Privatize the losses. These are the largest companies the world has ever seen, sitting on piles of assets like dragons. They have hyped AI endlessly to the point that everyone else is sick of it. If they haven't saved any money, and can't bail themselves out, they deserve bankruptcy.
Nobody who has ever run a business or knows about how a business runs would think that any business "sits on piles of assets like dragons." Very little cash sits idle, usually for anticipated operational expenditures with a reserve for unanticipated expenditures; the rest is invested, loaned, spent on building the business, or returned to investors. If OpenAI is holding $25 billion in cash, it's because they anticipate spending that much in the near future, not because they're sleeping atop it in a mountain cave.
If you owe the bank $100, that's your problem. If you owe the bank $100 million, that's the bank's problem. If you owe the bank $100 billion, that's country's problem.
Can you privatize the losses if the banks are investing, providing debt to and repackaging that debt into broader instruments?
I worry that one of the reasons that Anthropic has 500 billion in commitments and everyone is running to the credit markets is to make it so that if they fail everyone suffers.
Privatizing the losses doesn't mean the rest of us are going to escape without a scratch. It just means that we let fail the things that are going to fail.
What makes this so politically unpalatable is that having constantly bailed out the last few cycles, everyone is now operating on the assumption that they will be bailed out, so many entities who in a more perfect world would know better have done a lot of stupid things, like pensions that have a lot of bonds that will go bad if the AI bubble collapses.
Frankly we're looking at massive problems and upheaval at this point no matter what we do. But the advantage of not socializing the losses is that the signals of what needs to be changed go directly into the economy. Socializing the losses means that the signals get muted, confused, and mangled and people in the future end up doing the wrong things as a result.
Unfortunately, at this point I think the choice is between people clearly and forthrightly losing massive portions of their retirement as visible losses, or people deceptively losing massive portions of their retirement as the bailout creates massive inflation and their retirement is inflated until it has a fairly similar amount of losses. Unfortunately, it is a feature for the rulers that the bailouts obscure the losses and dampen the correct apportionment of blame, even as it destroys the underlying economy. Giving people clear, correct targets for their rage, the same process of correct pain apportionment the economy needs to become healthy again, would point people at the "wrong" clear, correct targets, and the clear, correct targets are the ones deciding whether or not a bailout occurs. Not just the literal lawmakers nominally making the call, but the entire system that makes the decision.
All the more reason to make them fail. FDIC should pay insurance and otherwise fail the banks. Capitalism cannot work if there isn't feedback that not watching for things like this will hurt you.
I actually liked the approach that happened with several large companies in 2008 - the stockholders get wiped out, the senior management is gone, but the depositors get protected.
Anthropic is not going to fail. Claude does amazing things and all the companies built on top of it are more likely to fail because at the end of the day, we can remake their value proposition by using Claude and a few command line tools…
If Anthropic has made promises that its wallet cannot keep (as it looks like), it will fail. Let me remind you that Anthropic isn't profitable yet, and there's no clear path that it will be.
The IPOs are for "socializing" the losses, in the sense that the losses are dumped on retail investors, who are the general populace. Especially if you consider big pension funds
A lot of comments read this is as fluff. And in some ways it is.
However, looking at the other side, I think what this might do is add some barrier or consideration when a bailout is considered.
The last time we did large scale bailouts (TARP[0], in 2008), It was rushed through and ratified in around two weeks following the sudden collapse of Lehman.
Any measures we can take to ensure that we don't fall into a situation where there is only one choice is sensible.
This doesn’t actually require anything like a law, it’s basically a verbal promise. Or am I misreading?
If AI companies were to go bankrupt, it’s possible they’d be nationalized. I don’t see any other feasible option besides that or a bailout (or one but it’s effectively the other, e.g. “nationalized” but the employees including leaders are the same, or bailed out but already subservient to the military). I think the best case is for the compute to be made effectively public: giving every citizen free access, since we’re paying taxes that prevent their bankruptcy.
You are reading this exactly correct. It's the most government-coded kind of resolution: it sounds impressive, but means nothing. It is kinda nice to get that narrative out there though, we shouldn't bail out companies, we should let them fail and re-organize into companies that can work. "Bankrupt" doesn't necessarily have to mean you disappear and go away, it can mean you shuffle people and equipment around and get a new name and logo that makes more sense.
It’s a resolution, so pure fluff. That said, a prohibition on bailing out this sector in particular might make it possible to claw back anything this regime illegally gives.
There are almost certainly bank execs who are looking at Anthropic's IPO and thinking "if it crashes, we'll get bailed out". Saying "no you won't" now, before the IPO, when it has a chance to affect peoples' behavior, is a worthwhile thing to do.
Then the bill will never become relevant and it’ll just be a waste of time and you can laugh at it in hindsight. And I’m sure none of these companies will lobby against it, since it’s just tilting at windmills and won’t mean anything.
The investment to date has been ridiculous and undisciplined. It’s very hard to understand the valuations of these companies. If this wasn’t an election year I think there would be more oversight and critical review. But we have a very unserious party in charge right now.
It’s good that it’s being discussed. We’re tired of the wealth class privatizing the profits and leaving us to cover the costs.
I have no idea who would they even bailout. Investors? Why? What for? What would that change?
Bailing out banks made some sense so that they couldn't operate. Without the bailout there would be no more banks and a hundred million of people left on the ice.
If you want to keep AI operating you can just (temporarily?) ban investors from tearing down the AI companies they invested in for parts. You could say they are important for national security or something. I'm sure they are already earning enough to keep the lights on. Just not enout to give investors payback for insate capital expediture to set up.
Such ban wolud the governement zero, especially if you write the law well enough, so that it can't be succesfully challenged in court.
I agree with much of your sentiment, but if the lack of bailout (which let me reiterate, I agree with) leads to a recession, it will kill the D chances in 2028. They will apply this calculus to their political behavior, and (as is typical for the Ds) will do nothing.
The Democrats I suspect are planning on doing the exact opposite unfortunately. They were very wimpy on the bubbly ai stuff under Biden, why would they change when they don't get punished from their own constituents for going against their promises. Same party apparatus as a whole that existed under Kamala. We need a Trump style clean out of the Democratic party because the brand is as bad as the Republican brand at this point. I can't with surity say that the Democrats are going to do anything to reign in big tech, all I can hope for is that they will be slightly better than trump or if not that, no worse.
Gosh I really gotta get a real job again. The RLHF gig work has been great while I’ve been in school - flexibility, good pay - but I’ve known the whole time that it won’t last. Gig workers will be the last consideration when it all falls apart
absolutely ZERO Ai companies need bailouts. this is complete theater.
if you are reading this, please do some research on how much money inference prints? its not wildly spoken about because people don’t yap about printing money.
“Oh but it’ll all come crashing down” - say you don’t work at a lab or know anyone at a lab.
My greater annoyance is HN becoming annoying political in the past few years. Sure, everything is political now but the moderation siloed these posts better. Maybe the user base is too old now. Fresh blood went to Discord or X.
Please do some research on the other costs involved in AI. Not a single AI company is making any profit and realistically are not close. Training models costs billions of dollars. There is no moat for these companies. If an open model comes out tomorrow that is better than Claude what happens to Claude's 1.5 trillion dollar valuation? What happens if revenue does not grow as fast as anticipated when Anthropic alone has 750 billion dollars in obligations? How do they pay their bills? What happens then.
The only company making money are hardware manufacturers who are throwing logs on to keep the fire burning through direct capital investments in their customers. Without Nvidia some of these neoclouds would already have failed.
I think the “it’ll come crashing down” argument comes from cheaper inference, not more expensive. The frontier labs have to pay for these massive training runs. If inference gets cheap enough that you can run it yourself, why run it on their side? Open weights are (much) less than a year behind.
Projecting that none of these AI companies will demand bailouts after the bubble crash is wildly speculative in the other direction. This has nothing to do with politics.
a) How courageous for them declaring wanting to ban something that isn't being proposed
b) if an AI bubble burst were to happen, the fallout would be severe, ie. hard to abstain for policy makers from giving the public ease of mind
c) All this proposal wants is to drive a wrench between the GOP and its new (?) populist voters, that don't want any favors given to any Big Tech oligarchs, regardless of it being AI or not.
d) this may work or not – neither party is going to abstain from bailing them out, anyways.
Signed, the Silent Majority
If you owe the bank $100, that's your problem. If you owe the bank $100 million, that's the bank's problem. If you owe the bank $100 billion, that's country's problem.
I worry that one of the reasons that Anthropic has 500 billion in commitments and everyone is running to the credit markets is to make it so that if they fail everyone suffers.
Ever is a stretch. Government shouldn’t be bailing out shareholders and senior managers.
What makes this so politically unpalatable is that having constantly bailed out the last few cycles, everyone is now operating on the assumption that they will be bailed out, so many entities who in a more perfect world would know better have done a lot of stupid things, like pensions that have a lot of bonds that will go bad if the AI bubble collapses.
Frankly we're looking at massive problems and upheaval at this point no matter what we do. But the advantage of not socializing the losses is that the signals of what needs to be changed go directly into the economy. Socializing the losses means that the signals get muted, confused, and mangled and people in the future end up doing the wrong things as a result.
Unfortunately, at this point I think the choice is between people clearly and forthrightly losing massive portions of their retirement as visible losses, or people deceptively losing massive portions of their retirement as the bailout creates massive inflation and their retirement is inflated until it has a fairly similar amount of losses. Unfortunately, it is a feature for the rulers that the bailouts obscure the losses and dampen the correct apportionment of blame, even as it destroys the underlying economy. Giving people clear, correct targets for their rage, the same process of correct pain apportionment the economy needs to become healthy again, would point people at the "wrong" clear, correct targets, and the clear, correct targets are the ones deciding whether or not a bailout occurs. Not just the literal lawmakers nominally making the call, but the entire system that makes the decision.
So plan on bailouts.
That’s what happened in 2023.
Who are you thinking of?
That’s what happened in 2023. In 2008, depositors almost got wiped out and a lot of senior managers and stockholders got bailouts.
Isn't that what the IPOs are for?
However, looking at the other side, I think what this might do is add some barrier or consideration when a bailout is considered.
The last time we did large scale bailouts (TARP[0], in 2008), It was rushed through and ratified in around two weeks following the sudden collapse of Lehman.
Any measures we can take to ensure that we don't fall into a situation where there is only one choice is sensible.
[0] https://home.treasury.gov/data/troubled-assets-relief-progra...
If AI companies were to go bankrupt, it’s possible they’d be nationalized. I don’t see any other feasible option besides that or a bailout (or one but it’s effectively the other, e.g. “nationalized” but the employees including leaders are the same, or bailed out but already subservient to the military). I think the best case is for the compute to be made effectively public: giving every citizen free access, since we’re paying taxes that prevent their bankruptcy.
Check out this excellent article for more background: https://www.theatlantic.com/ideas/2026/10/bubble-bailout-tru...
It’s good that it’s being discussed. We’re tired of the wealth class privatizing the profits and leaving us to cover the costs.
Bailing out banks made some sense so that they couldn't operate. Without the bailout there would be no more banks and a hundred million of people left on the ice.
If you want to keep AI operating you can just (temporarily?) ban investors from tearing down the AI companies they invested in for parts. You could say they are important for national security or something. I'm sure they are already earning enough to keep the lights on. Just not enout to give investors payback for insate capital expediture to set up.
Such ban wolud the governement zero, especially if you write the law well enough, so that it can't be succesfully challenged in court.
So there will be a huge pressure. Trump Jr. and the Kushners are invested in all sorts of bubbly assets, too.
The Democrats should do this however, and if they win the midterms keep their promises for once and block everything. Then 2028 is easy.
Promises in the wind.
If you're lucky enough to be in this field and you're not managing your finances, that's not the governments problem to fix
if you are reading this, please do some research on how much money inference prints? its not wildly spoken about because people don’t yap about printing money.
“Oh but it’ll all come crashing down” - say you don’t work at a lab or know anyone at a lab.
My greater annoyance is HN becoming annoying political in the past few years. Sure, everything is political now but the moderation siloed these posts better. Maybe the user base is too old now. Fresh blood went to Discord or X.
The only company making money are hardware manufacturers who are throwing logs on to keep the fire burning through direct capital investments in their customers. Without Nvidia some of these neoclouds would already have failed.
b) if an AI bubble burst were to happen, the fallout would be severe, ie. hard to abstain for policy makers from giving the public ease of mind
c) All this proposal wants is to drive a wrench between the GOP and its new (?) populist voters, that don't want any favors given to any Big Tech oligarchs, regardless of it being AI or not.
d) this may work or not – neither party is going to abstain from bailing them out, anyways.