This is a common myth, inference is not typically run at a loss, despite claims. It’s only a loss if you include staff and ongoing training costs. They could lock in their models now and be profitable if they wanted to.
Comment on Vibe management
i078@europe.pub 6 hours agoGiven the ai companies are running at al loss, it’s fair to assume which if these is likely
pennomi@lemmy.world 5 hours ago
adb@lemmy.ml 5 hours ago
Yes, and let’s also not count all the investments in infrastructure because you know… like training and staff it’s not a real cost that’s essential to the business.
Anyways, you wouldn’t happened to have heard that from Anthropocene or OpenAI?
Somehow we don’t have any actual indisputable numbers (I wonder why) but it is actually quite controversial and some of those who have done deep research on the subject are saying inference IS run at a loss and it might not get profitable ever.
pennomi@lemmy.world 5 hours ago
We do have numbers from comparably sized Chinese models.
Yes, every AI company is bleeding money, they’re not healthy in any way. But inference by itself is profitable, based on everything that we know.
Inference + amortizing the training costs is NOT profitable, which is what most people are talking about.
youcantreadthis@quokk.au 4 hours ago
Those people disagree with me and what I want are therefore wrong
akwd169@sh.itjust.works 5 hours ago
“Inference is not typically run at a loss”
Bro thats called cherry picking
Businesses work on cash in cash out
Right now AI companies make way less cash than they spend overall when you dont include investments
Furthermore, most people use a free version of AI and would stop using it if it cost them anything
Explain how to pivot to profit when the investments dry up, were all waiting
pennomi@lemmy.world 5 hours ago
I’m not saying they’re healthy, I’m saying that inference is the one profitable part of their business.
They’re all going to die because training costs dwarf the inference, and training doesn’t generate ANY revenue.
rambling4491@feddit.online 4 hours ago
Furthermore, most people use a free version of AI and would stop using it if it cost them anything
Do companies tend to use the free version too?
akwd169@sh.itjust.works 2 hours ago
Do rhetorical questions add to the conversation?
Rentlar@lemmy.ca 5 hours ago
You know that wouldn’t happen. Which AI company wants to be the one that says, “we’re happy with where the model is at right now” and stops throwing cash into the boiler of the investor hype train and let their competitors exceed them in real or imagined metrics? Clearly firms like Anthropic have to rely on circus marketing tricks like “This model is too dangerous for the general public to see! Ooooh scary! Coming Soon!”, and they can’t do that without continuous training.
For you and I, the offline models aren’t too bad for getting little side projects started, but for major AI firms, the ongoing training cost for the next model and the one after that has become ingrained into the operating model.
pennomi@lemmy.world 5 hours ago
I’m aware! I’m not saying they are healthy in any way. I’m just correcting that specific misinformation, because truth is important.
These companies are fucked if they keep operating the way they currently are, and I strongly suspect it’s going to pop like the dotcom bubble, but worse.
LurkingLuddite@piefed.social 5 hours ago
It’s both.
FinjaminPoach@lemmy.world 4 hours ago
Precisely. The question then is, which one is the main driver? I think it does fall on energy cost/the ridiculous scale of infrastructure they’ve decided is required to sustain AI companies.
Conclusion (for a luddite) is that One could cripple AI companies if simply prevented them from finishing their data centre every time. Goodness, it’s like a RTS strategy game where you have to build a monument to win the game.
If the other one is the main driver of this, purely an inflated profit margin, it indicates that AI is already collapsing and they’re desperately trying to scrape more venture capital off the back of the businesses that haven’t clued-in the how ineffective AI usually is.