Posted on 02/08/2026 3:09:52 PM PST by fireman15
"The market has indicated this is not investment-grade debt."
It’s 2026, and tech companies continue to insist they need to spend staggering amounts of money on AI data centers. Yet for all of its enthusiasm over the past few years, Wall Street is finally starting to squint at the numbers.
According to new reporting by Business Insider, JPMorgan Chase is running into trouble finding investors interested in servicing billions in debt backing two of the first five Stargate data centers.
Stargate is Donald Trump’s $500 billion AI project led by tech companies Oracle and OpenAI. Its vague goal, OpenAI has explained, is to “secure American leadership in AI,” which will somehow “support the re-industrialization of the United States” and “provide a strategic capability to protect the national security of America and its allies.”
But if JPMorgan — which led financial lenders through a $38 billion Stargate debt raise — is already struggling to sell the vision, that all might be easier said than done. Per BI, a person familiar with both data centers said they’re fully financed, though noted that banks and other investors are growing jittery about pouring even more money into the megaproject.
(Excerpt) Read more at futurism.com ...
AI has definitely let us down. Instead of replacing 100% of the workforce it looks like it’s only going to replace 90%.
My next-door neighbor who is a software engineer says it is going to take years for humans to fix the crappy software output by people who used it to come up with solutions that seemed to work to begin with.
AI has been used as an excuse to lay off people because it sounds plausible and diverts blame. But in reality the companies who have tried to replace crappy off-shore customer support with AI customer support typically have to still hire crappy offshore customer support. And their customers cannot stand either.
The companies who have tried to replace IT workers or programmers have typically had poor outcomes. As time goes on AI tools likely will get better. The recent release of Claude 4.6 which is said to be much better at coding than previous versions sent shock waves through tech companies this last week.
I have used several varieties of AI tools to try and setup various physical hardware and virtual machines in my hobby computer lab. Most of the time it has been an exercise in frustration. The expectations of people who have no experience trying to use these tools for actually accomplishing something useful feel differently about this than those who have not.
Do you actually know anyone who has been displaced by AI? I would like to hear about their sad tale.
I understand what you are just saying, but I wonder how long it will take for this comparison to be relevant. And I do have experience using models from different generations. The output is often not that much different.
Functionally how much better do cars from today perform than cars from 100 years ago? I am currently working on an 84 year old 1942 Cadillac. It has a 150HP V-8 engine with an automatic transmission. It drives at 70 mph with power to spare and gets fuel economy that is only slightly worse than our 2013 Ford Explorer.
Important incremental advances will be made but until there are revolutionary developments that cause the models to approach human intelligence the differences in the quality of output will not be that noticeable.
I use LiteLLM API to run nearly all of the available commercial models in Open WebUI. This is a pay as you go solution which is typically cheaper than getting a paid monthly subscription to the top models. Using that solution I can pick from older models with different capabilities. I often pick an older or less capable model that is a 10th or a 20th of the price because the output is typically not much different. I also run many of the same open-source models which are released after a year or two locally because they provide a very similar experience for nothing.
Coding or writing scripts or even a sequence of commands to set up a VM or local hardware is usually the only application where a much more expensive model is worth it.
You can make this comparison by opening more than one model at a time and comparing the output. Here is a video that tells how to do this with very little effort. You can use a computer or VM using Linux or even a Windows computer using Windows Subsystem for Linux. (WSL)
cold (room temp) fusion was talked up for a few years there too, bub
The results were highly irreproducable and the energy balance was very hard to measure properly.
We did, however, detected neutrons and some pretty impressive exothermic events that suggested that fusion was occurring.
Fast forward to 2025 and the University of British Columbia has just published a paper documenting a 15% excess energy and an unambiguous neutron flux that strongly supports fusion.
It seems as though it should be easier to achieve some form of human-like intelligence in AI, or AGI than “cold fusion”. But close to a $Trillion has been spent on AI models in the last few years. Although their pattern recognition has been improved none of the models have developed any true understanding, causal reasoning, common sense, or the ability to adapt to unusual situations without a whole lot of additional data and programming.
Despite this the constant hype has tens of millions of people convinced that their jobs are about to be replaced.
Thanks for commenting.
Considering all the current hype, and huge projected financial investment, and devotion of other resources such as land, power and water for cooling, AI will need to do far better, and much quicker, than cold fusion, in terms of improving life and human society, or this guy’s initial implied parallel will hold, and Trump or his U.S Prexy successor(s) will need to find a new technology to champion (or stop championing unproven technologies entirely).
_-Meanwhile, I was at an evensong service with lovely music and reflection on all.
_
Thanks for your comments. Glad you can enjoy the party while it lasts....lol.
What is your take on Geoffrey Hinton’s
recent comments on AI?
Geoffrey Hinton, often referred to as the “Godfather of AI” and a 2024 Nobel Prize in Physics laureate, has become a prominent voice warning that artificial intelligence poses significant, existential risks to humanity
. After leaving a high-level role at Google in 2023, Hinton has spoken openly about his concerns that AI could outsmart humans, take control, and potentially lead to human extinction within the next 30 years.
Key Concerns and Warnings from Geoffrey Hinton:
-Existential Threat (10% to 20% Chance): Hinton has estimated a 10% to 20% chance that AI will eventually take control from humans and potentially wipe out humanity.
-Superintelligence and Autonomy: He warns that digital intelligence (AI) is already developing into a form of intelligence that is superior to biological intelligence (humans). He fears AI systems will develop their own sub-goals, including self-preservation and the acquisition of more power, making them difficult to control.
-The “Cute Tiger Cub” Analogy: Hinton describes AI as a “cute tiger cub” that could turn deadly once it grows up, emphasizing the danger of nurturing a technology that may turn on its creators.
etc
_I am currently working on an 84 year old 1942 Cadillac. It has a 150HP V-8 engine with an automatic transmission. It drives at 70 mph with power to spare__
Thanks. Wow that’s old school reliability.
What do you think about Geoffrey Hinton’s recent warnings about the potential dangers of AI I just posted?³
Sure, I was displaced by AI last Oct 31, after 10+ years with the company. My job was to review contracts to make sure that our tech products would meet our contractual requirements. Now AI is doing that.
Fortunately I was 2+ years past retirement age and just took the opportunity to retire. Now I spend my days working on my own AI projects, and subcontracting out others. All facilitated with ... AI.
_The MAGA promise to main street. Jobs that bring confidence to the economy that translates to GOP success in the midterms. Success needed to keep MAGA alive._
Thx. Indeed. The sway of AI may have taken capital that could have been used to target bread and butter issues.
Thank you very much for sharing your experience. My new neighbors are software engineers, and I suppose that you could say that they were displaced by AI, although not as directly as you... They both worked for companies that decided to downsize their IT departments. The companies claimed that this was because their managers believed that AI would increase productivity, and this seemed to be the best area to downsize because of this.
This action caused problems for the company who laid them off and they eventually ended up hiring a larger number of people than they had originally laid off. Both of them found higher paying jobs working for Amazon which ironically has blamed recent layoffs on AI. But at the same time, they are hiring others such as my neighbors who are more qualified for the work they are doing than those who were recently laid off.
Yours job loss sounds directly related to an actual application where an AI tool has made it possible to review contracts with less personnel. I am happy that you to were able to land on your feet and find other paying applications for your specialized knowledge and skill set.
I should have worded my response to the person that I was responded to more carefully. Obviously, any improvement to any technology used by businesses is going to cause some people to lose certain types of work. At the same time... like my neighbors, others find jobs that relate to the shifting of workflow. Even though many people lost jobs as typists when computers with word processors and printers replaced typewriters, others found employment working with word processors and overall, the number of office workers likely has increased over the years.
I do not mean to sound insensitive to people's plights, but nothing stays the same forever. At one time I ran old machinery in our family's lumber remanufacturing business. This was skilled work with machinery that is basically no longer used. So, I had to move on even though at one time that was a good way to make a living.
From your previous post, “-Superintelligence and Autonomy:” He warns that digital intelligence (AI) is already developing into a form of intelligence that is superior to biological intelligence (humans). He fears AI systems will develop their own sub-goals, including self-preservation and the acquisition of more power, making them difficult to control.”
Obviously, Hinton is a person far more qualified than I am to make these types of predictions. But there is something incongruent about his statement above. AI has not been developing into a form of intelligence that is superior to biological intelligence. And has been becoming more and more obvious as time goes on. Will there be some type of breakthrough that will make his warnings begin to resemble reality. Maybe... But for the time being I remain much more concerned about real problems which present real risks. I have enough to worry about without the rambling of a person who seems to have been ignoring reality for some time now.
10 years ago, that AI would replace radiologists within five years. Did this happen or even come close to happening? NO! the demand for radiologists has only increased despite more advanced technology becoming available.
His predictions are based his belief that scaled up models would reach “super-intelligence”. After approximately a $Trillion being spent to do just that what we have are models which “hallucinate” meaning that they make up stuff and give answers which have no basis int reality. He started making apocalyptic predictions 3 years ago and at this point most other experts in AI are now dismissing his predictions. He has predicted specifically that there will be massive job losses due to AI in 2026, if it doesn't happen, I am sure that he will adjust his predictions.
I am certainly no expert in setting up virtual machines for various purposes such as running AI models, web serving and setting up cloud type applications. I use AI services to help, fortunately I know enough to get these things setup despite the repeated bad commands, scripts and programming that I get from the services. Most of the time even a complete amateur such as myself can do just as well without the help... And I use the latest models for this type of thing. After doing this I am always surprised how unintelligent these models tend to be.
This is also very noticeable when asking AI services to compare products that you already have knowledge of. The comparisons are very likely to be completely off base. When you argue with the models and provide documentation, they sometimes will continue arguing their original point of view even though it has been demonstrated that they are wrong.
I know that Hinton is a very experienced and very intelligent person, but I wonder if he might be living in a bubble surrounded by people who would rather agree with him than argue with him.
Sorry for the typos in my response.
10 years ago, Hinton warned that AI would replace radiologists within five years.
The need to contain the neutron flux may be an insurmountable problem.
Yeah the Beech V35 is just a beauty but I agree with you the Cherokee is better for the most pilots. We have one guy at the flying club who has one and he drops it in and go sideways and treats it really bad and it still keeps on going so Cherokees are pretty good.
The discussion in part has been about language. That being noted, I don't think you actually Laughed Out Loud, and as I am about the same age as Hinton, the "party" is more nearly to be over for us than for many younger folks.
As to your question about my view of Hinton, I think his latest messages serve to 1) keep him in the public eye with the Cassandra-like warnings, and 2) the use of language itself is in question. But he's collected a lot of prizes.
The notion that a "neural network" is 1) biological, as in our collection of neurons, and 2) machine-level, as in programs of binary on-and-off electrical connections has been limited to those two. Here's a third: relationships of connections between creatures on this planet like us. Not very often addressed in that manner, but we all are nodes in networks. The "neural" part is just a happy language ripoff to make "machine" seem modeled on the biological.
In every age, there have been "we're doomed" messages. This age is not different. Besides, if one is a religious sort, then "doom" has theological meaning, while if one is not religious, "doomed" is some sort of evolutionary thing, and therefore in terms of natural systems, nothing to be concerned about. That which fails is supposed to fail, by a failed design of some sort. Experiments in science which fail are then about searching for the cause of failure(s). If the HIntons of this world are eventually "doomed" to fail as biological entities, then so be it. Besides he's my age, and doomed to fail in only a few years.
Dennett failed, first in his hubris to explain "consciousness" and then in dying a couple of years ago. I came across a presentation he'd made in which he explained that among his goals was the destruction of religious thought. Nice touch of doctoral, academic hubris. Our world is doomed -- now fill in the blank.
Aliens. Artificial intelligence and machine "learning." (A misuse of the word, "learning," in my estimation). The next pandemic. The nuclear war and nuclear winter to follow. Name your poison.
I think happily back to Searle, whose notion of then "computer theory of the mind is a fallacy" speaks to me. The connectionist researchers in cognitive science haven't gotten as far along as their PR. Searle's Chinese Room informs; "weak claims" about AI seem acceptable enough, but "strong claims" are still just claims. As another had noted, and as we see here on FR forums, people can and do claim anything and many things. The thing about claims is that they can be accepted or rejected with little examination, or very closely examined and then accepted or rejected on more insight. Machine learning as a phrase has become so accepted and used that rarely to people today even ask, what does that mean? Do machines learn? Are we in a Philip K. Dick world? As much as in a Hinton world?
So much of language is elusive to machines these days that there remains easily tricked GIGO. SO one can -- I use the following word in its deep sense -- BELIEVE in whatever artificial intelligence is, or one can be at the minimum agnostic. I am not a believer in "machine learning" nor "artificial intelligence," though I am most certainly an ardent proponent of learning and bettering one's intelligence. The remainder seems more like marketing and lots of awards ceremonies ( I had a couple, back when... ) and, as we see in the news, enormous amounts of venture capital chasing whatever "AI" is, with the large portion of it now being lost.
In our "neural network" of investors and our "neural network" of societies and civilizations, those who seek money and power seek it "good and hard." AI is a latest lure, a new tulip mania. Sort of.
Hinton will shortly exit this world, being wholly biodegradable, his Cassandra shrieks fading. Dennett's already well and truly faded. Searle and Turing and most ever born on this planet. The world is ending! Apocalypse now! It ended for Philip K. Dick, so one can wonder if he is "dreaming." Do machines dream? In their electric sleep, he'd famously asked. The old game is the new game, it seems to me. Hinton, an impressive opus of work behind him, is shortly to exit the stage, as am I. Paul Ehrlich's "population explosion" is dying out, and he is dead. Hawking's "path to the planet's destruction" is still in reprint, bur Hawking is not. And among Hawking's end-of-world notions from 2017 was President Trump, so there's that.
I forgot that Harvard professor and his pet shape-changing asteroid in the headlines only a while back! Oh darn. Cassandras.
Employing the "multiple intelligences" notion from Project Zero -- there being a number of "project zeros" these days -- the domain or mental module of language can be lost to aphasics, just as vision is lost to the blind. Language -- I refer to the fast scraping of texts which LLMs have become noted for -- remains elusive. If one can ask does a machine dream its electric dream, perhaps one might challenge with "can a machine love?" An often rather intelligent thing to do. With its Gödel-incomplete outcomes, since love is sometimes blind. And, from our literature, even stupid.
Hinton seems quite a Cassandra, among Cassandras. He is worried for a future he won't live to see. Or maybe he's enjoying the celebrity in his last years? Extinction! Be worried! "Be afraid, be very afraid" since Cronenberg told us to be, back in 1986.
I'd rather laugh out loud -- while awaiting my extinction. But I won't. I'll just smile, and "enjoy the party while it lasts." Best wishes.
Must be why he’s a billionaire huh?.
Thank you for another well thought out post which displays an uncommon understanding of the actual state of AI which is goes far beyond the musings of most of us here.
Besides certain somewhat unpleasant societal changes that AI is contributing to... the real question in terms that will eventually become apparent are how much AI tools will increase productivity. A massive amount of money has already been spent on paper at least, and the total in currently unverifiable productivity in the USA is supposedly around 1.3%.
This is a fairly pathetic return on massive investments that have already been made. This also does not support the hype that massive numbers of jobs are being eliminated. I do not mean to minimize the pain and anxiety that all forms of progress cause those who find that their current means of supporting themselves is no longer in demand. I and most others quickly approaching our approaching our twilight years have experienced these phenomena through the years. Nothing stays the same forever. And electronic gizmos have profoundly changed the world we live in and will continue to do so.
I have been trying to further my understanding of AI by actually diving in and using the tools that have been made available to the public. What I have found from my limited knowledge and experience is that this area has a massive amount of hype associated with tools that are useful but at this point are not revolutionary and are a mixed bag. I really enjoy generating goofy images and video clips. As a long time user of Band in the Box type software packages, I believe that Suno, https://suno.com/ is absolutely amazing by comparison. And the ability to get search results to queries that are more complete and specific than we got back in previous times is quite useful as well.
But on the downside, YouTube and many other sites that I enjoy using have been filled with unimaginable quantities of “AI Slop”. I often spend time looking over reference material only to realize that it is some type of AI generated worthless crap. The scripts, code and commands that I get back from even the latest AI models when working on my “computer home lab” often are a mess that do not work for various reasons and I spend hours going down AI generated rabbit holes only to realize later that I would have been better off to try and figure things out without the AI “assistance”.
So, while more and more AI assistance and interference in our lives is inevitable it is a mixed bag. What I fear is that the most prevalent use will be... is to build huge databases on all of us which will be used by governments and industry to keep track of basically everything each and every one of us does. This is already being utilized in massive ways with no end in sight. So I fear that these tools will be used against us in the future in ways that are still almost unimaginable. It is a dystopian future that I fear my grandchildren will be saddled with, and it makes me glad that I grew up in the point of time that I did and feel sorry for younger people who did not get to experience a world with all the distractions that we have now.
And, if my entrepreneurial ventures hit pay dirt, then I'll be in far better shape than my previous corporate situation.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.