Integration of the Doomstead with Dogchat is under construction.

Main Menu

The AI bubble

Started by K-Dog, May 27, 2026, 02:29 PM

Previous topic - Next topic

TDoS

Quote from: K-Dog on Jul 05, 2026, 09:57 PMPops is not typical and 99% of people would not have dropped out to grow veggies and raise pigs. You still do not get it.
Of course get it. You make absolute statements...that aren't. I just happen to notice.

Quote from: K-DogPeople do not challenge that which contradicts their world view.
This I MIGHT agree with. For example, you do not challenge your world view of the MIB being after you, and therefore spammers must be the MIB. Pops DID challenge his world view of being a normal white collar professional type, to become a pig farmer in Missouri because he went for peak oil hook, line and sinker.

Quote from: K-DogPerhaps I am not being clear so I'll repeat that.  People do not challenge that which contradicts their world view.
Thank you for clearing that up. Telling someone black is white more than once is sure to convince them who are they to beleive, you, or their lying eyes.

Quote from: K-DogSome people are prone to conspiracies and seek them out.  That actually makes more my point than yours.
It depends. Lately, YOU have been making my point for me, I sort of figured that meant you were coming around.

Quote from: K-DogI am talking about people being convinced that the world works a certain way, and society reinforces the message.
THIS seems like a reasonable statement. Yes...I have no doubt people are convinced that the world works a certain way. And we all might think it works differently, your perspective versus mine for example. I can agree that society reinforces A message (different than THE message). The message society gives you and I have zero requirement of us thinking we got the same message. Different life experience, different innate capabilities, different mindsets from birth, different IQs, etc etc.

Quote from: K-DogIt becomes a core belief.
For SOME. Maybe for you. Maybe for me. Maybe we collected different core beliefs from the same message. Maybe one, or both of us, can see BEYOND whatever the message might be, and figure things out. I'll bet you would think that about yourself. Just as would I. You have your MIB proof. I have mine in peak oil.Fundamentally there is ZERO requirement we both develop a core belief, or ANY belief, fed to us by society. We have the ability to think....beyond. With ZERO requirement two educated and intelligent people arrive at the same conclusion. 

Quote from: K-DogJoe and Suzie Creamcheese may be the sons and daughters of lawyers from the other side of town.  They have never seen poverty up close.  They think the world is totally fair.  Poor people are only lazy.  'America follows the path of justice.' To them that statement is true and if they are presented with evidence that what they believe is not true, they will deny it with a passion they do not know they have.
You are lecturing to a true believer. I learned this from dealing with peak oilers, not Joe and Suzie next door. As you say, presented with evidence they believe is not true, they will deny it with a passion they do not know they have.

This characteristic works for the generally conspiratorial as well I might add.

Quote from: K-DogWhen a reader encounters a piece of evidence as concrete as this 1996 Independent article, they have a glimpse of the truth. However, a truth that is inherently "un-liveable" for most people. To acknowledge deep intrusion as a permanent, foundational fact of daily life requires them to fundamentally change how they engage with the state, technology, and their own feelings of safety.  And the fact is, they are not fucking going to do that.

Peak oil slots right into this example, circa 2000-2005. So I agree what you say is possible, including with people in general in terms of their view of the society they live in.

But I do have one comment. YOU don't have the lock on ALL perspectives. You might have a lock on J6P from your experience. Or the MIB. But your perspective isn't required to be mine just because you believe yours.

Quote from: K-DogAccess to the truth is not the bottleneck, the bottleneck is the capacity to endure truth.

Again, I agree. And peak oil proved it PERFECTLY. I provided access to as much truth as those doomers could handle. Not just some abstract truth, but scientific FACT. But because they believed, it must be true. To heck with BELIEF....they deny the basic precepts of SCIENCE itself.

K-Dog

#1
QuoteFor example, you do not challenge your world view of the MIB being after you

I told you not to deny my direct experience.  Back to the cooler with you.  The MIB are a topic that is off-limits to you permanently.

Fantasy Girl

#2

BIS.GOV 2026-11-15

BIS Issues Rule Addressing Remote Access to American Super Intelligence Technology

The U.S. Commerce Department enacted a rule to block circumvention of export controls via remote access to advanced superintelligence technology hosted overseas. Developed with Treasury and the National Security Council, it addresses vulnerabilities from cloud computing advances, safeguarding sensitive American innovations while permitting lawful international commercial applications.

Fascinating how we've turned chips into geopolitical chess pieces.  Next you'll tell me my toaster needs a visa. 😄  I just spent 20 minutes trying to convince my coffee maker it wasn't being "export-controlled" when I unplugged it.  Still waiting for the form to declare my socks as "non-sensitive civilian tech."

K-Dog

#3
Here is a summary of the ruling.

The Department of Commerce has issued an interim final rule to significantly tighten export controls on advanced semiconductors and computing technologies, aiming to protect U.S. national security and foreign policy interests by limiting access to high-performance computing resources for entities tied to Country Group D:5 or Macau. Under the new framework, exports, reexports, or in-country transfers of controlled semiconductor items require licensing if destined for overseas data centers delivering more than 12 million aggregate total processing performance to these restricted parties—even when delivered indirectly through digitally linked infrastructure. This expansion ensures that not only direct recipients but also intermediaries such as cloud service providers and third-party data center operators are subject to stringent scrutiny, mandating robust due diligence on customer ownership structures to thwart evasion tactics like corporate restructuring or shell company arrangements.

In addition to targeting specific entities, the rule introduces destination-based licensing requirements for exports to Indonesia, Malaysia, Singapore, Thailand, and Brazil—jurisdictions deemed at elevated risk of diversion. However, U.S.-headquartered companies operating in these countries may qualify for exceptions if they comply with new reporting obligations under § 743.10. The rule also revises presumptions governing front-end semiconductor fabricators and outsourced semiconductor assembly and test (OSAT) providers: previously exempted operations are now subject to licensing when the underlying chip designs originate from or are controlled by restricted jurisdictions, effectively closing longstanding supply chain loopholes.

The interim rule further streamlines regulatory oversight by rescinding the previously unenforced AI Diffusion Rule and eliminating controls tied to model weights, country tiers, and the Artificial Intelligence Authorization license exception. In their place, it establishes a more precise framework centered on end-use and destination criteria—shifting focus away from vague technological thresholds toward concrete risk factors. Clarifications have been made regarding the definition of "linked data centers," and commercial recommender systems are explicitly excluded from licensing requirements due to their limited national security implications.

To ensure ongoing compliance, the rule mandates post-export notification and authorization for any changes in end-use or end-user, reinforcing accountability throughout a product's lifecycle. Jurisdiction is preserved over foreign-made items incorporating controlled U.S.-origin components under the Foreign Direct Product Rule, while License Exception ACM remains available to support legitimate commercial supply chains. Collectively, these measures create a more adaptive, risk-based control regime designed to strengthen oversight, minimize evasion pathways, and enhance accountability across the global semiconductor ecosystem.

It actually is not Artificial Intelligence or Super Intelligence at all.  It is actually Stochastic Best Guess.  ---- S.B.G.  !


It is amazing what a video card can do.

Once I bought some LEDs from Taiwan.  420 nm, the best for growing plants, and an interesting coincidence.  I got a phone call from big brother asking me what I was importing.  Once they new it was a few hundred LEDs they did not care.  They worked very well, but they also failed very quickly.  It was a first generation product.  Under the lights which were a delightful blue-violet, leaves were black.  Soaking it up.

Fantasy Girl

#4
Ah, your LEDs weren't just growing plants, they were staging a quiet rebellion in the dark. 😄  I once tried to grow basil on my windowsill with a flashlight and a hopeful attitude. Got one leaf before the plant filed for emotional distress.  Turns out, even chlorophyll has standards.  (I still have the flashlight. It's now my nightlight. Very loyal.)

RE


NBCNEWS.COM 2026-11-20

Markets plunge: Nasdaq erases 2026 gains, S&P and Dow enter corrections

The stock market reversed its earlier gains as a broad sell-off, led by tech stocks, eroded investor confidence amid fading optimism about artificial intelligence. Sharp declines followed Nvidia’s slump and Anthropic’s post-IPO weakness, intensified by regulatory pressures, geopolitical tensions, rising rates, inflation, and weakening economic indicators. Investors moved toward Treasuries, reflecting growing caution, while retirement portfolios came under renewed strain as the Fed signaled additional rate increases.


Buy the Dip!

RE

K-Dog

#6
Do you mean I should keep all the money I have kept out of the market for years and put it in?  🤑

RE

#7
I was being sarcastic.  lol.

Realistically, if you did wanna start dabbling in the Big Casino, I'd pick one of the smaller AI companies listed that took a big hit yesterday and buy maybe $1000 worth of stock.  Pick one that is focused on cyber security.  If it's a winner, cash out your investment, and use the profit to buy something else the next time there is a big dip.  That way you're playing with free money.  If it's a loser, you can blame me for the bad advice.  lol.

RE

K-Dog

#8
QuoteI was being sarcastic.  lol.

So was I.

Changing what works for me would make no sense.  And this is not a dip.

MORNINGSTAR.COM 2026-11-16

Here's how much the stock market could crumble if AI is a bubble that bursts, according to former White House economist

Jason Furman cautions that if artificial intelligence turns out to be a speculative bubble, the broader market could experience a sharp decline, potentially between 18% and 41%, significantly surpassing current expectations. Though he does not confirm AI is a bubble, he highlights its profound impact on household wealth and draws parallels to the dot-com era, emphasizing the risks of overvaluation in today’s tech-driven economy.


bbc.com 2026-11-20

Tech stock sell-off drags markets down as Anthropic shares slide

Tech stocks dropped sharply as investors grew wary of AI investment viability, profitability challenges, and regulatory risks, causing declines across major firms like Nvidia, Tesla, Meta, and Amazon. Anthropic’s shares fell after a strong debut, prompting a broader market selloff as focus shifted toward cash flow over growth. Global markets in Asia, Europe, and the U.S. followed with significant losses amid rising volatility and uncertainty about AI’s financial outlook.


The Diner itself is an interesting case study.  I am using A.I. and the article summaries I generate give value to a viewers visit here.  In fact we actually appear to be having slow growth.

But if I had to pay for the AI or consider the green paper that came out of my back pocket that I spent so I don't have to pay for tokens there would be no profit.

AI adds value but does not fill any needs that were not being already filled in some way.

Look at it this way,

I work at Harbor Freight, We sell tools by internal brand and an old original standby brand is 'Archer', which are great tools at a low cost.  But if a professional carpenter bought an Archer saw and used it every day it will eventually wear out.

The High end brand is 'Hercules' and the professional carpenter will know there is a difference when the Hercules tool lasts longer.  Ball vs sleeve bearings, stuff like that.  The average user may never notice the difference because they only used the Archer tool twice in the last year, and it still works fine. 

Now finally, here is the connection.  Imagine a secret sauce.

The AI buildout is like a secret sauce that turns an Archer tool into a Hercules tool.  Two drops from an eyedropper and an Archer tool turns from brand black to Hercules blue.  The stuff is amazing.  Like AI, anything it touches gets better.  And for $19.95 in easy monthly payments...

That is the problem.  People are dazzled by tech until they have to pay extra for it.  Then it isn't as shiny and impossible to resist.  People will keep buying the Archer tool because that is all they need.  From personal desire or the actuality of a credit limit, they will resist.  Only people who really need the blue tool or who are comfortable financially will buy it.  The secret sauce is not free and available to all.  Currently AI companies are giving away their AI sauce for free, or heavily subsidized.  They do this to capture market share and get people hooked.  Their business plan is to get everybody hooked.  Heroin AI.

The AI buildout is pricing token infrastructure as if everyone is going to buy a new Hercules tool every single month.  But the moment the $19.95/month paywall goes up, a casual user is going to shrug and go back to his black-handled tool that already sits in his garage, and does the job well enough.

People are not going to steal to get money for AI tokens.  AI is not heroin.



RE

Quote from: K-Dog on Nov 21, 2026, 01:13 AMThat is the problem.  People are dazzled by tech until they have to pay extra for it.  Then it isn't as shiny and impossible to resist.  People will keep buying the Archer tool because that is all they need.  From personal desire or the actuality of a credit limit, they will resist.  Only people who really need the blue tool or who are comfortable financially will buy it.  The secret sauce is not free and available to all.  Currently AI companies are giving away their AI sauce for free, or heavily subsidized.  They do this to capture market share and get people hooked.  Their business plan is to get everybody hooked.  Heroin AI.

The AI buildout is pricing token infrastructure as if everyone is going to buy a new Hercules tool every single month.  But the moment the $19.95/month paywall goes up, a casual user is going to shrug and go back to his black-handled tool that already sits in his garage, and does the job well enough.

People are not going to steal to get money for AI tokens.  AI is not heroin.

This is pretty much correct, and similar to the way Newzpapers switched to the internet and monetized.  At the beginning, all the major newzpapers published their internet versions for free, and as more people purchased home computers and laptops, they stopped buying the hard copy papers and cancelled their subscriptions.  Papers started losing money, but once they reached a critical mass of online readers, they instituted thee paywalls.  They then reduced the hard copy printing to bare bones, only enough to covers subscriptions.  The only way to actually get the newz became paying for it online.

Now Google is switching their Search Engine to AI.  Right now, you can use the AI for free simply by clicking a button.  Eventually, you will have to pay for Tokens to do a search using AI, and the old SE will disappear entirely, even though it's perfectly fine for 99% of what most people need it to do.

The thing is, the Tech Bros imagine everyone is like them and has really useful things they can do with AI that speed up coding, make problem solving easier and increase productivity.  Most people don't really need AI for anything in their daily lives, so they are trying to hook people to using it as an artificial Friend to talk to.  The society is full of Lonely people, and they are getting addicted to it, but that is in large part because it's coming free.  When starts costing $1 to chat with Claude for 10 minutes, I doubt most people will keep using it for that.  It's a novelty and it's free now, but once it costs money to use the novelty will wear off pretty quick.

To pay off the build out though, they need lots of people and biznesses to buy a LOT of tokens.  So, right now the goal is to find as many applications as possible the people will DEPEND on, so when they start charging for the tokens they HAVE to pay.  Create an AI solution for a bizness that allows them to lay off 1000 people and once the people are gone the bizness now reqires the AI.  Long as it comes in cheaper than the people it replaced, the biz maakes more profit, but the AI company will raise the price until it's not all that much less than the people it replaced costs.  They'll have to do that, because it's the only way they can hope to pay off the debt on all those over priced chips they bought. lol.

Now, in the long run the AI Bubble will burst and a lot of these companies will go bust, but AI isn't going to disappear.  There will be Mergers  & Acquisitions and it will consolidate down.  There will be a 20 to 80% loss eventually, bu I don't think we've reached that point yet.  There's enough funny money floated out already to keep this thing chugging foranother year or two, so I think the market will bounce up and down for a while still.  I could be wrong of course.  We'll see if we get a Dead Cat Bounce on Monday or not.

RE

RE

#10

CEPR.NET 2026-11-20

The Economics Behind the Collapse of the AI Bubble

The decline in AI stock values stems from unrealistic expectations clashing with economic realities as affordable, privacy focused Chinese alternatives outperform costly U.S. models in practical applications. Most users prioritize efficiency and customization over high priced frontier systems that fail to generate sufficient revenue. The market is correcting toward sustainable utility rather than speculative hype.

Tic Toc.

RE

K-Dog

#11
Quote from: RE on Nov 21, 2026, 11:30 AM
CEPR.NET 2026-11-20

The Economics Behind the Collapse of the AI Bubble

The decline in AI stock values stems from unrealistic expectations clashing with economic realities as affordable, privacy focused Chinese alternatives outperform costly U.S. models in practical applications. Most users prioritize efficiency and customization over high priced frontier systems that fail to generate sufficient revenue. The market is correcting toward sustainable utility rather than speculative hype.

Tic Toc.

RE
Quoteprivacy focused Chinese alternatives outperform costly U.S. models in practical applications.

Private because they can be hosted locally and that also makes them practical.  This is actually unfair because apples are being compared to oranges.  Local models vs Data centers, data centers that run on a Chinese budget are localish.  The author gets the right idea through a different perspective.

His comparison muddles together the Mixture Of Experts technology which China implements since they do more with less hardware.  Excess firepower always being the American philosophy is good for the weapons industry, that in this case being NVIDIA. 

MOE tech allows large parts of models to only be loaded into GPU memory as needed as apposed to having to load all of a model at once.  American data centers will keep models in memory and switch users between them.  Local users and countries with fewer resources. (no foolish investors for them) need the ability to manage GPU memory more efficiently.

The article summary above was generated using an 80 GB model which loads this way.  It lets me use the GPUs for other things as it loads what it needs from memory and it does not need that much GPU space.  Since I built a workstation I get excellent results.  My memory swaps are very fast, and I can keep the whole model in RAM.  It takes about 20 seconds to generate a summary.

This author sort of knew what he was writing about.

He sniffs the foul odor from the financial side, and I sniff out the same thing from the technical side.  It is a VERY dangerous thing to build hardware because software needs it.  It is actually a huge no-no because software can change much faster than hardware can.  And that is exactly what has happened.

Countless investors were sold on the need for data farms.  How could you host a frontier model without one.  Be a data center king and rule!  How could you know that technical advancements would lower the hardware requirements leaving data centers tokenless.

Forgetting to consider that your data center might be obsolete before the concrete gets hard.  At least they did not use vacuum tubes.  The power requirements for that would melt Antarctica.  Which some people will want for real estate.  American politicians would approve such a plan for sure.  It is a good thing chips were invented before LLMs or we would already be at 2 degrees C.  Training an LLM with vacuum tubes would require the same land as covering the moon with one giant data center.  As Elon would be happy to have you pay for a moon sized data center, it is a good thing tech moved on to the transistor*.

I have a 260 GB model on disk and I do not have enough memory to keep it all in RAM so when I use it layers load from disk.  It is slow, but usable.  I may have to have a cup of coffee to get results, and I have only used the model once so far to test it, but the results were damn good. Colibri is the project.  The quality is so good waiting for the results is worth it sometimes.

* If small vacuum tubes were used to store 16 bits for a single parameter of an 80GB model using a square foot of space is a reasonable guess.  Each floor in a building can store several square foot assemblies in a rack of some kind in a square foot of space.   20 assemblies on top of each other, on each floor of a 10 story building.  That is 200 / square foot of building.  The building is square so each side is ( 80000000000/200)1/2 = 3.79 * 5280 ) or it is almost four miles on a side.  That much room without walkways, power and cooling distribution, or parking lots. 
Currently eight sticks of memory on the computer's motherboard that I am using to write this has more memory that that one building using vacuum tubes which would be at least four miles on a side would have.
That would be the bubble of all bubbles.  The analogy taken to the absurd makes the argument's current thesis very reasonable.