I finally got around to watching that a year or two ago. That scene is among several that are just remarkable bits of work, but somehow the whole film itself felt like a slog.
I feel like that's the problem generally with late career Scorsese. The last of his films that felt consistently interesting and free of flab was, to me, Casino, though he got close for me with The Aviator and The Departed.
This is going to end badly — and soon. It’s ironic that catastrophic forgetting (i.e., the inability to perform continuous learning) and hallucinations (i.e., the failure to recognize when a prediction is unfounded) won’t be the causes of the crash, but rather greed and stupidity.
All clearly designed to cash in on the speculative value being given to AI companies.
Strategically I thought OpenAI's deal of getting 10% of AMD for driving their stock price to $600 was a pretty clever way of creating $97 Billion from nothing - effectively paying for the GPUs they'd purchase.
At the same time I would've thought this insider pump and cash-in strategy would be somehow illegal, but I guess anything goes with this administration.
These sorts of shenanigans happen under whatever administration happens to be in office at the end of a long business cycle and near the frothy peak of a hype cycle. And that frothy peak can be quite tall and last for years.
Anecdotally, many people at OpenAI, Nvidia, Oracle, etc., sincerely believe their company's own hype.
They remind me of the story about the oil prospector in Warren Buffett's year-end 1985 letter to Berkshire Hathaway shareholders:
> An oil prospector, moving to his heavenly reward, was met by St. Peter with bad news. “You’re qualified for residence”, said St. Peter, “but, as you can see, the compound reserved for oil men is packed. There’s no way to squeeze you in.” After thinking a moment, the prospector asked if he might say just four words to the present occupants. That seemed harmless to St. Peter, so the prospector cupped his hands and yelled, “Oil discovered in hell.” Immediately the gate to the compound opened and all of the oil men marched out to head for the nether regions. Impressed, St. Peter invited the prospector to move in and make himself comfortable. The prospector paused. “No,” he said, “I think I’ll go along with the rest of the boys. There might be some truth to that rumor after all.”
People are bringing up the AMD deal but isn't that giving it too much credence? The deal hasn't had any material consequences yet apart from stock market fluctuations. The bigger problem for me is that AMD doesn't seem like is going to be a player of note in the AI sphere. So this deal like many other big money AI deals look like optics to me.
> AMD doesn't seem like is going to be a player of note in the AI sphere.
I thought AMD is well positioned in the inference space? They have high VRAM, somewhat high connectivity, already shipping pods that are pretty ok for inference? Training is still dominated by nvda and their CUDA moat, but there's an increased need for scalable inference, and that should fit AMD well. Am I wrong in thinking that?
I notice that when I ask ChatGPT a question the answers I get back seem more verbose than they were a year ago. Where a 1-2 line response would be sufficient GPT delivers a sprawling essay. Overall the value of the answer has probably gone down, but tokens are up.
You can quite easily ask it to summarize the result in a sentence or paragraph. LLMs have no other way to compute than write text and the more text they write the more compute they do. You only care about the final output.
SBF lacked funds to ride it out. If they survived for a bit longer, Bitcoin would have surged and they'd keep going. AMD, Nvidia etc all have income and funds to survive.
This is just about finances unless you're implying there's some crime here too.
Here is money printing scheme that looks to be at work:
Initial situation:
* Big corp M has X$ in cash where X is huge
Big Corp M invest X$ in AI startup O, with a provision that O needs to use most of the money to buy cloud infrastructure from M to power AI models.
End situation:
* Big corp M has X$ wort of shares in O, the value of which will rapidly grow
* Big corp M cloud division has ~X$ in extra revenue
The deal automatically turns X$ into ~2X$ in books.
Rinse and repeat with next round deals and next AI startups. The big corps are reporting increased cloud divisions revenue from AI spent, but it is their own investment money flowing back to cloud divisions.
> The deal automatically turns X$ into ~2X$ in books.
No it doesn't. The revenue it gets back is valued only at a fraction, because it's only worth its profit. Revenue != profit.
And your "the value of which will rapidly grow" is doing all the work here. That's not guaranteed. It might collapse as well.
It's not money printing at all. It's tying up cash long-term in exchange for a much smaller amount of profit short-term. Which is risky but entirely normal.
To be precise I wrote 'in books' which record also revenue, not just profits. Increasing cloud revenue is one of the things that drives big corps share price up, the missing bit is that this revenue comes from big corps themself. The growth investors mantra is that as long as revenue is increasing rapidly, they don't care much about today's profits, this is why so many unprofitable companies have crazy high valuation.
> Two weeks ago, Nvidia Corp. agreed to invest as much as $100 billion in OpenAI to help the leading AI startup fund a data-center buildout so massive it could power a major city. OpenAI in turn committed to filling those sites with millions of Nvidia chips. The arrangement was promptly criticized for its “circular” nature.
That's not a "circular" deal, that's a... regular deal.
I genuinely don't understand the criticism here. Every business deal is -- I'll do something for you, you do something for me, and we'll both be better off. That's the free market.
Regardless of whether AI is a bubble, this is just business as usual.
The other day I was helping somebody choose some funds to invest in. The majority of "global" or "index" funds we looked into either directly or indirectly had the largest share of holdings in Nvidia, Microsoft, Meta, Alphabet, AMD. Often these were about 50-70% of their portfolio. The returns for the past year look great but the crash is going to be devastating and widespread because so many investments link back to this bubble.
There are other companies than the magnificent 7. Other boring companies with boring dividends and solid earnings that have nothing to do with tech. Ulta. Pepsi. Chevron. But you sound pretty smart so I’m sure you had a good reason to invest in non us markets.
There are plenty of non-US index funds, like EU, UK, Japan and others. There are also indices that track smaller companies rather than just the S&P500 or Nasdaq.
Or diversify in both directions - small US, and big and small international funds.
It just we’ve seen the same thing for over four decades. I was getting stock certificates mailed to me from IBM in the 1980s when the savings and loan scandal rocked the market in 87. Then Enron. Then the internet bubble. Then the subprime meltdown. It just happens every decade or so when “too good to be true” runups happen.
Welcome to how economies work. You only worry when you can’t see where the ends of the circle meet up. Ponzi schemes are notable specifically due primarily to the fact that they’re so directional and one-way. If you can’t pinpoint the escapement and see how it’s feeding into the winding of some other spring, walk away.
There is some circular financing going on, but AI accelerationists think this will be offset by demand, value, and adoption in businesses. Hence these deals are warranted for the incoming demand.
I think I saw Altman saying there's a global shortage of compute just now so this may address it. I'm not sure how much is actual user demand though and how much just investors wanting to pile into AI startups.
Welcome to how economies work. You only worry when you can see where the ends of the circle meet up. Ponzi schemes are notable specifically due primarily to the fact that they’re so directional and one-way.
This incest investing is pretty telling. The whole economy is about to get burned- doubly so with DJTs tariffs and bonds not looking like the safe haven they usually are during turbulent times.
“We’re now locked into a particular version of the market and the future where all roads lead to big tech,” says Amba Kak, co-executive director of the AI Now Institute, which studies AI development and policy. Indeed, the success of major stock indexes—and perhaps your 401(k)—is resting on the continued growth of AI: Meta, Amazon, and the chipmakers Nvidia and Broadcom have accounted for 60 percent of the S&P 500’s returns this year."
Upvote because can someone explain to someone as dense as me, whether or not this is likely to make some likely AI bubble worse? Is this just how industry allocates capital?
Part of what's concerning here is that the deals are conditional. OpenAI must meet XYZ conditions before cash/stock/etc is transferred, and the conditions are pretty hard to meet.
The money between OpenAI, Nvidia, Oracle, AMD is not circulating. There is no cashflow, only future commitments that may (and quite likely will) collapse. Yet the stock market & media react as if it's a sure thing. Even in the criticisms of these deals, the hype is affirmed.
This is the same problem as Enron's accounting, minus the fraud. (No need for fraudulent accounting when people simply don't read the fine print.)
The money is allocated by institutional investors. They buy the stock, the companies trade their more valuable stocks amongst themselves for various deals. If institutional investors stop investing then the flow of deals stops & the bubble pops. There is nowhere else the institutional investors could park their money other than tech so I don't think the bubble is going to pop any time soon. But infinite growth is obviously a logical & physical impossibility so eventually there will be a correction but whoever says they know exactly when that will happen is lying & they'd be better off buying lottery tickets to cash in on their ability to predict the future.
Is it worse than the dot-com bubble? I remember everyone and their mom who knew HTML could get hired, and there were way more companies that went IPO during the dot-com bubble, like theglobe.com.
AI is a bubble, but is it worse than the dot-com bubble or the real estate bubble in 2008?
> Is it worse than the dot-com bubble? I remember everyone and their mom who knew HTML could get hired, and there were way more companies that went IPO during the dot-com bubble, like theglobe.com.
IDK, the number of companies and employees might be smaller, but the valuations and comp packages are so much bigger that I suspect there's more money in the bubble this time around, just a bit less spread out.
Personally think this makes a tech bubble a contagion for other parts of the financial industry - especially if institutional investors take the "easy" trade that everyone is doing and add leverage to it.
Now once these folks don't get 800B per year in revenue and the money runs out, all of the banks go as well. But don't worry - they'll get bailed out with our money...
So far OpenAI hoovered - (minus) $7 Billion in first quarter 2025. $11.2 billion OPEX (compute, sales/marketing) in exchange for $4.3bn revenue. I wonder if revenue sharing agreement with Microsoft means Microsoft gets to pay for half of that :-)
The MI450 is the pets.com from the 90s of AI hardware. It's bad.
The bubble now isn't even about LLMs, but has affected the entire stack. Even hardware companies are offering rubbish for the sake of prop'ing up their own valuation.
CUDA had decades of improvements, we may have AGI (tm) earlier than what it will take AMD to fix their broken software.
MI450 doesn't work well at all, OpenAI would rather take the deal and overspend in a project to try to use it for their software stack, which is incompatible with it.
Just for the sake of proping up valuations, with no "market competitiveness" guidance. Just for moving money.
Funnily enough, Google is ahead of all of this, as they can make their on TPUs and have their integrated stack. They don't need to generate fake economic activityu.
On the surface, the circular financing seems worrying but once you dig into it, it seems quite benign, tbh. NVIDIA is handing out GPUs into OpenAI, in return for stock/future revenue, whichever way you want to slice it. OpenAI has wonky unit economics, yes, but they're growing at ~100% YoY, so it all checks out, if you ask me.
The part that I think may have some credence is that it may be a way to prop up huge profit margins that may persist for longer than they would otherwise in lieu of further commoditization and competition.
If NVIDIA has a stake in the company, they are less likely to do something like start brewing inference chips in house with the help of the foundry partner and a provider of vanilla chip designs. The company also gets a huge cash injection that is somewhat contingent on not doing that, and hey, they have a fresh pile of cash for cutting edge chips so whatever, right? My first thought was that these deals had more to do with that than anything else.
That aspect may end up being a bit illusory in the end. But then again, Nvidia has been proven to be quite skilled at building out and defending their ecosystem, sometimes through ruthless means, so maybe it persists (and I'm not sure that's a good thing). China certainly isn't a threat to throw a wrench in this situation... the entire US geopolitical complex will ensure that is the case.
The weird part is the _impression_ that $100B is being paid to a hardware company for chips, when in fact zero dollars are being paid to a hardware company and instead they are relying on public markets to spike the stock price based on potential future orders as if the demand is real.
The only reason this works is if we are the useful idiots buying up the stock.
How are these "circular deals" different than Boeing (and other heavy civilian equipment makers) providing financing on purchases? My point: Is this simply clutching pearls over the AI/LLM bubble, or is there some substance to this concern?
And another: How does it compare against the new breed of Bitcoin treasuries, like Strategy (MSTR.OQ)? As I understand, Strategy uses secondary offerings to continuously see more shares, then uses the capital to buy more Bitcoin.
This happens also with automotive companies. Peugeot has its own bank, which lends money to people buying peugeots. The big difference: is not circular. The bank is a bank, the OEM building cars build cars.
The bank makes money by lending and cashing interest.
The car maker by selling for profit.
The money goes from bank to OEM, and not the way around, no circle.
We are fully in the grifting stage of society. No one wants to work anymore and everyone wants to invest in imaginary outcomes (cryptocoins (most of which no one uses with little to no real use cases) and now AGI (science fiction)). Once the stock market crashes within an already low trust brittle society I imagine that there will be catastrophic outcomes. Maybe I am just a doomer who needs to touch grass but maybe not.
I sense a business opportunity here: betting on stocks! Maybe we can have a central market where people can exchange their money for bets. Maybe even sell their bet off before the due date.
We can call it a "stock market", or maybe even a "stock exchange". I expect I'll be able to get a patent on it if I prefix the name with "online", or suffix it with "on the internet".
https://archive.is/dvKkB
There is a beautiful scene from "The Wolf of Wall Street", in which Matthew McConaughey outshines Leonardo DiCaprio.
https://www.youtube.com/watch?v=Y4iBdIq0aaY
I think it serves as a good commentary.
I finally got around to watching that a year or two ago. That scene is among several that are just remarkable bits of work, but somehow the whole film itself felt like a slog.
I feel like that's the problem generally with late career Scorsese. The last of his films that felt consistently interesting and free of flab was, to me, Casino, though he got close for me with The Aviator and The Departed.
And plop! goes the bubble.
This is going to end badly — and soon. It’s ironic that catastrophic forgetting (i.e., the inability to perform continuous learning) and hallucinations (i.e., the failure to recognize when a prediction is unfounded) won’t be the causes of the crash, but rather greed and stupidity.
Has greed and stupidity ever not been the underlying cause of a financial bubble?
All clearly designed to cash in on the speculative value being given to AI companies.
Strategically I thought OpenAI's deal of getting 10% of AMD for driving their stock price to $600 was a pretty clever way of creating $97 Billion from nothing - effectively paying for the GPUs they'd purchase.
At the same time I would've thought this insider pump and cash-in strategy would be somehow illegal, but I guess anything goes with this administration.
These sorts of shenanigans happen under whatever administration happens to be in office at the end of a long business cycle and near the frothy peak of a hype cycle. And that frothy peak can be quite tall and last for years.
it'd be illegal if they bought on the public market... but they bought directly from AMD, so nothing to see here, move along. music is still playing.
Anecdotally, many people at OpenAI, Nvidia, Oracle, etc., sincerely believe their company's own hype.
They remind me of the story about the oil prospector in Warren Buffett's year-end 1985 letter to Berkshire Hathaway shareholders:
> An oil prospector, moving to his heavenly reward, was met by St. Peter with bad news. “You’re qualified for residence”, said St. Peter, “but, as you can see, the compound reserved for oil men is packed. There’s no way to squeeze you in.” After thinking a moment, the prospector asked if he might say just four words to the present occupants. That seemed harmless to St. Peter, so the prospector cupped his hands and yelled, “Oil discovered in hell.” Immediately the gate to the compound opened and all of the oil men marched out to head for the nether regions. Impressed, St. Peter invited the prospector to move in and make himself comfortable. The prospector paused. “No,” he said, “I think I’ll go along with the rest of the boys. There might be some truth to that rumor after all.”
Source: https://berkshirehathaway.com/letters/1985.html
Feels like some GPD hack where you could make it fake grow by just handing money around in a loop between actors.
A 0.1% tax on all money transfers (or any other transfers of value from entity A to entity B) would end this...
People are bringing up the AMD deal but isn't that giving it too much credence? The deal hasn't had any material consequences yet apart from stock market fluctuations. The bigger problem for me is that AMD doesn't seem like is going to be a player of note in the AI sphere. So this deal like many other big money AI deals look like optics to me.
> AMD doesn't seem like is going to be a player of note in the AI sphere.
I thought AMD is well positioned in the inference space? They have high VRAM, somewhat high connectivity, already shipping pods that are pretty ok for inference? Training is still dominated by nvda and their CUDA moat, but there's an increased need for scalable inference, and that should fit AMD well. Am I wrong in thinking that?
That's what I have read as well but the ocean's distance between Nvidia and AMD in terms of market share of AI chips does not reflect that.
This is fueled by the parabolic token usage. If token usage continues to grow parabolic, then these are the most genius deals made.
If token usage declines because of high prices, look at what happened to SBF.
I notice that when I ask ChatGPT a question the answers I get back seem more verbose than they were a year ago. Where a 1-2 line response would be sufficient GPT delivers a sprawling essay. Overall the value of the answer has probably gone down, but tokens are up.
You can quite easily ask it to summarize the result in a sentence or paragraph. LLMs have no other way to compute than write text and the more text they write the more compute they do. You only care about the final output.
I do. It typically goes on to write a preamble about why it gave a long answer before finally providing a summary. Token stuffing.
> look at what happened to SBF
SBF lacked funds to ride it out. If they survived for a bit longer, Bitcoin would have surged and they'd keep going. AMD, Nvidia etc all have income and funds to survive.
This is just about finances unless you're implying there's some crime here too.
> SBF lacked funds to ride it out. If they survived for a bit longer, Bitcoin would have surged and they'd keep going.
Yes and then maybe he'd never have been arrested and jailed for all his crimes.
What is the economy if not a bunch of much larger overlapping circles of different sizes?
And I guess the smaller the cluster, the larger the bubble?
Here is money printing scheme that looks to be at work:
Initial situation:
* Big corp M has X$ in cash where X is huge
Big Corp M invest X$ in AI startup O, with a provision that O needs to use most of the money to buy cloud infrastructure from M to power AI models.
End situation:
* Big corp M has X$ wort of shares in O, the value of which will rapidly grow
* Big corp M cloud division has ~X$ in extra revenue
The deal automatically turns X$ into ~2X$ in books. Rinse and repeat with next round deals and next AI startups. The big corps are reporting increased cloud divisions revenue from AI spent, but it is their own investment money flowing back to cloud divisions.
> The deal automatically turns X$ into ~2X$ in books.
No it doesn't. The revenue it gets back is valued only at a fraction, because it's only worth its profit. Revenue != profit.
And your "the value of which will rapidly grow" is doing all the work here. That's not guaranteed. It might collapse as well.
It's not money printing at all. It's tying up cash long-term in exchange for a much smaller amount of profit short-term. Which is risky but entirely normal.
Nothing "money printing" about it.
To be precise I wrote 'in books' which record also revenue, not just profits. Increasing cloud revenue is one of the things that drives big corps share price up, the missing bit is that this revenue comes from big corps themself. The growth investors mantra is that as long as revenue is increasing rapidly, they don't care much about today's profits, this is why so many unprofitable companies have crazy high valuation.
> Two weeks ago, Nvidia Corp. agreed to invest as much as $100 billion in OpenAI to help the leading AI startup fund a data-center buildout so massive it could power a major city. OpenAI in turn committed to filling those sites with millions of Nvidia chips. The arrangement was promptly criticized for its “circular” nature.
That's not a "circular" deal, that's a... regular deal.
I genuinely don't understand the criticism here. Every business deal is -- I'll do something for you, you do something for me, and we'll both be better off. That's the free market.
Regardless of whether AI is a bubble, this is just business as usual.
Anyone remember the subprime crisis?
The other day I was helping somebody choose some funds to invest in. The majority of "global" or "index" funds we looked into either directly or indirectly had the largest share of holdings in Nvidia, Microsoft, Meta, Alphabet, AMD. Often these were about 50-70% of their portfolio. The returns for the past year look great but the crash is going to be devastating and widespread because so many investments link back to this bubble.
I completely withdrew from the US market last Nov 5th.
It's plainly obvious where all of this is going.
well, this means you've missed out on everything since then, wouldn't a simple stop order have been a reasonable alternative on this?
the US dollar is down over 10% since they elected a fascist, no stop order for that.
There are other companies than the magnificent 7. Other boring companies with boring dividends and solid earnings that have nothing to do with tech. Ulta. Pepsi. Chevron. But you sound pretty smart so I’m sure you had a good reason to invest in non us markets.
I'm a non-american and I don't trust their government as a steward any longer.
I've diversified into international markets that are more resistant to the whims of a dictator and his feeble attempts at monetary policy.
Also any gains in dollar denominated assets should include the >10% haircut the USD has taken since the fascist coup.
Where did you put your funds? bonds are screwed but I'm thinking gold
There are plenty of non-US index funds, like EU, UK, Japan and others. There are also indices that track smaller companies rather than just the S&P500 or Nasdaq.
Or diversify in both directions - small US, and big and small international funds.
this is going to be bigger, mostly because the leadership vaccume in the US Gov is all too happy for it to get very big before it goes bang.
Why does "Web of circular deals" sounds like a scam?
Because it is.
Every day a new deal involving OpenAI.
Every day one or more article wondering if/when the AI boom will transform into a bursting bubble.
Maybe people are crying wolf or misreading the situation.
My gut feeling is that yes, this is indeed a bubble and behind closed doors the CEOs are in panic mode, weirdly repeating past and well known mistakes
I think you could classify the recent deals as bubble inflating. We are probably a way off from it bursting.
It just we’ve seen the same thing for over four decades. I was getting stock certificates mailed to me from IBM in the 1980s when the savings and loan scandal rocked the market in 87. Then Enron. Then the internet bubble. Then the subprime meltdown. It just happens every decade or so when “too good to be true” runups happen.
“Web of circular deals”
Welcome to how economies work. You only worry when you can’t see where the ends of the circle meet up. Ponzi schemes are notable specifically due primarily to the fact that they’re so directional and one-way. If you can’t pinpoint the escapement and see how it’s feeding into the winding of some other spring, walk away.
There is some circular financing going on, but AI accelerationists think this will be offset by demand, value, and adoption in businesses. Hence these deals are warranted for the incoming demand.
I think I saw Altman saying there's a global shortage of compute just now so this may address it. I'm not sure how much is actual user demand though and how much just investors wanting to pile into AI startups.
“Web of circular deals”
Welcome to how economies work. You only worry when you can see where the ends of the circle meet up. Ponzi schemes are notable specifically due primarily to the fact that they’re so directional and one-way.
This incest investing is pretty telling. The whole economy is about to get burned- doubly so with DJTs tariffs and bonds not looking like the safe haven they usually are during turbulent times.
“We’re now locked into a particular version of the market and the future where all roads lead to big tech,” says Amba Kak, co-executive director of the AI Now Institute, which studies AI development and policy. Indeed, the success of major stock indexes—and perhaps your 401(k)—is resting on the continued growth of AI: Meta, Amazon, and the chipmakers Nvidia and Broadcom have accounted for 60 percent of the S&P 500’s returns this year."
Is this the proverbial writing on the wall then?
Upvote because can someone explain to someone as dense as me, whether or not this is likely to make some likely AI bubble worse? Is this just how industry allocates capital?
Part of what's concerning here is that the deals are conditional. OpenAI must meet XYZ conditions before cash/stock/etc is transferred, and the conditions are pretty hard to meet.
The money between OpenAI, Nvidia, Oracle, AMD is not circulating. There is no cashflow, only future commitments that may (and quite likely will) collapse. Yet the stock market & media react as if it's a sure thing. Even in the criticisms of these deals, the hype is affirmed.
This is the same problem as Enron's accounting, minus the fraud. (No need for fraudulent accounting when people simply don't read the fine print.)
These deals certainly make the bubble look larger because people are double-counting revenue. They also seem to be triggering extreme investor FOMO.
> They also seem to be triggering extreme investor FOMO.
Bubble's only bad if you get out at the wrong time.
Only for you as an individual, from an economic and societal perspective a bursting bubble is never good!
It’s not the bursting that is bad. It’s the bubble that is bad. The bursting is unpleasant, but good.
The money is allocated by institutional investors. They buy the stock, the companies trade their more valuable stocks amongst themselves for various deals. If institutional investors stop investing then the flow of deals stops & the bubble pops. There is nowhere else the institutional investors could park their money other than tech so I don't think the bubble is going to pop any time soon. But infinite growth is obviously a logical & physical impossibility so eventually there will be a correction but whoever says they know exactly when that will happen is lying & they'd be better off buying lottery tickets to cash in on their ability to predict the future.
>AI bubble worse?
Is it worse than the dot-com bubble? I remember everyone and their mom who knew HTML could get hired, and there were way more companies that went IPO during the dot-com bubble, like theglobe.com.
AI is a bubble, but is it worse than the dot-com bubble or the real estate bubble in 2008?
> Is it worse than the dot-com bubble? I remember everyone and their mom who knew HTML could get hired, and there were way more companies that went IPO during the dot-com bubble, like theglobe.com.
IDK, the number of companies and employees might be smaller, but the valuations and comp packages are so much bigger that I suspect there's more money in the bubble this time around, just a bit less spread out.
Personally think this makes a tech bubble a contagion for other parts of the financial industry - especially if institutional investors take the "easy" trade that everyone is doing and add leverage to it.
Now once these folks don't get 800B per year in revenue and the money runs out, all of the banks go as well. But don't worry - they'll get bailed out with our money...
If memory serves, it's shit like this that sank Global Crossing: creative accounting, I think it was called at the time.
clearly openAI is the top dog here who can hoover all the money , but it's not a public company
So far OpenAI hoovered - (minus) $7 Billion in first quarter 2025. $11.2 billion OPEX (compute, sales/marketing) in exchange for $4.3bn revenue. I wonder if revenue sharing agreement with Microsoft means Microsoft gets to pay for half of that :-)
I do wonder a bit about those finances. Losing $7bn on 4bn revenue so let's invest another $1000bn capex in this wonderful business.
The MI450 is the pets.com from the 90s of AI hardware. It's bad.
The bubble now isn't even about LLMs, but has affected the entire stack. Even hardware companies are offering rubbish for the sake of prop'ing up their own valuation.
CUDA had decades of improvements, we may have AGI (tm) earlier than what it will take AMD to fix their broken software.
MI450 doesn't work well at all, OpenAI would rather take the deal and overspend in a project to try to use it for their software stack, which is incompatible with it.
Just for the sake of proping up valuations, with no "market competitiveness" guidance. Just for moving money.
Funnily enough, Google is ahead of all of this, as they can make their on TPUs and have their integrated stack. They don't need to generate fake economic activityu.
HN community predictions so far are
2000s: cloud bubble 2010s: crypto bubble 2020s: AI bubble
Seems modern economy works different. I missed cloud and crypto and dont wana miss AI so im investing about 20% of my income.
Circular trading is a common technique used in Ponzi schemes to create the illusion of trading activity and to attract more investors.
> remains largely unproven as an avenue for profit-making
That seems like a bit of a stretch. How much money is OpenAI making?
Pretty sure they're losing money on the products that they actually sell. All the cash they have is coming from investors (ie hype).
On the surface, the circular financing seems worrying but once you dig into it, it seems quite benign, tbh. NVIDIA is handing out GPUs into OpenAI, in return for stock/future revenue, whichever way you want to slice it. OpenAI has wonky unit economics, yes, but they're growing at ~100% YoY, so it all checks out, if you ask me.
> wonky unit economics, yes, but they're growing at ~100% YoY
Isn't this "losing money on each sale but making it up in volume"? Sounds like I heard that before and it did not turn out well.
The part that I think may have some credence is that it may be a way to prop up huge profit margins that may persist for longer than they would otherwise in lieu of further commoditization and competition.
If NVIDIA has a stake in the company, they are less likely to do something like start brewing inference chips in house with the help of the foundry partner and a provider of vanilla chip designs. The company also gets a huge cash injection that is somewhat contingent on not doing that, and hey, they have a fresh pile of cash for cutting edge chips so whatever, right? My first thought was that these deals had more to do with that than anything else.
That aspect may end up being a bit illusory in the end. But then again, Nvidia has been proven to be quite skilled at building out and defending their ecosystem, sometimes through ruthless means, so maybe it persists (and I'm not sure that's a good thing). China certainly isn't a threat to throw a wrench in this situation... the entire US geopolitical complex will ensure that is the case.
The weird part is the _impression_ that $100B is being paid to a hardware company for chips, when in fact zero dollars are being paid to a hardware company and instead they are relying on public markets to spike the stock price based on potential future orders as if the demand is real.
The only reason this works is if we are the useful idiots buying up the stock.
yes but their growth isn't driven by actual profits. It's a bet.
How are these "circular deals" different than Boeing (and other heavy civilian equipment makers) providing financing on purchases? My point: Is this simply clutching pearls over the AI/LLM bubble, or is there some substance to this concern?
And another: How does it compare against the new breed of Bitcoin treasuries, like Strategy (MSTR.OQ)? As I understand, Strategy uses secondary offerings to continuously see more shares, then uses the capital to buy more Bitcoin.
This happens also with automotive companies. Peugeot has its own bank, which lends money to people buying peugeots. The big difference: is not circular. The bank is a bank, the OEM building cars build cars.
The bank makes money by lending and cashing interest.
The car maker by selling for profit.
The money goes from bank to OEM, and not the way around, no circle.
I do not know the specifics of Boeing though.
Didn't Boeing's deals turn out to be bad?
Cyclical growth
Really? https://news.ycombinator.com/item?id=45511368
My brain now reads Sam's face as an "AI generated" watermark.
We are fully in the grifting stage of society. No one wants to work anymore and everyone wants to invest in imaginary outcomes (cryptocoins (most of which no one uses with little to no real use cases) and now AGI (science fiction)). Once the stock market crashes within an already low trust brittle society I imagine that there will be catastrophic outcomes. Maybe I am just a doomer who needs to touch grass but maybe not.
how long until this thing pops?
my money is on 18 months
18 days
I sense a business opportunity here: betting on stocks! Maybe we can have a central market where people can exchange their money for bets. Maybe even sell their bet off before the due date.
We can call it a "stock market", or maybe even a "stock exchange". I expect I'll be able to get a patent on it if I prefix the name with "online", or suffix it with "on the internet".
Even better, form pairs of stocks and make them fight! Bet on the outcome using any liquid electronic currency, such as CS skins!
How about giving Nouveau some actual support so you're not screwing over everybody using Linux
They are doing that now. https://www.phoronix.com/news/NVK-Vulkan-Red-Hat-NDA-Docs