OpenAI is generating $13B a year in revenue. Let’s be generous and say $20B. They’ve signed commitments to spend something like $1.4 trillion on compute. An asset that to date has proven to have a hyper-depreciation cycle.
Someone has to come up with $1.4 trillion in actual cash, fast, or this whole thing comes crashing down. Why? At the end of all this circular financing and deals are folks that actually want real cash (eg electricity utilities that aren’t going to accept OpenAI shares for payment).
If the above doesn’t freak you about a bit at how bonkers this whole thing has become then you need a reality check. “Selling ads” on ChatGPT ain’t gonna close that hole.
> Someone has to come up with $1.4 trillion in actual cash, fast, or this whole thing comes crashing down.
These deals aren't for 100% payment up front. The deals also include stock, not just cash. So, no, they do not need to come up with $1.4 trillion in cash quickly.
This AWS deal is spread over 7 years. That's $5.4 billion per year, though I assume it's ramping up over time.
> At the end of all this circular financing and deals are folks that actually want real cash (eg electricity utilities that aren’t going to accept OpenAI shares for payment).
Amazon's cash on hand is on the order of $100 billion. They also have constant revenue coming in. They will not have any problem accepting OpenAI shares and then paying electricity bills with cash.
These deals are also being done in the open with publicly traded companies. Investors can see the balance sheets and react accordingly in the stock price.
Interestingly, it looks like there is a move away from financing these data centers with tech company cash-on-hand and moving to Special Purpose Vehicles over the past 18 months or so. So now there is a lot more debt involved in funding DC's than equity, in ways that are a sudden change to what was largely a funded-by-equity process at the beginning of 2024.
The one I found best documented (1) is a Meta's SPV to fund their Hyperion DC in Louisiana, which is a deal that is 80% financed by private credit firm Blue Owl. There is a lot of financial trickery to getting the SPV to be counted by the ratings agencies as debt belonging to a different entity that does not count against Meta's books but treated by the market as basically something that Meta will back. But xAI's Memphis DC is also a SPV, and Microsoft is doing that as well. I'm not sure about AMZN, but that we're starting to see that from their competitors suggests they will also be going to this way.
I saw this, and honestly, it's kind of silly. We all know what's going on, so why do the credit ratings agencies play dumb to this kind of financial engineering? Why don't they just say "actually no, we all know that's debt and it's owned by Meta so we will consider it when rating their credit."?
If the market collapses I think Meta can technically just walk away and they lose access to those data centers (which they no longer want anyways) and the SPV is stuck holding $X of assets with $>X liabilities and the issues of the credit are on the hook but not Meta.
And investors are fine being on the hook because they get a higher return from the SPV bonds than Meta bonds. (risk adjusted it's probably the same return).
Because, to quote from The Big Short, “if we don’t give them the rating they want they’ll just walk down the street and go to [the other ratings agency].”
>These deals are also being done in the open with publicly traded companies. Investors can see the balance sheets and react accordingly in the stock price.
I'm no expert on the specifics of the circular financing we're seeing here so the rest of what you wrote might be true, but I know enough about how Wall Street and the world in general works to know that closing with this as a defense shows an incredible naivete that makes me question everything else you have said.
Indeed, a comment above linked to Matt Levine's newsletter on the off-books debt that is showing up instead as things like JVs, and here's another Bloomberg Reporter, Carmen Arroyo, covering it from a more journalistic angle:
Exactly. Enron was a publicly traded company doing weird circular financing stuff. It was all in the open for anyone who cared to look. Just no one did until the music stopped...
We’re a bit too far if we assert this. The weird circular Enron stuff wasn’t all in the open, was by wholly owned subsidiaries, and the downfall was massive trading losses that could no longer be hidden by shuttling money to and from subsidiaries at the right time. A hole in a balance sheet is quite different from a purchase done by financing, thus “circular financing” when applied to both means “things we worry about that involve payments between 2 entities”
No need for all that, the idea OpenAI is committed to $1.4 trillion in pay is a Ed Zitron-sourced number where he calculates $400B based on a number he made up for how much a gigawatt costs, and the trillion figure by multiplying further by claiming every deal is for 2026 and will be repeated over next N years.
“OpenAI CEO Sam Altman sounded exasperated when Altimeter Capital founder—and OpenAI shareholder—Brad Gerstner asked him the question that Gerstner said was ‘hanging over the market’: how a company generating $13 billion in revenue this year would pay for the $1.4 trillion in computing capacity that Altman has said the company is on the hook for.
‘Brad, if you want to sell shares, I’ll find you a buyer…I just—enough,’ Altman said on Gerstner’s podcast.”
If OpenAI continues on their current revenue growth trajectory, they should be larger than AWS by 2027. Burning 2x revenue to grow that fast is not really a concern beyond your continued ability to attract financing. Given the trajectory of inference cost, it unlikely that they would fail to reach profitability.
The big question would be how much of this revenue is unjustifiably circular, and how much of it is extractable - but those are questions for when the growth slows. Im certain every supplier has ways to back out of these commitments if the finances look shaky.
It doesn't freak me out and it's actually completely rational. If both OpenAI and AMZN expect real rates to keep rising while inflation spirals out of control, this deal makes a lot of sense for both of them. They're just duration hedging.
> If both OpenAI and AMZN expect real rates to keep rising while inflation spirals out of control, this deal makes a lot of sense for both of them. They're just duration hedging
It can’t be the same hedge on both sides of the trade.
Let's actually be generous and assume that all parties involved did the math and some due diligence and are not just idiots. If we try that approach, what could that plausibly tell us about a situation where OpenAI has struck deals with not one, but basically all the major chip/infra providers?
The history of bubbles strongly suggests this is precisely evidence of a bad decision, not a good one. For a bubble to exist and be sustained everyone needs to get on board with things that wouldn’t normally make any sense.
The $1.4T commitment is spread over multiple years. Let's assume 4 -- then that's $350B/year. Coincidentally, Google had $350B in revenue in 2024 (and projected to be ~$400B in 2025).
It's certainly possible to imagine OpenAI eventually generating far more revenue than Google, even without anything close to AGI. For example, if they were to improve productivity of 10% of the economy by 10% and capture a third of that value for themselves, that would be more than enough. Alternatively, displacing Google as the go-to place for search and selling ads against that would likely generate at least Google levels of revenue. Or some combination of both.
Is this guaranteed to happen? Of course not. But it's not in "bonkers" territory either.
> The $1.4T commitment is spread over multiple years. Let's assume 4
The Amazon deal is actually spread over 7 years. Other deals have different terms, but also spread over multiple years.
Deals like these have cancellation terms. OpenAI could presumably pay a fee and cancel in the future if their projections are too high and they don't need some of the compute from these deals.
The deals also include OpenAI shares. The deals are being made with companies that have sufficient revenue or even cash on hand to buy the compute and electricity.
The claim above that someone needs to come up with $1.4 trillion right now or everything will collapse isn't grounded in any real understanding of these deals. It's just adding up numbers and comparing them to a single annual revenue snapshot.
I don’t think the OP is saying $1.4 trillion cash is needed “right now.” The point being made is simply that with all the circular deals and financing for this to make sense OpenAI does need to generate $1.4 trillion in cash that can eventually work its way through the economy to pay for all of this. Hype and inflated valuations can be built on numbers on paper but real business are built on cash flow. The OP is simply calling out the lack of cash flow.
Even under the most bullish cases for AI the real $ requires here looks iffy at best.
I think we all know that a big part of the angle here is to keep the hype going until there’s a liquidity event, folks will cash out and then at the like they won’t care what happens.
Google is under existential threat. In that case, OpenAI has a very legitimate trillion dollar case for carving out a piece of Google.
Search engines were never a user friendly app to begin with. You had to know how to search well to get comprehensive answers, and the average person is not that scrupulous. Google’s product is inferior, believe it or not. There will be nothing normal about seeing a list of search results pretty soon, so Google literally has a legacy app out in the wild as far as facts are concerned.
So imagine that, Google would have to remove Search as they know it (remove their core business) and standup a app that looks the same as all the new apps.
People might like one AI persona more than others, which means people will seek out all types of new apps. LLMs is the worst thing that could have ever happened to Google quite frankly.
I find it more likely that the entire "second" level of software companies are in OpenAI's cross hairs more so than Google. Salesforce, ServiceNow, Intuit, DocuSign, Adobe, Workday, Atlassian, and countless others are easier to pick off than Google.
Not every kid born in the last five years will know Google as a verb as we do. They’ll be adults in 15 years, which is a paltry investment timeline for the type of Black Swan event we’re talking about, which AI is.
I don’t disagree with you entirely, but I’d argue the second level apps are harder to chase because they get so specialized.
Death of Google (as everyone knows it today) is a tricky one. It seems impossible to believe at this exact moment. It can sit next to IBM in the long run, no shame at all, amazing run.
Very true. I rarely find myself "Googling" anymore. I'd rather just ask ChatGPT. Even if the enshittification (ads, etc.) will happen down the line, at least we'll have an absolutely awesome product (like Google was to Yahoo) for 5-10 years.
OpenAI is at the very least worth at least half as much as Google. I foresee Google becoming like IBM, and these new LLM companies being the new generation of tech companies.
1. There’s no indication that AI is capable of doing so.
2. Outside of software, inventions have to be turned into physical things like power plants. That doesn’t happen overnight and is expensive.
3. The industry is already going through a power revolution in the form of battery + solar and it’s going to take a while for a new technology to climb the learning curve enough to be competitive.
You'll have your shot at shorting oAI soon apparently. I'm in a lot of these threads on the bull side, and I'll say - please be careful if you do, and only short what you can afford to lose. I'm sure the stock will be crazy volatile, but I don't see signs of anything unsustainable in oAI's ops right now, with the sole exception of increasing training spend using investor money. We're not in a good position outside the company to know if that will pay off. The parts we do know about, inference, users, growth, revenue growth and net income, are all generationally significant, and make shorting really risky.
I think the main issue with your theory is that it's $38B in today's dollars. In the 1970s we saw a lot less independence between the Fed and White House and as a consequence severe inflation. Trillions of dollars of liabilities is not going to sound so bad after 4 years of double-digit inflation ...
Also, IIUC the guys in The Big Short would've lost everything if the government stepped in sooner since the banks controlled the price of the CDSs and could've maintained the incorrect price if they had a bunch of extra cash.
> Also, IIUC the guys in The Big Short would've lost everything if the government stepped in sooner since the banks controlled the price of the CDSs and could've maintained the incorrect price if they had a bunch of extra cash.
Yeah. "Markets can remain irrational longer than you can remain solvent."
Why does it matter if everyone else knows or cares?
If you were actually the guys from the big short and you have strong conviction, you should short the market (literally like the guys from big short) and get really rich.
Money is the language they understand, so hit them where it hurts.
People always talk about shorting like it’s an efficient and reliable way to make money being right when everyone else is wrong. But it isn’t.
When you go long, you can still make money by being “sort of right” or “obliquely right” or “somewhat wrong but lucky”or by just collecting dividends if the market stays irrational long enough. If you short something you have to be exactly right (both about what will happen and precisely when) or your money will end up in the hands of the people you’re betting against. It’s not a symmetrical thing you can just switch back and forth on.
> A central theme of the discussion was the staggering demand for computational power. Gerstner highlighted OpenAI’s reported commitment of $1.4 trillion for compute over the next five years, questioning how a company with reported revenues of $13 billion could manage such an outlay.
> Altman pushed back forcefully. “First of all, we’re doing well more revenue than that. Second of all, Brad, if you want to sell your shares, I’ll find you a buyer,” he quipped. He expressed profound confidence in the company’s trajectory. “We do plan for revenue to grow steeply. Revenue is growing steeply. We are taking a forward bet that it’s going to continue to grow.”
This seems to be just the tip of the iceberg; what about the rest?
The main question I have with all of these deals: how much of the deal is OpenAI actually required to buy, versus how much of it is an option for OpenAI to buy? Because if you tot up all of these numbers, it's something like 10× current annual revenue that OpenAI is signing deals for, and if OpenAI is actually committing to all of that spend... there is a serious cash crunch looming. But if OpenAI is merely optioning to spend up to that much, and only has to commit to a tenth of those numbers, well, that's not as threatening to OpenAI as a going concern.
This does all smell a bit like when WeWork was buying up seemingly every available office for rent. When it came time to actually pay for said offices… oops.
> When it came time to actually pay for said offices… oops
I was at WeWork around the time of its downfall. I have a lot of opinions about how that place was ran, but I can assure you pre-pandemic they were buying up every office space because they were filling them with tenants. Not paying for offices was a result of tenants not paying due to the pandemic.
WeWork was taking on long term liability commitments and paying for them with short term revenue commitments. One bad thing and poof. Everybody in the commercial real estate market saw this coming.
OpenAI maybe in the same situation, committed to spending $1.4T while enjoying a good revenue year this year but then One Bad Thing and poof.
And these GPUs aren’t sitting idle either. Nobody is questioning that they “need” the compute, it’s the lack of a viable business model to pay for all this long term that has folks worried.
That’s the same as what happened when WeWork was buying up office space pre-pandemic and then using handwavy nonsense like “Community Adjusted EBITDA” as part of the smoke and mirrors to pretend like there was an actual business there.
The pandemic expedited the pain, but the business model was broken and folks called BS long before Covid hit.
I know that "ads" is the popular assumption for how you're supposed to make money, but most sites don't really make all that much money with ads. Facebook and Google aren't rich from selling ads, they're rich from all of the ad infrastructure and quite frankly their ability to follow you around on the internet.
Besides, how are ads on ChatGPT supposed to work? If some student is asking it to write their paper for them, is ChatGPT going to stop in the middle of it and go "Hey, you know what sounds good right now? A nice bowl of soup..." Although admittedly that would make for some hilarious proof of people using AI for things they shouldn't...
People regularly ask ChatGPT for product recommendations of all kinds, explicit and implicit.
ChatGPT will also probably be selling ad infrastructure to inject ads just like Google injects ads into search. They probably will pay out little to websites that include the “ChatGPT” widget to integrate ChatGPT with their site that also has ads.
Right now the barriers are technical for injecting ads into AI responses.
That directly tanks the entire value of AI though.
As an advanced research engine, knowing it will reliably only recommend you sponsored products means it's worthless - and worse it will be primed to advocate for sponsored products.
Then the whole thing becomes a scam engine, because check out what Facebook ads look like today.
That’s like claiming that “as an advanced search engine, knowing that Google promoted sponsored products and reranks its algorithms accordingly sometimes means it’s worthless”
Regardless of if that’s true, it’s clearly still a huge business opportunity. And you point out Facebook ads are a scam yet they bring in $164B/year and growing. Regardless of value judgement, there’s clearly a lot of money to chase.
OpenAI is positioned to be something like Adwords is/was. Think free LLM for your app/website/store or as the backend for products like character.ai. That will let them vacuum up a ton of user data.
Plus like Google search they have a ton of organic traffic. Chatgpt has replaced Google search as my starting point to investigate anything. Lots of that is related to things where I will eventually spend money
I don't think you'd want to do this at the training weight level. That could lead to wildly inappropriate references to your product, potentially to the determinant of your brand.
"While we're on the topic of self-harm, did you know the ABC Co Truck has the highest safety rating?"
> Facebook and Google aren't rich from selling ads, they're rich from all of the ad infrastructure and quite frankly their ability to follow you around on the internet.
Which then begs the question if I was the user: "Am I being served an ad or is ChatGPT taking in a bunch of product information and giving me an objective answer to what I'm looking for?"
This would create a ton of hesitation to use this for product recommendations if I knew ChatGPT wasn't using its extensive input for products and reviews and coming back with an objective answer for me.
I guess at this point would we even know the difference? Is it possible this is already happening?
I am also in the camp believing they will sell ads the second they find a viable way (churn worth it, base infrastructure for it built, enough people trusting ai with product recommendations...)
I think the queries will fall into profitable (product recommendations) and non profitable (writing an essay or code) just the way they do for Google. Probably former will have a generous free tier and latter will be largely paywalled. I don't know how they'll do that, but I imagine they'll find some way
It's a mass consumer (software) product and they need new revenue venues and ads have a history of working well. Even Spotify, Netflix, Amazon Prime, ... Companies that historically don't have the ad infrastructure of Google or Facebook have increasingly profitable ad tiers
WW was able to fill those offices by charging well below market prices, because they were VC-funded so growth was more important than profitability. OpenAI is doing the same. 800m users, the vast majority of whom are free users who won't convert to paid.
It's actually more like 100x their current revenue; they stated last week[0] that they have spending commitments for $1.4T of compute.
Or, well, they stated that the TCO of the compute they have commitments for is $1.4T, which is a somewhat strange phrasing. I assume it's due to it being a mix of self-owned vs. rental compute, and what they mean is the TCO to OpenAI rather than the TCO to the owner of the compute.
The real revenues of small and mid-sizes tech companies might, though. Which might be funny (in a Road Runner & Wile E. Coyote way) but only after considerable time and distance. Saying that, I think of those guys from The Big Short being reprimanded for celebrating their prediction of a market crash.
The financial impact if the whole AI space loses even 50% of its current "valuation" will be huge. The financial impact of the whole AI space continuing at its current velocity is... More of whatever is going on now?
I'd be happy if the industry/stock market proves me wrong, but I can't see this ending any other way than with a major crash that makes the dot-com boom seem like a minor blimp.
I lived through the first dot com bubble and bust and it was pretty nasty. I was working for a telecom company at the time. The building we were in were chocked full of bright eyed, bushy tailed startups who were pushing the edge on all kinds of things. Less than a year later, they were all gone.
We used to have lunch at the bar across the street and just about once or twice a week for several months, we'd walk in and there would be a table with about 15-20 people sitting around drinking and reminiscing about how they were going to change the world.
A lot of developers I know just completely left the industry and never came back.
If this crash exceeds that one? We're in for some seriously tough times.
I really don't know with this. By that I mean all logic says that, yes this is absolutely a bubble. But the markets have been irrational for a very long time now, just look at Tesla stock and it wild valuations for years now as an example. This could go on for a lot longer than anyone would think reasonable.
Does OpenAI have $38 billion to buy this? Does AWS have sufficient free cash flow to pay for this “infrastructure” investment they speak of?
Recent analysis shows AWS is burning through Amazon’s free cash on AI buildouts which is very concerning if the bubble pops, leaving Amazon holding the bag of invested capital not making returns.
Amazon is a bit late to the party on these headlines, and lots of unanswered questions about what’s really going on here.
oAI is growing rapidly at over $1bn a month in revenues, maybe as much as $2bn if you read between the lines from Sam's interviews. Over 7 years will they see $38bn of revenue demand against inference? ABSOLUTELY. Is it an incredibly good trick to get access to that much infrastructure without having to run a datacenter while you pilot the fastest growth ever consumer tech company? I'd say it is. Others might disagree
i think this shows Amazon filling a need Microsoft couldn't. They probably tried, but decided to use Amazon cloud services after reviewing infrastructure needs.
Everyone is. The grids always had a problem supplying enough power to even modestly sized commercial datacenters. This is what kicked off the trend of companies building their own datacenters in obscure cities that used to house large steel mills or other power hungry businesses.
I remember when everyone was racing to produce "datacenter in a shipping container" solutions. I just laughed because apparently nobody actually bothered to check if you could actually plug it in anywhere.
I don't think that's super surprising though. Nvidia has deals with OpenAI as well, so not using Trainium might have as much to do with OpenAI keeping Nvidia happy, as with Trainium not being on-par.
They just didn’t like the chips is the most logical answer. Particularly given AWS has been doing everything they can to pump up interest, and this huge PR release doesn’t even mention it at all. That omission speaks volumes.
But that feels weird combined with this. You can buy OpenAI API access which is served off of AWS infrastructure, but you can't bill for it through AWS? (I mean, lots of companies work like that. but Microsoft is betting that a lot of people move regular workloads to Azure so they can have centralized billing for inference and their other stuff?)
I would imagine they couldn't offer models through Bedrock. I think this means training and traditional computing workloads for their products (such as the workspaces for Codex cloud)
Only 38 billion dollars? I thought we are in the age of triple digit billion sums by now in the LLM space. What is that paltry change worth for, a couple seconds worth of OpenAI daily expenses?
This looks quite concerning imo. We all know this is a bubble. When the transformer implosion happens, you can be sure that OpenAI will be ground zero. All these investors feeding OpenAI and all these adjacent companies exposing themselves to OpenAI will suffer huge losses. Everyone is chasing growth so hard that they are making questionable choices regarding returns from a far future that may never come. And let's be clear, the future that is going to pay this off is a future where this tech or a direct successor to this tech brings about a level of general learning skills and autonomy that should be pretty close to a third revolution. Anything else is massive loves for all of these companies.
Nah this is a repeat of Google's early days. They built storage at such a scale that it was hard for anyone else to compete in anything that required storage like email.
OpenAI is doing the same with compute. They're going to have more compute than everyone else combined. It will give them the scale and warchest to drive everyone else out. Every AI company is going to end up being a wrapper around them. And OpenAI will slowly take that value too either via acquisition or cloning successful products.
But is OpenAI building that compute or are they renting it?
OpenAI and Anthropic are signing large deals with Google and Amazon for compute resources, but ultimately it means that Google and Amazon will own a ton of compute. Is OpenAI paying Amazon's cap ex just so Amazon can invest and end up owning what OpenAI needs over the long term?
For those paying Google, are they giving Google the money Google needs to further invest in their TPUs giving them a huge advantage?
Practically, it doesn't matter like it didn't matter for Google that storage got many orders of magnitude cheaper. By the time training a novel LLM and serving it to a billion users is trivial in the way that providing 1GB of email storage is today there will be other moats. They'll have decades of user history and a monitization framework that will be hard to overcome.
Google is a viable competitor here.
Everyone else is missing part of the puzzle. They theoretically could compete but they're behind with no obvious way of catching up.
Amazon specifically is in a position similar to where they were with mobile. They put out a competing phone but with no clear advantage it flopped. They could put out their own LLM but they're late. They'd have to put out a product that is better enough to overcome consumer inertia. They have no real edge or advantage over OpenAI/Google to make that happen.
Theoretically they could back a competitor like Anthropic but what's the point? They look like an also ran these days and ultimately who wins doesn't affect Amazon's core businesses.
FB seems to have figured it out finally and their stock took a huge hit for the investment of infra. Also, despite being behind in sota models and huge human capital investments for research, I believe they are benefiting greatly from oai and the likes.
Every image/video/text post on a meta app is essentially subsidized by oai/gemini/anthropic as they are all losing money on inference. Meta is getting more engagement and ad sales through these subsidized genai image content posts.
Long term they need to catch up and training/inference costs need to drop enough such that each genai post costs less than net profit on the ads but they’re in a great position to bridge the gap.
The end of all of this is ad sales. Google and Meta are still the leaders of this. OpenAI needs a social engagement platform or it is only going to take a slice of Google.
Do they? Doesn’t big corporations just buy CoPilot from Microsoft where they already have a license for Office, Teams, GitHub, Visual Studio, Azure etc.?
It's absolutely no longer about the data. We produce millions of new humans a year who wind up better at reasoning then these models but don't need to read the entire contents of the Internet to do it.
A relatively localized, limited lived experience apparently conveys a lot that LLM input does not - there's an architecture problem (or a compute constraint).
I started working in 1997 at the height of the dot com bubble. I thought it would go on forever but the second half of 2000 and 2001 was rough.
I know a lot of people designing AI accelerator chips. Everyone over 45 thinks we are in an AI bubble. It's the younger people that think growth is infinite.
I told them to diversify from their company stock but we'll see if they have listened after the bubble pops
You are stating a lot of things as fact that aren't really supported. We don't know this is a bubble, we don't know that there will be a transformer implosion, whatever that means, we don't know that OpenAI would ground zero if this is a bubble and it pops, etc..
No one ever knows before these things happen. These predictions are obviously always conjecture, they can’t be stated as fact, ever— at best you can give some supporting evidence often based on similar prior art
There’s been some buzz around the official opening of the Grand Egyptian Museum, which I visited last month. That project took 1.1 to 1.2B USD. Double its original budget estimate but still the museum looks fantastic and it feels, tangibly, like it’s worth a billion.
In contrast with all the money spent on AI, it just feels like monopoly money. Where’s the monument to its success? We could’ve built flying cars or been back to the moon with this much money.
I've been using AI and so did many people I knew. It's resulted in tangible difference in my life.
It's much less likely that I'd drive a flying car and there is 0 chance that I would be the one going to the moon if we spent the equivalent money on those things instead.
Me too, once you’ve had a lot of practice with it (like anything) and know how to mitigate some of its weaknesses, then it’s a superpower.
I currently pay 200 USD a month for AI, and my company pays about 1,200 USD for all employees to use it essentially unlimited - and I get AT LEAST 5x the return on value on that, I would happy multiply all those numbers by 5 and still pay it.
Domain knowledge, bug fixing, writing tests, fixing tests, spotting what’s incomplete, help visualising results, security review generation for human interpretation, writing boilerplate, and simpler refactors
It can’t do all of these things end to end itself, but with the right prompting and guidance holy smokes does it multiply my positive traits as a developer
> and I get AT LEAST 5x the return on value on that
You make $800 by paying OpenAI $200? Can you please explain how your the value put in is 5x and how I can start making $800 more a month?
> holy smokes does it multiply my positive traits as a developer
But it’s not you doing the work. And by your own admission, anyone can eventually figure it out. So if anything you’ve lost traits and handed them to the Llm. As an employee you’re less entrenched and more replaceable.
No they just had a great quarterly earnings report that day. Try to put the absolute minimal effort into checking the obvious before you revert to conspiracy.
I wouldn't be surprised if a few weeks later, Amazon was investing $38B in NVIDIA for new server processors... like nothing to see here folks, totally not a giant circular bubble...
We all know this is going to pop at some point but I personally think Amazon would be a good investment for the next 6 months or so. They just did layoffs, can ride the OpenAI hypetrain, and they beat recent earnings. If they beat next earnings with a lower headcount and mention "AI" a couple times, I think their stock price easily goes to $300.
Not financial advice, obviously, but that's my personal outlook. I've said it before: Alphabet is probably the safest play long term as they haven't been infected by any NVIDIA or OpenAI deals (yet)
OpenAI is an existential threat to Google. They're both solving the same problem for consumers. Something like "there's information on the Internet that I want but don't know where it is". Almost all of Google's value derives from being the default solution to that problem. They can't give up that space to OpenAI. And, at least IMHO, it's obvious LLMs are the future in this area.
The other side to that coin is monetization. Google is dominant there as well. OpenAI can't yield that space to Google because it's how the value is extracted from the consumer.
At this point these announcements are more PR posturing than anything that makes financial sense for shareholders.
I do worry what the other side of this looks like when the circular feedback loop driving hype up eventually reverses and drives things down with amplifying effect.
The negative drumbeat has started, and the narrative is changing to become critical. Every new fad reaches a saturation point, and people's attention spans are short. Already, I see folks piling onto the quantum bandwagon as if LLMs are passe.
Here, the clouds have pulled a trick to inflate their revenues with their own cashflows, and have not been punished yet for it by shareholders - except meta which is getting asked some difficult questions.
Came to post something with a similar sentiment. It's going to have broad market effects when the circle dealing doesn't work out as intended. OpenAI is losing so much money every quarter that they will undoubtedly have to raise prices I would think and people may not be willing to pay them.
If it is a bubble, why are huge companies like Amazon and Microsoft feeding it? What do they get by doing this nonstop? All I hear is that this is a bubble on the internet, yet big companies are making these huge deals. Don't they see what we, the uspeasants, see with their highly paid finance departments?
Big companies are made out of people, and those people are just as fallible as everyone else… and more so, in fact. They’re prone to group-think and suppressing bad news.
The existence proof is, well, every financial crisis ever. Start with the housing bubble and ask, why did huge banks whose entire job was financial modeling get caught up in it? What makes companies today immune to those same types of decision-making errors?
https://archive.is/6DtPq
OpenAI is generating $13B a year in revenue. Let’s be generous and say $20B. They’ve signed commitments to spend something like $1.4 trillion on compute. An asset that to date has proven to have a hyper-depreciation cycle.
Someone has to come up with $1.4 trillion in actual cash, fast, or this whole thing comes crashing down. Why? At the end of all this circular financing and deals are folks that actually want real cash (eg electricity utilities that aren’t going to accept OpenAI shares for payment).
If the above doesn’t freak you about a bit at how bonkers this whole thing has become then you need a reality check. “Selling ads” on ChatGPT ain’t gonna close that hole.
> Someone has to come up with $1.4 trillion in actual cash, fast, or this whole thing comes crashing down.
These deals aren't for 100% payment up front. The deals also include stock, not just cash. So, no, they do not need to come up with $1.4 trillion in cash quickly.
This AWS deal is spread over 7 years. That's $5.4 billion per year, though I assume it's ramping up over time.
> At the end of all this circular financing and deals are folks that actually want real cash (eg electricity utilities that aren’t going to accept OpenAI shares for payment).
Amazon's cash on hand is on the order of $100 billion. They also have constant revenue coming in. They will not have any problem accepting OpenAI shares and then paying electricity bills with cash.
These deals are also being done in the open with publicly traded companies. Investors can see the balance sheets and react accordingly in the stock price.
Interestingly, it looks like there is a move away from financing these data centers with tech company cash-on-hand and moving to Special Purpose Vehicles over the past 18 months or so. So now there is a lot more debt involved in funding DC's than equity, in ways that are a sudden change to what was largely a funded-by-equity process at the beginning of 2024.
The one I found best documented (1) is a Meta's SPV to fund their Hyperion DC in Louisiana, which is a deal that is 80% financed by private credit firm Blue Owl. There is a lot of financial trickery to getting the SPV to be counted by the ratings agencies as debt belonging to a different entity that does not count against Meta's books but treated by the market as basically something that Meta will back. But xAI's Memphis DC is also a SPV, and Microsoft is doing that as well. I'm not sure about AMZN, but that we're starting to see that from their competitors suggests they will also be going to this way.
1: By the invaluable Matt Levine, here: https://www.bloomberg.com/opinion/newsletters/2025-10-29/put... but the other major companies have their own SPV's
I saw this, and honestly, it's kind of silly. We all know what's going on, so why do the credit ratings agencies play dumb to this kind of financial engineering? Why don't they just say "actually no, we all know that's debt and it's owned by Meta so we will consider it when rating their credit."?
IIUC, they ignore it because they're supposed to.
If the market collapses I think Meta can technically just walk away and they lose access to those data centers (which they no longer want anyways) and the SPV is stuck holding $X of assets with $>X liabilities and the issues of the credit are on the hook but not Meta.
And investors are fine being on the hook because they get a higher return from the SPV bonds than Meta bonds. (risk adjusted it's probably the same return).
Because in credit ratings game, the customer is paying to get their bonds rated. Therefore the customer is always right.
Because, to quote from The Big Short, “if we don’t give them the rating they want they’ll just walk down the street and go to [the other ratings agency].”
Does that make any sense? No.
>These deals are also being done in the open with publicly traded companies. Investors can see the balance sheets and react accordingly in the stock price.
I'm no expert on the specifics of the circular financing we're seeing here so the rest of what you wrote might be true, but I know enough about how Wall Street and the world in general works to know that closing with this as a defense shows an incredible naivete that makes me question everything else you have said.
Indeed, a comment above linked to Matt Levine's newsletter on the off-books debt that is showing up instead as things like JVs, and here's another Bloomberg Reporter, Carmen Arroyo, covering it from a more journalistic angle:
https://www.bloomberg.com/news/articles/2025-10-31/meta-xai-...
Exactly. Enron was a publicly traded company doing weird circular financing stuff. It was all in the open for anyone who cared to look. Just no one did until the music stopped...
We’re a bit too far if we assert this. The weird circular Enron stuff wasn’t all in the open, was by wholly owned subsidiaries, and the downfall was massive trading losses that could no longer be hidden by shuttling money to and from subsidiaries at the right time. A hole in a balance sheet is quite different from a purchase done by financing, thus “circular financing” when applied to both means “things we worry about that involve payments between 2 entities”
No need for all that, the idea OpenAI is committed to $1.4 trillion in pay is a Ed Zitron-sourced number where he calculates $400B based on a number he made up for how much a gigawatt costs, and the trillion figure by multiplying further by claiming every deal is for 2026 and will be repeated over next N years.
“OpenAI CEO Sam Altman sounded exasperated when Altimeter Capital founder—and OpenAI shareholder—Brad Gerstner asked him the question that Gerstner said was ‘hanging over the market’: how a company generating $13 billion in revenue this year would pay for the $1.4 trillion in computing capacity that Altman has said the company is on the hook for.
‘Brad, if you want to sell shares, I’ll find you a buyer…I just—enough,’ Altman said on Gerstner’s podcast.”
https://www.theinformation.com/articles/ilya-saw-mira-murati...
If OpenAI continues on their current revenue growth trajectory, they should be larger than AWS by 2027. Burning 2x revenue to grow that fast is not really a concern beyond your continued ability to attract financing. Given the trajectory of inference cost, it unlikely that they would fail to reach profitability.
The big question would be how much of this revenue is unjustifiably circular, and how much of it is extractable - but those are questions for when the growth slows. Im certain every supplier has ways to back out of these commitments if the finances look shaky.
https://xkcd.com/605/
It doesn't freak me out and it's actually completely rational. If both OpenAI and AMZN expect real rates to keep rising while inflation spirals out of control, this deal makes a lot of sense for both of them. They're just duration hedging.
> If both OpenAI and AMZN expect real rates to keep rising while inflation spirals out of control, this deal makes a lot of sense for both of them. They're just duration hedging
It can’t be the same hedge on both sides of the trade.
The obvious answer is that they are going to IPO
I hope so just so I can watch the funny line graph of people burning money.
This circular game is wholly dependent on OpenAI's ability to access public funds via IPO.
The proportion of the utilities involved are a fraction of 1.4T.
Let's actually be generous and assume that all parties involved did the math and some due diligence and are not just idiots. If we try that approach, what could that plausibly tell us about a situation where OpenAI has struck deals with not one, but basically all the major chip/infra providers?
> Let's actually be generous and assume that all parties involved did the math and some due diligence and are not just idiots
Economic history strongly suggests this would be a bad assumption.
The history of bubbles strongly suggests this is precisely evidence of a bad decision, not a good one. For a bubble to exist and be sustained everyone needs to get on board with things that wouldn’t normally make any sense.
The $1.4T commitment is spread over multiple years. Let's assume 4 -- then that's $350B/year. Coincidentally, Google had $350B in revenue in 2024 (and projected to be ~$400B in 2025).
It's certainly possible to imagine OpenAI eventually generating far more revenue than Google, even without anything close to AGI. For example, if they were to improve productivity of 10% of the economy by 10% and capture a third of that value for themselves, that would be more than enough. Alternatively, displacing Google as the go-to place for search and selling ads against that would likely generate at least Google levels of revenue. Or some combination of both.
Is this guaranteed to happen? Of course not. But it's not in "bonkers" territory either.
> The $1.4T commitment is spread over multiple years. Let's assume 4
The Amazon deal is actually spread over 7 years. Other deals have different terms, but also spread over multiple years.
Deals like these have cancellation terms. OpenAI could presumably pay a fee and cancel in the future if their projections are too high and they don't need some of the compute from these deals.
The deals also include OpenAI shares. The deals are being made with companies that have sufficient revenue or even cash on hand to buy the compute and electricity.
The claim above that someone needs to come up with $1.4 trillion right now or everything will collapse isn't grounded in any real understanding of these deals. It's just adding up numbers and comparing them to a single annual revenue snapshot.
I don’t think the OP is saying $1.4 trillion cash is needed “right now.” The point being made is simply that with all the circular deals and financing for this to make sense OpenAI does need to generate $1.4 trillion in cash that can eventually work its way through the economy to pay for all of this. Hype and inflated valuations can be built on numbers on paper but real business are built on cash flow. The OP is simply calling out the lack of cash flow.
Even under the most bullish cases for AI the real $ requires here looks iffy at best.
I think we all know that a big part of the angle here is to keep the hype going until there’s a liquidity event, folks will cash out and then at the like they won’t care what happens.
> if they were to improve productivity of 10% of the economy by 10% and capture a third of that value for themselves, that would be more than enough
This is “if we get 1% of the market” logic.
Google is under existential threat. In that case, OpenAI has a very legitimate trillion dollar case for carving out a piece of Google.
Search engines were never a user friendly app to begin with. You had to know how to search well to get comprehensive answers, and the average person is not that scrupulous. Google’s product is inferior, believe it or not. There will be nothing normal about seeing a list of search results pretty soon, so Google literally has a legacy app out in the wild as far as facts are concerned.
So imagine that, Google would have to remove Search as they know it (remove their core business) and standup a app that looks the same as all the new apps.
People might like one AI persona more than others, which means people will seek out all types of new apps. LLMs is the worst thing that could have ever happened to Google quite frankly.
I find it more likely that the entire "second" level of software companies are in OpenAI's cross hairs more so than Google. Salesforce, ServiceNow, Intuit, DocuSign, Adobe, Workday, Atlassian, and countless others are easier to pick off than Google.
Not every kid born in the last five years will know Google as a verb as we do. They’ll be adults in 15 years, which is a paltry investment timeline for the type of Black Swan event we’re talking about, which AI is.
I don’t disagree with you entirely, but I’d argue the second level apps are harder to chase because they get so specialized.
Death of Google (as everyone knows it today) is a tricky one. It seems impossible to believe at this exact moment. It can sit next to IBM in the long run, no shame at all, amazing run.
Very true. I rarely find myself "Googling" anymore. I'd rather just ask ChatGPT. Even if the enshittification (ads, etc.) will happen down the line, at least we'll have an absolutely awesome product (like Google was to Yahoo) for 5-10 years.
OpenAI is at the very least worth at least half as much as Google. I foresee Google becoming like IBM, and these new LLM companies being the new generation of tech companies.
token usage is growing exponential at all providers.
it will grow even more with the next generation of models.
[dead]
> electricity utilities that aren’t going to accept OpenAI shares for payment
What if AI invents fusion power?
(Thanks for the downvotes I wanted to keep my karma at 69)
1. There’s no indication that AI is capable of doing so.
2. Outside of software, inventions have to be turned into physical things like power plants. That doesn’t happen overnight and is expensive.
3. The industry is already going through a power revolution in the form of battery + solar and it’s going to take a while for a new technology to climb the learning curve enough to be competitive.
4. What if AI gives us all a pony?
What if ChatGPT invents the Matrix? Electricity problem solved.
> Thanks for the downvotes
“Please don't comment about the voting on comments. It never does any good, and it makes boring reading.”
https://news.ycombinator.com/newsguidelines.html
Is there a reason you’re posting so often on this thread? Everyone gets your point.
Fair enough. I guess I’m just like those guys at the investors conference in The Big Short and can’t believe what I’m seeing.
You'll have your shot at shorting oAI soon apparently. I'm in a lot of these threads on the bull side, and I'll say - please be careful if you do, and only short what you can afford to lose. I'm sure the stock will be crazy volatile, but I don't see signs of anything unsustainable in oAI's ops right now, with the sole exception of increasing training spend using investor money. We're not in a good position outside the company to know if that will pay off. The parts we do know about, inference, users, growth, revenue growth and net income, are all generationally significant, and make shorting really risky.
I think the main issue with your theory is that it's $38B in today's dollars. In the 1970s we saw a lot less independence between the Fed and White House and as a consequence severe inflation. Trillions of dollars of liabilities is not going to sound so bad after 4 years of double-digit inflation ...
Also, IIUC the guys in The Big Short would've lost everything if the government stepped in sooner since the banks controlled the price of the CDSs and could've maintained the incorrect price if they had a bunch of extra cash.
> Also, IIUC the guys in The Big Short would've lost everything if the government stepped in sooner since the banks controlled the price of the CDSs and could've maintained the incorrect price if they had a bunch of extra cash.
Yeah. "Markets can remain irrational longer than you can remain solvent."
https://en.wikipedia.org/wiki/Michael_Burry had an investor panic and nearly lost everything. He was right, but he nearly got the timing wrong.
Why does it matter if everyone else knows or cares?
If you were actually the guys from the big short and you have strong conviction, you should short the market (literally like the guys from big short) and get really rich.
Money is the language they understand, so hit them where it hurts.
People always talk about shorting like it’s an efficient and reliable way to make money being right when everyone else is wrong. But it isn’t.
When you go long, you can still make money by being “sort of right” or “obliquely right” or “somewhat wrong but lucky”or by just collecting dividends if the market stays irrational long enough. If you short something you have to be exactly right (both about what will happen and precisely when) or your money will end up in the hands of the people you’re betting against. It’s not a symmetrical thing you can just switch back and forth on.
did the price of NVIDIA made sense to you 2 years ago, when a lot of people were screaming it's in an obvious bubble?
if no, and you thought it was a bubble, does that price of NVIDIA from 2 years ago (not from today) makes sense to you now?
https://founderboat.com/interviews/2025-11-01-openai-sam-sat...
> A central theme of the discussion was the staggering demand for computational power. Gerstner highlighted OpenAI’s reported commitment of $1.4 trillion for compute over the next five years, questioning how a company with reported revenues of $13 billion could manage such an outlay.
> Altman pushed back forcefully. “First of all, we’re doing well more revenue than that. Second of all, Brad, if you want to sell your shares, I’ll find you a buyer,” he quipped. He expressed profound confidence in the company’s trajectory. “We do plan for revenue to grow steeply. Revenue is growing steeply. We are taking a forward bet that it’s going to continue to grow.”
This seems to be just the tip of the iceberg; what about the rest?
The main question I have with all of these deals: how much of the deal is OpenAI actually required to buy, versus how much of it is an option for OpenAI to buy? Because if you tot up all of these numbers, it's something like 10× current annual revenue that OpenAI is signing deals for, and if OpenAI is actually committing to all of that spend... there is a serious cash crunch looming. But if OpenAI is merely optioning to spend up to that much, and only has to commit to a tenth of those numbers, well, that's not as threatening to OpenAI as a going concern.
This does all smell a bit like when WeWork was buying up seemingly every available office for rent. When it came time to actually pay for said offices… oops.
> When it came time to actually pay for said offices… oops
I was at WeWork around the time of its downfall. I have a lot of opinions about how that place was ran, but I can assure you pre-pandemic they were buying up every office space because they were filling them with tenants. Not paying for offices was a result of tenants not paying due to the pandemic.
WeWork was taking on long term liability commitments and paying for them with short term revenue commitments. One bad thing and poof. Everybody in the commercial real estate market saw this coming.
OpenAI maybe in the same situation, committed to spending $1.4T while enjoying a good revenue year this year but then One Bad Thing and poof.
And these GPUs aren’t sitting idle either. Nobody is questioning that they “need” the compute, it’s the lack of a viable business model to pay for all this long term that has folks worried.
That’s the same as what happened when WeWork was buying up office space pre-pandemic and then using handwavy nonsense like “Community Adjusted EBITDA” as part of the smoke and mirrors to pretend like there was an actual business there.
The pandemic expedited the pain, but the business model was broken and folks called BS long before Covid hit.
>it’s the lack of a viable business model to pay for all this long term that has folks worried.
They're going to sell ads at the moment people are looking to buy stuff. It's the single most viable business model we've ever seen.
I know that "ads" is the popular assumption for how you're supposed to make money, but most sites don't really make all that much money with ads. Facebook and Google aren't rich from selling ads, they're rich from all of the ad infrastructure and quite frankly their ability to follow you around on the internet.
Besides, how are ads on ChatGPT supposed to work? If some student is asking it to write their paper for them, is ChatGPT going to stop in the middle of it and go "Hey, you know what sounds good right now? A nice bowl of soup..." Although admittedly that would make for some hilarious proof of people using AI for things they shouldn't...
People regularly ask ChatGPT for product recommendations of all kinds, explicit and implicit.
ChatGPT will also probably be selling ad infrastructure to inject ads just like Google injects ads into search. They probably will pay out little to websites that include the “ChatGPT” widget to integrate ChatGPT with their site that also has ads.
Right now the barriers are technical for injecting ads into AI responses.
That directly tanks the entire value of AI though.
As an advanced research engine, knowing it will reliably only recommend you sponsored products means it's worthless - and worse it will be primed to advocate for sponsored products.
Then the whole thing becomes a scam engine, because check out what Facebook ads look like today.
That’s like claiming that “as an advanced search engine, knowing that Google promoted sponsored products and reranks its algorithms accordingly sometimes means it’s worthless”
Regardless of if that’s true, it’s clearly still a huge business opportunity. And you point out Facebook ads are a scam yet they bring in $164B/year and growing. Regardless of value judgement, there’s clearly a lot of money to chase.
OpenAI is positioned to be something like Adwords is/was. Think free LLM for your app/website/store or as the backend for products like character.ai. That will let them vacuum up a ton of user data.
Plus like Google search they have a ton of organic traffic. Chatgpt has replaced Google search as my starting point to investigate anything. Lots of that is related to things where I will eventually spend money
> ChatGPT going to stop in the middle of it and go "Hey, you know what sounds good right now? A nice bowl of soup..."
Google/facebook do that today, because the content they're showing is created pre-ad, and the ads have to be injected after the fact.
With AI- the content is being generated in the same place that the ads are being injected, which allows us to be much more subtle about it.
How much do you think a car company would pay for to put special training weight on their marketing materials? I would guess big money
I don't think you'd want to do this at the training weight level. That could lead to wildly inappropriate references to your product, potentially to the determinant of your brand.
"While we're on the topic of self-harm, did you know the ABC Co Truck has the highest safety rating?"
Those ads would also get mixed into content the advertisers would probably not appreciate.
> Facebook and Google aren't rich from selling ads, they're rich from all of the ad infrastructure and quite frankly their ability to follow you around on the internet.
https://openai.com/index/introducing-chatgpt-atlas/
> Besides, how are ads on ChatGPT supposed to work?
"How do I do XYZ?" "Product ABC can do XYZ for you."
Which then begs the question if I was the user: "Am I being served an ad or is ChatGPT taking in a bunch of product information and giving me an objective answer to what I'm looking for?"
This would create a ton of hesitation to use this for product recommendations if I knew ChatGPT wasn't using its extensive input for products and reviews and coming back with an objective answer for me.
I guess at this point would we even know the difference? Is it possible this is already happening?
Would you use an LLM knowing it will never answer honestly and any recommendation is likely a sponsored one?
I am also in the camp believing they will sell ads the second they find a viable way (churn worth it, base infrastructure for it built, enough people trusting ai with product recommendations...)
I think the queries will fall into profitable (product recommendations) and non profitable (writing an essay or code) just the way they do for Google. Probably former will have a generous free tier and latter will be largely paywalled. I don't know how they'll do that, but I imagine they'll find some way
It's a mass consumer (software) product and they need new revenue venues and ads have a history of working well. Even Spotify, Netflix, Amazon Prime, ... Companies that historically don't have the ad infrastructure of Google or Facebook have increasingly profitable ad tiers
I was simply addressing the implication that WeWork was just buying up office space for fun and not paying for it.
WW was able to fill those offices by charging well below market prices, because they were VC-funded so growth was more important than profitability. OpenAI is doing the same. 800m users, the vast majority of whom are free users who won't convert to paid.
won't? i doubt that. can't tell you the last time i paid $20/mo for any non-business sub, but i've paid for some combo of openai+claude for last year
It's actually more like 100x their current revenue; they stated last week[0] that they have spending commitments for $1.4T of compute.
Or, well, they stated that the TCO of the compute they have commitments for is $1.4T, which is a somewhat strange phrasing. I assume it's due to it being a mix of self-owned vs. rental compute, and what they mean is the TCO to OpenAI rather than the TCO to the owner of the compute.
[0] https://x.com/sama/status/1983584366547829073
That’s absolutely insane.
I get that folks are now just engaged in “keeping up with the Jones’” FOMO behavior but none of this is making any sense.
The real massive revenues that the big tech companies are having aren't going to disappear just because OpenAI goes away.
The real revenues of small and mid-sizes tech companies might, though. Which might be funny (in a Road Runner & Wile E. Coyote way) but only after considerable time and distance. Saying that, I think of those guys from The Big Short being reprimanded for celebrating their prediction of a market crash.
The financial impact if the whole AI space loses even 50% of its current "valuation" will be huge. The financial impact of the whole AI space continuing at its current velocity is... More of whatever is going on now?
I went to 10× revenue because I figured the ~$1 trillion was over something like a decade, rather than over a year.
And so the bubble grows.
I'd be happy if the industry/stock market proves me wrong, but I can't see this ending any other way than with a major crash that makes the dot-com boom seem like a minor blimp.
I lived through the first dot com bubble and bust and it was pretty nasty. I was working for a telecom company at the time. The building we were in were chocked full of bright eyed, bushy tailed startups who were pushing the edge on all kinds of things. Less than a year later, they were all gone.
We used to have lunch at the bar across the street and just about once or twice a week for several months, we'd walk in and there would be a table with about 15-20 people sitting around drinking and reminiscing about how they were going to change the world.
A lot of developers I know just completely left the industry and never came back.
If this crash exceeds that one? We're in for some seriously tough times.
I can't see a prediction like this as anything other than a description of what would be maximally emotionally satisfying for you personally.
I really don't know with this. By that I mean all logic says that, yes this is absolutely a bubble. But the markets have been irrational for a very long time now, just look at Tesla stock and it wild valuations for years now as an example. This could go on for a lot longer than anyone would think reasonable.
Doesn’t this mean that some percentage of Microsoft’s investment is being given directly to their competitor? This feels like a bubble moment.
They sure seem to be writing a lot of cheques. Hope they all cash ok because if they don’t it’ll suck the entire tech industry down with it
Those cheques absolutely are going to bounce at some point and will bring the entire tech industry down. At least on the stock market.
Does OpenAI have $38 billion to buy this? Does AWS have sufficient free cash flow to pay for this “infrastructure” investment they speak of?
Recent analysis shows AWS is burning through Amazon’s free cash on AI buildouts which is very concerning if the bubble pops, leaving Amazon holding the bag of invested capital not making returns.
Amazon is a bit late to the party on these headlines, and lots of unanswered questions about what’s really going on here.
No, they don’t have the money at the bottom of their burning pit of other peoples cash.
oAI is growing rapidly at over $1bn a month in revenues, maybe as much as $2bn if you read between the lines from Sam's interviews. Over 7 years will they see $38bn of revenue demand against inference? ABSOLUTELY. Is it an incredibly good trick to get access to that much infrastructure without having to run a datacenter while you pilot the fastest growth ever consumer tech company? I'd say it is. Others might disagree
Amazon is a bit late to the announcements party on this front so this comes across as a bit “hey guys, us too!”
Lots of questions on if this makes sense, and highly likely Amazon never gets $38B cash from OpenAI out of this.
>"hey guys, us too!"
In what context? This isn't fashion, being the 2nd mover has benefits which often outweigh the costs.
i think this shows Amazon filling a need Microsoft couldn't. They probably tried, but decided to use Amazon cloud services after reviewing infrastructure needs.
MSFT is power constrained.[0]
[0] https://www.tomshardware.com/tech-industry/artificial-intell...
Everyone is. The grids always had a problem supplying enough power to even modestly sized commercial datacenters. This is what kicked off the trend of companies building their own datacenters in obscure cities that used to house large steel mills or other power hungry businesses.
I remember when everyone was racing to produce "datacenter in a shipping container" solutions. I just laughed because apparently nobody actually bothered to check if you could actually plug it in anywhere.
Amazon still does this. The customers aren’t saas companies in the bay though. It’s militaries, and they’re sent to FOBs.
"It's only when the tide goes out that you know who's been swimming naked." - Warren Buffett
This bubble is one for the history books !
I don’t know but this feels more and more like jumping by stepping one’s own feet…
What is the end goal? GPT-5 wasn't even a step up from o3.
ChatGPT has 800 million weekly users but only 10 million are paying.
GPT 5 codex is definitely a step up from o3. I could use o3 still for chat, but prefer 5-thinking. Do you still use o3?
Millions of people run out of money? Govt: "Too bad, have fun being homeless"
A few billion dollar businesses run out of money due to negligence and greed? Govt:"THEY ARE JUST WITTLE GUYS WHO NEED HELP"
Reading between the lines of Trainium left out of the announcement says they tried it, weren’t impressed, and wanted NVidia chips instead.
I don't think that's super surprising though. Nvidia has deals with OpenAI as well, so not using Trainium might have as much to do with OpenAI keeping Nvidia happy, as with Trainium not being on-par.
Didn’t stop OpenAI doing a mega deal with AMD.
They just didn’t like the chips is the most logical answer. Particularly given AWS has been doing everything they can to pump up interest, and this huge PR release doesn’t even mention it at all. That omission speaks volumes.
Ah that's true, I missed that AMD announcement.
This is crazy, satya punching the air rn
I don't think they care that much. They just aren't able to keep up with the infra that OpenAI wants.
https://www.tomshardware.com/tech-industry/artificial-intell...
Will OpenAI models be available in Bedrock?
Last week's Microsoft deal said Azure was still exclusive: https://openai.com/index/next-chapter-of-microsoft-openai-pa...
But that feels weird combined with this. You can buy OpenAI API access which is served off of AWS infrastructure, but you can't bill for it through AWS? (I mean, lots of companies work like that. but Microsoft is betting that a lot of people move regular workloads to Azure so they can have centralized billing for inference and their other stuff?)
It says
> Non-API products may be served on any cloud provider.
I am not sure if Bedrock counts. There are 2 OpenAI models already there: https://aws.amazon.com/blogs/aws/openai-open-weight-models-n...
I would imagine they couldn't offer models through Bedrock. I think this means training and traditional computing workloads for their products (such as the workspaces for Codex cloud)
Only 38 billion dollars? I thought we are in the age of triple digit billion sums by now in the LLM space. What is that paltry change worth for, a couple seconds worth of OpenAI daily expenses?
Obviously the plan is to lock Anthropic and others out of AWS.
All financial analysis misses the point. They just need to buy enough time and compute to out-last the competition.
All bets are off if China find a way of reducing processing power by 50% or more.
If only killing off grok and Gemini was so easy...
This looks quite concerning imo. We all know this is a bubble. When the transformer implosion happens, you can be sure that OpenAI will be ground zero. All these investors feeding OpenAI and all these adjacent companies exposing themselves to OpenAI will suffer huge losses. Everyone is chasing growth so hard that they are making questionable choices regarding returns from a far future that may never come. And let's be clear, the future that is going to pay this off is a future where this tech or a direct successor to this tech brings about a level of general learning skills and autonomy that should be pretty close to a third revolution. Anything else is massive loves for all of these companies.
Nah this is a repeat of Google's early days. They built storage at such a scale that it was hard for anyone else to compete in anything that required storage like email.
OpenAI is doing the same with compute. They're going to have more compute than everyone else combined. It will give them the scale and warchest to drive everyone else out. Every AI company is going to end up being a wrapper around them. And OpenAI will slowly take that value too either via acquisition or cloning successful products.
But is OpenAI building that compute or are they renting it?
OpenAI and Anthropic are signing large deals with Google and Amazon for compute resources, but ultimately it means that Google and Amazon will own a ton of compute. Is OpenAI paying Amazon's cap ex just so Amazon can invest and end up owning what OpenAI needs over the long term?
For those paying Google, are they giving Google the money Google needs to further invest in their TPUs giving them a huge advantage?
Practically, it doesn't matter like it didn't matter for Google that storage got many orders of magnitude cheaper. By the time training a novel LLM and serving it to a billion users is trivial in the way that providing 1GB of email storage is today there will be other moats. They'll have decades of user history and a monitization framework that will be hard to overcome.
Google is a viable competitor here.
Everyone else is missing part of the puzzle. They theoretically could compete but they're behind with no obvious way of catching up.
Amazon specifically is in a position similar to where they were with mobile. They put out a competing phone but with no clear advantage it flopped. They could put out their own LLM but they're late. They'd have to put out a product that is better enough to overcome consumer inertia. They have no real edge or advantage over OpenAI/Google to make that happen.
Theoretically they could back a competitor like Anthropic but what's the point? They look like an also ran these days and ultimately who wins doesn't affect Amazon's core businesses.
FB seems to have figured it out finally and their stock took a huge hit for the investment of infra. Also, despite being behind in sota models and huge human capital investments for research, I believe they are benefiting greatly from oai and the likes.
Every image/video/text post on a meta app is essentially subsidized by oai/gemini/anthropic as they are all losing money on inference. Meta is getting more engagement and ad sales through these subsidized genai image content posts.
Long term they need to catch up and training/inference costs need to drop enough such that each genai post costs less than net profit on the ads but they’re in a great position to bridge the gap.
The end of all of this is ad sales. Google and Meta are still the leaders of this. OpenAI needs a social engagement platform or it is only going to take a slice of Google.
How is Anthropic an also-ran when they lead the enterprise market?
Do they? Doesn’t big corporations just buy CoPilot from Microsoft where they already have a license for Office, Teams, GitHub, Visual Studio, Azure etc.?
> OpenAI is doing the same with compute.
No, it’s Amazon that’s doing this. OpenAI is paying Amazon for the compute services, but it’s Amazon that’s building the capacity.
Every AI company is going to end up being a wrapper around them.
the race is for sure on: https://menlovc.com/perspective/2025-mid-year-llm-market-upd...
Google, too, has a lot of compute. Not to mention the chips to power the compute.
And they own the compute, as opposed to renting some of it. And they have the engineers to utilise that compute.
If only everyone in the world had compute in their pockets or on their desk…
seems like a flawed assumption when the cost of tokens -> 0
Pretty sure this "compute is the new oil" thesis fell flat when OAI failed to deliver on GPT-5 hype, and all the disappointments since.
It's still all about the (yet to be collected) data and advancements in architecture, and OAI doesn't have anything substantial there.
It's absolutely no longer about the data. We produce millions of new humans a year who wind up better at reasoning then these models but don't need to read the entire contents of the Internet to do it.
A relatively localized, limited lived experience apparently conveys a lot that LLM input does not - there's an architecture problem (or a compute constraint).
Like in politics, all they care about is getting out before shtf and pass the bag to the next sucker while making $$$ in the meantime
Are we at the Pets.com stage of the bubble yet?
I started working in 1997 at the height of the dot com bubble. I thought it would go on forever but the second half of 2000 and 2001 was rough.
I know a lot of people designing AI accelerator chips. Everyone over 45 thinks we are in an AI bubble. It's the younger people that think growth is infinite.
I told them to diversify from their company stock but we'll see if they have listened after the bubble pops
loves is typo for losses I assume?
You are stating a lot of things as fact that aren't really supported. We don't know this is a bubble, we don't know that there will be a transformer implosion, whatever that means, we don't know that OpenAI would ground zero if this is a bubble and it pops, etc..
No one ever knows before these things happen. These predictions are obviously always conjecture, they can’t be stated as fact, ever— at best you can give some supporting evidence often based on similar prior art
These deal numbers have lost all meaning for me.
There’s been some buzz around the official opening of the Grand Egyptian Museum, which I visited last month. That project took 1.1 to 1.2B USD. Double its original budget estimate but still the museum looks fantastic and it feels, tangibly, like it’s worth a billion.
In contrast with all the money spent on AI, it just feels like monopoly money. Where’s the monument to its success? We could’ve built flying cars or been back to the moon with this much money.
I've been using AI and so did many people I knew. It's resulted in tangible difference in my life.
It's much less likely that I'd drive a flying car and there is 0 chance that I would be the one going to the moon if we spent the equivalent money on those things instead.
Me too, once you’ve had a lot of practice with it (like anything) and know how to mitigate some of its weaknesses, then it’s a superpower.
I currently pay 200 USD a month for AI, and my company pays about 1,200 USD for all employees to use it essentially unlimited - and I get AT LEAST 5x the return on value on that, I would happy multiply all those numbers by 5 and still pay it.
Domain knowledge, bug fixing, writing tests, fixing tests, spotting what’s incomplete, help visualising results, security review generation for human interpretation, writing boilerplate, and simpler refactors
It can’t do all of these things end to end itself, but with the right prompting and guidance holy smokes does it multiply my positive traits as a developer
> I currently pay 200 USD a month for AI
> and I get AT LEAST 5x the return on value on that
You make $800 by paying OpenAI $200? Can you please explain how your the value put in is 5x and how I can start making $800 more a month?
> holy smokes does it multiply my positive traits as a developer
But it’s not you doing the work. And by your own admission, anyone can eventually figure it out. So if anything you’ve lost traits and handed them to the Llm. As an employee you’re less entrenched and more replaceable.
How certain are you that you need to pay $200/mo/seat for this value?
>it feels, tangibly, like it’s worth a billion
Lot of feeling going on in this comment, but that's not really how money works.
What percent of the world is going to set foot in this museum in the next few years? What percent has used or will use AI tools?
I’d rather useful AI tools than a flying car or someone else going to the moon.
But I agree that the numbers are increasingly beyond reasonable comprehension
[dead]
LOL Sam Conman trying to rope everybody in.
Official post: https://openai.com/index/aws-and-openai-partnership/ (https://news.ycombinator.com/item?id=45799021)
Looks like someone knew about this on Thursday, gotta love those over efficient market trends.
Thursday was their last earnings call.
No they just had a great quarterly earnings report that day. Try to put the absolute minimal effort into checking the obvious before you revert to conspiracy.
Great, can I buy hallucinated products now?
How much is Amazon 'investing' in OpenAI for this deal?
I wouldn't be surprised if a few weeks later, Amazon was investing $38B in NVIDIA for new server processors... like nothing to see here folks, totally not a giant circular bubble...
We all know this is going to pop at some point but I personally think Amazon would be a good investment for the next 6 months or so. They just did layoffs, can ride the OpenAI hypetrain, and they beat recent earnings. If they beat next earnings with a lower headcount and mention "AI" a couple times, I think their stock price easily goes to $300.
Not financial advice, obviously, but that's my personal outlook. I've said it before: Alphabet is probably the safest play long term as they haven't been infected by any NVIDIA or OpenAI deals (yet)
OpenAI is an existential threat to Google. They're both solving the same problem for consumers. Something like "there's information on the Internet that I want but don't know where it is". Almost all of Google's value derives from being the default solution to that problem. They can't give up that space to OpenAI. And, at least IMHO, it's obvious LLMs are the future in this area.
The other side to that coin is monetization. Google is dominant there as well. OpenAI can't yield that space to Google because it's how the value is extracted from the consumer.
reverse that. Google is the existential threat to openAI. Google can price tokens to make openAI never profitable.
Google gave birth to openai by trying to bottle up the gpt genie.
At this point these announcements are more PR posturing than anything that makes financial sense for shareholders.
I do worry what the other side of this looks like when the circular feedback loop driving hype up eventually reverses and drives things down with amplifying effect.
The negative drumbeat has started, and the narrative is changing to become critical. Every new fad reaches a saturation point, and people's attention spans are short. Already, I see folks piling onto the quantum bandwagon as if LLMs are passe.
Here, the clouds have pulled a trick to inflate their revenues with their own cashflows, and have not been punished yet for it by shareholders - except meta which is getting asked some difficult questions.
Came to post something with a similar sentiment. It's going to have broad market effects when the circle dealing doesn't work out as intended. OpenAI is losing so much money every quarter that they will undoubtedly have to raise prices I would think and people may not be willing to pay them.
> At this point these announcements are more PR posturing than anything that makes financial sense for shareholders.
corporate would like you to find the difference between these two photos
If it is a bubble, why are huge companies like Amazon and Microsoft feeding it? What do they get by doing this nonstop? All I hear is that this is a bubble on the internet, yet big companies are making these huge deals. Don't they see what we, the uspeasants, see with their highly paid finance departments?
Big companies are made out of people, and those people are just as fallible as everyone else… and more so, in fact. They’re prone to group-think and suppressing bad news.
The existence proof is, well, every financial crisis ever. Start with the housing bubble and ask, why did huge banks whose entire job was financial modeling get caught up in it? What makes companies today immune to those same types of decision-making errors?
It is not a panic if you are the first out…
At this point it is getting beyond parody
[dead]
[dead]
All these stories should have the dollar amounts in quotes as they are farcical.