this post was submitted on 03 Dec 2025
551 points (98.4% liked)
Technology
77084 readers
2446 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
But the profit absolutely can materialize because it is useful.
Right now the problem is hardware / data center costs, but those can come down at a per user level.
They just need to make it useful enough within those cost constants which is 100% without a doubt possible, it's just a matter of can they do it before they run out of money.
Edit: for example, nvidia giving OpenAI hardware for ownership helps bring down their costs, which gives them a longer runway to find that sweet spot.
The current machine learning models (AI for the stupid) rely on input data, which is running out.
Processing power per watt is stagnating. Moors law hasn't been true for years.
Who will pay for these services? The dot com bubble destroyed everyone who invested in it. Those that "survived" sprouted off of the corpse of that recession. LLMs will probably survive, but not in the way you assume.
Nvidia helping openAI survive is a sign that the bubble is here and ready to blow.
Thats part of the equation, but there is still a lot of work that can be done to optimize the usage of the llms themselves, and the more optimized and refined they are, the cheaper it becomes to run, and you can also use even bigger datasets that weren't feasible before.
I think there's also a lot of room to still optimize the data in the data set. Ingesting the entire worlds information doesn't lead to the best output, especially if you're going into something more factual vs creative like a LLM trained to assist with programming in a specific language.
And people ARE paying for it today, OpenAI has billions in revenue, the problem is the hardware is so expensive, the data centeres needed to run it are also expensive. They need to continue optimizing things to narrow that gap. Open AI charges $20 USD/month for their base paid plan. They have millions of paying customers, but millions isn't enough to offset their costs.
So they can
This is so early that they have room to both improve 1 and 2.
But like I said, they (and others like them) need to figure that out before they run out of money and everything falls apart and needs to be built back up in a more sustainable way.
We won't know if they can or can't until they do it, or it pops.
Oh ffs … stfu.