Here's How the AI Crash Happens; The Atlantic

[edhopper]

https://www.theatlantic.com/technology/2025/10/data-centers-ai-crash/684765/?gift=nwn-guseqS6cY1kVeEKZAdaSthhtnZy-pwftXDjsd3E&utm_source=copy-link&utm_medium=social&utm_campaign=share

Here is where the bubble dynamics get complicated. Tech firms don’t want to formally take on debt—that is, directly ask investors for loans—because debt looks bad on their balance sheets and could reduce shareholder returns. To get around this, some are partnering with private-equity titans to do some sophisticated financial engineering, Paul Kedrosky, an investor and a financial consultant, told us. These private-equity firms put up or raise the money to build a data center, which a tech company will repay through rent. Data-center leases from, say, Meta can then be repackaged into a financial instrument that people can buy and sell—a bond, in essence. Meta recently did just this: Blue Owl Capital raised money for a massive Meta data center in Louisiana by, in essence, issuing bonds backed by Meta’s rent. And multiple data-center leases can be combined into a security and sorted into what are called “tranches” based on their risk of default. Data centers represent an $800 billion market for private-equity firms through 2028 alone. (Meta has said of its arrangement with Blue Owl that the “innovative partnership was designed to support the speed and flexibility required for Meta’s data center projects.”)

In this way, the data-center financing ends up being a real-estate deal as much as an AI deal. If this sounds complicated, it’s supposed to: The complexity, investment structure, and repackaging make exactly what is going on hard to parse. And if the dynamics also sound familiar, it’s because not two decades ago, the Great Recession was precipitated by banks packaging risky mortgages into tranches of securities that were falsely marketed as high-quality. By 2008, the house of cards had collapsed.

A Bubble by any other name....

[Celerity]

1. dupe

https://www.democraticunderground.com/100220761906

DU search shows the first OP

[edhopper]

5. Great minds...

[Callie1979]

2. Biggest difference in '08 is that Meta, et.al. have tons of money.

In '08, we had loans being given to people who shouldve never gotten them finally go into default.
The biggest threat here is that AI turns out to be a bust & those data centers got under & along with it, those fancy rents. Right now, I'd ay that a pretty low risk, unless we build WAY too MANY data centers.

[edhopper]

6. That is just Meta

the other AI companies don't have streams of revenue to offset these expenses. And even Meta stock is over inflated with imaginary AI returns.

[Callie1979]

10. But META's P/E is about 28; not really inflated compared to the broader market

And that PE isnt forward looking.
As for the others, buyer beware if there aren't earnings to support them today

[yardwork]

9. I don't think that poor people were the main drivers of the 2008 crash.

Wealthy people took out huge loans for over-priced properties.

I'm not an expert on this but I doubt that poor people with small mortgages drove the crash. The math doesn't work.

If these data center loans turn out to be overpriced it could cause a crash.

[Callie1979]

12. Housing defaults were the biggest driver & being "poor" wasnt required.

People overextending WAS. Banks gave out loans that you'd NEVER get today. Yes, SOME "poor" were part of the problem; "NINJA" loans actually happened. "No job no income no assets". I knew people who fell into it; they'd "qualify" for 300k so they simply HAD to buy a 300k house even though they couldn't afford it (in my area, 300k would buy you a BIG home). Then add in the fraudsters who began the whole thing; inflating prices, flipping to each other with higher prices only months apart, & getting loans they never paid & tanking entire neighborhood values. All preceded by too-low rates for too long.
The first corporations who fell went down because they bought those shitty loans & backed the mortgage companies & banks who dealt them out; they also went down. But dont worry, the heads of those banks just went out & started new ones!! And the buyers of the failed ones only took the BEST loans; the rest were called due! I saw that happen too.

[Bernardo de La Paz]

17. It was agnostic about poor/rich. The problem was layers of wrapping that disguised the true value of the securities.

The securities were not secure.

[paleotn]

14. We may already have built way too many data centers.

AI isn't playing out as the productivity driver it was hyped to be. In fact, it's become the butt of jokes. And the problem isn't the tech giants. It's the hedge fund, shadow banking system that's leveraged to the hilt, and investors globally who've bought securities based solely on Meta and Amazon's willingness to keep paying leases on all those data centers they may not need and may never need.

When it becomes uneconomical for Meta, Amazon, Microsoft, et al., to continue those leases, for whatever reason, the music stops. If those giants decided to keep paying ( that being the lesser of two evils ) it's a massive, unproductive financial drain on their bottom lines. And the securities based on that stream of lease payment income become zombie investments. Collapsed in value but not gone to zero. Simply waiting for some other economic shock to figuratively shoot them in the head and wipe out bottom feeding speculators.

And all of this DOES NOT occur in a vacuum. You'd think that after the sub prime collapse, people would finally, FINALLY realize that everything in the global financial system is interconnected in one way or another. Financially, if a butterfly flaps its wings in Singapore, the S&P might just collapse in New York and an Icelandic bank goes belly up. So much we should have learned, should have firewalled, should have regulated, but didn't. And once again, it will be average people who pay the highest price. It sure as shit won't be Jeremy Iron's character in Margin Call or Zuckerberg the cyborg.

[leftstreet]

24. +1

[FakeNoose]

36. They all think they're going to find the pot o' gold at the end of the rainbow

As long as they can stay in the game with someone else's money.

Yes it's going to crash all right.

[FakeNoose]

3. Thanks for the gift link

[AnnaLee]

4. Can air or ether be far behind? ntp

[edhopper]

7. Well people do pay ridiculous prices

for water.

[Callie1979]

13. And there ARE "oxygen bars"!

https://totalsculpting.com/what-are-oxygen-bars-elk-grove/

"At an oxygen bar, individuals sit comfortably while inhaling concentrated oxygen through a nasal cannula or mask. The oxygen used is typically enriched with flavors or scents like eucalyptus, lavender, or citrus, enhancing the sensory experience and promoting relaxation."

[Celerity]

15. modern oxygen bars have been around since before I was born (1996) and talked about since 1776

https://en.wikipedia.org/wiki/Oxygen_bar

In 1776, Thomas Henry, an apothecary and Fellow of the Royal Society of England speculated tongue in cheek that Joseph Priestley’s newly discovered dephlogisticated air (now called oxygen) might become "as fashionable as French wine at the fashionable taverns". He did not expect, however, that tavern goers would "relish calling for a bottle of Air, instead of Claret."

Another early reference to the recreational use of oxygen is found in Jules Verne's 1870 novel Around the Moon. In this work, Verne states:

Do you know, my friends, that a curious establishment might be founded with rooms of oxygen, where people whose system is weakened could for a few hours live a more active life. Fancy parties where the room was saturated with this heroic fluid, theaters where it should be kept at high pressure; what passion in the souls of the actors and spectators! what fire, what enthusiasm! And if, instead of an assembly only a whole people could be saturated, what activity in its functions, what a supplement to life it would derive. From an exhausted nation they might make a great and strong one, and I know more than one state in old Europe which ought to put itself under the regime of oxygen for the sake of its health!

Modeled after the "air stations" in polluted downtown Tokyo and Beijing, the first oxygen bar (the O2 Spa Bar) opened in Toronto, Canada, in 1996. The trend continued in North America and by the late 1990s, bars were in use in New York, California, Florida, Las Vegas and the Rocky Mountain region. Customers in these bars breathe oxygen through a plastic nasal cannula inserted into their nostrils. Oxygen bars can now be found in many venues such as nightclubs, salons, spas, health clubs, resorts, tanning salons, restaurants, coffee houses, bars, airports, ski chalets, yoga studios, chiropractors, and casinos. They can also be found at trade shows, conventions and corporate meetings, as well as at private parties and promotional events.

https://o2spabar.com/about-us/

On March 2, 1996, O2 Spa Bar threw a launch party to celebrate the opening of the world’s first retail oxygen bar. The party took place in their gorgeous 1,600 square-foot retreat located in the heart of Toronto. While patrons lined up inside O2 Spa Bar to experience the benefits of 99.9% pure oxygen, the media with their satellite trucks outside beamed live reports and interviews with the proud co-owners. Needless to say, the launch party was an overwhelming success.

[Bernardo de La Paz]

18. Nitwits go to oxygen bars & then consume tons of anti-oxidant supplements. . . . nt

[dutch777]

8. Somehow, whoever holds it, the debt needs to be serviced. My problem is these tech folks get bailed out by taxpayers.

Silicon Valley Bank goes bust and instead of inflicting some pain that teaches lessons, they get bailed out. Until private and major interests start actually getting hurt, it will be the little folks and taxpayers that get hurt and see the national debt on their backs rise.

[hunter]

20. I can easily imagine a government bail out...

... as these data centers are converted from frivolous and unprofitable AI applications to incredibly intrusive government surveillance of our private lives.

[jfz9580m]

11. I hope it happens before it does more damage

I am glad people are fighting this. If this was for a verifiably worthwhile science it would be one thing. It is true that I am a layperson. But I am not particularly conspiracy minded. I am certainly not a Luddite. However, going by the people driving this (those sleazy industry leaders primarily), it seems like an endeavour propelled by charlatans rather than scientists. I could be wrong (hedge ;-/). But as of now I don’t view this as a positive development - not rolled out this way anyway.

A couple of people I know (an academic researcher and a student) were telling me that department heads are aggressively pushing AI on students and researchers. And then I ask “Is it any good? Is it useful?” The answer so far is “No. It’s garbage”. Talk about mindless trend following..

https://www.nbcnews.com/news/us-news/maryland-farmers-data-centers-extension-cord-power-lines-rcna233324

Farmers fight an 'extension cord' for data centers
Tensions over a proposed 67-mile transmission line have rocked Maryland's farming communities, as the nation’s largest grid operator says data centers are driving power demands.

For more than a year, tensions over a plan to construct a 67-mile transmission line across three Maryland counties — Baltimore, Frederick and Carroll — have rocked the state’s farming communities. Landowners have staked signs along the highway declaring, “No eminent domain for corporate gain.” A court battle has escalated into threatening social media posts and heated confrontations between farmers and land surveyors. Local officials from all three counties have opposed the project, and some have pleaded with state and federal officials to intervene.

“People are scared,” said Steve McKay, a councilman in Frederick County. “It brings people to tears. It makes others angry — the notion that this out-of-state firm is going to come in and potentially be given the authority by eminent domain to take land.”

The $424 million Maryland Piedmont Reliability Project is an effort by PJM Interconnection, which operates all or part of the electric grid for 13 states, including Maryland, to address climbing power demands amid the growth of energy-hungry data centers and an artificial intelligence boom.

“Without these upgrades, Maryland may face severe grid congestion, threatening both affordability and reliability for ratepayers,” the Public Service Enterprise Group (PSEG), the New Jersey-based utility company responsible for building the line, said in a statement.

[Bernardo de La Paz]

16. It's a decent article but does not deliver on the title promise. I read it in Celerity's thread

It lays out lots of background well, but was not very enlightening, for me.

It is worth the read.

[edhopper]

27. I think this paragraph lays it out

The people who are paying attention to this cycle are getting anxious. On a scale from one to 10, the AI-bubble concern is: people posting memes of Christian Bale’s character from The Big Short, squinting in disbelief at his computer monitor. If tech stocks fall because of AI companies failing to deliver on their promises, the highly leveraged hedge funds that are invested in these companies could be forced into fire sales. This could create a vicious cycle, causing the financial damage to spread to pension funds, mutual funds, insurance companies, and everyday investors. As capital flees the market, non-tech stocks will also plummet: bad news for anyone who thought to play it safe and invest in, for instance, real estate. If the damage were to knock down private-equity firms (which are invested in these data centers) themselves—which manage trillions and trillions of dollars in assets and constitute what is basically a global shadow-banking system—that could produce another major crash.

For now, money is still pouring into the AI industry. But there’s also something circular about these investments. To wit: OpenAI has agreed to pay $300 billion to Oracle for new computing capacity, Oracle is paying Nvidia tens of billions of dollars for chips to install in one of OpenAI’s data centers, and Nvidia has agreed to invest up to $100 billion in OpenAI as it deploys Nvidia chips. Attempts to illustrate these circular investments have produced a series of byzantine charts that one software engineer referred to on X as “the technocapital hyperobject at the end of time.” The consensus seems to be that although this is legal, it likely cannot go on forever.

[Bernardo de La Paz]

30. Yes, that is one way it might happen. Thanks . . . nt

[edhopper]

31. I think the article isn't predictive

(despite the title) but rather descriptive of what is happening.

It will crash, because too many companies that will not see profits in AI are investing too much into it. AI will be here, but so are houses and the internet.
When? That is impossible to know. I am largely out of the market and will wait for the crash to go back in. (If he doesn't take America down with him).

[Bernardo de La Paz]

32. It might not crash. It is a bubble, but might deflate in an orderly way, like Meta dropping 11% this week

... while the other Mag 7 / tech stocks held up generally.

Just a guess, but perhaps it will deflate in a way intermediate between crash and orderly. Like you I'm out. I'm looking into ways to do some put options for the decline, writing a program to help guide me (I would be cautious). But mostly I'm conserving cash to buy after some kind of bottom seems to be established. Might easily be a couple of years before buying back in.

[edhopper]

34. I am not sure when the last Bubble

deflated rather than crashed. But considering the equity put into AI a crash is more likely. All this looks highly leveraged, and when that falls, investors must sell other holdings to make up for it and you have a crash.
So far, since Trump took over, I do not see any rationality from Wall Street.

[SunSeeker]

44. Taxpayers would probably end up bailing out the scumbag private equity firms like we did the banks and insurers in 2008.

It makes me want to

[Borogove]

19. Can't happen soon enough.

[bif]

21. DUPE

https://www.democraticunderground.com/100220761906

[Joinfortmill]

22. Be careful where you invest your money, peeps.

[3Hotdogs]

23. So if you have index funds, buy puts as a hedge (insurance)?

[edhopper]

26. 80% of the run up this year4

is AI related companies.

[Bernardo de La Paz]

33. Cautiously, after much study

I'm out, but I'm looking into puts and other option strategies. I'm writing a program to help guide me. I want to do it cautiously, conservatively, but there is money to be made on the bear.

Mainly conserve cash to be able to buy into the bottom once it seems to be established. Warren Buffet did not invest in the 2000 dot.com bubble, rode it out. Right now he's somewhat more in cash than he usually is.

[GreatGazoo]

35. Learn to use stop loss

or set your downside limit mentally and then execute that trade yourself when the market declines by >6% or whatever you are comfortable with.

Index funds will replace losers with winners so for those whose horizon is 2+ years the only risk in not selling on the way down is how long you have to wait for the inevitable bounce.

In 1996 people said the internet was a fad. In 1999 it was "a bubble". In 2002 they said "told you so". QQQ was $120 at the height of the dot com era -- it is $630 today. Up ~400% in 20 years.

Most of the people repeating this "AI is bubble" thing don't understand how AI makes money. Comparing it to the mortgage backed securities of 2007 is wrong for many reasons but mostly because AI is tech, IP and infrastructure but the 2008 crash was just aggressive lending. Deep Seek didn't pop "the bubble" because it is not a bubble. It is military spending, the new arms race, and therefore will not be allowed to fail. The government is taking a direct stake in the key companies. Index funds backed by military spending, tariffs and outright warfare are socialism for the rich.

Same for oil. https://markets.businessinsider.com/commodities/oil-price?type=wtiWTI is good buy anytime it slips under $59 because the US government will always take steps quickly to boost it back over $60. We saw this 10/16 ($57.46) and then it pops 10/21 to $61.79 when the Russia sanctions are announced and Venezuela gets hit. It is not profitable to pump Texas crude when the price dips under $60 and the US is, as the saying goes, 'basically just an oil company with a huge military.'

[edhopper]

37. But the billions into Data Centers

and the cost to run them is highly leveraged infrastructure where the revenue doesn't pay for the investment.
It's not that AI will go away, it's that the amount invested does not give a return. Some will go onto to be big companies, but many will fail, similar to the dot com bubble.
The AI companies are not making money. Most are operating at a loss. This will eventually kill many of them. That is a crash.

[GreatGazoo]

40. Some of the infrastructure entities

just rolled from crypto to AI. There is also cloud computing so that is three different customer bases for the same infrastructure.

MARA has locked in electrical power at $0.04 per kwh -- that guarantees a profit even if they wind up simply reselling the power. WULF is in a similar position and can do AI, cloud or crypto. SMCI likewise is well positioned with or without AI.

Outside of infrastucture, the software plays are looking good -- PLTR will have massive cash flow since it basically a software company. On a slightly smaller but still well into the billions play, Figma (FIG) with a $24-bil valuation is poised to take down Adobe, a $142 bil company.

HOND, UUUU and USAR seem guaranteed to grow even if AI was outlawed. Cheap green power doesn't need AI to be profitable.

Agree that there will be a shake out but AI is here to stay and many of these companies are guaranteed in the ways that Leonard Cohen sang about.

>The AI companies are not making money< Yes they are. PLTR has an 80% profit margin for 2024. APPL GOOGL, MSFT are all AI players that have been in the black for decades. NVDA made $3.54 per share. Trades at 57 P/E but it's growing and, my main point, it has the entire US government behind it.

[edhopper]

41. Appl, Google MSFT,Meta

Are not making money from AI, but their stocks are rising like they are.
They cannot keep pouring money into division that don't make a profit.
Again, the Internet did not go away in the dot com crash. But the sector was highly inflated.
Just as AI is today.

[Tetrachloride]

25. Lawyers and their henchmen accountants. or vice versa

[Bluetus]

28. The key point: Meta, Amazon et al will have no responsibility for the data center bankruptcies

These rent agreements will allow them to walk away, and that $800 billion capital investment is somebody else's problem.

By putting most of the risk into 3rd party real estate bundles, Meta, Amazon et al preserve the big profits for themselves and pass the greatest risks onto suckers. If AI is a screaming success, then the real estate bundles will get some profit from the rents, but no big upside. And if the bubble bursts, the big guys walk away clean and the real estate investors end up with vast data centers they can't put to any other productive use.

[Martin Eden]

29. My greatest concern with AI has nothing to do with economic impact

It has everything to do with the immense power of information concentrated in a handful of tech billionaires at the top -- combined with the capability of AI to invade privacy and produce deep fakes to manipulate public perception far beyond the relatively primitive means of rightwing talk radio and Faux News.

Concentrated wealth is inherently corrupt and dangerous to representative democracy.

Concentrated wealth and information domination can lead us into a 21st century Orwellian dystopia.

[lame54]

38. The Big Short - great break down of 2008 housing collapse...

This sounds very similar

[snot]

39. Actually, fwiw,

the worst of the losses from the 2008 GFC were not due to the securitized mortgages but to the credit derivative liablities layered on top of that.

It was the derivates that made the GFC losses orders of magnitude greater than those in the S&L crisis in the 1980's.

[SamuelTheThird]

42. Powell just said AI isn't a bubble

But it's not like he could say the opposite

[edhopper]

43. And the folks that ignored the Housing Bubble agree

https://finance.yahoo.com/news/powell-says-unlike-dotcom-boom-210422880.html

Makes me feel better.

He also said AI has killed job creation.

https://finance.yahoo.com/news/jerome-powell-says-ai-hiring-163037152.html

So how is there not going to be a crash?