Note that the Fed has a $6.7tn balance sheet [1]. (This is a silly comparison. But still fun.)
The real comparison: Nvidia's $500+ billion of investments and commitments [2] is substantially more than any easing the Fed has done in the same time [3]. Monetarily, Nvidia is creating a lot of money in our economy.
The good news: I have seen no evidence Nvidia has borrowed against its stock or otherwise linked its equity value to these commitments. Its stock could crash without causing–as long as its cash flows continue–a credit crisis through its investments and commitments.
“as long as it’s cash flow
continues” is doing a lot of optimistic heavy lifting. The whole premise of the circular financing worry is that Nvidia sits in the middle of all the guarantees made to companies like OpenAI. If any of those companies become insolvent, Nvidia is on the hook for it.
Also Nvidia isn’t really creating money. The 500B number is third party capital that already exists (BX, Apollo, etc).
Compute has already lost value for me. Six months ago I thought you needed a 1T+ model to be useful coding. Now I am able to get by just fine with a 27b model.
I see two factors converging to cause a collapse of this house of cards:
1. People are realizing that what they need isn't more general intelligence, it's more specialization. A small but well tuned coding model, a small but well tuned customer service model, a small but well tuned document explorer.
2. Specialized hardware - TPUs and NPUs - especially coming out of china. The latest GLM model was trained and runs on Huawei hardware. Nvidia is only worth so much because they are the biggest and best provider of the kind of compute needed to run llms, but the export bans mean china has a lot of incentive to topple that monopoly.
The amount of compute we need to do the things llms do is falling rapidly, the number of people who can provide that compute is rising.
> People are realizing that what they need isn't more general intelligence, it's more specialization. A small but well tuned coding model...
It’s not quite as simple as that. Several studies have shown the opposite: models trained on more diverse knowledge tend to cross-pollinate across domains. So a more generalized model can actually perform better than a specialized one.
That’s why you’re not seeing tons of tiny models (one for Python, one for Pascal, one for Rust, etc).
This is definitely the position of the big ai companies.
But it doesn't match my experience. Qwen3.8 27b is clearly smarter at coding than MANY bigger models. gpt-oss-120b for example, is almost 4x the size, and performs way worse at coding tasks.
It's clear to me that you can build small models that work well at specific tasks.
Python vs Rust is probably too fine grained a way to build a model. Coding in general seems like a better target.
There will always be a place for large generalist models, no doubt. But I think that place is much smaller than the big ai companies are counting on.
I make heavy use of smaller local models on a daily basis (Qwen3-VL for auto-captioning images, Gemma3:27b for some translation work, etc.). Gemma3:27b is a good example of a very capable general purpose multimodal model and has handled almost everything I've thrown at it from sentiment analysis to documentation writing.
I suppose I was drawing a distinction between specialized and general intelligence versus small and large. I don’t think those are necessarily mutually exclusive.
Problem is conflict of interest: the studies are mostly from the providers of the biggest models, or someone who received free tokens to do the research.
I think what will keep the industry afloat, all else failing, is the surveillance industry! Nothing like a fat reoccurring cheque from the government to check if little Jimmy is committing thought crime!
Sure, but it would be actively bad to make the code more complex simply because we have machinery that helps us deal with the complexity. A big part of how people assess the models' coding capability is whether they create needless, incidental complexity.
OpenAI dropped sora because it was costing them ridiculous amounts of money and earning them very little. They determined that the market can't support the cost of generating video.
Without a material change in the market (more buyers, vastly cheaper generation), it's unlikely a different company could make that work. More buyers isn't likely to happen, so that leaves vastly cheaper generation - something that would cause nvidia's value to collapse if it happened.
oAI isn't anywhere near close to fucked as long as their models are head and shoulders above even the very best open models in terms of tool calling and rock solid stability/reliability for agents/coding harnesses. Which, they are right now and we'll see if open models actually catch up in that regard. Even the "best" open models pale in comparison with tool calling and general "prompt and go do something else for an hour" reliability that we have with GPT models. With GPT models, streaming rarely stops unexpectedly. You almost never have to constantly nudge them along, etc. Granted with open models all of this can vary depending on the provider, and perhaps open models/protocols/APIs/harnesses aren't well enough aligned, but OpenAI models just seem to work without constant (or hardly any) wrinkles and with almost any harness/agent.
>oAI isn't anywhere near close to fucked as long as their models are head and shoulders above even the very best open models in terms of tool calling and rock solid stability/reliability for agents/coding harnesses
That's already not the case today. If you sat me in front of an LLM and told me to figure out if I'm working with K3 or Astra, I could probably do it, but it would take some work to be certain.
If you reshuffle your argument, and apply the same facts you get to a similar conclusion but with a drastically different spin.
> it's more specialization
China, constrained by hardware, and talent (not to slight the Chinese, but they are limited to domestic resources - and much of the US effort is very international). They did, what the Chinese do, and optimized the process of production, and drastically lowered the cost of development of their models. Cheeper to build, cheaper to run is just good economics.
Meanwhile in the us, we have open AI doing "experiments" - it looks like the costs around the hugging face hack are going to be about the same as China would spend on building out one of their smaller efforts (several million dollars). (Depending on whos numbers you trust, the fact that I can even make this claim should make you raise an eyebrow).
Go back to the 80s' and "expert systems" - most people will tell you that for their time, they were amazing, and useful. People would have loved to have more of them but they were so cost prohibitive that we all but abandoned them for serious use. The US frontier labs seem to have forgotten this lesson and their calls to "slow down" look like an excuse to "cut the waste so we can move to making money".
The problem is that if OpenAI doesn't want the compute nobody does. All of these companies' demand for compute are correlated. It isn't likely that OpenAI will want less compute in isolation. Furthermore, the circular financing structure means that if OpenAI buys less chips it means that Nvidia has less money to give to say Anthropic to buy more chips and suddenly the exponential growth that circular financing has allowed to grow runs in reverse.
> if OpenAI can’t use the compute, someone else can.
The problem with that is that OpenAI can only afford to pay for the compute because they are burning investor money (and so are most of OpenAI's biggest clients). They are losing billions. If they stop burning money, nobody else will be there to pay for that compute at OpenAI's cost.
Sure, somebody will probably be able to use these GPUs, they just won't be able to pay nearly as much for them as OpenAI does.
In reality, it's just nowhere near worth as much as OpenAI pays for it. Inflating the cost of compute is part of the problem caused by the circular financing, and if (or maybe when) OpenAI goes, the price of compute will go with them.
Devil's advocate: OpenAI not being able to use compute is highly correlated to many other AI companies not being able to find a meaningful use of this compute.
Failure to take into consideration those kind of correlations ("If my biggest client isn't able to buy it, I would be able to find someone else who will") is one of the principle causes why many risk models turned out to be garbage during the Great Financial Crisis.
But I also doubt Nvidia is on the hook if OpenAI just no longer wants the compute. I bet they are only on the hook if OpenAI cannot pay for it (is insolvent in some way).
I also have to bring up that OpenAI has already spat out an inference chip that beats Nvidia on flops per watt. So they could potentially not need the compute while other ai companies do.
> if OpenAI can’t use the compute, someone else can
This is the big point IMO since I have never given $1 to OpenAI but I subscribe to Vidu and Typecast, and have given money to Kling, Hailou, and even Gemini in the form of Google Workspace.
So these other guys have products and use cases, which OpenAI has never been able to crack beyond ChatGPT. And ChatGPT was never worth paying for, IMO.
If OpenAI dies, it's not because there is no market for the technology (which is all NVIDIA cares about), it's more that OpenAI doesn't know how to run a relevant technology company.
They were given everything, not just NVIDIA's billions of dollars and credit backing but all the first-mover advantage, all the respect and credibility early on, so it's really sad to see them unable to develop interesting products and turn a profit in a space they helped pioneer, while so many others are making money with the tech all around them.
NVIDIA is fine. The technology will continue to improve and NVIDIA will stay at the center. OpenAI is fucked - knew it when they retired Sora to focus on text-to-text and coding (a largely solved problem).
Do you use coding agents? Just curious bc from my experience using coding agents, open ai’s codex is neck and neck with anthropic’s claude code if not ahead. I wouldn’t agree that OpenAI hasn’t done anything since ChatGPT since codex is my daily driver for software engineering
It's kinda nuts to me how people can act like Claude is lightyears ahead of OpenAI models. Sure, it's one thing to simply have a preference or claim that Claude does some things better, but in reality they are both about as effective at doing the same job. I've long preferred GPT models because they know better how to shut up and don't seem to overthink as much as Claude, but I'm under no illusions that if OpenAI went belly-up then I couldn't do my job essentially the same with Claude. GPT models have also clearly improved over time in terms of programming. There haven't been any "big bangs" necessarily, but it's really not hard to give the same task to 5.3 and 6 and see which one has better output.
All the "frontier" AI companies *are* currently insolvent. They have never been anything other than cash burning machines.
The only way they keep the lights on and the doors open is by borrowing money --- and epic amounts of it. If those operating the cash spigot decide to turn it off, all AI companies will likely be similarly affected --- and so will Nvidia.
OpenAI expects to burn through more cash between 2024 and 2029 than Uber, Tesla, Amazon and Spotify did - combined - before those companies started making money
To make things worse, hardware prices have spiked, due to AI companies.
Fairly sure data center construction costs are also going up (they require so many resources that everything is constrained at the moment, especially electricity production).
So I don't understand in what world these frontier AI companies can somehow become profitable. The basic tech they're using is basically the same. Yes, around the edges there are a lot of things that can be done, and were done, like caching, batching, mixture of experts, etc, but basically everyone has done all of that by now, and they're still losing money.
So:
Total costs going up a lot - revenues per unit not increasing proportionally, if anything, Chinese models are forcing those down.
How does that math work out to profits? I don't see it.
Or about as bad, after trillions of dollars in investments over multiple years, let's say the entire frontier AI sector has a total profit of $20bn by 2030. In what world does that make sense? Assuming they can scale that total profit to $100bn in 2035 without investing another cent from 2027 to 2035 (utterly ridiculous), the return on investment would happen in roughly 20 years.
> “as long as it’s cash flow continues” is doing a lot of optimistic heavy lifting
It's not. It's stating a condition. For traditional banks, a stock crash can independently trigger a failure.
> whole premise of the circular financing worry is that Nvidia sits in the middle of all the guarantees made to companies like OpenAI. If any of those companies become insolvent, Nvidia is on the hook for it
Sorry, I meant revenues. If Nvidia's revenues stay stable, these commitments aren't a problem. Even if the stock price crashes.
> Nvidia isn’t really creating money
It absolutely is. Similar to the way banks create money [1]. The commitments support credit that wouldn't exist without it.
> The good news: I have seen no evidence Nvidia has borrowed against its stock or otherwise linked its equity value to these commitments.
Nvidia doesn't need it. It funds other companies. They do this thing that Nvidia doesn't do. It shows up on their balance sheets and Nvidia just gets to claim the valuation of the investment on its balance sheet.
> how much did the fed sign up to backstop then? Surely it’s more than physical currency
The Fed doesn't backstop physical currency. That is issued by the Treasury (specifically, the Mint). The FDIC backstops bank deposits; the U.S. guarantees is obligations.
The Fed doesn't properly "backstop" anything. It's the lender of last resort–if you have a Treasury or other good collateral, it will loan you money against it. It's a financial regulator. And it regulates interest rates (i.e. the price of money) to influence inflation and employment.
The only backstops the Fed truly makes are to banks, by guaranteeing to always stand ready to lend against Treasuries and other good collateral.
That isn't a backstop, it's a direct obligation. It's also, like, not a real one? You can't redeem notes for specie. The term originates from when you could redeem dollars for metal. The Fed did play a role in backstopping that guarantee.
> know they don’t backstop jack squat but in practice there is a fed put
Sure. That isn't the same as a backstop. Backstops are hard–the Fed can't turn away an eligible borrower at the discount window. The Fed put is soft–the Fed will let market participants fail to send a message.
Going back to the top, it is incorrect to say the Fed backstops M2. This wouldn't be a footnote in a central-banking discussion, it would be something that would get called out as a screwup.
> How about what’s the amount from banks that the fed would willing lend as a last resort?
Infinity. The Fed mainly accepts Treasuries as collateral, but it can and has expanded the definition of good collateral in crises. There is no legal or frankly practical limit on how much money the Fed can create. Its only constraint is ultimately political. (Which is in turn mostly a function of inflation and employment and I guess now social media vibes.)
If you're looping this back to Nvidia, yes, I never claimed Nvidia has more lending capacity than the Fed. What I said was it's interesting that in practice, Nvidia appears to have created more money (if we're being pedantic, M3 which turns into M1) than the Fed has in that time. The Fed wasn't particularly trying to ease financial conditions in that time, so this is more of a curiosity tied to the title than a statement of capability.
The Fed can never default on any dollar-denominated debt. There is no similar currency that Nvidia can create ad infinitum.
That said, the number I think you're looking for in respect of the Fed is $30 to 40 trillion. It's about U.S. GDP. And it's also about U.S. bank and money-market assets plus the Fed's balance sheet. If every American bank failed, this could be the amount of money the Fed would have to create.
That said, Treasury running out and e.g. defending the euro-yen could easily increase that cross section in practice.
> There is no similar currency that Nvidia can create ad infinitum.
It can create obligations to provide future GPUs in return for present money.
Yes at some point people might start to question, but what are the true hard limits there, especially once SPVs and such start to get involved to shuffle things off the books?
It would also follow that by increasing the money supply significantly they’re also contributing to inflation a great deal correct? (Given the rest of the economy is not growing at near the same rate as the AI industry)
Maybe a dumb question but how is NVIDIA increasing the total supply of money? Only the fed can actually order more money to be "created". Private companies can only work within the existing supply, that is, their reserves, no?
All debt is money. Anybody can create money, the trick is getting other people to accept it.
Nvidia is vendor financing its output.
An ai company order $100m of GPUs. Nvidia delivers and holds onto that debt as an asset - like a bank loan.
The production company uses AI to create better plant and purchases $100m of AI tokens to do so. The ai company hold that debt like a bank loan
Nvidia requests $100m of production based on its $100m of orders. The production company holds that debt like a bank loan.
You now have a monetary loop. Take a single $10 bank deposit and Nvidia pays the production company, who pays the ai company who pays Nvidia. Run that round the circle a few million times and everybody has been paid.
> Only the fed can actually order more money to be "created".
If you go to a bank and get a loan, that is literally money that did not exist before you got a loan. People think that you are borrowing money that somebody else put in the bank, but that's not true. Banks can lend out a lot more money than people put into them.
Any time sometime makes a loan at a bank, that money is created. An accompanying debt is also created. It's like matter and antimatter. And when the debt is repaid, the matter and antimatter disappear again.
NV gives out a $100 to Party A, who puts it in their bank.
Bank takes $90 of that deposit (assuming 10% fractional reserve rule, no idea what the actual number is), and loans it out to party B, who pays it into either the same or another bank. Same rules apply -- except now it's down to $81 being loaned out, and so on and so forth, until that 100$ generated $1000 in total bank deposits.
edit: of course, it's never actually directly like this, a lot of other factors are involved, maybe the money is spent, maybe no one wants to borrow it, etc etc -- so it's more complicated but that's I think what they mean
Yup. Reserve requirements are functionally obsolete and never worked particularly well in the first place. Capital and liquidity requirements are far more robust and fine tuned.
That was my intuition at first too, but the original comment specified that they weren't borrowing all this money they're spending. The article also says how this is part of NVIDIA's strategy to enable demand, not create it, so supposedly these investments into their customers are actually going straight to paying for things.
Even if this money eventually gets loaned out eventually by one of NVIDIA's customers putting it into a bank, it isn't NVIDIA inflating the money supply, it's the borrowers, no? Or is this an ineffective way to look at things?
> There is no such thing as fractional reserve banking
Yes, there is. We just changed how we measure the fraction from a crude one like a reserve requirement (which takes zero account of asset quality or funding source) to finer and more-robust ones like capital and liquidity reqirements.
Banks still have to hold reserves. And those required reserves constrain their lending and thus the amount of money they can create. The limits just aren't the old-school reserve requirement.
If they’re effectively guaranteeing $500B in loans that adds close to $500B to M1, basically, that banks were not otherwise providing or loaning - at least that was my calculation.
Every form of lending that is specified via currency increases the supply.
If I give you GPUs worth $1bn, but take 100m payments for 11 years, then during that time you can use your other mony to buy other things that arent GPUs
If we stop after the 11 years and dont make new loans, the supply has shrunk back
Private banks increase money supply by lending. If 10 people deposit $1000 in a bank, it can loan $9000 to an 11th person. Now the economy has $19000 total.
I learned about this concept in college macroeconomics. I asked this exact question and the TA said “yeah I guess repaying debt is like destroying money” as if they had never thought of that before. The idea of lending money increasing the money supply is definitionally true.
> when debt is wiped out through bankruptcy that inflation remains
Bankruptcy is deflationary. The same way credit creation makes money bankruptcy (and any other reduction of debt, including through repayment) destroys it. It's why financial crises were often followed by deflation in gold-based economies.
It's a simplification to help people understand, but this is in the spirit of how things work because the value in the economy is not the money, but the goods and services that get created in the economy as a consequence of it. Most constructive uses of financial instruments in the markets (stocks, bonds, mutual funds, etc) are about efficient reallocation of money to enable value creation while balancing different risks, and people who provide this money indirectly benefit from this value creation via interest, dividends, selling stock at a higher price, etc.
Now to expand GP's example (still simplified):
- A borrows $100k money to pay B toward building a house. B puts $100k in their bank.
- C borrows $90k from B's bank toward building a house to pay D. D puts $90k in their bank.
- etc
So, houses were created (or other services were provided), and that's the real multiplicative factor. If banks loan out 90% of the cash stored (i.e. keep 10% in reserve [1]), the multiplicative factor of value creation in the economy is 10x the original amount of cash deposited in the first bank.
Now, if all of us withdrew our savings at once or sold all our stocks at once, we would have an economic shock analogous to that which resulted the Great Depression. That's why for banks, we have FDIC insurance - to mitigate such a panic so that money can serve its value-multiplicative role when it's not being actively used for anything else by the person owning the money. That's also why a positive (but low) inflation was originally considered economically healthy - so that people put their money in banks/market rather than under their mattresses gradually losing value. When interest rates are low, that encourages people to put their money into riskier (non-FDIC-insured) investments with higher growth potential, like a balanced portfolio of stocks/bonds/etc to avoid losing value to inflation, resulting in more economic growth.
But at that point in time, there's 19k in money. And future repayments of that loan back to the bank are less valuable to it than that current value figure. Because a bank can do a lot more shenanigans with that loan figure than it can with just the deposits.
> Private banks increase money supply by lending. If 10 people deposit $1000 in a bank, it can loan $9000 to an 11th person
It's the other way around. When a bank loans someone $1,000, they create a $1,000 deposit (their liability) and a $1,000 asset (their loan). Loans create deposits.
The Treasury can mint coin. But that's basically negligible in modern economies.
Ummm. No. I suggest you research how balance sheets work.
Unfortunately this kind of thinking is why so many people seem to think the big AI labs are totally killing it the second they make a “profit” on inference. Yes if you ignore the balance sheet all looks fine. Unfortunately companies go bankrupt because of their balance sheets, not operating profits and losses. You can make money on the direct COGS on every transaction and still be bankrupt.
You should research economics.
That $9000 can build a house that wouldn’t have existed otherwise. Then it gets paid back. $10000 in the bank and a $9000 house.
Banks create money when issuing a loan. This is how fractional reserve banking works. They lend money they don't have (most of). This is institutionalized fraud, and it's been standard operating procedure for centuries.
I’m not worried about the lack of reserve, i’m worried about the money shell game where private companies can drive inflation or deflation whichever serves their profit margins best.
The 2008 global financial crisis was a result of this, so not a made up worry.
They aren’t; the parent comment is incorrect. It’s safer to say Nvidia is encouraging the money that already exists to be deployed on AI buildouts.
But everyone is now chasing the same opportunity (AI and its dependencies like hardware and power) that will drive prices higher in those sectors until supply responds (or demand disappears).
Regardless of NVIDIA and LLM/AI, the claim that inflation is caused directly, or without-fail, by an increase in money supply - is not well founded. A significant money supply increase may very well have a tiny or possibly even negative price-increasing impact - depending on how money is supplied, to which elements and under what conditions.
It’s a fair point. Definitely depends on the how. I was figuring that adding $500B to a hot part of the economy while the rest of the economy shrinks might nudge a bit toward inflationary tendency, but at this point it’s hard to say what tenets of economics actually hold since the entire concept of “rational actors” went into the dustbin :)
Key difference is that these loans, which do increase the money supply and create inflation, are on average productive and profitable and thus deflationary. Quantitative easing is just printing money and often goes towards repaying bad debts, which are unproductive and thus not deflationary, so the inflation (increase in money supply) does not outweigh the deflation (creating of goods)
> that’s a pretty load-bearing as long as its cash flows continue. The two things are surely correlated
It's an important difference. In the GFC, the value of AAA-rated tranches fell. With the benefit of hindsight, we know they continued paying. They were directly leveraged, however, so mark-to-market losses caused firms to fail.
Nvidia stock crashing shouldn't have a similar effect to these commitments. If someone else has massively levered their Nvidia position, they'll obviously blow up. But Nvidia could survive a good deal of equity-market tumult in a way a bank could not.
Related but in specific ways. Stock is often priced in anticipation of growth. If NVDA could meet its credit obligations while its real profit stayed flat, the two would diverge, at least for awhile. A large amount of NVDA’s current cash flow is likely purchase contracts with a fixed multi-year term, which further smooths out the impact of, say, a stock crash following a couple of quarters of terrible earnings.
Now, whether many things NVDA has invested in with expectation of repayment or earnings would be able to repay or appreciate in a market environment where Nvidia’s stock was crashing? That’s another question entirely.
"Uh, that’s a pretty load-bearing as long as its cash flows continue. The two things are surely correlated."
Ppl have made the prediction of it being a bubble or unsustainable since 2022. At this point, it's hard to say these people have credibility anymore. Ai is big enough, much like Google in 2005 or Facebook/Social Network in 2010 or apps in 2015, that it's an institution unto itself. It's not going to just crash as so many are expecting and have been wrong the past 4 years about.
I've always found it interesting when corporations start acting like public institutions. When traditionally philosophical, social contract ideas apply to things like corporate governance. Or like here, where private structures get powerful and important enough to resemble government structures.
The ideas we deal with when we discuss society and organization aren't exclusive to government, they relate to human nature in general. I wonder if in the future we will have more discussion of power and how to organize it in corporations, similar to what we discuss today about government.
Well the key difference making any superficial similarities fall apart is Nvidia does not have neutral economy-wide goals of maintaining small+stable inflation, near-full employment and stabilizing financial institutions like the Fed does. Nvidia is entirely self interested in protecting their own shareholder value (that includes the incestuous web of investments ultimately ending up spent on their GPUs).
The structure of the Fed is setup the way it is to limit the sort of self-serving, myopic political micromanaging that could be damaging to the economy at large. And, unlike a beneficiary of rapid growth like Nvidia, has (historically) tried to identify potential indicators warning of unsustainable bubbles that could lead to financial contagion and tries to mitigate that risk using the limited monetary tools available and their public soapbox.
A similar decision making structure would potentially be very undesirable to Nvidia shareholders as caution over long time horizons would likely produce what they would consider an excessively conservative, defensive strategy to avoid putting too much air into the bubble too quickly (at the expense of their valuation).
Corporations certainly don't have the same goals government does, but human nature applies universally.
I'm not saying that corporations should have exactly the same rules and structure as government does, but perhaps many of the ideas used to design governments can be borrowed.
In a country without religion, banner or ideology to unite the people in current-and-coming turbulent times, the bet is made on "unite under money, or have no money left"
the way to fight it is to be principled even in front of cheaper options - and to support others like you
The American Revolution was very much a discussion about power. Of course, it wasn't just a discussion (we had to fight a war to defend the new form of government), but it was driven by the same concerns about the distribution of power.
Yeah its soooo interesting! Totally not dystopian, just soooo interesting and fascinating to ponder these scenarios in which corporations hold equal power to national governments!
Just such a curious scenario to let your mind wander about, how society would look like in these scenarios!
/s
I'm honestly so sick of the suspense of disbelief on this site, how is this more "interesting" to you, than the absolute sheer terror you should feel about going back to feudalism and serfdom? A typical western national state ensures that you have basic rights as a human being and aren't exploited to the death by non-government entities.
I'm not advocating for corporations to have equal power with governments, though. I think that's generally a terrible idea.
Looking at any organization with power and people involved, perhaps we can use the same ideas that traditionally apply to government in more places. That's all I'm saying.
This is not some mindless intellectual exercise to distract from how things are. In fact, as a proposal for how to reform corporate power, it's the opposite.
(Also, even though I think corporate power should be limited to a very specific arm of society, and that we shouldn't encourage more corporate power, you are greatly exaggerating. Feudalism and the abolition of human rights are not right around the corner.)
I wonder when they will give up on the gaming market because that could take down several publishers and developers. I really don't think it's an if question but a when because it almost feels like an afterthought at this point (they removed the standalone gaming revenue report from the financial reports this summer). Also I don't think AMD and Intel is capable "to step in" to replace them.
AMD would step in, they where behind nvidia but they've been closing that gap for a while and the 9070XT is the current value king (in this fucked up market) for mid-high gaming and you can actually buy them.
Demand for Nvidia cards has outstripped supply even on the mid-high cards specifically because they do better with local models than AMD cards with the same VRAM do broadly.
AI has completely broken the PC gaming market (and PC/Laptop market more broadly but gaming is really hit hard because it's the exact components that matter for both that overlap).
Nvidia had the mind share among gamers but so did Intel once, inertia only lasts so long they've been thoroughly intent on burning that to the ground for a while, back to the post 1080's
If RDNA5 is good (and 4 was it closed the gap on RT) they'll been in a solid place to take the spot if Nvidia do cede the ground.
I have a 7900XTX the last flagship card AMD did (about equal to a 9070XT for raster but 24GB VRAM not 16GB) and it has been and is a stellar card for gaming (let down only if you care about RT and the games I play don't have it).
Flawless under Linux, weaker on the AI behind nvidia but it runs Qwen surprisingly well and I didn't and don't care too much about that except to poke it occasionally.
Cryptocurrency mining had it broken before that. PC gaming's golden years are behind us, as AI, even if not LLMs, seems like it's not going away, so the gaming market is now the niche hobby while AI takes center stage.
RIP. My first gaming ~GPU~ (we called them 3d accelerators back then) was a Diamond Monster II with a 3dfx Voodoo 2 chip.
Oh shit, I had file deleted that. I remember now. I think I used that with the first N64 emulator, I think Project something to play Turok the first person dinosaur hunter game. Life before the internet was better I think.
The Glide API was the only hardware acceleration in there for a good while before it got OpenGL. I remember using a third party wrapper or something to make it work on whatever it was I had in 1999.
AMD and Intel would be HAPPY to replace them, and plenty capable.
Game consoles like PS5 have been AMD for a few generations.
Steamdeck/Steam Machine are AMD.
But really, Nvidia has no reason to leave gaming behind. They can just start dialing back their ambition on the gaming side and providing GPUs that aren't too useful for inference or training. All gaming needs is stability so that devs can aim for something. Games looked great 20 years ago and they'll look great 20 years from now, as long as developers know what they are building for.
> Also I don't think AMD and Intel is capable "to step in" to replace them.
It's kind of weird. nVidia kind of has the PC market cornered, but AMD has had the last couple of generations of Xbox and Playstation. Also, they power the Steam Deck/Machine, and Valve has been contributing a lot of AMD graphics features into the Linux drivers. There is a world where AMD (and maybe even Linux on AMD specifically) becomes the de facto standard for gaming.
Maybe this is silly of me, but maybe gaming would enjoy an era of hardware upgrades being rather unviable so the focus turns to optimization and aesthetic.
It will be interesting times but I don't think anyone will enjoy a plateau because it's expensive. Even replacing existing hardware is expensive now. Hardly enjoyabe.
A retraction from the gaming market wouldn't necessarily mean they'd still even produce the current crop of products. I could (and I would argue would) involve a complete shuttering of the GeForce brand, halting current production. On the assumption that Intel and AMD would follow, that wouldn't be an end of upgrades, but an end to the market.
Gaming is a larger market than Hollywood. Maybe it becomes a distraction for Nvidia, but someone will step in. That might actually be a good thing and why Nvidia wont do that: it creates an under served market in which newcomers can cut their teeth.
It’s not silly. games use far more hardware than they really need. It also pushes out release dates of aggressive console schedules like ps6 because even if it’s a massive upgrade and you have IP locked into your console, no one is going to pay $4000 to play a game like wolverine.
> I really don't think it's an if question but a when because it almost feels like an afterthought
I feel your reasoning is very weird. Are they losing money by selling consumer GPU? Just because the profit isn't that much compared to AI it doesn't mean that it's negative, and for-profit companies are not known for leaving money on the table. Apple doesn't reveal how much Apple TV+ makes for them either but I don't see it be gone anytime soon.
NVidia is limited by the number of chips they can produce.
If you can fab 1000 chips, and can sell some for $500 and some for $80000 what are you going to do?
The game GPU is at once profitable, but causes them to give up far more profits than they're gaining from it.
They're maintaining the game market to have multiple markets and not go all in, but it's strategic hedging at this point. When NVidia makes a gaming GPU instead of a data center GPU they are leaving money on the table in the short term since they're constrained at the fab level.
This almost seems like we need a strategic reserve for semiconductors- kinda like we have it for food, to prevent suppliers from throwing it away when they are suddenly able to sell something much more lucrative.
The CHIPS Act in the US did set aside a reserve for older processes used for automobile, defense, and industrial semiconductors - but that's not gaming GPUs that used the previous process node.
Nvidia is limited by the number of engineers they have. It might turn out that the AI market is so lucrative that it’s best to reallocate their gaming-focused engineers to AI.
Apple TV is at least a growth market for them, whereas gaming is sort of capped and clearly a tiny piece of nvidia’s revenue atm.
Which is distinctly different both in purpose and technology than the previous versions. DLSS 5 is no longer about frame improvements. It’s about increasing graphical fidelity with active AI rendering.
DLSS 5 is trying to relight and retexture the scene using AI. DLSS 4 is just trying to take a lower quality image and upscale it using AI
That seems a bit overzealous, most consoles[0] run on AMD chips today.
Both PS 5 and Xbox are based on AMD APUs and both serve the AAA market quite well. GTA 6, Assassin's Creed and CoD are probably good enough indicators that the performance is enough, even if there is always room for more (as PC ports show). The PC market will also probably be fine even if stagnation in perfomance gains has been creeping in for a few years now.
[0]: except Nintendo which relies on NVIDIA although their APU there focuses more on efficiency than top performance.
A Chinese company like CXMT will surely fill that unaddressed market the way CXMT is doing for memory. It would be rather unwise for nvidia to leave the door to the market open. AMD is positioned well to grow right now due to their non-Apple sillicon unified ram hardware coming soon.
Question: wouldn’t the fab be the actual bottleneck? In other words, why wouldn’t TSMC make more NVIDIA chips instead of AMD ones? I assume they’ll just do whichever is paying more, so if NVIDIA chips are better, gamers would be willing to pay more for them and in turn TSMC will be willing to make more of them?
It's still a 15B market for Nvidia, it's not nothing. But I'm curious how they're going to turn Rubin into a gaming GPU. I think at this point consumer GPU upgrades are going to be AI-related upgrades that happen to also help raster capabilities. Blackwell was already kind of a dud on performance uplift from Ada beyond the new LLM features.
Hermes still makes saddles that are somewhat reasonably priced compared to their handbags.
It's still profitable, it's their original raison d'etre, and there's no real reason for them to stop even if it its rounding error on their regular business.
Probably will never, ever see an Nvidia card with >32GB of VRAM though unless they start making dies that lack LLM performance like the gimped ethereum mining cards.
You must not be paying attention, they already have.
Intel is going nowhere but we all knew that anyways.
And again, you must not be paying attention, AMD is doing exactly what they said they would. No flagship for RDNA4 (just like RDNA2), RDNA5 flagship (10900 XT) coming right on schedule
I think they'll keep the gaming market alive for a while because renting gaming hardware from the cloud (GeForce NOW) is very congruous with AI keeping consumer hardware prices sky-high.
It continues modern trend of chow companies don't want consumers to truly own anything. Finance a car, pay a monthly subscription fee for heated seats, rent a phone, stream a movie, get rid of physical media, rent a GPU.
But if GeForce NOW doesn't take off, and they get convinced that the AI bubble will not pop, I could see them pulling a Micron and ending their consumer product lines.
> I really don't think it's an if question but a when
Nvidia still will ship gaming products. The upcoming RTX Spark laptop APUs are still gaming-capable - we also have Blackwell gaming GPUs and the Nvidia-powered Nintendo Switch 2.
People echoed this sentiment during the crypto mining crunch, and we still got gaming hardware designs after that blew over. One of CUDA's core value props is the consumer market, and Nvidia probably won't surrender it unless hardware becomes unreasonably scarce.
I don't think crypto is comparable. They barely made some dedicated crypto GPUs that market was always fickle due to ASCIs.
Look at the nvidia revenue breakdown chart, the AI boom looks quite different.
The GPU crunch came because cards like the 3060 were extremely cheap and could outrun most sub-$1000 ASICs at the time. The dedicated crypto GPUs were too-little too-late; hundreds of thousands of ordinary CUDA-capable GPUs had already been repurposed for mining by the time they launched.
> Also I don't think AMD and Intel is capable "to step in" to replace them
AMD has powered 2 generations each of Sony and Xbox consoles, Steam deck and shops a ton of GPUs especially if you count APUs. And then Intel literally ship more GPUs than Nvidia and AMD combined.
The gaming market doesn't need Nvidia. Especially as AAA is cratering.
They may allocate different number of resources every year based on market conditions but they'll never give up on gaming, that would be extremely silly.
>Nvidia’s financial engineering is partly a response to its biggest customers’ transformation into rivals. “Hyperscalers”, tech giants such as Amazon, Google, Meta and Microsoft, account for roughly half of Nvidia’s revenue
Companies just don't want to pay Jensen's tax. Hyperscalers might still pay Jensen's tax for LLM training but for inference. you don't have to. they are also betting on their own chip for training to replace Nvidia.
this is Nvidia panicking and doing vendor fiance to Neoclouds and buying Hugging Face. even none hyperscalers like Meta is betting on its own chip for AI inference.
Cracks are starting to appear. Open Ai and Anthropic are publicly asking for slowdown in AI research. Translation: We see this technology not being any more useful than what it is now, no AGI is coming, and the first one to accept this and slow down the dollar burn rate will incur the wrath of the market. So let's say this big boogey man technology will end all life on earth and we all slow down together. Bonus points if we can lobby for this to be included in the national security bucket. Then US govt can bail us out. Yay!
If Nvidia is the bank, they should be starting to sweat a bit. It's not often companies ask for a voluntary slowdown.
Realistically, they're worse than a central bank, because they can't exactly expand supply monotonically like a normal central bank. Nor do they realistically control rates.
Okay, but at some point these investments need to start turning profits; the financing NVDA has arranged is temporary, and private credit needs returns at some point. The overinvestment in AI will lead to a downturn in the capital cycle.
Quite a lot better than people originally thought. There is a LOT of demand for 5+ year old Ampere on the secondary market, and hyperscalers are still holding on to many of them since they can’t acquire new compute fast enough.
The Japanese economy and yen carry trade is a close second.Rising oul prices and reduced output due to the conflicts in the Middle East could filter through to increasing yields on Japanese debt. In turn, the yen interventions have to continue to keep it lower than 160, which seems to be the psychological barrier for the yen carry trade.
and it doesn’t disappear [1], much like law of conservation
at any given time there is a finite amount of it most easily
observed in currencies’ relative price to another currency, and its movement between owners causes distortions in asset prices
most people I’m around and most trading indicator developers I’ve seen don’t seem to conceptualize the flow of value around the economy. so it feels niche and maybe visualizing this mental model can help many people
[1] central banks can functionally delete money in their bond purchase and roll off programs
I think we can go even further and say that IBM, Zeiss, several Japanese companies, and even the US Department of Energy are the actual foundation since ASML is largely just an integrator of many different technologies they license from elsewhere.
Weird to think we are currently living in the ‘unlimited free Ubers because you recommended a friend or 2’ phase of this new technology.
Imagine if running fable costs you what it actually costs to run fable. A lot of vibe coders (and just proper software engineers) are gonna be very sad if that comes to pass.
This is what I’m not getting - we’re spending more than the global gross sales of all software on the planet on AI, how much more software needs to be sold and for how much more to recoup?
Imagine a scenario where the AI bubble bursts and AI companies and neoclouds go bankrupt en masse, and then a huge rebound occurs when AI has a delayed takeoff. Nvidia ends up with a massive amount of compute on its hands from its backstop deals, and it also owns assets from failed companies when profits start to grow. New startups running using Hugging take the place of OpenAI and Anthropic when their compute assets are divided between survivors like Nvidia, Microsoft, Alphabet, Meta.
If/when there is an AI crash, any number of small startups can buy compute for the price of electricity without anyone wanting to buy them. That is when the real innovation happens. The top of the hype cycle is usually more about getting rich quick and buying and shutting down competition.
As the breakneck growth slows, the multiples will compress, and the companies will pull back on spending accordingly as they watch their stocks decline and investors demand more conservative spending behavior.
The Internet didn't stop expanding in 2000-2001. Everything got drastically larger over the following two decades. The multiples on earnings did implode for ~15 years however. MSFT stock for one example went nowhere during that time and compressed down to a near single digit PE.
AI will be a minimum of 10x larger in most every regard 20 years out. That has nothing to do with shorter-term multiples given to these companies in relation to the hyper fast growth they have been riding early in the boom.
> worth around $5.4trn
Note that the Fed has a $6.7tn balance sheet [1]. (This is a silly comparison. But still fun.)
The real comparison: Nvidia's $500+ billion of investments and commitments [2] is substantially more than any easing the Fed has done in the same time [3]. Monetarily, Nvidia is creating a lot of money in our economy.
The good news: I have seen no evidence Nvidia has borrowed against its stock or otherwise linked its equity value to these commitments. Its stock could crash without causing–as long as its cash flows continue–a credit crisis through its investments and commitments.
[1] https://www.federalreserve.gov/monetarypolicy/bst_recenttren...
[2] https://www.sec.gov/Archives/edgar/data/1045810/000104581026...
[3] https://www.federalreserve.gov/monetarypolicy/bst_recenttren...
“as long as it’s cash flow continues” is doing a lot of optimistic heavy lifting. The whole premise of the circular financing worry is that Nvidia sits in the middle of all the guarantees made to companies like OpenAI. If any of those companies become insolvent, Nvidia is on the hook for it.
Also Nvidia isn’t really creating money. The 500B number is third party capital that already exists (BX, Apollo, etc).
Yea, but they would have to become insolvent in a way that makes compute lose value.
The reason Nvidia is comfortable making these deals is because if OpenAI can’t use the compute, someone else can.
Granted OpenAI going insolvent likely means a drop in the value of compute…
Compute has already lost value for me. Six months ago I thought you needed a 1T+ model to be useful coding. Now I am able to get by just fine with a 27b model.
I see two factors converging to cause a collapse of this house of cards:
1. People are realizing that what they need isn't more general intelligence, it's more specialization. A small but well tuned coding model, a small but well tuned customer service model, a small but well tuned document explorer.
2. Specialized hardware - TPUs and NPUs - especially coming out of china. The latest GLM model was trained and runs on Huawei hardware. Nvidia is only worth so much because they are the biggest and best provider of the kind of compute needed to run llms, but the export bans mean china has a lot of incentive to topple that monopoly.
The amount of compute we need to do the things llms do is falling rapidly, the number of people who can provide that compute is rising.
> People are realizing that what they need isn't more general intelligence, it's more specialization. A small but well tuned coding model...
It’s not quite as simple as that. Several studies have shown the opposite: models trained on more diverse knowledge tend to cross-pollinate across domains. So a more generalized model can actually perform better than a specialized one.
That’s why you’re not seeing tons of tiny models (one for Python, one for Pascal, one for Rust, etc).
This is definitely the position of the big ai companies.
But it doesn't match my experience. Qwen3.8 27b is clearly smarter at coding than MANY bigger models. gpt-oss-120b for example, is almost 4x the size, and performs way worse at coding tasks.
It's clear to me that you can build small models that work well at specific tasks.
Python vs Rust is probably too fine grained a way to build a model. Coding in general seems like a better target.
There will always be a place for large generalist models, no doubt. But I think that place is much smaller than the big ai companies are counting on.
I think we’re in agreement.
I make heavy use of smaller local models on a daily basis (Qwen3-VL for auto-captioning images, Gemma3:27b for some translation work, etc.). Gemma3:27b is a good example of a very capable general purpose multimodal model and has handled almost everything I've thrown at it from sentiment analysis to documentation writing.
I suppose I was drawing a distinction between specialized and general intelligence versus small and large. I don’t think those are necessarily mutually exclusive.
Problem is conflict of interest: the studies are mostly from the providers of the biggest models, or someone who received free tokens to do the research.
I think what will keep the industry afloat, all else failing, is the surveillance industry! Nothing like a fat reoccurring cheque from the government to check if little Jimmy is committing thought crime!
That's unless the code produced in the future is much more complex than today's.
Sure, but it would be actively bad to make the code more complex simply because we have machinery that helps us deal with the complexity. A big part of how people assess the models' coding capability is whether they create needless, incidental complexity.
Assuming it’s all going to be vibe coded garbage, yeah it will be much more complex. Like a toddler writing a symphony.
You're not considering video which OpenAI opted out of when they retired Sora.
Generative video requires significantly more computing power and energy than generative text.
OpenAI is fucked, compute is still needed, it's just them that isn't.
OpenAI dropped sora because it was costing them ridiculous amounts of money and earning them very little. They determined that the market can't support the cost of generating video.
Without a material change in the market (more buyers, vastly cheaper generation), it's unlikely a different company could make that work. More buyers isn't likely to happen, so that leaves vastly cheaper generation - something that would cause nvidia's value to collapse if it happened.
Minimax H3 works pretty great and you can run it on a 3090.
There would also need to exist sufficient demand for video, which hasn’t happened yet.
oAI isn't anywhere near close to fucked as long as their models are head and shoulders above even the very best open models in terms of tool calling and rock solid stability/reliability for agents/coding harnesses. Which, they are right now and we'll see if open models actually catch up in that regard. Even the "best" open models pale in comparison with tool calling and general "prompt and go do something else for an hour" reliability that we have with GPT models. With GPT models, streaming rarely stops unexpectedly. You almost never have to constantly nudge them along, etc. Granted with open models all of this can vary depending on the provider, and perhaps open models/protocols/APIs/harnesses aren't well enough aligned, but OpenAI models just seem to work without constant (or hardly any) wrinkles and with almost any harness/agent.
>oAI isn't anywhere near close to fucked as long as their models are head and shoulders above even the very best open models in terms of tool calling and rock solid stability/reliability for agents/coding harnesses
That's already not the case today. If you sat me in front of an LLM and told me to figure out if I'm working with K3 or Astra, I could probably do it, but it would take some work to be certain.
They need the right harness and either your help it auto produces in time enough content to further improve.
If you reshuffle your argument, and apply the same facts you get to a similar conclusion but with a drastically different spin.
> it's more specialization
China, constrained by hardware, and talent (not to slight the Chinese, but they are limited to domestic resources - and much of the US effort is very international). They did, what the Chinese do, and optimized the process of production, and drastically lowered the cost of development of their models. Cheeper to build, cheaper to run is just good economics.
Meanwhile in the us, we have open AI doing "experiments" - it looks like the costs around the hugging face hack are going to be about the same as China would spend on building out one of their smaller efforts (several million dollars). (Depending on whos numbers you trust, the fact that I can even make this claim should make you raise an eyebrow).
Go back to the 80s' and "expert systems" - most people will tell you that for their time, they were amazing, and useful. People would have loved to have more of them but they were so cost prohibitive that we all but abandoned them for serious use. The US frontier labs seem to have forgotten this lesson and their calls to "slow down" look like an excuse to "cut the waste so we can move to making money".
The problem is that if OpenAI doesn't want the compute nobody does. All of these companies' demand for compute are correlated. It isn't likely that OpenAI will want less compute in isolation. Furthermore, the circular financing structure means that if OpenAI buys less chips it means that Nvidia has less money to give to say Anthropic to buy more chips and suddenly the exponential growth that circular financing has allowed to grow runs in reverse.
> if OpenAI can’t use the compute, someone else can.
The problem with that is that OpenAI can only afford to pay for the compute because they are burning investor money (and so are most of OpenAI's biggest clients). They are losing billions. If they stop burning money, nobody else will be there to pay for that compute at OpenAI's cost.
Sure, somebody will probably be able to use these GPUs, they just won't be able to pay nearly as much for them as OpenAI does.
In reality, it's just nowhere near worth as much as OpenAI pays for it. Inflating the cost of compute is part of the problem caused by the circular financing, and if (or maybe when) OpenAI goes, the price of compute will go with them.
> investor money
But that’s the point. Investors believe investment in AI will pay off.
Right, "believe".
Devil's advocate: OpenAI not being able to use compute is highly correlated to many other AI companies not being able to find a meaningful use of this compute.
Failure to take into consideration those kind of correlations ("If my biggest client isn't able to buy it, I would be able to find someone else who will") is one of the principle causes why many risk models turned out to be garbage during the Great Financial Crisis.
That is why I added the last line.
But I also doubt Nvidia is on the hook if OpenAI just no longer wants the compute. I bet they are only on the hook if OpenAI cannot pay for it (is insolvent in some way).
I also have to bring up that OpenAI has already spat out an inference chip that beats Nvidia on flops per watt. So they could potentially not need the compute while other ai companies do.
> they would have to become insolvent in a way that makes compute lose value
They would have to go insolvent in a way that hits Nvidia revenue. Those are related by distinct factors, a difference that may matter in a crisis.
> if OpenAI can’t use the compute, someone else can
This is the big point IMO since I have never given $1 to OpenAI but I subscribe to Vidu and Typecast, and have given money to Kling, Hailou, and even Gemini in the form of Google Workspace.
So these other guys have products and use cases, which OpenAI has never been able to crack beyond ChatGPT. And ChatGPT was never worth paying for, IMO.
If OpenAI dies, it's not because there is no market for the technology (which is all NVIDIA cares about), it's more that OpenAI doesn't know how to run a relevant technology company.
They were given everything, not just NVIDIA's billions of dollars and credit backing but all the first-mover advantage, all the respect and credibility early on, so it's really sad to see them unable to develop interesting products and turn a profit in a space they helped pioneer, while so many others are making money with the tech all around them.
NVIDIA is fine. The technology will continue to improve and NVIDIA will stay at the center. OpenAI is fucked - knew it when they retired Sora to focus on text-to-text and coding (a largely solved problem).
Do you use coding agents? Just curious bc from my experience using coding agents, open ai’s codex is neck and neck with anthropic’s claude code if not ahead. I wouldn’t agree that OpenAI hasn’t done anything since ChatGPT since codex is my daily driver for software engineering
It's kinda nuts to me how people can act like Claude is lightyears ahead of OpenAI models. Sure, it's one thing to simply have a preference or claim that Claude does some things better, but in reality they are both about as effective at doing the same job. I've long preferred GPT models because they know better how to shut up and don't seem to overthink as much as Claude, but I'm under no illusions that if OpenAI went belly-up then I couldn't do my job essentially the same with Claude. GPT models have also clearly improved over time in terms of programming. There haven't been any "big bangs" necessarily, but it's really not hard to give the same task to 5.3 and 6 and see which one has better output.
All the models converged, in every single generation since 2022.
I've paid more money to OpenAI than I've spent on all other products I pay for combined the past 5 years.
Granted OpenAI going insolvent
All the "frontier" AI companies *are* currently insolvent. They have never been anything other than cash burning machines.
The only way they keep the lights on and the doors open is by borrowing money --- and epic amounts of it. If those operating the cash spigot decide to turn it off, all AI companies will likely be similarly affected --- and so will Nvidia.
OpenAI expects to burn through more cash between 2024 and 2029 than Uber, Tesla, Amazon and Spotify did - combined - before those companies started making money
https://www.morningstar.com/news/marketwatch/20251205243/thi...
To make things worse, hardware prices have spiked, due to AI companies.
Fairly sure data center construction costs are also going up (they require so many resources that everything is constrained at the moment, especially electricity production).
So I don't understand in what world these frontier AI companies can somehow become profitable. The basic tech they're using is basically the same. Yes, around the edges there are a lot of things that can be done, and were done, like caching, batching, mixture of experts, etc, but basically everyone has done all of that by now, and they're still losing money.
So:
Total costs going up a lot - revenues per unit not increasing proportionally, if anything, Chinese models are forcing those down.
How does that math work out to profits? I don't see it.
Or about as bad, after trillions of dollars in investments over multiple years, let's say the entire frontier AI sector has a total profit of $20bn by 2030. In what world does that make sense? Assuming they can scale that total profit to $100bn in 2035 without investing another cent from 2027 to 2035 (utterly ridiculous), the return on investment would happen in roughly 20 years.
It's capitalism run amuck --- and on an epic scale.
China is the one that is really in the driver's seat here. They have the opportunity and the ability to nullify/wipe out our huge investment in AI.
> “as long as it’s cash flow continues” is doing a lot of optimistic heavy lifting
It's not. It's stating a condition. For traditional banks, a stock crash can independently trigger a failure.
> whole premise of the circular financing worry is that Nvidia sits in the middle of all the guarantees made to companies like OpenAI. If any of those companies become insolvent, Nvidia is on the hook for it
Sorry, I meant revenues. If Nvidia's revenues stay stable, these commitments aren't a problem. Even if the stock price crashes.
> Nvidia isn’t really creating money
It absolutely is. Similar to the way banks create money [1]. The commitments support credit that wouldn't exist without it.
[1] https://www.bankofengland.co.uk/-/media/boe/files/quarterly-...
> The good news: I have seen no evidence Nvidia has borrowed against its stock or otherwise linked its equity value to these commitments.
Nvidia doesn't need it. It funds other companies. They do this thing that Nvidia doesn't do. It shows up on their balance sheets and Nvidia just gets to claim the valuation of the investment on its balance sheet.
It can't go tits up!
M2 is $21 trillion, which is what the fed signed up to backstop. How much did NVDA sign up to backstop?
> which is what the fed signed up to backstop
No, it's not. M2 includes things like traveler's cheques and money-market funds.
Ok how much did the fed sign up to backstop then? Surely it’s more than physical currency
> how much did the fed sign up to backstop then? Surely it’s more than physical currency
The Fed doesn't backstop physical currency. That is issued by the Treasury (specifically, the Mint). The FDIC backstops bank deposits; the U.S. guarantees is obligations.
The Fed doesn't properly "backstop" anything. It's the lender of last resort–if you have a Treasury or other good collateral, it will loan you money against it. It's a financial regulator. And it regulates interest rates (i.e. the price of money) to influence inflation and employment.
The only backstops the Fed truly makes are to banks, by guaranteeing to always stand ready to lend against Treasuries and other good collateral.
“Federal reserve note”
You are taking this too literally anyway. I know they don’t backstop jack squat but in practice there is a fed put.
Have a conversation with me, don’t be a pedant.
If you are saying they will lend last resort against treasuries you should know there are $40 trillion of those outstanding…
How about what’s the amount from banks that the fed would willing lend as a last resort?
> “Federal reserve note”
That isn't a backstop, it's a direct obligation. It's also, like, not a real one? You can't redeem notes for specie. The term originates from when you could redeem dollars for metal. The Fed did play a role in backstopping that guarantee.
> know they don’t backstop jack squat but in practice there is a fed put
Sure. That isn't the same as a backstop. Backstops are hard–the Fed can't turn away an eligible borrower at the discount window. The Fed put is soft–the Fed will let market participants fail to send a message.
Going back to the top, it is incorrect to say the Fed backstops M2. This wouldn't be a footnote in a central-banking discussion, it would be something that would get called out as a screwup.
> How about what’s the amount from banks that the fed would willing lend as a last resort?
Infinity. The Fed mainly accepts Treasuries as collateral, but it can and has expanded the definition of good collateral in crises. There is no legal or frankly practical limit on how much money the Fed can create. Its only constraint is ultimately political. (Which is in turn mostly a function of inflation and employment and I guess now social media vibes.)
If you're looping this back to Nvidia, yes, I never claimed Nvidia has more lending capacity than the Fed. What I said was it's interesting that in practice, Nvidia appears to have created more money (if we're being pedantic, M3 which turns into M1) than the Fed has in that time. The Fed wasn't particularly trying to ease financial conditions in that time, so this is more of a curiosity tied to the title than a statement of capability.
Wasn’t trying to dick measure nvda vs fed. More thinking thru guarantees vs assets for the two institutions
> guarantees vs assets for the two institutions
The Fed can never default on any dollar-denominated debt. There is no similar currency that Nvidia can create ad infinitum.
That said, the number I think you're looking for in respect of the Fed is $30 to 40 trillion. It's about U.S. GDP. And it's also about U.S. bank and money-market assets plus the Fed's balance sheet. If every American bank failed, this could be the amount of money the Fed would have to create.
That said, Treasury running out and e.g. defending the euro-yen could easily increase that cross section in practice.
> There is no similar currency that Nvidia can create ad infinitum.
It can create obligations to provide future GPUs in return for present money.
Yes at some point people might start to question, but what are the true hard limits there, especially once SPVs and such start to get involved to shuffle things off the books?
It would also follow that by increasing the money supply significantly they’re also contributing to inflation a great deal correct? (Given the rest of the economy is not growing at near the same rate as the AI industry)
Maybe a dumb question but how is NVIDIA increasing the total supply of money? Only the fed can actually order more money to be "created". Private companies can only work within the existing supply, that is, their reserves, no?
All debt is money. Anybody can create money, the trick is getting other people to accept it.
Nvidia is vendor financing its output.
An ai company order $100m of GPUs. Nvidia delivers and holds onto that debt as an asset - like a bank loan.
The production company uses AI to create better plant and purchases $100m of AI tokens to do so. The ai company hold that debt like a bank loan
Nvidia requests $100m of production based on its $100m of orders. The production company holds that debt like a bank loan.
You now have a monetary loop. Take a single $10 bank deposit and Nvidia pays the production company, who pays the ai company who pays Nvidia. Run that round the circle a few million times and everybody has been paid.
Rinse and repeat.
If the debt cancels out doesn't this mean that there was no debt ?
I’m guessing it doesn’t “cancel out” due to interest.
> Only the fed can actually order more money to be "created".
If you go to a bank and get a loan, that is literally money that did not exist before you got a loan. People think that you are borrowing money that somebody else put in the bank, but that's not true. Banks can lend out a lot more money than people put into them.
> Only the fed can actually order more money to be "created"
No. Most money in modern economies is created by private parties [1].
[1] https://www.bankofengland.co.uk/-/media/boe/files/quarterly-...
Any time sometime makes a loan at a bank, that money is created. An accompanying debt is also created. It's like matter and antimatter. And when the debt is repaid, the matter and antimatter disappear again.
NV gives out a $100 to Party A, who puts it in their bank.
Bank takes $90 of that deposit (assuming 10% fractional reserve rule, no idea what the actual number is), and loans it out to party B, who pays it into either the same or another bank. Same rules apply -- except now it's down to $81 being loaned out, and so on and so forth, until that 100$ generated $1000 in total bank deposits.
edit: of course, it's never actually directly like this, a lot of other factors are involved, maybe the money is spent, maybe no one wants to borrow it, etc etc -- so it's more complicated but that's I think what they mean
0%. Zero percent is the actual reserve rule. https://www.stlouisfed.org/bank-supervision/reserve-administ...
Yup. Reserve requirements are functionally obsolete and never worked particularly well in the first place. Capital and liquidity requirements are far more robust and fine tuned.
That was my intuition at first too, but the original comment specified that they weren't borrowing all this money they're spending. The article also says how this is part of NVIDIA's strategy to enable demand, not create it, so supposedly these investments into their customers are actually going straight to paying for things.
Even if this money eventually gets loaned out eventually by one of NVIDIA's customers putting it into a bank, it isn't NVIDIA inflating the money supply, it's the borrowers, no? Or is this an ineffective way to look at things?
There is no such thing as fractional reserve banking. The multiplier is a myth.
Quite why this persists when the Bank of England debunked it in 2014 [0] is anybody’s guess.
Just another of those concepts that is neat, plausible and wrong.
[0]: https://www.bankofengland.co.uk/quarterly-bulletin/2014/q1/m...
> There is no such thing as fractional reserve banking
Yes, there is. We just changed how we measure the fraction from a crude one like a reserve requirement (which takes zero account of asset quality or funding source) to finer and more-robust ones like capital and liquidity reqirements.
Banks still have to hold reserves. And those required reserves constrain their lending and thus the amount of money they can create. The limits just aren't the old-school reserve requirement.
They don’t constrain the quantity of lending. They only change the price.
Liability side controls don’t work.
I think at the top level between Govt and Industry and understanding has been reached that AI industry will be backstopped
If they’re effectively guaranteeing $500B in loans that adds close to $500B to M1, basically, that banks were not otherwise providing or loaning - at least that was my calculation.
Every form of lending that is specified via currency increases the supply.
If I give you GPUs worth $1bn, but take 100m payments for 11 years, then during that time you can use your other mony to buy other things that arent GPUs
If we stop after the 11 years and dont make new loans, the supply has shrunk back
Private banks increase money supply by lending. If 10 people deposit $1000 in a bank, it can loan $9000 to an 11th person. Now the economy has $19000 total.
The $9000 has to be paid back, and then some. I sure hope you aren't an accountant.
But for the duration, there is more money. This isn’t some crank theory, it’s orthodox economics: https://en.wikipedia.org/wiki/Fractional-reserve_banking
I learned about this concept in college macroeconomics. I asked this exact question and the TA said “yeah I guess repaying debt is like destroying money” as if they had never thought of that before. The idea of lending money increasing the money supply is definitionally true.
> the TA said “yeah I guess repaying debt is like destroying money” as if they had never thought of that before
They shouldn't have been a TA. Modern money is destroyed in three ways: through taxation, defaults and the extinguishing of debts.
And when debt is wiped out through bankruptcy that inflation remains.
> when debt is wiped out through bankruptcy that inflation remains
Bankruptcy is deflationary. The same way credit creation makes money bankruptcy (and any other reduction of debt, including through repayment) destroys it. It's why financial crises were often followed by deflation in gold-based economies.
It's a simplification to help people understand, but this is in the spirit of how things work because the value in the economy is not the money, but the goods and services that get created in the economy as a consequence of it. Most constructive uses of financial instruments in the markets (stocks, bonds, mutual funds, etc) are about efficient reallocation of money to enable value creation while balancing different risks, and people who provide this money indirectly benefit from this value creation via interest, dividends, selling stock at a higher price, etc.
Now to expand GP's example (still simplified):
- A borrows $100k money to pay B toward building a house. B puts $100k in their bank.
- C borrows $90k from B's bank toward building a house to pay D. D puts $90k in their bank.
- etc
So, houses were created (or other services were provided), and that's the real multiplicative factor. If banks loan out 90% of the cash stored (i.e. keep 10% in reserve [1]), the multiplicative factor of value creation in the economy is 10x the original amount of cash deposited in the first bank.
Now, if all of us withdrew our savings at once or sold all our stocks at once, we would have an economic shock analogous to that which resulted the Great Depression. That's why for banks, we have FDIC insurance - to mitigate such a panic so that money can serve its value-multiplicative role when it's not being actively used for anything else by the person owning the money. That's also why a positive (but low) inflation was originally considered economically healthy - so that people put their money in banks/market rather than under their mattresses gradually losing value. When interest rates are low, that encourages people to put their money into riskier (non-FDIC-insured) investments with higher growth potential, like a balanced portfolio of stocks/bonds/etc to avoid losing value to inflation, resulting in more economic growth.
[1]: https://en.wikipedia.org/wiki/Fractional-reserve_banking
And thus $9k of <something they got that $9k worth of value for> is injected into the economy, either assets sold or work performed.
Eventually
Which is, you know, the entire risk that people are worried about.
But at that point in time, there's 19k in money. And future repayments of that loan back to the bank are less valuable to it than that current value figure. Because a bank can do a lot more shenanigans with that loan figure than it can with just the deposits.
> Private banks increase money supply by lending. If 10 people deposit $1000 in a bank, it can loan $9000 to an 11th person
It's the other way around. When a bank loans someone $1,000, they create a $1,000 deposit (their liability) and a $1,000 asset (their loan). Loans create deposits.
The Treasury can mint coin. But that's basically negligible in modern economies.
Ummm. No. I suggest you research how balance sheets work.
Unfortunately this kind of thinking is why so many people seem to think the big AI labs are totally killing it the second they make a “profit” on inference. Yes if you ignore the balance sheet all looks fine. Unfortunately companies go bankrupt because of their balance sheets, not operating profits and losses. You can make money on the direct COGS on every transaction and still be bankrupt.
You should research economics. That $9000 can build a house that wouldn’t have existed otherwise. Then it gets paid back. $10000 in the bank and a $9000 house.
Banks create money when issuing a loan. This is how fractional reserve banking works. They lend money they don't have (most of). This is institutionalized fraud, and it's been standard operating procedure for centuries.
But the fraction to be kept in reserve has been zero for 4-5 years.
Almost like the concept is complete bunkum.
It’s been zero in the UK for hundreds of years.
that.. doesn't make it better
It was replaced by other mechanisms. It’s not literally zero any kind of reserves.
I’m not worried about the lack of reserve, i’m worried about the money shell game where private companies can drive inflation or deflation whichever serves their profit margins best.
The 2008 global financial crisis was a result of this, so not a made up worry.
They aren’t; the parent comment is incorrect. It’s safer to say Nvidia is encouraging the money that already exists to be deployed on AI buildouts.
But everyone is now chasing the same opportunity (AI and its dependencies like hardware and power) that will drive prices higher in those sectors until supply responds (or demand disappears).
Regardless of NVIDIA and LLM/AI, the claim that inflation is caused directly, or without-fail, by an increase in money supply - is not well founded. A significant money supply increase may very well have a tiny or possibly even negative price-increasing impact - depending on how money is supplied, to which elements and under what conditions.
It’s a fair point. Definitely depends on the how. I was figuring that adding $500B to a hot part of the economy while the rest of the economy shrinks might nudge a bit toward inflationary tendency, but at this point it’s hard to say what tenets of economics actually hold since the entire concept of “rational actors” went into the dustbin :)
well, M2 money supply is increasing with or without AI industry
Key difference is that these loans, which do increase the money supply and create inflation, are on average productive and profitable and thus deflationary. Quantitative easing is just printing money and often goes towards repaying bad debts, which are unproductive and thus not deflationary, so the inflation (increase in money supply) does not outweigh the deflation (creating of goods)
Its stock could crash without causing–as long as its cash flows continue–a credit crisis through its investments and commitments
Uh, that’s a pretty load-bearing as long as its cash flows continue. The two things are surely correlated.
> that’s a pretty load-bearing as long as its cash flows continue. The two things are surely correlated
It's an important difference. In the GFC, the value of AAA-rated tranches fell. With the benefit of hindsight, we know they continued paying. They were directly leveraged, however, so mark-to-market losses caused firms to fail.
Nvidia stock crashing shouldn't have a similar effect to these commitments. If someone else has massively levered their Nvidia position, they'll obviously blow up. But Nvidia could survive a good deal of equity-market tumult in a way a bank could not.
Related but in specific ways. Stock is often priced in anticipation of growth. If NVDA could meet its credit obligations while its real profit stayed flat, the two would diverge, at least for awhile. A large amount of NVDA’s current cash flow is likely purchase contracts with a fixed multi-year term, which further smooths out the impact of, say, a stock crash following a couple of quarters of terrible earnings.
Now, whether many things NVDA has invested in with expectation of repayment or earnings would be able to repay or appreciate in a market environment where Nvidia’s stock was crashing? That’s another question entirely.
Indeed. https://www.groundbrkr.com/p/the-second-derivative-why-no-on...
> load-bearing
I'm worried I'm going to start picking up claudisms, and then accused of being AI.
"Uh, that’s a pretty load-bearing as long as its cash flows continue. The two things are surely correlated."
Ppl have made the prediction of it being a bubble or unsustainable since 2022. At this point, it's hard to say these people have credibility anymore. Ai is big enough, much like Google in 2005 or Facebook/Social Network in 2010 or apps in 2015, that it's an institution unto itself. It's not going to just crash as so many are expecting and have been wrong the past 4 years about.
I've always found it interesting when corporations start acting like public institutions. When traditionally philosophical, social contract ideas apply to things like corporate governance. Or like here, where private structures get powerful and important enough to resemble government structures.
The ideas we deal with when we discuss society and organization aren't exclusive to government, they relate to human nature in general. I wonder if in the future we will have more discussion of power and how to organize it in corporations, similar to what we discuss today about government.
Well the key difference making any superficial similarities fall apart is Nvidia does not have neutral economy-wide goals of maintaining small+stable inflation, near-full employment and stabilizing financial institutions like the Fed does. Nvidia is entirely self interested in protecting their own shareholder value (that includes the incestuous web of investments ultimately ending up spent on their GPUs).
The structure of the Fed is setup the way it is to limit the sort of self-serving, myopic political micromanaging that could be damaging to the economy at large. And, unlike a beneficiary of rapid growth like Nvidia, has (historically) tried to identify potential indicators warning of unsustainable bubbles that could lead to financial contagion and tries to mitigate that risk using the limited monetary tools available and their public soapbox.
A similar decision making structure would potentially be very undesirable to Nvidia shareholders as caution over long time horizons would likely produce what they would consider an excessively conservative, defensive strategy to avoid putting too much air into the bubble too quickly (at the expense of their valuation).
Corporations certainly don't have the same goals government does, but human nature applies universally.
I'm not saying that corporations should have exactly the same rules and structure as government does, but perhaps many of the ideas used to design governments can be borrowed.
USA is trying to become corporatocracy
In a country without religion, banner or ideology to unite the people in current-and-coming turbulent times, the bet is made on "unite under money, or have no money left"
the way to fight it is to be principled even in front of cheaper options - and to support others like you
To be clear: I'm not trying to advocate for more corporate power.
The governments of today weren't formed out of "discussions of power"
The American Revolution was very much a discussion about power. Of course, it wasn't just a discussion (we had to fight a war to defend the new form of government), but it was driven by the same concerns about the distribution of power.
Yeah its soooo interesting! Totally not dystopian, just soooo interesting and fascinating to ponder these scenarios in which corporations hold equal power to national governments!
Just such a curious scenario to let your mind wander about, how society would look like in these scenarios!
/s
I'm honestly so sick of the suspense of disbelief on this site, how is this more "interesting" to you, than the absolute sheer terror you should feel about going back to feudalism and serfdom? A typical western national state ensures that you have basic rights as a human being and aren't exploited to the death by non-government entities.
You're catastrophizing and misreading the original comment in an uncharitable way.
I'm not advocating for corporations to have equal power with governments, though. I think that's generally a terrible idea.
Looking at any organization with power and people involved, perhaps we can use the same ideas that traditionally apply to government in more places. That's all I'm saying.
This is not some mindless intellectual exercise to distract from how things are. In fact, as a proposal for how to reform corporate power, it's the opposite.
(Also, even though I think corporate power should be limited to a very specific arm of society, and that we shouldn't encourage more corporate power, you are greatly exaggerating. Feudalism and the abolition of human rights are not right around the corner.)
Lighten up
I wonder when they will give up on the gaming market because that could take down several publishers and developers. I really don't think it's an if question but a when because it almost feels like an afterthought at this point (they removed the standalone gaming revenue report from the financial reports this summer). Also I don't think AMD and Intel is capable "to step in" to replace them.
AMD would step in, they where behind nvidia but they've been closing that gap for a while and the 9070XT is the current value king (in this fucked up market) for mid-high gaming and you can actually buy them.
Demand for Nvidia cards has outstripped supply even on the mid-high cards specifically because they do better with local models than AMD cards with the same VRAM do broadly.
AI has completely broken the PC gaming market (and PC/Laptop market more broadly but gaming is really hit hard because it's the exact components that matter for both that overlap).
Nvidia had the mind share among gamers but so did Intel once, inertia only lasts so long they've been thoroughly intent on burning that to the ground for a while, back to the post 1080's
If RDNA5 is good (and 4 was it closed the gap on RT) they'll been in a solid place to take the spot if Nvidia do cede the ground.
I have a 7900XTX the last flagship card AMD did (about equal to a 9070XT for raster but 24GB VRAM not 16GB) and it has been and is a stellar card for gaming (let down only if you care about RT and the games I play don't have it).
Flawless under Linux, weaker on the AI behind nvidia but it runs Qwen surprisingly well and I didn't and don't care too much about that except to poke it occasionally.
Cryptocurrency mining had it broken before that. PC gaming's golden years are behind us, as AI, even if not LLMs, seems like it's not going away, so the gaming market is now the niche hobby while AI takes center stage.
RIP. My first gaming ~GPU~ (we called them 3d accelerators back then) was a Diamond Monster II with a 3dfx Voodoo 2 chip.
> Diamond Monster II with a 3dfx Voodoo 2 chip
Oh shit, I had file deleted that. I remember now. I think I used that with the first N64 emulator, I think Project something to play Turok the first person dinosaur hunter game. Life before the internet was better I think.
The Glide API was the only hardware acceleration in there for a good while before it got OpenGL. I remember using a third party wrapper or something to make it work on whatever it was I had in 1999.
AMD and Intel would be HAPPY to replace them, and plenty capable.
Game consoles like PS5 have been AMD for a few generations. Steamdeck/Steam Machine are AMD.
But really, Nvidia has no reason to leave gaming behind. They can just start dialing back their ambition on the gaming side and providing GPUs that aren't too useful for inference or training. All gaming needs is stability so that devs can aim for something. Games looked great 20 years ago and they'll look great 20 years from now, as long as developers know what they are building for.
> Also I don't think AMD and Intel is capable "to step in" to replace them.
It's kind of weird. nVidia kind of has the PC market cornered, but AMD has had the last couple of generations of Xbox and Playstation. Also, they power the Steam Deck/Machine, and Valve has been contributing a lot of AMD graphics features into the Linux drivers. There is a world where AMD (and maybe even Linux on AMD specifically) becomes the de facto standard for gaming.
Well AMD does have the Steam Machine now, they really like doing console integrations. Shame it won't really sell given the pricing situation.
> It's kind of weird. nVidia kind of has the PC market cornered, but AMD has had the last couple of generations of Xbox and Playstation
I have been getting the vibes that Sony is positioning themselves to back out of videogames. I don't think we're going to see a PS6.
There will be a PS6 and a NextBox… it’s just that Sony is cracking down on what some idiot MBAs have decided are “unrealized monitary streams”.
Maybe this is silly of me, but maybe gaming would enjoy an era of hardware upgrades being rather unviable so the focus turns to optimization and aesthetic.
It will be interesting times but I don't think anyone will enjoy a plateau because it's expensive. Even replacing existing hardware is expensive now. Hardly enjoyabe.
A retraction from the gaming market wouldn't necessarily mean they'd still even produce the current crop of products. I could (and I would argue would) involve a complete shuttering of the GeForce brand, halting current production. On the assumption that Intel and AMD would follow, that wouldn't be an end of upgrades, but an end to the market.
Gaming is a larger market than Hollywood. Maybe it becomes a distraction for Nvidia, but someone will step in. That might actually be a good thing and why Nvidia wont do that: it creates an under served market in which newcomers can cut their teeth.
Much of the gaming market is phone games however, not traditional consoles or PCs.
Edit: about half in fact! The rest is shared between the other traditional gaming types.
This feels so insane to me, because I don't think I know anyone who plays phone games
I think we would see an increased push towards cloud gaming. That's probably what Nvidia would want.
It’s not silly. games use far more hardware than they really need. It also pushes out release dates of aggressive console schedules like ps6 because even if it’s a massive upgrade and you have IP locked into your console, no one is going to pay $4000 to play a game like wolverine.
We wouldn't be here today if the video game industry hadn't continued pushing the envelope for decades.
They're advertising DLSS5 just now though.
> I really don't think it's an if question but a when because it almost feels like an afterthought
I feel your reasoning is very weird. Are they losing money by selling consumer GPU? Just because the profit isn't that much compared to AI it doesn't mean that it's negative, and for-profit companies are not known for leaving money on the table. Apple doesn't reveal how much Apple TV+ makes for them either but I don't see it be gone anytime soon.
NVidia is limited by the number of chips they can produce.
If you can fab 1000 chips, and can sell some for $500 and some for $80000 what are you going to do?
The game GPU is at once profitable, but causes them to give up far more profits than they're gaining from it.
They're maintaining the game market to have multiple markets and not go all in, but it's strategic hedging at this point. When NVidia makes a gaming GPU instead of a data center GPU they are leaving money on the table in the short term since they're constrained at the fab level.
This almost seems like we need a strategic reserve for semiconductors- kinda like we have it for food, to prevent suppliers from throwing it away when they are suddenly able to sell something much more lucrative.
The CHIPS Act in the US did set aside a reserve for older processes used for automobile, defense, and industrial semiconductors - but that's not gaming GPUs that used the previous process node.
An alternative example would be Apple & the Mini 12 & 13. They made money on those, too, but here we are.
Nvidia is limited by the number of engineers they have. It might turn out that the AI market is so lucrative that it’s best to reallocate their gaming-focused engineers to AI.
Apple TV is at least a growth market for them, whereas gaming is sort of capped and clearly a tiny piece of nvidia’s revenue atm.
Which is distinctly different both in purpose and technology than the previous versions. DLSS 5 is no longer about frame improvements. It’s about increasing graphical fidelity with active AI rendering.
DLSS 5 is trying to relight and retexture the scene using AI. DLSS 4 is just trying to take a lower quality image and upscale it using AI
Thank for explaining what DLSS5 is! I don't think it has anything to do with "Nvidia is going to give up gaming market" narrative though.
AAA gaming is cooked, and AMD/Intel is plenty able to support indie to AA needs.
That seems a bit overzealous, most consoles[0] run on AMD chips today.
Both PS 5 and Xbox are based on AMD APUs and both serve the AAA market quite well. GTA 6, Assassin's Creed and CoD are probably good enough indicators that the performance is enough, even if there is always room for more (as PC ports show). The PC market will also probably be fine even if stagnation in perfomance gains has been creeping in for a few years now.
[0]: except Nintendo which relies on NVIDIA although their APU there focuses more on efficiency than top performance.
Kushner and MBS bought Electronic Arts for $55 billion, so at least there will be a bailout or instructions to Nvidia to continue gaming cards.
How so? I have a 5070ti but from what I've read the 9070xt keeps up well enough. As long as you don't need the cuda or dlss AMD is a good option
A Chinese company like CXMT will surely fill that unaddressed market the way CXMT is doing for memory. It would be rather unwise for nvidia to leave the door to the market open. AMD is positioned well to grow right now due to their non-Apple sillicon unified ram hardware coming soon.
> I wonder when they will give up on the gaming market because that could take down several publishers and developers.
It really shouldn't: the most money in games are those games that don't require high end GPUs.
Question: wouldn’t the fab be the actual bottleneck? In other words, why wouldn’t TSMC make more NVIDIA chips instead of AMD ones? I assume they’ll just do whichever is paying more, so if NVIDIA chips are better, gamers would be willing to pay more for them and in turn TSMC will be willing to make more of them?
It's still a 15B market for Nvidia, it's not nothing. But I'm curious how they're going to turn Rubin into a gaming GPU. I think at this point consumer GPU upgrades are going to be AI-related upgrades that happen to also help raster capabilities. Blackwell was already kind of a dud on performance uplift from Ada beyond the new LLM features.
That 15B also includes people buying 5090s for local LLMs.
And it’s maybe 5-10% of their revenue at lower profit margins.
Consumer cards just don’t matter very much to nVidia anymore.
In 2020 it was half of their revenue.
Hermes still makes saddles that are somewhat reasonably priced compared to their handbags.
It's still profitable, it's their original raison d'etre, and there's no real reason for them to stop even if it its rounding error on their regular business.
Probably will never, ever see an Nvidia card with >32GB of VRAM though unless they start making dies that lack LLM performance like the gimped ethereum mining cards.
I kind of feel like this topples the house of cards a bit. What else is all the visual genAI tech for besides consumer entertainment?
Fraud?
You must not be paying attention, they already have.
Intel is going nowhere but we all knew that anyways.
And again, you must not be paying attention, AMD is doing exactly what they said they would. No flagship for RDNA4 (just like RDNA2), RDNA5 flagship (10900 XT) coming right on schedule
I think they'll keep the gaming market alive for a while because renting gaming hardware from the cloud (GeForce NOW) is very congruous with AI keeping consumer hardware prices sky-high.
It continues modern trend of chow companies don't want consumers to truly own anything. Finance a car, pay a monthly subscription fee for heated seats, rent a phone, stream a movie, get rid of physical media, rent a GPU.
But if GeForce NOW doesn't take off, and they get convinced that the AI bubble will not pop, I could see them pulling a Micron and ending their consumer product lines.
Eh, AMD makes a lot of video game console SoCs. If you look at the steam hardware survey most people are running ancient computers.
> I really don't think it's an if question but a when
Nvidia still will ship gaming products. The upcoming RTX Spark laptop APUs are still gaming-capable - we also have Blackwell gaming GPUs and the Nvidia-powered Nintendo Switch 2.
People echoed this sentiment during the crypto mining crunch, and we still got gaming hardware designs after that blew over. One of CUDA's core value props is the consumer market, and Nvidia probably won't surrender it unless hardware becomes unreasonably scarce.
I don't think crypto is comparable. They barely made some dedicated crypto GPUs that market was always fickle due to ASCIs. Look at the nvidia revenue breakdown chart, the AI boom looks quite different.
https://ourworldindata.org/data-insights/nvidias-revenue-fro...
> that market was always fickle due to ASCIs.
The GPU crunch came because cards like the 3060 were extremely cheap and could outrun most sub-$1000 ASICs at the time. The dedicated crypto GPUs were too-little too-late; hundreds of thousands of ordinary CUDA-capable GPUs had already been repurposed for mining by the time they launched.
> Also I don't think AMD and Intel is capable "to step in" to replace them
AMD has powered 2 generations each of Sony and Xbox consoles, Steam deck and shops a ton of GPUs especially if you count APUs. And then Intel literally ship more GPUs than Nvidia and AMD combined.
The gaming market doesn't need Nvidia. Especially as AAA is cratering.
They may allocate different number of resources every year based on market conditions but they'll never give up on gaming, that would be extremely silly.
Archive: http://archive.today/kt50V
Thank you
>Nvidia’s financial engineering is partly a response to its biggest customers’ transformation into rivals. “Hyperscalers”, tech giants such as Amazon, Google, Meta and Microsoft, account for roughly half of Nvidia’s revenue
Companies just don't want to pay Jensen's tax. Hyperscalers might still pay Jensen's tax for LLM training but for inference. you don't have to. they are also betting on their own chip for training to replace Nvidia.
this is Nvidia panicking and doing vendor fiance to Neoclouds and buying Hugging Face. even none hyperscalers like Meta is betting on its own chip for AI inference.
Cracks are starting to appear. Open Ai and Anthropic are publicly asking for slowdown in AI research. Translation: We see this technology not being any more useful than what it is now, no AGI is coming, and the first one to accept this and slow down the dollar burn rate will incur the wrath of the market. So let's say this big boogey man technology will end all life on earth and we all slow down together. Bonus points if we can lobby for this to be included in the national security bucket. Then US govt can bail us out. Yay!
If Nvidia is the bank, they should be starting to sweat a bit. It's not often companies ask for a voluntary slowdown.
The frontier companies asking for regulation is somehow one of their less insane marketing ploy to date
Nvidia owes us money.
Because vendor financing in tech to keep a bubble going never ended badly.
Sadly it seems like some haven’t watched the end of the last movie on this subject.
It’s Different This Time(tm)
This is a good headline and point.
Realistically, they're worse than a central bank, because they can't exactly expand supply monotonically like a normal central bank. Nor do they realistically control rates.
Okay, but at some point these investments need to start turning profits; the financing NVDA has arranged is temporary, and private credit needs returns at some point. The overinvestment in AI will lead to a downturn in the capital cycle.
Nvidia will quake the stock market.
I'm curious what the actual rate of replacement/useful life of these GPUs running AI inference 24/7 is.
If these cards burn out in less than the ~5 years of depreciation that accounting puts them at, well then there will be problems.
Quite a lot better than people originally thought. There is a LOT of demand for 5+ year old Ampere on the secondary market, and hyperscalers are still holding on to many of them since they can’t acquire new compute fast enough.
https://www.cnbc.com/video/2026/08/24/making-old-gpus-new-ag...
Market correction will happen. Banks go down and close during recessions. Hope these people are wise enough to see through these effects.
Should there be a LIBOR for 1GB VRAM set each morning?
https://www.dramexchange.com/
Considering that LIBOR was discontinued in 2023 after a scandal of manipulation and fraud, YES! It's back to the future.
CME is launching futures contracts for GPU compute: https://www.cmegroup.com/markets/energy/power/compute-future...
The Japanese economy and yen carry trade is a close second.Rising oul prices and reduced output due to the conflicts in the Middle East could filter through to increasing yields on Japanese debt. In turn, the yen interventions have to continue to keep it lower than 160, which seems to be the psychological barrier for the yen carry trade.
I view the market as celestial objects influencing each other through a lattice similar to spacetime
Everything influences each other with varying gravitational pull
At one point the mental model was more like a web, but spacetime with mass matches the model more closely
Money is a lot like mass, it has it's own gravity.
and it doesn’t disappear [1], much like law of conservation
at any given time there is a finite amount of it most easily observed in currencies’ relative price to another currency, and its movement between owners causes distortions in asset prices
most people I’m around and most trading indicator developers I’ve seen don’t seem to conceptualize the flow of value around the economy. so it feels niche and maybe visualizing this mental model can help many people
[1] central banks can functionally delete money in their bond purchase and roll off programs
More like investors/AWS right?
Funding companies under the condition that they use their infra.
Also called: Buying customers.
I thought a central bank would be like: China is the world's largest official creditor and holds the highest foreign exchange reserves.
That's what makes you a naturally forming central bank.
Interesting juxtaposition with Dario's/Anthropic's 'we must pace the frontier' missive today
If you read between the lines, it was mostly about China.
And what is TSMC in this analogy?
> And what is TSMC in this analogy?
The mint?
A platform: aka the foundation under the bank.
The material cement that allows chips to exist above it.
And the platform is made of time: ours.
If TSMC is a platform, then why doesn't it have an App Store?
They do, you only get access to it if you're comissioning chips.
ASML is the foundation. Without them, none of this would exist.
I think we can go even further and say that IBM, Zeiss, several Japanese companies, and even the US Department of Energy are the actual foundation since ASML is largely just an integrator of many different technologies they license from elsewhere.
We can go even further, all of the mining companies! Or even further, and the dead stars that exploded!
“Our age of silicon is collapsing” - Christopher Nolan
Idk… government bonds or something?
What’s upstream of the central bank? …Congress?
The currency press operator and manufacturer.
do they lend money to their customers to buy their own chips?
Being the central bank of anything is certainly not a compliment. I'd take it as an insult
Weird to think we are currently living in the ‘unlimited free Ubers because you recommended a friend or 2’ phase of this new technology.
Imagine if running fable costs you what it actually costs to run fable. A lot of vibe coders (and just proper software engineers) are gonna be very sad if that comes to pass.
This is what I’m not getting - we’re spending more than the global gross sales of all software on the planet on AI, how much more software needs to be sold and for how much more to recoup?
The economics only make sense if they can replace a huge number of workers with AI.
But that reminds me of the story of the union rep telling Ford: “good luck getting your machines to buy your cars”
(Fyi Ford took note and started paying his workers enough that they’d buy his cars)
If Nvidia is a bank, it's an Islamic Bank. They don't take interest (usury), they share profits and risk.
https://en.wikipedia.org/wiki/Islamic_banking_and_finance
Imagine a scenario where the AI bubble bursts and AI companies and neoclouds go bankrupt en masse, and then a huge rebound occurs when AI has a delayed takeoff. Nvidia ends up with a massive amount of compute on its hands from its backstop deals, and it also owns assets from failed companies when profits start to grow. New startups running using Hugging take the place of OpenAI and Anthropic when their compute assets are divided between survivors like Nvidia, Microsoft, Alphabet, Meta.
If/when there is an AI crash, any number of small startups can buy compute for the price of electricity without anyone wanting to buy them. That is when the real innovation happens. The top of the hype cycle is usually more about getting rich quick and buying and shutting down competition.
> They don't take interest (usury)
Nvidia booked $496 million in interest income in Q2 alone [1].
[1] https://www.sec.gov/Archives/edgar/data/1045810/000104581026... page 15
Paywall?
yep, but click on the archiv.ph link. There you can read the article for free.
How many top signals like this article do you need to see before you exit the market?
Let’s look to the past:
https://www.history.com/articles/1929-stock-market-crash-war...
As the saying goes. The market can stay irrational longer than you can stay solvent.
This phrase has reached the point of semantic satiation in my mind.
Maybe THAT'S the real recession indicator.
What indication do you see that we will see a slow down in model capabilities or AI use?
As the breakneck growth slows, the multiples will compress, and the companies will pull back on spending accordingly as they watch their stocks decline and investors demand more conservative spending behavior.
The Internet didn't stop expanding in 2000-2001. Everything got drastically larger over the following two decades. The multiples on earnings did implode for ~15 years however. MSFT stock for one example went nowhere during that time and compressed down to a near single digit PE.
AI will be a minimum of 10x larger in most every regard 20 years out. That has nothing to do with shorter-term multiples given to these companies in relation to the hyper fast growth they have been riding early in the boom.
Where did you exit to?
just cashout at the peak, hedge fund manager probably
hardware? yes Nvidia is software? Google is. AI = Data Data = Google
This reads like a SoftBank slide deck.
Reading this makes me yearn for the ability to downvote comments