only one. Suddenly this rocket trip which had lots of passengers is completely empty, with no one to fund the trip. This so-called stabilizing force has killed the trip.
There used to be a time when people would invest over the course of decades and enjoy the dividends they received. But these days it’s all just about the short term rewards.
A company can get into the sights of "infinity believers" even if they try not to. Infinityists will honestly believe that the stock is a very good buy and whenever a "stability lover" has to sell, they will outbid the next stabilityist. Price goes up, it becomes more visible to infinityists and so on. Eventually infinityist will outnumber stabilityist and management will enter "or die trying" mode on promises they never made. Or be replaced.
OpenAI not being able to IPO I think is the huge indicator that we are in denial of.
The tide is starting to go out and we are starting to see who is swimming naked.
Jensen and the Jensenverse as the villain of the bubble is the prestige of this bubble's magic trick. That is a show stopper.
OpenaAI as Myspace I think is the second act, yet to come.
Sorry what is this in reference to?
Anyway, ethics has no place in business. If it did, profit would be zero.
It's funny how the problem of Chinese models ended up affecting Nvidia and the US AI industry in general. Just not in the way it was predicted.
That’s just a bear market. Bubbles pop, otherwise we wouldn’t call them bubbles. If you call every downturn a bubble it ceases to be meaningful, and we’d have to come up with a better term for actual bubbles.
All of the investable assets will collapse at the same time. Inflation will evaporate cash. Bonds will worth nothing (if the us gov bothers to repay them).
I think part of a bubble is knowing the problems but participating anyway. Over time then the problems start to feel normal and while things are good you rationalize why maybe the problems aren't such a big deal after all.
Between the circular financing and behaviour some analysts call “channel-stuffing”, it does seem that with Jensen Huang gone, the music is perhaps about to stop.
It's basically the entire Anthropic IPO model too right? Start with the model, then get into productivity, design, and slowly other businesses and companies with your AI. I agree with another post here that Nvidia didn't seem particularly AGI pilled, but they are certainly DC pilled.
All told there have over $500 Billion in equity deals and "future commitments" like above since July and even more after that.
Even for a $4 Trillion company, that's still quite a lot.
IMO Jensen's affairs might have been leaked because of him being out of control and making too many deals like this. The board might have wanted to fire him, but not say the real reason.
Since these firms (neoclouds and AI startups) will get consolidated if Nvidia puts too much money into them, I assume the majority of their funds are coming from elsewhere.
So about a third, a sixth, and a twelth of their annual revenue respectively.
I think the larger concern is that both the revenue and agreements depend on Nvidia maintaining sales, market share, and capacity TSMC/etc... If they run into trouble on any of those fronts, their revenue and assets could take a hit from multiple directions.
> Corporate development-led AI equity investments alone exceeded $40 billion in just the first four months of 2026
The accusation in the headline is that this activity contributed to the general shortage of GPUs which would, if true, increase their pricing power and profit margins even on chips sold to other customers. So I don't think it's a direct comparison to their equity deals.
Is this the only reason Nvidia made money? No, obviously. But is it adding fuel to the fire? If there is a bubble will this make the fallout worse? Seems likely.
More importantly this suggests a second plausible reason the board might have wanted to wrest control of the company away from him. (The other hypothetical one being that he was doing illegal stuff.)
My theory is that once new CEO is found, especially if he comes from the outside, he will be highly incentivized to denounce past sins and disclose as much as possible to start with a clean sheet. I would bet that this is one of the main reasons the stock is tanking. House of cards is about to collapse, and that may be the trigger.
Very much this. If you read the article it mentions that the current acting CEO was against investing into startups and disagreed with Jensen Huang.
How about companies who heavily discount their product to startups to support new uses for their products because it could create new demand?
I’d say “is Nvidia the only investor in this company?” If not then clearly other investors feel like the startup is an investment.
It’s a hell of an allegation to say it’s a pump and dump scheme since those same behaviors can just be smart business practice.
At best, financing like that means any honest competitor that doesn't get on the honey train is at massive market disadvantage.
Kinda similar to when niche gets flooded by VC money. Now anyone with better product will get out-competed by people that can sell at cost or below it because they can ride the VC honey train till the competitors are extinct and they have captured the market
I imagine the shortsellers and put option holders like that.
As a result they never seriously pursued frontier research for open models - which could have easily absorbed any excess compute they planned for.
On the other side their customers are rapidly becoming their competitors and will happily eat up spare capacity at TSMC if nvidia can’t fulfill its obligations.
:D
Investing in tooling so more people get on their stuff is direct return, but having yet another AI model is of little value aside from being a test bed for their AI tooling
https://fortune.com/2026/09/09/markets-agi-nvidia-singularit...
If it isn't AGI, it's at least something that shows that "AGI" is not a useful label. You wouldn't argue that we don't have Artificial General Flying because planes can't flap their wings.
Have you ever seen a bird? Lol
In any case that's kind of my point. The existence of small niches that bird are good at that planes can't do doesn't mean that we can't fly.
The existence of small niches like writing prompts and counting letters in strawberry that AI doesn't do doesn't mean that our jobs are safe.
People think "we don't have AGI yet because AI still can't do X, so we're safe from Y" without realising that Y doesn't really require X.
Nope, that was Fable.
Here’s the original article: https://www.wsj.com/tech/jensen-huang-colette-kress-nvidia-c...?
Seems like a great way to cash out at the top
I can likely reveal a whole lot of your biases and upend this phony notion we can't falsify over screen that you're pure in your neutrality
Trying to deny your real biological existence and inherent character flaws that come with is vanity
A neutral person would not have an opinion on negative or positive tone.
See Futurama for a comedic dramatization of the logic of the neutral character
If you cannot wipe efficiently, you better make sure nobody else can too so that no-one can call you out on that.
But so far as we can tell, American proprietary models are more compute efficient than open models? They're also practically cheaper, if you factor subscriptions into account. Does that mean we should be decrying all the GPUs used for GLM, when they can be put to better use for claude or chatgpt models?
That wasn't my personal opinion, just guessing the mindset behind those who want to prevent competition.
Funny enough that seems to be the exact stance of OpenAI/sama.
https://www.tomshardware.com/tech-industry/data-centers/open...
Clouds are offering previous gen GPU capacity just fine
The more often you do it, and the greater the margin becomes due to squeeze, the dumber it is.
You get less and less of their company each time and you've already peaked because future funding rounds can't keep being valued higher.
The trick only works once but everyone thinks it's a perpetual motion machine.
Nvidia did a great job at engineering a central bank-style AI funding bubble for as long as possible. Unfortunately, Cisco did the exact same thing.
Before, there was denial about the circular financing bubble, now look at them believing that a crash is coming.
Unlike that weird technology "the web", the basis of the .com bubble. Turned out to have no value in it!
So all those vendor financed startup may die, taking the VC funds with them, but that is irrelevant to whether the tech stays around or not.
You then followed up by pointing out a way in which it's exactly the same as the .com bubble. One might even say it's their biggest common ground.
The problem comes in if my idea doesn't end up generating a return. NVDA has not only sold me my GPUs, but it's also leveraged itself to my return. Repeat this across many businesses, none of which are generating returns yet. This is why when/if this unwinds it's going to be pretty catastrophic. As Charlie Munger said, "There are only three ways a smart person can go broke: liquor, ladies, and leverage".
They're just another rank monopolist. We should celebrate their fall, maybe there will be room for some real competition.
How are you going to get more competition when the entry cost is so high? The only competition will be the Chinese willing to dump endless cash into this but they are not doing it out of the goodness of their hearts either and it will still take years. If NVIDIA starts to falter then in the short term things will get even more painful for consumers as AMD becomes the new crummy company people love to hate.
At which point customers are buying for hardware price/performance or performance/watt rather than because they care which vendor they're getting it from.
And then you'd still have the biggest companies making the highest performance hardware, but you could also get something from the RISC-V people which had e.g. half or a quarter the performance of the enterprise GPUs but the same amount of VRAM, for less money.
People seemed to be programmed to be sycophants or at least willing victims to rich and powerful people and companies, even as they royally screw them and empty their pockets. I never understood it. As the song goes: we'd make great pets.
The barriers to entry are not capital costs they are compatibility moats. Wake up and smell the coffee.
I also had the displeasure to work with nvidia. After years of pushing gaming cards on us for compute, the years I spent working with nvidia was full of conversations where they would list silly excuse after excuse as to why we should use the expensive enterprise-grade cards instead. All the while, we begged them to unlock driver features that were available to Quadro cards but not GeForce cards, features as simple as hardware acceleration pass-through on a remote desktop connection. It took them a full-on epidemic, where remote work became super common, to allow some of those features to GeForce cards.
So it's really on brand that they tried creating a fake scarcity to keep their prices up. I don't envy the situation Jensen is in but I won't be sorry for him.
That's no longer the case.
Connected. Complete lack of integrity.
If he cheats on his wife, and on his customers, he'll probably cheat on his shareholders. Hence having to quit before he's pushed.
But customers having to pay more due to NVIDIA milking them - now that's something that really would require mandatory jail time for all NVIDIA folks involved here, IF true and verified in court (right now we can only speculate, even if we have extremely convincing proof; the courts would have to come to such a conclusion at the end of the day; and for this, the court systems have to work. I have severe doubts the courts work in the USA right now. It seems a billionaires-controlled court. Why is the whole Epstein network not investigated? Why is only Ghislaine in jail? A single person master-minded +5000 sex slaves over some 20 years or so? How is that even logistically possible?).
It’s the task of every capitalist business in existence to “milk” their customers for the maximum amount of money they can for any given product. Everyone in business would be in jail by your metric.
Such a complete failure of the market that NVIDIA hasn’t had any real competition. All of the vendors abandoned PC gamers though. Though the price choke point is memory, not cores.
Which makes me curious, if AMD did another HBM-based card like the Vega series, would that RAM be cheaper now than GDDR?
They do not. SR-IOV is only available on Arc Pro cards.
https://www.intel.com/content/www/us/en/support/articles/000...
https://www.phoronix.com/news/Intel-SR-IOV-Only-For-Arc-Pro
many have asked why AMD doesn't make its GPUs cheaper.
lisa su and jensen huang not competing with each other on price and being related is as much of a coincidence as wuhan having a virology institute experimenting on COVID-19 and being the center of the COVID-19 outbreak
Wasn’t this debunked years ago?
[0]: https://www.dw.com/en/covid-pandemic-likely-unleashed-by-lab...
In any case, I'd wager no non-Chinese intelligence agency has the evidence to ascertain COVID-19 origin one way or the other. Both the initial confidence in natural origin and now this new certainty in declaring the lab leak hypothesis a fact seem nonsensical to me.
https://usrtk.org/wp-content/uploads/2026/11/ODNI-3-November...
> CIA now joins DOE and FBI in assessing the more likely cause of the first human COVID-19 infection was a research-related incident.
> CIA has low confidence in this judgment
> Given CIA's low confidence, credible new intelligence reporting or open-source information could prompt CIA to change its assessment.
> Four agencies and the National Intelligence Council (NIC) continue to assess that natural contact with an animal infected with SARS-CoV-2—or a closely related progenitor virus—was the most likely cause of the first human COVID-19 infection.
It's interesting that the BND has higher confidence in its assessment but not sufficient to support your statement that the lab leak hypothesis has "been confirmed by agencies around the world"
Weirdly, Youtube has labelled that video as "AI".
Which is kinda hard to imagine, seeing as it was released 10 years ago.
Nah.
From what I've heard from folks who used to do bricks and mortar distribution, that 30% is _way_ better than what you'd get cut out in that system. [0] From the perspective of the person publishing a video game, you also get waaaaaaaaaaay more for your cut with Valve than you'd get in a B&M distribution. _So_ much more. The same is true for the purchaser and player of video games.
[0] If for no other reason [1] than the fact Valve will sell your game forever, unless some government or powerful payment processor very strongly demands that it not be sold.
[1] ...and there are many other reasons...
Yes, they could drop the price for indies but they do have progressive tiering, at top you pay 20% to them and by revenue split it's much closer to taking 20% cut on average than 30%. But yes, indies are being a bit screwed there.
... if not for the fact every single cheaper platform will still get you massively less sales just because they can't compete on feature, let alone popularity of Steam.
Also the 30% funded ability to play games at all on my Linux box so I'm fine with paying the extra.
Companies price based on perceived value and customer willingness to pay, not on cost to produce.
HP weren't alone. About the same time, modifying one of IBM's line printers to the more expensive, faster model was done by changing out a couple of gears.
Have they been pushy? Hell yes, I guess everyone have a different line on this. But on the scale of business partnership and leaving money for its ecosystem, Nvidia is far from the worst. Just look at Apple, or in other industry, Walmart. ( Note: Steve Jobs used to be pretty good at that, the same as Jensen. It was actually Tim Cook who pushed this to previously unseen level )
>And then completely shafting gamers when crypto and AI markets opened up?
I believe Jensen was the one who continue to push capacity in Geforce for consumer market when they could have done 100% AI. And there just isn't enough capacity for everyone. And had AI and Crypto not been that popular the R&D cost of amortised over GPU would have been far higher on leading edge node with Big GPU design, often pushed to Reticle limit. Not only have I recounted exactly zero appreciation on any technical forums, consumer simply want their GPU to be faster, cheaper and widely available every single time regardless of its cost, when it was Nvidia who is pushing the whole GPU industry forward.
You are in a comment thread related to an article about Huang intentionally engineering that shortage by investing into AI startups. Why do you need to defend a billionaire?
And if they didn't wanted dual use "problem" they could've chilled on AI cores in their chips and went all in on 3D performance.
Nvidia is a business but them having a (near) monopoly has absolutely been horrible for gamers.
If this is truly a problem, it's her job to say no to the CEO. They need to clean house at the top and hire someone like Johny Srouji to right the ship.
What is 'curious' is how everyone on this doesn't recognize that these programs are extremely common.
Every large company has some kind of vc, bd or developer program for this kind of thing.
Free Google credits are similar - they are 'giving companies money to buy google services'.
What do you think that is?
It's not cash, but it's economically equivalent if not financially.
The service is consumed, companies get 'value' they have to declare, Google gets the write off.
An engine maker makes money selling engines for energy production.
Then someone invents cars.
The engine maker sees a 10x bigger market, but they have cash and the car inventor does not.
The engine maker funds the car inventor, to create a big distribution channel.
Some things are larger in proportion.
There's nothing wrong with this as long as investors are aware of the structure.