1194
Crucial is shutting down — because Micron wants to sell its RAM and SSDs to AI companies instead
(www.theverge.com)
this post was submitted on 03 Dec 2025
1194 points (99.4% liked)
Technology
77084 readers
1270 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related news or articles.
- Be excellent to each other!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, this includes using AI responses and summaries. To ask if your bot can be added please contact a mod.
- Check for duplicates before posting, duplicates may be removed
- Accounts 7 days and younger will have their posts automatically removed.
Approved Bots
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
AI was never meant to benefit the working class in any capacity.
Its a great rule of thumb that if you see oligarchs hype up something and push for it to be everywhere, its a BAD fucking thing.
Resource drain of LLMs inescapably makes them tools availiable only to big players. They are ideal in the way they are naturally gated. Making them mandatory == giving these select companies and people power over everything. And not only oligarchs' promotion, but the whole situation of them being given for free or cheap at a huge loss gives one an idea that there's a lot to milk from it's growing adoption.
But that's completely not true! Like, not a single thing you said is even slightly correct!
LLMs are relatively cheap to run - at small scales. You can run an LLM on your own computer right now. It won't be super fast, it won't have super skills, but you can run it, and you can train it yourself.
Massive LLMs like ChatGPT require tremendous resources precisely because they are not just tools available only to big players. Everybody on the planet has access to them - for free. The only actual difference there is between running an LLM locally and through a provider is that you get better speed and (sometimes, depending on context) better training through a provider.
As for "there's a lot to milk from its growing adoption" - maybe? Probably? Who knows? That's the "magic" of the AI bubble we're experiencing right now - the big players keep saying that it will "make work and money obsolete", that "anyone will be able to do anything", that "a time of post-scarcity approaches", and a billion other bullshit marketing slogans like that. But the reality is that nobody has yet figured out how to make money on that thing.
Right now, the only reason it's "growing", is because of the weird and probably illegal circular financing that's going on at the very top - Nvidia invests in OpenAI, which invests in Oracle, which invests in Nvidia - and so on. No money is actually being made or (often) even changing hands, but everyone can now show they've received a lot of investment which pumps up their stock prices. The only reason this hasn't popped yet is probably because the main investing parties are using tonnes of cash they had stored.
Growing adoption means nothing. It's a marketing tool for them to keep shareholders happy while they keep a literal investing circlejerk going, every now and again inviting another player into the fold.
Your private LLM will have nothing to compete against the big guys though. A cute hobby project but nothing of economic value.
You're not training it from scratch, though. There are people, enthusiasts, doing it for you. I can fire up LM Studio and browse through thousands of models to then have a conversation with, or have them write stories, etc., etc.
As for "nothing of economic value" - that's, again, just plain misunderstanding what AI can be used for. Corridor Crew - a VFX team publishing on YouTube - used self-trained AI to boost their film making options. For example, to copy the "bullet time" effect from The Matrix, they were able to use around a dozen cameras instead of hundreds, and then used AI to create the "in between" frames.
How does that have "no economic value", mate?