Market signals on an impending AI bust are broader than just Oracle’s woes.
For example, Amazon just had a challenging bond offering where the market is clearly starting to seriously question the ROI on all this money being pumped into AI buildout. That does not bode well at all for AI-only companies without broader cash flow from other businesses. And when the cash dries up this whole thing comes crashing down like a house of cards.
> Market signals on an impending AI bust are broader than just Oracle’s woes.
It's worse than that - I believe that Oracle is one of the (many) companies right now that, if their AI experimentation fails, will stop the music, and everyone will be running for a chair.
Oracle is one of a few foundational components in the circular-investing group of AI companies. If they fail to make their commitments they're the first domino to fall.
You think the hedge funds selling SPY options don't have this priced in already? Of course, you can still make money on this bet, just like you can win money at a roulette table, but unless you think have some special insight that hedge/quant funds don't have, buying options should be negative EV.
I was at the ophthalmologist for the second time in two weeks - my new prescription wasn't quite right, new lenses should be here this week.
All that to say: I had to move my focus around a bit and re-read "...pumped into AI buildout." several times, because I thought I was reading Ed Zitron :D
A bunch of press on this today you can look up. Demand on the offering was much lower than expected and what materialized in prior rounds. Amazon had to sweeten the deal to get the money loaned.
Low bid to cover ratio - it's rare for bond auctions to out and out fail (that would be fairly disastrous), but you can have an auction where they successfully sell all the bonds they were trying to sell but with much less demand than they were hoping for.
That's not a good sign and it's a blatant red flag for the market
IMHO these signals have more to do with the market than AI. They aren't finding AI to be have less ROI than before - they are requiring higher ROI than before, because there is less money remaining to be invested.
Managing the total amount of money so that investment bubbles peter out before they get excessively big is supposed to be the central bank's job.
> The exit plan is to offload overpriced shares, that they paid billions for, onto the public market. If they don't IPO, those investors get nothing.
I keep seeing these unsubstantiated claims. They’re out to get us and just pump and dump on public markets!
Yet, before they IPO they have to go around and do what? Who sets the IPO price? Who buys the shares? If the shares tank, the valuation of the company goes down and locked up shares lose value. It’s not really in anyone’s interest for IPOs or investments to fail and while pump-and-dump schemes certainly exist they are not the norm. The conspiracy theory level of distrust and cynicism is not healthy and makes one a very poor investor.
If individual investors are buying shares and getting blown up, that’s their problem. Invest and due your own research. Broad market funds exist and have so for decades. Most financial advisors even will put you in to those funds and corporate 401k plans while increasingly allowing for more investment flexibility (freedom is good) default and educate employees by default on target date funds and index funds. There is a wealth of information out there.
My point was that there is no ROI until the investors exit!
IMO, those shares are overpriced even at private investment levels, but my opinion is still irrelevant to the fact that there is no ROI until the investors exit!
With a couple million dollars, you can buy many many articles on the financial times and barron's. With a couple friends, you can get other friends in pension funds to allocate into you. With other friends, you can get beneficial messaging from all sorts of public and private channels. Banks and funds can pump your offerings for something in return if you went to the right bar mitvah. Of course this only lasts for some time, but if Billy the boomer and the Korean teachers pension fund bought in, you are already half way there.
Information is only relevant in the long term, in the short term the stock market is about FRIENDSHIP.
The bond market is measuring the risk of repayment though not the success ROI of the dollars invested by the company (that impacts the stock price but not so much the bond price). The bond markets are hiccuping on AI because there’s growing concern that these loans simply won’t get repaid.
This may be related to the commonly-held fallacy of "cash on the sidelines". Cash is always on the sidelines. Cash is not created or destroyed by buying and selling stocks or bonds. Cash is simply handed from one party to another, but the cash has to be held by somebody.
Without even getting into the practical vs. theoretical of Fed dual mandate (funding deficits), even the most uncharitable take on modern CBs wouldn't suggest this.
Kinda cool to be at a point in the hype cycle where the capital markets are almost exhausted due a to a speculative bubble, pushing up yield demand. Move over tulip mania.
> No of course there isn't enough capital for all of this. Having said that, there is enough capital to do this for a at least a little while longer. -- Gil Luria (Managing Director and Analyst at D.A. Davidson)
When we tried to do a pilot with their cloud we couldn't even sign-up. None of the corporate credit cards were accepted.
In addition to that the form basically only worked in Edge. We emailed support, they changed something on the backend. It still did not work. We gave up.
In retrospective that was a very clear warning sign that their priorities were misguided. I'm glad we did not waste any further time and effort on them.
Oracle Cloud sometimes feels like an elaborate prank that I'm not in on. I know people and companies on AWS (obviously), Azure, Google Cloud, Hetzner, CloudFlare's various PaaS offerings, etc., but I can't name a single thing running on Oracle Cloud. Somebody out there is clearly using but I'll be damned if I know who it is.
I know at least one of those only uses Oracle for internal/HR "cloud" purposes, while their main customer-facing business is on AWS. Not sure about the others, but when I think of a business using "Oracle cloud" I don't interpret it as just their marketing/HR.
Good to know it's not only problematic on the free tier. I wanted to sign up to get the free credits but couldn't finish the setup. I tried again now and it accepted/charged my card ($1 verification test) but then after the account was created it said I need a credit card?
For the longest time they were a piñata for free compute with people making multiple accounts for their free ARM instance, but with the AI crunch they're clamping down.
I'm guessing they don't care if actual business gets caught up in that because from their POV actual business comes from an account manager, and self-serve is just them cargo culting AWS/GCP
True. The linked article's title says that. I wonder if that was a typo by the OP or one of those HN quirks where the title was automatically changed when it shouldn't have been.
This site is shady as hell. You try to decline marketing in their pop-up and it hides maybe a 100 providers and expects you to click each one individually.
Is it me or do none of the AI companies have a "moat" in the Ben Grahmm sense.
I use their services, but I frankly don't care who provides it. I'll chase the chepest/best and have no issue switching from one to another.
The only moat I can see is Microsoft providing its services to companies in its Azure system. Nervous IT departments probably like that it's not leaving their control if Bob in the SAP team spins up some AI crap.
I've been thinking for a while, there's not real winners here except the incumbent technology providers. Hear me out: all models are converging towards the same level, gains are getting smaller and harder to come by. The models are commodities nothing more.
This is the leap, nobody really wants to front a model for someone else. If i build an agent, or a service that requires a model, I'd prefer to push the model onto someone else, preferably at no cost. This is a leap as I'm sure right now, most people / businesses are thinking actually i do want to own / front the model.
However, if you accept the leap the easiest way to do this is to make the model the users problem.
From a business point of view that makes things really easy, from a customer point of view, they simply have to accept whatever their vendor of choice is pushing down their throats.
So as a business I build for whatever model Google makes available to android, and whatever model windows bundles, and whatever model Apple bundles, and, excluding the long tail of Chinese vendors and Linux (sorry, its always left out) and that's it, problem solved, and the customer picks up the tab for the tokens
You may not care, but a lot of people I know care what brand chat bot they use personally,. usually it's tied to trust and reputation more than anything else. People are fickle.
I think anthropic with its enterprise strategy and google
with its integration in everything have a bit of a moat.
But I switched from ChatGPT to Claude 3 months ago because my account was down for like 6 hours. I haven’t used it since. It’s too easy to switch away from chatbots on a whim. There is no moat for that.
Observationally, for people that /aren't/ using models to code but to just do their white-collar job, claude.ai /is/ AI, now. The entire perspective for how to use AI is through claude skills, claude projects, claude cowork, etc. They've massively won the corp buy-in at the moment I believe.
I guess I’m thinking a lot of companies seem to be getting Claude code subscriptions. It usually takes some time and effort for an org to switch away from one solution. In the meantime a lot of workflows get more and more tied to Claude in particular.
It’s not much of a moat, but it’s more than a lot of orgs have.
AWS and Google at least own their own hardware (Trainium and TPUs, respectively). It's a moat in the sense that designing, building, and deploying your own chips at scale is quite a feat and not easily replicated. The vertical integration will allow them to continue to be profitable once the models get good enough and competitors' prices race to the bottom. Google has Gemini; AWS may not deploy its own models (yet?), but that's not necessarily a losing position, as long as the market is able to run models sourced elsewhere on Trainium and the price is right.
Isn't specialized hardware also a big risk? GPUs are more amenable to any big changes that may happen in the next 5, 10 years of AI research. Maybe we won't even be talking about LLMs anymore. Maybe matrix multiplication won't even be the main primitive.
The moat is shifting from technology to access to proprietary training data. It doesn't matter how good your LLM platform is if you don't have good data to feed the training run. Public Internet data and published media is already mined out. Now the frontier LLM vendors have shifted to licensing proprietary data that's locked up behind corporate firewalls, and even hiring human domain experts specifically to create new training content in target verticals. You'll see the effects of this next year, although it might not be obvious to those who mostly only use LLMs for coding tasks in popular programming languages for which there was already a lot of training data.
> Now the frontier LLM vendors have shifted to licensing proprietary data that's locked up behind corporate firewalls, and even hiring human domain experts specifically to create new training content in target verticals.
That's a losing proposition for any token provider - it's expensive and slow, and when you're done everyone with money to rent a last-gen H100 is going to distill your "closed" model anyway.
The adoption of standards like skills and agent setup helps a ton. Nobody wants to be locked into an AI vendor like with cloud systems in general. And companies can't hold on to the #1 spot across multiple areas for very long, so users are even more motivated to move their process and stack between coding tools and AI companies behind them like Claude code.
Google has a bit of a Network Effect going... my vehicle got an OTA update to use Gemini. Between that, search, storage, and the YT Premium bundle it was enough to convince me to float a subscription.
If/when open-weight models do catch up (i.e. become the dominant product in demand), Amazon transitions from a middle-man to the supplier with the best economies of scale.
Great business either way. You could even draw an analogy to Linux/OSS & the origins of AWS. They started as basically an infra middle-man for other people’s technology. But as the core tech commoditized, they transitioned into selling their own higher level services at scale—like Bedrock.
> I use their services, but I frankly don't care who provides it. I'll chase the chepest/best and have no issue switching from one to another.
For the hyperscalers, there is an ease of remaining in the Azure/AWS/GCP fabric from a data provenance perspective, particularly for regulated industries or large, risk-averse enterprises. There's also, of course, a certain network egress tax in most cases.
That's fine, but your inexperience with large companies that are MS's bread and butter doesn't really give you any credibility here. It's the standard for a reason.
AMD is held back by their interconnect and firmware disadvantage compared to nvidia. They’ve been trying really hard to create their own cuda, but rocM and HIP still aren’t very popular especially for research.
And their repeated refusal to either implement CUDA or reimplement everyone's CUDA libraries on their own platform. They say that AMD never misses a chance to miss a chance.
I thought that Nvidia's moat was more in CUDA? Hardware is hard but we've already seen other companies like Google design neural processors with compute efficiency close to Nvidia.
Campaign? They're friends of the administration, and the US is firmly in the kleptocracy stage now (the last wrungs of democracy are about to be undone this Thursday evening).
They'll give a bribe to Trump, they'll offer up 5% of the stock to Chairman Trump as the People's Stock now that the US is basically a bizarre oligarchy form of communism, and Oracle will be declared a state enterprise that cannot lose money.
The super rich simply do not fail, and they utterly control every aspect of the US now, exactly as the people apparently wanted.
Nah, they tried the invulnerable thing. They tried really, really hard to avoid the "chaos in the White House" firings from the first term. Noem wrecked it.
> They'll give a bribe to Trump, they'll offer up 5% of the stock to Chairman Trump as the People's Stock now that the US is basically a bizarre oligarchy form of communism, and Oracle will be declared a state enterprise that cannot lose money.
A little bit dangerous for a US administration (any US administration) to do a bailout of unloved companies just before a midterm.
Not that Trump won't do it, just saying that he'll think twice about it if he wants to hold on to the power that the American people have given him. It's one thing to boast that he can shoot someone in the street and the public won't care, quite another to tell the masses that he's funding their upcoming unemployment using their tax money :-)
The Republican strategy has moved away from swaying public opinion for a while now. Now their strategy is to manipulate voting maps, intimidate voters and suppress votes in areas likely to vote against them.
The Iran war is unpopular, but they did it anyway.
There's still a massive compute crunch, I know the opencode guys had been struggling to get capacity, Claude effectively lowered it's limit till the SpaceX deal, Google is struggling.
Their competitors eat them. I would not be surprised to see Oracle's cloud business get absorbed by IBM or Microsoft. Maybe Amazon. The extra DC capacity is valuable to a couple companies right now.
Whatever happens, I can assure you that the Ellisons will remain multi-billionaires and the American taxpayer will manage to end up poorer, courtesy of their friend in the White House.
There is AI data center overcapacity already. The KOSPI crashed last week, and it's a leading indicator for the cyclical hardware industry. It already had been that indicator in the 2000 bubble.
I don't know what possessed Ellison to ruin a functioning company, but it will be interesting if he gets a margin call for ORCL's other debt exposures, which are Ellison's massive loans against his ORCL stock.
I remember one thing that struck me when skim reading that the first time:
it only "works" if the government actively does everything in its power to support the boom. No restrictions on new power sources, on pylons and transformers, on new factories to make power sources and compute, on data centres.
This was never going to be the world we live in.
Still surprised by the admin actively punishing politically incorrect power supplies (renewables) and then starting a stupid war with Iran, but even without that nonsense, we were never going to see the US do a command economy pivot, and even if we had something would've broken like it usually does with noobs (and even most politicians are noobs) trying a command economy pivot.
The ability of Silicon Valley to hype itself up into a frenzy is unparalleled. Apparently nothing was learned from "blockchain for everything" and "we're going to live in the metaverse".
This is surprising to me. Judging by what appears to be the common sentiment here on HN - which is that AI inference is already profitable, and OpenAI is fairly valued by private markets.
Given that Oracle and Microsoft are major counterparties of OpenAI, it seems odd that their stocks have been performing so poorly recently. Can anyone square this circle for me?
The general fallacy of the “but inference is profitable” argument is that it tends to ignore all the costs of building and training the model. Given the fact that 1) that’s not trivial, and 2) the arms race underway means one can’t stop training, then it ruins the financial picture.
It’s like saying a new apartment building is “profitable” because the monthly income covers the monthly running costs, but ignoring the giant mortgage that covers the cost of building the building. That thinking is a good way to go bankrupt in real estate and a good way to go bankrupt in AI.
> The general fallacy of the “but inference is profitable” argument is that it tends to ignore all the costs of building and training the model. Given the fact that 1) that’s not trivial, and 2) the arms race underway means one can’t stop training, then it ruins the financial picture.
Or that it’s all hearsay and no one has released financials yet?
Well there is clearly also a lot of non-GAAP style “trust us bro” things going on too which generally boil down to “if you ignore all the reasons why we’re not profitable then we’re profitable.” It’s WeWork’s “community adjusted EBITDA” messaging repackaged.
If the company who holds the mortgage wanted to own the building, they would have just bought it themselves. They don't, for whatever reason, so to some extent they have an incentive to help their customer succeed.
That's why it's so hard to get a residential mortgage, for example. It's more of a partnership, with more mutual vulnerability, than most people think. Same thing seems to be true here.
Given what happened with xAI’s excess capacity lease to Anthropic, and Meta’s noises about doing the same, seems likely that the demand for inference will continue to slope upwards for a while. If I’m Oracle, I’m not worried about being able to utilize the data centers I’ve built for some price, almost certainly a profitable one.
I’m guessing, though, that Oracle made their capital investments on assumptions of a higher price & return. Possibly because it wasn’t clear when these decisions were made how much competition OpenAI would have at the frontier.
I don’t think this math is all that hard. Capital markets have everything they need to start to figure it out, most especially a year or two of history to project forward.
Inference might be profitable, but it does not mean the profits of AI datacenters will rise in future. Open weight models and local AI already put the pressure on the AI datacenter profit margins, and local AI is set to become much more efficient in the future.
Market signals on an impending AI bust are broader than just Oracle’s woes.
For example, Amazon just had a challenging bond offering where the market is clearly starting to seriously question the ROI on all this money being pumped into AI buildout. That does not bode well at all for AI-only companies without broader cash flow from other businesses. And when the cash dries up this whole thing comes crashing down like a house of cards.
> Market signals on an impending AI bust are broader than just Oracle’s woes.
It's worse than that - I believe that Oracle is one of the (many) companies right now that, if their AI experimentation fails, will stop the music, and everyone will be running for a chair.
Oracle is one of a few foundational components in the circular-investing group of AI companies. If they fail to make their commitments they're the first domino to fall.
What's the best way to hedge against this, considering many of us have significant savings in the market?
A few puts on SPY dated a year or two out?
>A few puts on SPY dated a year or two out?
You think the hedge funds selling SPY options don't have this priced in already? Of course, you can still make money on this bet, just like you can win money at a roulette table, but unless you think have some special insight that hedge/quant funds don't have, buying options should be negative EV.
Gold maybe? (no investment advice)
What's the best way to hedge against this, considering many of us have significant savings in the market?
I dunno.
"The market can remain irrational longer than you can remain solvent"
I was at the ophthalmologist for the second time in two weeks - my new prescription wasn't quite right, new lenses should be here this week.
All that to say: I had to move my focus around a bit and re-read "...pumped into AI buildout." several times, because I thought I was reading Ed Zitron :D
And none of the major model makers (not counting SpaceX) have IPO'd yet
Google (and to a much lesser degree, Facebook)
Hi there, how do you know Amazon's bond offering was "challenging"? Curious to learn more. Thank you.
A bunch of press on this today you can look up. Demand on the offering was much lower than expected and what materialized in prior rounds. Amazon had to sweeten the deal to get the money loaned.
Low bid to cover ratio - it's rare for bond auctions to out and out fail (that would be fairly disastrous), but you can have an auction where they successfully sell all the bonds they were trying to sell but with much less demand than they were hoping for.
That's not a good sign and it's a blatant red flag for the market
IMHO these signals have more to do with the market than AI. They aren't finding AI to be have less ROI than before - they are requiring higher ROI than before, because there is less money remaining to be invested.
Managing the total amount of money so that investment bubbles peter out before they get excessively big is supposed to be the central bank's job.
> They aren't finding AI to be have less ROI than before - they are requiring higher ROI than before, because there is less money remaining.
What ROI? There was no return, and there currently isn't any return on investment, because those companies did not exit yet!
The exit plan is to offload overpriced shares, that they paid billions for, onto the public market. If they don't IPO, those investors get nothing.
ROI on bank loans to Oracle and corporate bonds. Those will have interest rates and returns.
If Oracle is highly leveraged or betting the farm on AI, then their credit worthiness goes down.
Alternatively, if money floating around to make loans is drying up, companies have to offer better terms to attract the dwindling supply
> The exit plan is to offload overpriced shares, that they paid billions for, onto the public market. If they don't IPO, those investors get nothing.
I keep seeing these unsubstantiated claims. They’re out to get us and just pump and dump on public markets!
Yet, before they IPO they have to go around and do what? Who sets the IPO price? Who buys the shares? If the shares tank, the valuation of the company goes down and locked up shares lose value. It’s not really in anyone’s interest for IPOs or investments to fail and while pump-and-dump schemes certainly exist they are not the norm. The conspiracy theory level of distrust and cynicism is not healthy and makes one a very poor investor.
If individual investors are buying shares and getting blown up, that’s their problem. Invest and due your own research. Broad market funds exist and have so for decades. Most financial advisors even will put you in to those funds and corporate 401k plans while increasingly allowing for more investment flexibility (freedom is good) default and educate employees by default on target date funds and index funds. There is a wealth of information out there.
> If the shares tank, the valuation of the company goes down and locked up shares lose value.
"Oh no, my $10B became $5B!"
They'll still be happy.
> If individual investors are buying shares and getting blown up, that’s their problem.
Having the general populace fleeced by bad actors is everyone's problem, eventually.
My point was that there is no ROI until the investors exit!
IMO, those shares are overpriced even at private investment levels, but my opinion is still irrelevant to the fact that there is no ROI until the investors exit!
With a couple million dollars, you can buy many many articles on the financial times and barron's. With a couple friends, you can get other friends in pension funds to allocate into you. With other friends, you can get beneficial messaging from all sorts of public and private channels. Banks and funds can pump your offerings for something in return if you went to the right bar mitvah. Of course this only lasts for some time, but if Billy the boomer and the Korean teachers pension fund bought in, you are already half way there.
Information is only relevant in the long term, in the short term the stock market is about FRIENDSHIP.
The bond market is measuring the risk of repayment though not the success ROI of the dollars invested by the company (that impacts the stock price but not so much the bond price). The bond markets are hiccuping on AI because there’s growing concern that these loans simply won’t get repaid.
> there is less money remaining.
In what sense?
This may be related to the commonly-held fallacy of "cash on the sidelines". Cash is always on the sidelines. Cash is not created or destroyed by buying and selling stocks or bonds. Cash is simply handed from one party to another, but the cash has to be held by somebody.
Challening bond offerings and higher yields can be a funtion of supply.
Downgrade of credit worthiness is different. That depends on how leveraged the company is
> is supposed to be the central bank's job.
What? No it's not, and never has been.
Without even getting into the practical vs. theoretical of Fed dual mandate (funding deficits), even the most uncharitable take on modern CBs wouldn't suggest this.
Kinda cool to be at a point in the hype cycle where the capital markets are almost exhausted due a to a speculative bubble, pushing up yield demand. Move over tulip mania.
https://en.wikipedia.org/wiki/Tulip_mania
> No of course there isn't enough capital for all of this. Having said that, there is enough capital to do this for a at least a little while longer. -- Gil Luria (Managing Director and Analyst at D.A. Davidson)
And they terminated 30k employees to achieve this?
https://www.forbes.com/sites/jonmarkman/2026/04/06/oracles-m...
When we tried to do a pilot with their cloud we couldn't even sign-up. None of the corporate credit cards were accepted.
In addition to that the form basically only worked in Edge. We emailed support, they changed something on the backend. It still did not work. We gave up.
In retrospective that was a very clear warning sign that their priorities were misguided. I'm glad we did not waste any further time and effort on them.
That is crazy. One of the main rules of business is to always make it as easy as possible for customers to give you money.
Oracle Cloud sometimes feels like an elaborate prank that I'm not in on. I know people and companies on AWS (obviously), Azure, Google Cloud, Hetzner, CloudFlare's various PaaS offerings, etc., but I can't name a single thing running on Oracle Cloud. Somebody out there is clearly using but I'll be damned if I know who it is.
> I can't name a single thing running on Oracle Cloud
CrowdStrike and Uber
> Hetzner
I don't know of any upper market EMEA customers on Hetzner. I've met Scaleway, OVHCloud, and even STACKIT users but never Hetzner.
zoom. uber. airbnb. openai. bunch of banks. samsung, apparently..
I know at least one of those only uses Oracle for internal/HR "cloud" purposes, while their main customer-facing business is on AWS. Not sure about the others, but when I think of a business using "Oracle cloud" I don't interpret it as just their marketing/HR.
They couldn't integrate a payment provider and expect to build out the data centers for AGI?
Uh, good luck guys.
Good to know it's not only problematic on the free tier. I wanted to sign up to get the free credits but couldn't finish the setup. I tried again now and it accepted/charged my card ($1 verification test) but then after the account was created it said I need a credit card?
For the longest time they were a piñata for free compute with people making multiple accounts for their free ARM instance, but with the AI crunch they're clamping down.
I'm guessing they don't care if actual business gets caught up in that because from their POV actual business comes from an account manager, and self-serve is just them cargo culting AWS/GCP
Title is inaccurate. They're BBB- now, not BBB.
True. The linked article's title says that. I wonder if that was a typo by the OP or one of those HN quirks where the title was automatically changed when it shouldn't have been.
I think there's an errant space in between the BBB and the - but yes, the title is wrong with that space
I would say that the more a company still has plenty of old-fashioned intangible positive corporate goodwill, the bigger the notch.
Wouldn't want to be negative at a time like this.
Here's hoping this screws up the collateralization of the Paramount takeover deal, and the whole thing unravels.
I hope not, that would further weigh them down.
S&P Global link: https://www.spglobal.com/ratings/en/regulatory/article/-/vie...
Good. F** oracle.
Bond rating is about financial solvency, not goodness.
Ed Zitron must be feeling quite validated :-)
he is correct on most counts and for the rest I lack the competence to vouch for or denounce his research. a rare sight!
This site is shady as hell. You try to decline marketing in their pop-up and it hides maybe a 100 providers and expects you to click each one individually.
This shady site is an established business created in 1949.[1]
[1] https://en.wikipedia.org/wiki/Heise_Group
So are all the other shady websites.
Is it me or do none of the AI companies have a "moat" in the Ben Grahmm sense.
I use their services, but I frankly don't care who provides it. I'll chase the chepest/best and have no issue switching from one to another.
The only moat I can see is Microsoft providing its services to companies in its Azure system. Nervous IT departments probably like that it's not leaving their control if Bob in the SAP team spins up some AI crap.
I've been thinking for a while, there's not real winners here except the incumbent technology providers. Hear me out: all models are converging towards the same level, gains are getting smaller and harder to come by. The models are commodities nothing more.
This is the leap, nobody really wants to front a model for someone else. If i build an agent, or a service that requires a model, I'd prefer to push the model onto someone else, preferably at no cost. This is a leap as I'm sure right now, most people / businesses are thinking actually i do want to own / front the model.
However, if you accept the leap the easiest way to do this is to make the model the users problem.
From a business point of view that makes things really easy, from a customer point of view, they simply have to accept whatever their vendor of choice is pushing down their throats.
So as a business I build for whatever model Google makes available to android, and whatever model windows bundles, and whatever model Apple bundles, and, excluding the long tail of Chinese vendors and Linux (sorry, its always left out) and that's it, problem solved, and the customer picks up the tab for the tokens
You may not care, but a lot of people I know care what brand chat bot they use personally,. usually it's tied to trust and reputation more than anything else. People are fickle.
I think anthropic with its enterprise strategy and google with its integration in everything have a bit of a moat.
But I switched from ChatGPT to Claude 3 months ago because my account was down for like 6 hours. I haven’t used it since. It’s too easy to switch away from chatbots on a whim. There is no moat for that.
> I think anthropic with its enterprise strategy and google with its integration in everything have a bit of a moat.
But... Anthropic doesn't have a moat. It's clear at this point that SOTA models are not a moat, and Opus 4.6-level (or GLM 5.2) is sufficient.
Google, though... they own the entire vertical, from the semiconductors to the end-user software. They may have a moat.
Observationally, for people that /aren't/ using models to code but to just do their white-collar job, claude.ai /is/ AI, now. The entire perspective for how to use AI is through claude skills, claude projects, claude cowork, etc. They've massively won the corp buy-in at the moment I believe.
I guess I’m thinking a lot of companies seem to be getting Claude code subscriptions. It usually takes some time and effort for an org to switch away from one solution. In the meantime a lot of workflows get more and more tied to Claude in particular.
It’s not much of a moat, but it’s more than a lot of orgs have.
AWS and Google at least own their own hardware (Trainium and TPUs, respectively). It's a moat in the sense that designing, building, and deploying your own chips at scale is quite a feat and not easily replicated. The vertical integration will allow them to continue to be profitable once the models get good enough and competitors' prices race to the bottom. Google has Gemini; AWS may not deploy its own models (yet?), but that's not necessarily a losing position, as long as the market is able to run models sourced elsewhere on Trainium and the price is right.
Isn't specialized hardware also a big risk? GPUs are more amenable to any big changes that may happen in the next 5, 10 years of AI research. Maybe we won't even be talking about LLMs anymore. Maybe matrix multiplication won't even be the main primitive.
The moat is shifting from technology to access to proprietary training data. It doesn't matter how good your LLM platform is if you don't have good data to feed the training run. Public Internet data and published media is already mined out. Now the frontier LLM vendors have shifted to licensing proprietary data that's locked up behind corporate firewalls, and even hiring human domain experts specifically to create new training content in target verticals. You'll see the effects of this next year, although it might not be obvious to those who mostly only use LLMs for coding tasks in popular programming languages for which there was already a lot of training data.
> Now the frontier LLM vendors have shifted to licensing proprietary data that's locked up behind corporate firewalls, and even hiring human domain experts specifically to create new training content in target verticals.
That's a losing proposition for any token provider - it's expensive and slow, and when you're done everyone with money to rent a last-gen H100 is going to distill your "closed" model anyway.
The adoption of standards like skills and agent setup helps a ton. Nobody wants to be locked into an AI vendor like with cloud systems in general. And companies can't hold on to the #1 spot across multiple areas for very long, so users are even more motivated to move their process and stack between coding tools and AI companies behind them like Claude code.
Vendor lock in cannot happen, or you're bankrupt.
Google has a bit of a Network Effect going... my vehicle got an OTA update to use Gemini. Between that, search, storage, and the YT Premium bundle it was enough to convince me to float a subscription.
> my vehicle got an OTA update to use Gemini
G. A. H.
edit: Y'all downvoters want genAI in your cars?!
Amazon Bedrock is probably middlemanning an insane amount of token consumption these days for the same reasons.
Is Bedrock a "middleman?" I believe that they run all inference inside of AWS data centers, on their own infrastructure.
Their new endpoint even promises zero operator access [0]
[0] https://aws.amazon.com/blogs/machine-learning/exploring-the-...
Sure, but fundamentally they’re acting as a distributor of someone else’s product in the form of the frontier models. That’s a classic middle-man.
No value judgement. I think this is a fantastic strategy.
Weights are worth far more than data centers.
Why?
Seems like open weight models keep catching up to state of the art within a few months, at most. Doesn’t seem like much of a moat to me.
If/when open-weight models do catch up (i.e. become the dominant product in demand), Amazon transitions from a middle-man to the supplier with the best economies of scale.
Great business either way. You could even draw an analogy to Linux/OSS & the origins of AWS. They started as basically an infra middle-man for other people’s technology. But as the core tech commoditized, they transitioned into selling their own higher level services at scale—like Bedrock.
> I use their services, but I frankly don't care who provides it. I'll chase the chepest/best and have no issue switching from one to another.
For the hyperscalers, there is an ease of remaining in the Azure/AWS/GCP fabric from a data provenance perspective, particularly for regulated industries or large, risk-averse enterprises. There's also, of course, a certain network egress tax in most cases.
Uhh. I actively and vocally avoid all things Microsoft. I see Microsoft and I immediately think buggy software with zero security.
That's fine, but your inexperience with large companies that are MS's bread and butter doesn't really give you any credibility here. It's the standard for a reason.
Can concur. I hate them with a passion, but corps love them, and I hate to say it... with good reason.
They're the only player in the Identity-Document-Email-VM-Storage space that's even remotely unified.
Nvidia has a moat. Hardware is hard. No one really competes with them for general compute
AMD Instinct is their direct competitor for compute and they are better per dollar, better per watt, and out competing on raw performance.
Only thing holding them back is fab capacity which nVidia keeps buying in bulk to keep them small.
Have you ever actually had anyone work with these chips? Developer ux on amd is terrible.
AMD is held back by their interconnect and firmware disadvantage compared to nvidia. They’ve been trying really hard to create their own cuda, but rocM and HIP still aren’t very popular especially for research.
And their repeated refusal to either implement CUDA or reimplement everyone's CUDA libraries on their own platform. They say that AMD never misses a chance to miss a chance.
I thought that Nvidia's moat was more in CUDA? Hardware is hard but we've already seen other companies like Google design neural processors with compute efficiency close to Nvidia.
General compute is also the worst solution to the problem.
Nvidia's entire business is dependent on Google not being able to make TPUs fast enough.
Oh great, good to know the shovel seller has the market cornered.
Now back to the conversation, do any of the gold miners have a moat? Or is this a race to the bottom?
Wasn't Tesla rated an F while it was in its hyper growth phase?
Most sensible Tesla valuation.
What happens when Oracle can't pay the interest on their loans?
They'll use their purchases of TikTok and Paramount to campaign for a bailout.
Campaign? They're friends of the administration, and the US is firmly in the kleptocracy stage now (the last wrungs of democracy are about to be undone this Thursday evening).
They'll give a bribe to Trump, they'll offer up 5% of the stock to Chairman Trump as the People's Stock now that the US is basically a bizarre oligarchy form of communism, and Oracle will be declared a state enterprise that cannot lose money.
The super rich simply do not fail, and they utterly control every aspect of the US now, exactly as the people apparently wanted.
The administration isn't fully immune to public opinion yet.
Public opinion seems immune to reality though.
If you look back to what he has been able to get away with, and still get re-elected, I'd say he is.
Nah, they tried the invulnerable thing. They tried really, really hard to avoid the "chaos in the White House" firings from the first term. Noem wrecked it.
> They'll give a bribe to Trump, they'll offer up 5% of the stock to Chairman Trump as the People's Stock now that the US is basically a bizarre oligarchy form of communism, and Oracle will be declared a state enterprise that cannot lose money.
A little bit dangerous for a US administration (any US administration) to do a bailout of unloved companies just before a midterm.
Not that Trump won't do it, just saying that he'll think twice about it if he wants to hold on to the power that the American people have given him. It's one thing to boast that he can shoot someone in the street and the public won't care, quite another to tell the masses that he's funding their upcoming unemployment using their tax money :-)
The Republican strategy has moved away from swaying public opinion for a while now. Now their strategy is to manipulate voting maps, intimidate voters and suppress votes in areas likely to vote against them.
The Iran war is unpopular, but they did it anyway.
They can sell the software business to broadcom.
The result would turn into that concentrated evil black lump from Time Bandits
They can rent out their AI infra to The Hyperscalers.
> They can rent out their AI infra to The Hyperscalers.
I can't tell if this is supposed to be sarcasm or not :-/
Aren't all the token providers right now over-provisioned? They aren't trying to use up all their capacity, they're selling it to one another.
Apart from SpaceXAI no?
There's still a massive compute crunch, I know the opencode guys had been struggling to get capacity, Claude effectively lowered it's limit till the SpaceX deal, Google is struggling.
https://x.com/thdxr/status/2024539643673211054
https://www.anthropic.com/news/higher-limits-spacex
https://finance.yahoo.com/technology/ai/articles/ai-demand-o...
I think the hyperscalers are smart enough to not let Oracle be their landlord.
Are they?
Anthropic is renting compute from a competitor, that also is known for their blackhat business practices.
Their competitors eat them. I would not be surprised to see Oracle's cloud business get absorbed by IBM or Microsoft. Maybe Amazon. The extra DC capacity is valuable to a couple companies right now.
They will ask tax payers for a bailout?
The lenders will then just report missed payments as revenue on their books.
Whatever happens, I can assure you that the Ellisons will remain multi-billionaires and the American taxpayer will manage to end up poorer, courtesy of their friend in the White House.
There is AI data center overcapacity already. The KOSPI crashed last week, and it's a leading indicator for the cyclical hardware industry. It already had been that indicator in the 2000 bubble.
I don't know what possessed Ellison to ruin a functioning company, but it will be interesting if he gets a margin call for ORCL's other debt exposures, which are Ellison's massive loans against his ORCL stock.
> I don't know what possessed Ellison to ruin a functioning company,
Same thing that drives all these execs of large companies - naked greed!
"If only we can fire all workers, imagine how profitable we'll be!"
They are attempting to set civilisation on fire with the intention of being on top when they no longer need humans.
Well it seems like he bought the “AGI is 2 years away” line. As did… pretty much everyone in Silicon Valley.
yeah.. https://ai-2027.com/
I remember one thing that struck me when skim reading that the first time:
it only "works" if the government actively does everything in its power to support the boom. No restrictions on new power sources, on pylons and transformers, on new factories to make power sources and compute, on data centres.
This was never going to be the world we live in.
Still surprised by the admin actively punishing politically incorrect power supplies (renewables) and then starting a stupid war with Iran, but even without that nonsense, we were never going to see the US do a command economy pivot, and even if we had something would've broken like it usually does with noobs (and even most politicians are noobs) trying a command economy pivot.
> yeah.. https://ai-2027.com/
That site is too funny :-)
> [mid-2026] But China is falling behind on AI algorithms due to their weaker models.
If/when the AI bubble pops, this website will be really funny. I guess it's already funny. This is what it shows for Apr 2026:
> Reliable Agent copies thinking at 13x human speed
Still waiting for a reliable agent to think at any speed.
The ability of Silicon Valley to hype itself up into a frenzy is unparalleled. Apparently nothing was learned from "blockchain for everything" and "we're going to live in the metaverse".
This is surprising to me. Judging by what appears to be the common sentiment here on HN - which is that AI inference is already profitable, and OpenAI is fairly valued by private markets.
Given that Oracle and Microsoft are major counterparties of OpenAI, it seems odd that their stocks have been performing so poorly recently. Can anyone square this circle for me?
The general fallacy of the “but inference is profitable” argument is that it tends to ignore all the costs of building and training the model. Given the fact that 1) that’s not trivial, and 2) the arms race underway means one can’t stop training, then it ruins the financial picture.
It’s like saying a new apartment building is “profitable” because the monthly income covers the monthly running costs, but ignoring the giant mortgage that covers the cost of building the building. That thinking is a good way to go bankrupt in real estate and a good way to go bankrupt in AI.
> The general fallacy of the “but inference is profitable” argument is that it tends to ignore all the costs of building and training the model. Given the fact that 1) that’s not trivial, and 2) the arms race underway means one can’t stop training, then it ruins the financial picture.
Or that it’s all hearsay and no one has released financials yet?
Well there is clearly also a lot of non-GAAP style “trust us bro” things going on too which generally boil down to “if you ignore all the reasons why we’re not profitable then we’re profitable.” It’s WeWork’s “community adjusted EBITDA” messaging repackaged.
If the company who holds the mortgage wanted to own the building, they would have just bought it themselves. They don't, for whatever reason, so to some extent they have an incentive to help their customer succeed.
That's why it's so hard to get a residential mortgage, for example. It's more of a partnership, with more mutual vulnerability, than most people think. Same thing seems to be true here.
That sentiment only seems to pop up in Anthropic / OAI threads, wonder why
Good question.
Given what happened with xAI’s excess capacity lease to Anthropic, and Meta’s noises about doing the same, seems likely that the demand for inference will continue to slope upwards for a while. If I’m Oracle, I’m not worried about being able to utilize the data centers I’ve built for some price, almost certainly a profitable one.
I’m guessing, though, that Oracle made their capital investments on assumptions of a higher price & return. Possibly because it wasn’t clear when these decisions were made how much competition OpenAI would have at the frontier.
I don’t think this math is all that hard. Capital markets have everything they need to start to figure it out, most especially a year or two of history to project forward.
HN has been split on this question, with both pro and con strongly and vigorously argued.
Inference might be profitable, but it does not mean the profits of AI datacenters will rise in future. Open weight models and local AI already put the pressure on the AI datacenter profit margins, and local AI is set to become much more efficient in the future.
I think those are just the loud minority. I wouldn't be surprised if they're like 20-30% if a poll were made here
in all our hearts they were always rated CCC
I can’t wait for Ai bubble to bust already. Maybe it will happen in October/November like the crypto hype.
Imagine if their acquisition of TikTok had gone through.
Wait, they don't own US TikTok? Who does?
TikTok USDS Join Ventures LLC owns 80%, ByteDance still owns a minority stake.
Oracle holds 15% & is the hosting provider, Silver Lake has a stake, MGX (UAE state backed firm) owns some as well.
But Oracle still manages the content recommendation algorithm and the infrastructure so I'd argue they still have the biggest impact on the platform.
I thought it was only 15% of the company.