this post was submitted on 18 Nov 2026
121 points (98.4% liked)

No Stupid Questions

4500 readers
13 users here now

There is no such thing as a Stupid Question!

Don't be embarrassed of your curiosity; everyone has questions that they may feel uncomfortable asking certain people, so this place gives you a nice area not to be judged about asking it. Everyone here is willing to help.


Reminder that the rules for lemmy.ca still apply!


Thanks for reading all of this, even if you didn't read all of this, and your eye started somewhere else, have a watermelon slice ๐Ÿ‰.


founded 4 years ago
MODERATORS
 

To be honest, I'm pretty sure I already know the reasons but I'd like to see other people's takes on it.

you are viewing a single comment's thread
view the rest of the comments
[โ€“] queermunist@lemmy.ml 4 points 3 days ago (1 children)

Notably, it doesn't make any money. Not the image generators, not the vibe coding, it basically has no use in a profit-driven sector.

That's why they're pushing it, though. They're trying very hard to find a way to make this huge investment pay off. Hasn't worked yet.

[โ€“] Omgpwnies@lemmy.world 1 points 3 days ago (1 children)

It does and it doesn't. Companies who make the bare metal hardware can still profit. Companies who use AI tools intelligently will profit.

OTOH, open models are getting better and better. Non-AI tech companies are already building out local capacity to run open models in-house instead of paying OpenAI/Anthropic/Microsoft/etc for models and compute time.

The bubble is in the fact that so much money has been pushed into these middlemen, and end-users are becoming more and more able to sidestep them. It's like the 'selling shovels in the gold rush' except they bought all these shovels, but the prospectors found that they could just buy the handle and blade from the stamping factory directly and make the shovel themselves for a fraction the price, so now nobody wants to buy the pre-made one.

When you look at it like that, then it makes sense why they're trying to allocate all the future manufacturing capacity from chip makers - they want to be the only one with the parts to make the shovels.

[โ€“] queermunist@lemmy.ml 1 points 3 days ago

It's probably useful to differentiate between local models and hyperscale models.

The hyperscalers do not make money and they never will. Companies can make money using hyperscale AI because they're able to buy tokens that are kept cheap to drive up demand, but the instant hyperscalers try to charge the real cost those businesses leave.