Aus diesem Grund sind die Bedenken hinsichtlich einer KI-Blase größer denn je
https://www.npr.org/2025/11/23/nx-s1-5615410/ai-bubble-nvidia-openai-revenue-bust-data-centers
Aus diesem Grund sind die Bedenken hinsichtlich einer KI-Blase größer denn je
https://www.npr.org/2025/11/23/nx-s1-5615410/ai-bubble-nvidia-openai-revenue-bust-data-centers
39 Kommentare
Apologies for the clickbait title from NPR.
>Circular deals raise even more concern
>Another aspect of the over-heated AI landscape that is raising eyebrows is the circular nature of investments.
>Take a recent $100 billion [deal](https://nvidianews.nvidia.com/news/openai-and-nvidia-announce-strategic-partnership-to-deploy-10gw-of-nvidia-systems) between Nvidia and OpenAI.
>Nvidia will pump that amount into OpenAI to bankroll data centers. OpenAI will then fill those facilities with Nvidia’s chips. Some analysts say this structure, where Nvidia is essentially subsidizing one of its biggest customers, artificially inflates actual demand for AI.
>“The idea is I’m Nvidia and I want OpenAI to buy more of my chips, so I give them money to do it,“ Kedrosky said. „It’s fairly common at a small scale, but it’s unusual to see it in the tens and hundreds of billions of dollars,“ noting that the last time it was prevalent was during the dot-com bubble.
Please pop. Please crash.
Please pop. Please crash.
Please pop. Please crash.
Please pop. Please crash.
There is also the fact that Sam Altman is publicly asking for a government bailout.
The key question is does everyone using AI today really need and want to pay for the service. Running these AI services is not cheap. I suspect that once the investors start to demand positive profit numbers things will go down fast.
’special purpose vehicle‘ I.e. laundering. When investors try to allay the public’s fear, there is fear.
Jensen/Nvidia got lucky so many times, it’ll run out eventually
Man….really looking forward to yet another tech bubble popping/crashing in my lifetime….lol
The bubble is real, the pop will be devastating.
What does their final AGI (but not really AGI) look like? Is it a tool that’s so good at solving problems that any business can and should use it in their day to day? If so, what’s stopping customers from using said tools themselves instead of paying for a middleman to do the same?
If they need folks to pay for tools, that by their nature, heavily reduce or completely eradicate the need for human services/middlemen, how long does it take for the value or services to tank to 0? ChatGPT is a great tool for folks who have enough knowledge in a specific subject and understand context, but they need a motivation or goal (which are likely monetary) to use the tool in the first place.
If the final version of AGI is being the best solution across the board, then folks will lack what are mostly financially driven needs, because all the work that AGI would help with would be rendered worthless, and customers would stop existing across multiple industries.
There are no concerns about whether it’s a bubble. It IS a bubble, we should be concerned about apprehending the ones responsible and mitigating the consequences.
Very tiresome to see these douchey tech bros strutting around like they’re God’s gift to humankind, especially when they haven’t gone nearly far enough to provide a convincing argument to everyday consumers about how this tech will be significant in their daily lives. Right now, it looks significant to well-off investors and not much else, and the vibe around it reeks of the same edgelord bitcoin bullshittery.
Bombarding tv with fart-sniffing, circle-jerky ads about how great AI is (oooh! I can look up recipes and make goofy art!), wrecking local ecosystems, driving up utility bills.. I can’t wait to see this nuisance bubble pop.
What an enlightening article with new information I haven’t seen 50 times this week!
It’s very simple: trillions invested and billions being burned without a clear path to profitablity. The only actual profit being made by the shovel/pick supplier.
That this is a bubble is very obvious to anybody who isn’t an AI fanatic on a religious level. Unfortunately, there are many such people in Silicon Valley that just keep inflating it.
I’d pay $20 /mo for them to stop developing AI. Maybe then we can have affordable GPU and Ram prices again.
I will NEVER fucking pay for AI. It’s about only good for summarizing shit, and whittling down things, but still accuracy is in the shitter.
Not many used the internet in the early stages. Quicker to go to the library for research. However things changed .m
It’s gonna pop!
Everyone here acting like AI is just a chatbot lmao. It’s about to be integrated into every conceivable product on earth. Follow the money.
I love how the news cycle is like 6 months late to what we’ve been saying since then
Does the article mention the reality that language is just a tool to communicate information and does not represent intelligence? Because they’re absolutely busted on their turbo scam. There is factually zero intelligence in an LLM.
I really dont think people are capable of seeing the reality here.
Whatever you mean by AI, LLMS, robots, self driving etc. We are so close to any number of these and there is truly no reason to believe we wont get there for ANY.
So it wont just stop. The big guys can play at all these at once. Until they lose everywhere the game isn’t over.
Someday someone will break through it all and instantly obsolete everyone else.
But that is the end. Not we go back to before this moment and forget about it all because its expensive.
Sam Altman just told VCs that there is no revenue and they have no idea how to generate revenue with AI. Their brilliant plan is to ask AGI how to generate ROI for the investors. Nervous laughter followed his statement.
[source](https://youtube.com/shorts/YYQOf4sL9AI?si=rEziS4TzQPTIlsFI)
Let that bitch pop. It’s made life for humans worse. No bailouts.
I can’t wait for this bubble to pop.
Im my opinion with these federal stats reports not being releases along with the Naritive that AI is going to replace a lot of the work force. I wouldn’t put it past this administration of using the AI narrative as a scape goat for mass layoffs due to a recession.
Shots gonna hit the fan and they need something to point the blame at. Trump has been in office almost a year and still blames Biden for things going wrong.
Past decade or so tech has just changed to grifting. I can’t remember if it was a YouTube vid or news article but it discussed how regular tech innovation slowed in the 2010s, didn’t have the big leaps of the 80s to 00s so these tech companies just went in on crypto and AI just snake oil and grifting is all they had
Stop with the „bubble“ nonesense. If all you know is LLMs then you yourself are in a bubble. AI is way more than that. Until you realize that, you’re only peddling mainstream media BS.
I see so many legitimate uses every day. I can’t wait for the “crash” to happen. I would really like to buy more stocks.
Not sure why people think AI is just going to disappear!
POP THE BUBBLE! POP THE BUBBLE!
From a tech standpoint, the only reason I sometimes use AI is because search engines now suppress stackoverflow.
When the people at the top are saying the normal rules of the economy no longer apply, run for the hills. Seen this happen before..
AI will never pop …….. I just roll my eyes anytime I hear this
People keep saying AI is just a bubble waiting to pop but honestly that’s not how i t works. AI is everywhere now in healthcare, manufacturing, logistics, and marketing. It’s not some passing trend, but a crucial tool helping businesses run better and faster.
The tech behind AI keeps getting better too. It’s solving problems that old computers can’t touch like drug discovery, climate modeling, and smart automation. So it’s deeply tied to real breakthroughs not just hype.
Plus the amount of money and effort going into AI innfrastructure from huge tech companies and governments is insane. That kind of investment means AI isn’t going anywhere anytime soon. It’s a long-term game for them.
In short AI is more than hype.It’s a foundation for future technology and industry that’s here to stay. It’s too important and too well-backed to just fizzl e out.
LLM’s are useful for systems where we don’t have to validate processes – and for those processes if sort of ok is good enough than that’s great, but for many systems you might need to have verifiable processes in place where we KNOW a certain output is the expected result given a certain set of conditions. This makes LLM’s a dubious choice, meaning LLM’s can often produce a bad result , with two kinds of bad „the answer is correct“ by not for an objectively valid reason, „the answer is wrong“ for for a reason that follows a certain logic that’s not objectively correct.
Difficulty in managing emergent behaviors – Completely separate and apart from that , many (if not all) LLM’s have a tendency to hallucinate/cheat, and the models are subject to a whole rainbow of tasks that make the models themselves unreliable at best and fundamentally unusable at worst – whether it’s systems that „cheat“ to arrive at a „desired“ output, or systems that cannot be easily inspected for how a particular answer was derived – in this way , separate from the process isn’t working correctly (as mentioned above), here , even when you have a correct result, you can’t know (let alone prove) that the answer was derived by way of a robust process.
First principles errors – Underscoring this is the intentional limited visibility or knowledge of how a particular model is trained – we see this with ideologically crippled models like Grok, which fetish certain idiosyncrasies of bad training/reinforcement. So Mr. Musk sought out to make important contributions to LLM types of models , mission accomplished – he proved that under trivial conditions these models are not trustable, except in limited cases where the entire training set and secondary cultivation and reinforcement of model paths and such are able to be shown – which can still lead back to the earlier mentioned problems of not being validatable or with processes that are provably robust.
The rats nest of inter-related companies and vanishingly small number of regulatory folks that can’t even get the „serious“ people to pay any attention whatsoever – risk if it’s discussed at all seems to fall into two categories – fuzz-up/disregard misbehaving portions of model developed and an utter disregard to be even willing to discuss/disclose the nature of the models – separate and apart from the troubles with the models themselves is a fundamental unwillingness to the very open-ness most firms start with.
Wait till these companies / investors want their profits… then those AI token prices are going to skyrocket. It’s going to be expensive then people’s salaries.
And when they say ‚Scales of economy’… each prompt cost $$, so more queries more $$ (not cheaper), and then the hardware for processing will also detoriate. It’s going to get hectic expensive…
If you can use AI to write an email, was it a necessary email in the first place?
On the article itself, there are some things that are nonsensical. Amazon, Meta, Google, MSFT are investing $400B… An amount that wou require $250/iPhone. Well, apple makes the iPhone, so why make that comparison? More like there are 400mm Microsoft office subscribers. So if MSFT was paying the whole thing you’d need to make $1000/user extra, and if spread out over 10 years, that is about $8/month/user to pay for the whole investment. Google is making bank on it, so is Meta.
GPUs in the cloud also aren’t just for LLMs. My tiny little company was using them to build old fashioned deep learning models years ago. There are many use cases for them that aren’t chat bots.
On the work front some of the professional use cases and productivity enhancements are amazing. I am a meh-level coder even though I run a software company. With Gemini and some good prompting I am on par with our best coders without AI
And that is what has me convinced. The business value is amazing and I let going to get better. I can have it very quickly do things I used to pay an EA or entry level hire for. Have it go read through 100 parents and summarize. Have it read these 19 academic papers.
And the real value for my business will be once I can set it free to start marketing. Compose and send 100 personalized emails. Have a trigger to track replies and then take action. Book a meeting with a person, send a quote, etc.
If 10 employees cost me a million a year and AI can someday replace 7 of them, then I could pay say $300k/year.
Don’t know what the outcome will be. I just know I made a ton of money in tech AI related stocks in the last 3 years. Taking profits and letting the speculators keep speculating.
Other than the people who have a stake in it, who is concerned?