As AI Companies Borrow Billions, Debt Investors Grow Wary (nytimes.com)
- Reference: 0180460115
- News link: https://news.slashdot.org/story/25/12/26/1737240/as-ai-companies-borrow-billions-debt-investors-grow-wary
- Source link: https://www.nytimes.com/2025/12/26/business/ai-debt-investors.html
Wulf Compute, a subsidiary of Bitcoin-miner-turned-data-center-operator Terawulf, raised $3.2 billion in mid-October at 7.75%, well above the 5.5% average yield for similarly rated issuers. Cipher Compute sold $1.7 billion in early November at just over 7%. CoreWeave, which rents data centers and installs computing systems for companies like OpenAI and Meta, raised $1.75 billion in July at 9%. The company's bonds have since fallen to around 90 cents on the dollar, pushing the effective yield above 12% -- nearly double the average for companies at its single-B rating level.
"We just have to be much more pessimistic and not buy into the hype," said Will Smith, a portfolio manager at AllianceBernstein. Construction delays and uncertain demand for AI computing power remain key concerns for lenders who, unlike equity investors, have no upside beyond getting their principal back.
[1] https://www.nytimes.com/2025/12/26/business/ai-debt-investors.html
Re: The Great AI Collapse of 2026 is beginning (Score:2)
Isn't that most boards already?
Re: (Score:2)
> It's a computer program FAKING intelligence.
I'd like to say this is on the one side true, and I can relate to the assumption about ML not being "real" intelligence, but on the other hand there are so many descissions we humans make based soley on the training and the weights in our own neural network, that to predict a general collapse is really short sighted. How much of our own life is actually "real" intelligence, and not just a neural network generating an answer to an input? One striking example where statistical methods are very good and extr
Re: (Score:2)
It doesn't matter if AI is "faking" intelligence, it still does many useful things. Just in the last month, I wrote my first Vue.js web app. GitHub Copilot shortened my learning curve *enormously*. When I wanted to do some thing that were off-the-beaten-path, like an intentional component circular reference, it came up with a solution that actually worked. Without AI, it would have taken me many more weeks to get my first app running.
The solution is IPOs (Score:2)
This has been in the smart news sources for a while. The solution is IPOs. Tap the equity investor market in exchange for cash. The equity markets are huge. Except all the trillionaires holding provate control of these companies don’t want to give up any control, deal with pesky shareholders, or meet transparency requirements. They want to borrow an infinite amount of money, offload the default risk, and keep the corporate growth for themselves.
Boo-hoo for the billionaires. I’m actually extr
Re: (Score:2)
What is the problem we are solving?
Investors growing weary, is a good, healthy thing in a frothy market. It helps bring common sense to the market.
Re: (Score:2)
or make due with the money they have
Pretty sure they're looking at door #3, "World's biggest bust-out".
Just look at all that money consumed by zero-marginal-product grannies. Surely that money would be better-employed on AI girlfriends for incels.
Ratings Company Nonsense (Score:2)
The debt rating companies are usually behind in evaluating the creditworthiness of a company.
Debt rating companies have two near (?) conflict of interests:
1) The debt rating companies get paid by the company issuing the debt
2) The debt rating companies want repeat customers and do not want to scare away other possible customers
The debt rating companies rarely update the credit rating of older debt issues as they do not earn a fee for doing so.
Looking at the interest rate spread to risk-free US Treasury bond
Cautious? But why? (Score:2)
AI is the answer to every question ever asked. Why would you hesitate giving loans?
Bubble (Score:4, Interesting)
So when does everyone think this AI bubble is going to burst? Personally, I plan to do some major shifts in my 401k come the new year as while I've seen some great gains lately I feel like we don't have much time left until the crash happens.
Re: (Score:3)
You are at least 100 years early on expecting this to happen. We still don't have AI replacing the majority of jobs. We don't even have AI call centers replacing the customer service people. So, while a lot of potential is there, AI isn't killing too many jobs YET. It will take at least 20-50 years before robotics will be replacing the workers at Amazon in the warehouses. So, right now, it's all hype based on the potential for AI, and yet, it's causing problems with water, electricity, and other t
Re:Bubble (Score:5, Insightful)
Bursting of the bubble will not be the end of money,
but money alone might be less worth,
and I would say when the bubble bursts,
we will have a bank run.
Because this bubble and inflated stocks are built on debt,
so its Lehman Bros. again .. or in this case "trust me ai Bro."
Food reserves, drinking water reserves, dish washer soap,
DIY energy and so on is important.
btw. (Score:2)
I forgot to mention the bubble will not burst,
it has already started to burst, so it will be a very quick trickle down.
My indicator of the beginning -> Oracle Stock, since October 2025
a very sharp decline.
Re: (Score:2)
I would say this is likely to happen.
It all depends on the judgement for a company if the company provides a "real" value - in the "old sense".
However these current companies with a real value core will shrink quite substantial.
I compared that just recently in a discussion with collegues to the "Deutsche Telekom"(*) from the 2000s, its also still around, but it has cost many people of their savings - due to their new found interest in the stock market.
Using the 2000s and the 2008s as a base scheme, many com
Re: (Score:2)
I think in 2026, pretty quickly
Re: Bubble (Score:1)
Around midterm elections. That will cause a chain reaction that will make the bubble pop.
Re: Bubble (Score:2)
I'd stay invested in the big players until small players start going bankrupt, I think. Companies like Google, Microsoft, etc. that act as cloud providers for small AI companies will continue to see money coming in right up until the end. And once AI pulls back, they are still left with valuable infrastructure.
Re: (Score:2)
Like physical bubbles, the "AI bubble" isn't a single, big bubble that will burst all at once. Rather, it's a vast number of small, medium-sized, and large bubbles, that will burst at many times along the timeline. Some are already bursting, like the slashdot story about Indian firms being pai to clean up AI messes.
Also like physical bubbles, there is some actual usefulness in the underlying soap. Not everything about AI will disappear into thin air. Like the dot-com bubble, the best of AI will remain long-