oldGPU wrote on 2026-10-31, 06:25:

Everyone has predicted the AI bubble popping for over a year.

I was around for the Dotcom bubble took it a few false starts to pop too. Thee AI bubble will be no different, might even take longer as there is more money involved backed up by trillion dollar companies.

When you see the bigger investors start to flee and see nVidia start to backpaddle in a big way by clawing back the money its dumped into AI, that's when you should really start watching the markets. When the likes of nVidia, Microslop, Oracle and Amazon start to get cold feet itll cause a mass exodus from the smaller investors. There will then be a small gap where the biggest investors will try their best to assure everyone that things are ok and they shouldn't be worried ...then you'll see the rats try to abandon the AI ship with the likes of Scam Altmann and the other AI bros trying to get their pay outs and walk away before the bottom drops out of it.

Safe to say itll be a longer process than the Dotcom bubble, you just need to pay attention to the movers and shakers in the AI investment world.

You can be certain that even after it pops AI will still be around, its far too late to shove that cat back in the bag, I don't know what form AI will take if and when the bubble pops but I can see local AI being used over massive DC based AI. Afterall we already have the weights and models which don't require massive hardware to run and there are many of the bigger AI models that can already be run at home with some caveats but nothing that cannot be overcome.

We just need to hope the PC hardware market can recover in a way that is favourable to home PCs again and we must never forget the corporations that abandoned the consumer market to chase the "AI Bros" money ...the likes of Micron and Nvidia for example.

There are fewer players in this bubble and a lot more money involved, when it crashes it will be fast and hard and even the biggest survivors will be scarred.

Hopefully some of the pieced together power plant upgrades will be repurposed for other industries since we need that done anyway.

The Ai bubble pop will cause a massive recession, but the global trade decoupling down the road will be much worse.

Unknown_K wrote on 2026-10-31, 08:08:

There are fewer players in this bubble and a lot more money involved, when it crashes it will be fast and hard and even the biggest survivors will be scarred.

Hopefully some of the pieced together power plant upgrades will be repurposed for other industries since we need that done anyway.

The Ai bubble pop will cause a massive recession, but the global trade decoupling down the road will be much worse.

Fewer visible ones and a thousand others that are connected but not as visible to the general masses. By longer to happen I mean itll take longer for all the players involved to hit that oh shit we dun fucked up wall due in part to that circular funding money involved propping up lack of profits.

It all depends I guess on just how much longer nVidia can keep bailing them out with cash.

IIRC, a couple of Texans tried to buy up all the silver in the USA back in the 1970s. For a time this made silver unaffordable, not just for jewelers, but for industrial manufacturers, which also use large quantities of silver.

This ended when the commodities exchanges, supported by the government, halted the purchase and sale of silver. The price then fell back to normal and the two clowns responsible were bankrupted.

Errius wrote on 2026-10-31, 20:59:

IIRC, a couple of Texans tried to buy up all the silver in the USA back in the 1970s. For a time this made silver unaffordable, not just for jewelers, but for industrial manufacturers, which also use large quantities of silver.

This ended when the commodities exchanges, supported by the government, halted the purchase and sale of silver. The price then fell back to normal and the two clowns responsible were bankrupted.

Manufactured scarcity and market manipulation are how quite a few large sects of industry make money. Without it they'd be bankrupt like the Texans. But when conglomerates do it, it's ok. If you or I did it, we'd be in jail.

Trashbytes wrote on 2026-10-31, 07:09:

I was around for the Dotcom bubble took it a few false starts to pop too. Thee AI bubble will be no different, might even take l […]
Show full quote

oldGPU wrote on 2026-10-31, 06:25:

Everyone has predicted the AI bubble popping for over a year.

I was around for the Dotcom bubble took it a few false starts to pop too. Thee AI bubble will be no different, might even take longer as there is more money involved backed up by trillion dollar companies.

When you see the bigger investors start to flee and see nVidia start to backpaddle in a big way by clawing back the money its dumped into AI, that's when you should really start watching the markets. When the likes of nVidia, Microslop, Oracle and Amazon start to get cold feet itll cause a mass exodus from the smaller investors. There will then be a small gap where the biggest investors will try their best to assure everyone that things are ok and they shouldn't be worried ...then you'll see the rats try to abandon the AI ship with the likes of Scam Altmann and the other AI bros trying to get their pay outs and walk away before the bottom drops out of it.

Safe to say itll be a longer process than the Dotcom bubble, you just need to pay attention to the movers and shakers in the AI investment world.

You can be certain that even after it pops AI will still be around, its far too late to shove that cat back in the bag, I don't know what form AI will take if and when the bubble pops but I can see local AI being used over massive DC based AI. Afterall we already have the weights and models which don't require massive hardware to run and there are many of the bigger AI models that can already be run at home with some caveats but nothing that cannot be overcome.

We just need to hope the PC hardware market can recover in a way that is favourable to home PCs again and we must never forget the corporations that abandoned the consumer market to chase the "AI Bros" money ...the likes of Micron and Nvidia for example.

I think ultimately local AI will be what provides the death knell. That's why nvidia and other companies are trying to keep powerful pcs out of the market and shove these chromebook style cloud AI laptops down our throats.

Right now AI is sucking in all the money in the world. But it can't suck forever. Sooner or later it's gonna blow. Reminds me of Spaceballs.

kingcake wrote on 2026-10-31, 21:33:

I think ultimately local AI will be what provides the death knell. That's why nvidia and other companies are trying to keep powerful pcs out of the market and shove these chromebook style cloud AI laptops down our throats.

The difference in system requirements of local LLMs and frontier models is staggering. That kind of hardware won't be affordable to private consumers for a long time.

RetroPCCupboard wrote on 2026-11-01, 07:05:

kingcake wrote on 2026-10-31, 21:33:

I think ultimately local AI will be what provides the death knell. That's why nvidia and other companies are trying to keep powerful pcs out of the market and shove these chromebook style cloud AI laptops down our throats.

The difference in system requirements of local LLMs and frontier models is staggering. That kind of hardware won't be affordable to private consumers for a long time.

I would question the long term usefulness of the Frontline models for the general public, your everyman would be better served by a locally ran model that is trained around them and their life and is able to cater its functionality towards bettering that life. These local models don't require the stupid levels of hardware the frontline models do, your average home PC can already run many of them at a level that is more than acceptable.

The K-boom is happening... Me so happy camper ! 😀

Trashbytes wrote on 2026-11-01, 07:43:

I would question the long term usefulness of the Frontline models for the general public, your everyman would be better served by a locally ran model that is trained around them and their life and is able to cater its functionality towards bettering that life. These local models don't require the stupid levels of hardware the frontline models do, your average home PC can already run many of them at a level that is more than acceptable.

Well, yeah, but depends what you want to use them for. Local image generation isn't very good as far as I am aware. Maybe I am wrong. Nor is programming. I haven't really tried the local ones for anything else. What do people use the local ones for?

I would think that most of the general public won't have a PC that can run one. Maybe PC gamers, but that's about it. Most people use their phones or a tablet now. Possibly a laptop if they do more than basic tasks.

RetroPCCupboard wrote on 2026-11-01, 12:21:

Trashbytes wrote on 2026-11-01, 07:43:

I would question the long term usefulness of the Frontline models for the general public, your everyman would be better served by a locally ran model that is trained around them and their life and is able to cater its functionality towards bettering that life. These local models don't require the stupid levels of hardware the frontline models do, your average home PC can already run many of them at a level that is more than acceptable.

Well, yeah, but depends what you want to use them for. Local image generation isn't very good as far as I am aware. Maybe I am wrong. Nor is programming. I haven't really tried the local ones for anything else. What do people use the local ones for?

I would think that most of the general public won't have a PC that can run one. Maybe PC gamers, but that's about it. Most people use their phones or a tablet now. Possibly a laptop if they do more than basic tasks.

Valid points but if you are looking to run one locally then you would also already know what itll need to run. As for what you would use it for .. pretty much the same things you currently use them for, many of the ones you can run locally currently have the ability to link back to the internet for weights and training if you allow it to. Image and media generation will always require a cloud based solution, I cant see that ever changing even if the bubble pops.

InterstateSeventyFix wrote on 2026-11-01, 05:30:

Right now AI is sucking in all the money in the world.

Imagine what Dr Sbaitso would have been capable of with the injection of a mere quadrillion dollars and its own electrical grid.

When the bubble pops the average person is going to become a billionaire. Invest in a wheelbarrow today so you can more easily cart your $1M bills to the fireplace tomorrow!

Big Pink wrote on 2026-11-01, 15:23:

Imagine what Dr Sbaitso would have been capable of with the injection of a mere quadrillion dollars and its own electrical grid.

Three pairs. Two words. One parity.
Bring your own questions. It's *chef's kiss*

Trashbytes wrote on 2026-11-01, 07:43:

I would question the long term usefulness of the Frontline models for the general public, your everyman would be better served by a locally ran model that is trained around them and their life and is able to cater its functionality towards bettering that life. These local models don't require the stupid levels of hardware the frontline models do, your average home PC can already run many of them at a level that is more than acceptable.

I have been running GPT4ALL on my main gaming rig (because it has a 4080 Super) and indeed it performs beautifully. I made it index all my papers, invoices etc in PDFs from Paperless NGX and I can query it in natural language. It is not perfect but acceptable.

I agree that the frontier models are perhaps not that useful for the general public. It'd probably be more efficient to have several, smaller but specialized models depending on the tasks...

Trashbytes wrote on 2026-11-01, 07:43:

RetroPCCupboard wrote on 2026-11-01, 07:05:

kingcake wrote on 2026-10-31, 21:33:

I think ultimately local AI will be what provides the death knell. That's why nvidia and other companies are trying to keep powerful pcs out of the market and shove these chromebook style cloud AI laptops down our throats.

The difference in system requirements of local LLMs and frontier models is staggering. That kind of hardware won't be affordable to private consumers for a long time.

I would question the long term usefulness of the Frontline models for the general public, your everyman would be better served by a locally ran model that is trained around them and their life and is able to cater its functionality towards bettering that life. These local models don't require the stupid levels of hardware the frontline models do, your average home PC can already run many of them at a level that is more than acceptable.

Yes this is correct. Local models can do most basic tasks that AI is actually useful for. You'll still need powerful cloud models for making hentai of your neighbor or whatever.

RetroPCCupboard wrote on 2026-11-01, 12:21:

Trashbytes wrote on 2026-11-01, 07:43:

I would question the long term usefulness of the Frontline models for the general public, your everyman would be better served by a locally ran model that is trained around them and their life and is able to cater its functionality towards bettering that life. These local models don't require the stupid levels of hardware the frontline models do, your average home PC can already run many of them at a level that is more than acceptable.

Well, yeah, but depends what you want to use them for. Local image generation isn't very good as far as I am aware. Maybe I am wrong. Nor is programming. I haven't really tried the local ones for anything else. What do people use the local ones for?

I would think that most of the general public won't have a PC that can run one. Maybe PC gamers, but that's about it. Most people use their phones or a tablet now. Possibly a laptop if they do more than basic tasks.

Man, you are out of touch. Even phones run and accelerate local models now.

kingcake wrote on 2026-11-01, 19:27:

for making hentai of your neighbor

🤣 so THAT's what it's for. I was using it all wrong. Boy I feel silly.

kingcake wrote on 2026-11-01, 19:28:

Man, you are out of touch. Even phones run and accelerate local models now.

I have tried using an LLM on my phone (Gemma 4) and it was useless. Context was tiny. The only positive was that it was faster than I thought it would be.

I have also tried larger models on my RTX 3080 Ti 12Gb and was also unimpressed. Local LLMs, for me, are a waste of time.

Even the frontier models aren't perfect. I have to use copilot at work for software development. The code it generates by default is poorly structured, verbose, overly defensive, fragile and not very readable. Don't get me wrong, it is helpful when guided, but I wouldn't trust it architect a full software solution itself. I also don't trust it to do only what I asked it to do, as often it spontaneously changes things you didn't ask it to.

I know nothing about the inner workings of AI LLMs etc, but I have been browsing various dev forums and it is fascinating to see the gearhead talk about hardware and the various versions of their models. Completely new language to me. People buying 10K Apple MacBook Pros with 8TB NVMe drives and 128GB of RAM to run 70B models. The newly refreshed Apple Mac Studio workstations that can climb to 30K geared with 16TB NVMe drives and 512GB RAM so they can run massive LLM models. Of course I got educated in other forums of how smaller models run lightning fast on 32GB 5090 cards. Last, but not least, I browsed through all the NVidia DGX Spark stations that you can attach as a breakout box to your current system. Gotta say, I'm lost in the wilderness in all this and the gear I see people using is certainly a rich man's game if you want powerful localized models (especially at the current prices of hardware).