Collapsar
pull everything inward with growing intensity, consuming energy and resources until only the core of true value (or nothing at all) remains, a new term Collapsital has been put forward to describe this.
In late 2025, a series of multi-billion-dollar deals in the artificial intelligence sector is causing déjà vu among industry veterans. Money, computer chips, and cloud credits are rotating in a closed loop among a handful of companies: Nvidia, OpenAI, Microsoft, Oracle, AMD, CoreWeave, xAI, and a few others. This has fueled a …
Elsewhere in these comments forums I'd been wondering how the plutocrats would continue to make money after they'd fired all the workers and reduced the masses to a hunter-gatherer existence but it has struck me in recent days they can continue their trillion dollar ways by simply exchanging shares with one another in some bizarre form of perpetual motion.
Or, per Douglas Adams, perhaps it will come to the following
MANAGEMENT CONSULTANT: Since we decided a few weeks ago to adopt leaves as legal tender, we have, of course all become immensely rich. [...] But, we have also run into a small inflation problem on account of the high level of leaf availability. Which means that I gather the current going rate has something like three major deciduous forests buying one ship's peanut. So, um, in order to obviate this problem and effectively revalue the leaf, we are about to embark on an extensive defoliation campaign, and um, burn down all the forests. I think that's a sensible move don't you?1
______________
1 Adams, D., The Hitch Hiker's Guide to the Galaxy (radio series), Episode 6, Scene 7.
Hundreds of billions of dollars involved, possibly several trillions ... But what is it actually buying? Nothing concrete, that's for sure.
The US economy is going to fold up faster than a used condom.
A couple dozen already extremely wealthy men will get much richer, and everybody else will be financially ruined when it all collapses.
It's not going to be pretty. I give it under a year. And probably decades to repair the damage.
The thing is that this IS in part supposed to be buying literal concrete. The people building the actual new data centers are going to want some more tangible than AI credits. Hopefully the power companies will too. Where is that money supposed to come from? Nearby a company that has revenue of $64 million that loses $18 million is going to build a $3 billion data center. I'm wondering where that money is going to come from.
Said data centers don't even exist on paper (CAD, whatever) at the moment. Building materials aren't part of the conversation yet. To say nothing of site purchase, cleanup, EPA approval, planning permission, and all the other little paper-pushing exercises that go into a project of this size and scope.
Quite frankly, I'll be surprised if ground gets broken before the AI bubble bursts.
The masters of this kind of financial "engineering" nonsense at Enron must be wishing they were still in business to profit from such idiocy.
Yep. It seems to me that it's pretty much the exact same vendor financing wheeze that helped trigger the dot.com implosion. I mean it works, after a fashion, until people twig that something is missing.. Like actual cash & revenues to keep the bubble inflated. I mean OpenAI does have the revenues to support their valuation, and pay for these transactions.. doesn't it?
This running around ever faster in ever smaller circles has reminded numerous people of that famous fable of Aesop.
The Tortoise and the Hare in AI Why the U.S. May Already Be Losing the Strategic Race
The story of AI’s global race is shifting from speed to sustainability, integration and systemic efficiency. While the U.S. sector sprinted ahead, China has quietly built a layered, resilient and cost-effective AI ecosystem. Its advantage is embedded in energy arbitrage, supply chain security, analogue chip dominance and open-source integration that reaches the developing world.
AI, the Tortoise, and the Hare
Further, ceding the field on AI safety and governance presents too wide a gap that China’s “AI Governance Action Plan” is primed to fill. Beijing called for convening governments, industry and civil-society around multilateral standards, capacity-building programs and joint R&D - while at WAIC not only did no Trump administration AI safety officials show, the only representative from a US AI lab even proposed an outright ban on open-source models, arguing “you wouldn’t give everyone a nuclear bomb.”
In short, building ever grander edifices reminds me of Stalin's plans to impress the world with the biggest, largest, and most useless buildings. Meanwhile, the Soviet economy was expertly run into the ground.
Almost all of the economic growth in the USA the previous year has been AI-related infrastructure spending. These are hundreds of billions of dollars which may never have a positive return-on-investment. Meanwhile, there is no growth in consumer spending indicating that the people are not benefiting from the growth. So, the GDP "growth" this year will likely be more than wiped out by the "disinvestment" the coming years.
It seems to work like this: every so often, rich people go mental and start pretending that things that are actually work fuck-all are now worth a fortune (AI, mortgage-backed securities). Then it all goes tits and the ordinary folks who had nothing to do with it end up suffering while the rich walk away largely unscathed. Regulations might be introduced to "prevent this happening again".
You start off burning your investors' money.
You then burn off your customers' money.
Next you burn off your suppliers' money.
If you replace the "investors" with "hydrogen", "customers" with "helium" and "suppliers" with "all elements lighter than iron," I think you can see how this all ends.
If every employed person in the US pays $100 month to AI companies, that's about $160B per year in revenue (not income). Maybe on a good day they'd get 5-6 points of margin, but that seems optimistic considering how many companies will be trying to recoup their $100B-200B per year investment at the same time. And GPU chips become obsolete in a couple of years, so the AI companies are not going to get that monkey off their backs any time soon. So, all together your AI companies are getting perhaps $10B in income and spending about $300-$500B per year in capital expenditures.
Exactly how long can this go on? I think the relevant quote is Ben Stein's "If something can't go on forever, it will stop."
As reported by Tom's Hardware, Satya Nadella said in a recent interview - “... you may actually have a bunch of chips sitting in inventory that I can’t plug in. In fact, that is my problem today. It’s not a supply issue of chips; it’s actually the fact that I don’t have warm shells [data centers] to plug into.”
This pretty much sounds like MS buying up memory and GPUs that aren't being used just to prevent any competition from developing. Including domestic competition.