Operai is increasingly blocked in a complex network of financing agreements with chips manufacturers. – Justin Sullivan/Getty Images
A version of this story appeared in the Nightcap Bulletin of CNN Business. To get it on your entrance tray, register for free here.
The nest of the Rat of associations of AI became a little more. And for those who suspect that everything is a bubble, well, the last entanglement is something else to worry.
Icymi: Operai has just reached an agreement with AMD, a chip designer who competes directly with Nvidia, who announced an investment of $ 100 billion in OpenAi less than a week ago. Nvidia, the world’s largest company by market value, It also competes with Oracle, which last month became the action of hot when also As reported, it reached a $ 300 billion agreement, guessed it, OpenAi.
According to the reports, OpenAI, the largest private company on the planet, with an assessment of half a trillion of dollars, is also deeply financially connected with Microsoft and Google (public companies that are quoted on the stock market, respectively, respectively).
If it seems that the industry is mainly a small handful of large companies that only exchange money and services from one side, congratulations: it includes the AI ​​financing machine.
According to the agreement announced on Monday, which sent the actions of the chip designer (AMD), 23%, AMD is effectively subsidizing the demand by issuing OpenAi orders for up to 160 million shares, said investor Paul Kedrosky on Monday. “That makes Openai a 10%shareholder: part of the client, a financial part, a transfer of risk of cash to shares, as well as making OpenAi the greatest AMD shareholder,” Kedrosky said.
As I wrote here last week, intertwined configurations, including some known as “Supplier Financing”, are common in the sparkling world of artificial intelligence. But the many layers of overlap, concentrated among companies whose shares have been supporting the US stock market, have hindered the remission of how much demand for AI is authentic hunger for customers and investors and how much of it is only the recycled capital to maintain the appearance of progress.
The financing of the suppliers, together with the duplication, triplication or quadrup of the assessments of the companies, is one of the non -flatter echoes that some analysts see when comparing the frenzy of today with the bubble of end points of the 90s. At that time, the giants of telecommunications equipment such as Cisco, Nortel and Lucent took to offer their clients to offer their customers They secured a sustained demand for telecommunications equipment.
Because this high -tech team was scarce, customers, many of them in the new companies that tried to build fiber optic cable, inflated their orders, contributing to an excess that left them staggering when, in 2001, it was clear that technological and telecommunications companies had overestimated demand. Team manufacturers like Cisco were left with an uncollectible debt while new companies started.
“We have all been witnessing elements of this behavior,” wrote Mike O’Rourke, head of the Jonestrading market, in a note last month. “The lessons of the bubble of the points-com are almost forgotten, but they echo the eternity,” he wrote, pointing out that the financing of the suppliers was “key” for the disappearance of Lucent, once the company of the world’s largest telecommunications equipment was billed, as Nvidia is now, as a “selection and shovel” in the emerging technological economy. (Lucent barely avoided bankruptcy in the 2002 recession and was sold to the French competitor Alcatel in 2006.)
There are not two exactly the same market cycles, of course, and the existence of supplier financing is not a guarantee that we are now.
“We see some important differences,” said Morgan Stanley analysts in a note on Monday. The main one: the big technological companies are in a much stronger financial position than many of the over-inflated actions of the ERA of the DOT-COM.
But putting the discussion about circular financing and whether or not it is profitable, it is worth looking at the product in the center of this whole expense: generative the generative driven by large or LLM language models. That’s where many other investors, analysts and academics shout that the emperor has no clothes.
The LLMs are popular, no doubt, and have grown in capacities in recent years. The proponents say that LLMS will change the world, someday. But their cases of use so far have not reached the exaggeration of “superintelligence” that AI defenders have promoted for a long time. A MIT survey of 300 companies found that 95% “is obtaining zero yield” in its AI investment, a result that briefly shook investors in August. Meanwhile, “Workslop”, the non -sensitive and useless presentations that spit for AI, have become its own headache in many white collar industries.
However, the disadvantages of AI are well documented, since chatbots have dragged some people to delusional spirals, sometimes mortal.
But let’s resume that discussion of the dollar and cents business again. Because a group of people begin to use Word B.
“I think this is the largest and most dangerous bubble that the world has seen,” Julien Garran de Macrostrategy Partnership wrote in a fairly pessimistic report on Friday. He estimates that a “bad capital allocation in the United States” directed by AI is 17 times larger than the bubble of points and four times larger than the 2008 real estate bubble.
Garran points out that the “majority of the LLM AI ecosystem” is losing money. “It’s just an explosion on the round trip”, buy and sell an asset again and again to create the appearance of high demand, different from the financing of suppliers, “of Nvidia that maintains the inflated bubble.”
To get more news and newsletters of CNN, create an account on CNN.com
(Tagstotranslate) Nvidia (T) Financing of the supplier
