this post was submitted on 08 May 2026
87 points (100.0% liked)
Technology
42905 readers
195 users here now
A nice place to discuss rumors, happenings, innovations, and challenges in the technology sphere. We also welcome discussions on the intersections of technology and society. If it’s technological news or discussion of technology, it probably belongs here.
Remember the overriding ethos on Beehaw: Be(e) Nice. Each user you encounter here is a person, and should be treated with kindness (even if they’re wrong, or use a Linux distro you don’t like). Personal attacks will not be tolerated.
Subcommunities on Beehaw:
This community's icon was made by Aaron Schneider, under the CC-BY-NC-SA 4.0 license.
founded 4 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
It's going to be bad, too. These companies are repportedly selling tokens for 1/10 - 1/15 the actual processing costs. Companies are already spending significantly more than they would junior developers (to generate an un-revewable amount of code). Raising peices by a factor of 20 will kill the whole project dead. They'll need to hold out until there's no one left who knows how to write anymore to survive that kind of price hike.
Except that it'll never work out that way. Open models are almost as good and cost a tiny fraction of the cost of the proprietary models. There are no moats to protect their business model. Anyone can come along and eat their lunch.
AI has no path to profitability since it's going to be commoditized. There isn't a big enough difference between individual models to justify the price premium of paying $100/million tokens when open models cost 10¢/million tokens.
And it gets even worse when you consider specialized models; the real future is likely going to be custom training models for specific use cases, trained on the company's data (and other data too, of course). A much smaller model can be much more successful on tasks it has been trained on. It'll cost a tiny fraction of the compute of a mega model to train and likely beat mega models on tasks within its training domain. And it can run entirely on the company intranet, so there are no real privacy/security concerns.
Right now, the big players are giving away their compute at cents on the dollar, so there's not much incentive to run local models. As soon as they start to push pricing to try to become profitable, companies will switch to in-house models.
OpenAI is doomed. I doubt they'll be relevant in a decade.