Is the Great AI meltdown imminent? [NSFW]
A $100 billion dollar deal that was propping up the industry just disappeared
The big economic news in AI 2025 was circular financing, like a deal in which Nvidia was supposedly going to put in $100 billion into OpenAI which would then buy a lot of Nvidia chips. No picture better captures last year’s vibe than this NSFW one (origins unknown) that was floating around on social media:
The big economic news in AI 2026 is that all that is poised to collapse.
Significantly, over the last few days Nvidia has apparently pulled back on its $100B promise; they are now hinting at something like a more modest $20B.
That’s a huge problem for OpenAI, given its immense many billions a year burn rate.
The Guardian rightly called all this out this morning:
The Guardian article is, for the most part, great. But I do have one small objection.
On the plus side, it concisely explains what circular financing means (“This was a circular arrangement through which the chipmaker would supply the ChatGPT developer with huge sums of money that would largely go towards the purchase of its own chips”). And the article rightly points out how these kinds of deals a lot of people nervous (“it is this type of deal that has alarmed some market watchers, who detect a whiff of the 1999-2000 dotcom bubble in these transactions.”)
Furthermore, aptly, if a little understated, the authors write “That a $100bn deal between two of the most crucial players in AI appears to have evaporated over a weekend is unsettling”.
My only objection is to the second part of the subheadline, “Apparent collapse of Nvidia–OpenAI tie-up raises questions about circular funding and who will bear the cost of AI’s expansion”. The first part is dead on; the apparent collapse of a marquee $100B deal between two immense and central players does indeed raises a ton of questions.
But the second clause, the one about who will bear the cost of the expansions, leaves out another, very real possibility. If this deal collapses, possibly bringing down OpenAI, the multitrillion dollar expansion that so many were anticipating might well not happen at all.
it’s a huge huge negative signal that any other prospective investor has to take very very seriously
The degree and intensity with which this lot has managed to put off the inevitable collapse of their empire is quite remarkable. The collapse is inevitable due to their product being intrinsically incapable of doing what the majority of investors think it is capable of: True cognition, versus clever but extremely unreliable retrieval of largely stolen intellectual property.
Even this week, NVIDIA's CEO has been blabbering somewhat incoherently -- and in language worthy of a Heaven's Gate devotee -- about how important it is for big investors and entire governments to recognize that LLM AI is still an itty-bitty, poor baby that folks must coddle with trillions more in investments to help the poor thing reach its full potential. After factoring out the deception of casual intellectual property theft, that full potential still seems, even after several years of promising a correction, to be nothing more than the astonishing ability to count arms and legs correctly.
To paraphrase an old saying, why has this scam "Boldly gone where no scam has gone before"? That's easy: What they created, starting in particular with the Attention mechanism, is the full and global automation of the confidence artist: the ability to bring up just the right facts and with just the right confidence to persuade others, pretty please, to hand over their bank accounts just one more time, since this time, surely, without doubt, their latest contribution will finally result in the release of the promised fountains of gold.