There’s no way that the long term billing for AI is tokens. It’s not budgetable, it’s arbitrary, and eventually people will figure out ways to either game it or evade it.
It’s much more likely that they get crammed full of ads and become useless than it is that they become profitable charging via token IMO.
If they really wanted to, they could charge for “GPU time” directly. Usage gets a bit complicated with batching and context caching, but its definitely a metric that could be precisely derived.
Don’t get me wrong though. That whole business model is doomed.
The plan is for them to make agentic AI and charge a yearly salary per agent. They are obviously not close to that goal, or you would hear more about it.
There is applications and ways to make money. But competition isn’t hard to come by and profit will be under expectations.
So my expectations is “skill-based programming” will become too expensive as a long-term solution and people who went that route will quickly become uncompetitive.
Pretty much all cloud compute is charged by usage (CPU time, storage, ingress, egress, class X api calls, etc), and is notoriously unpredictable. Businesses deal with it because, despite the unpredictability, it is cheaper than any other model could viably be offered at.
Consumer facing products will likely adopt different models; but “AI” is not a product, and most AI products will be bought by businesses.
There’s no way that the long term billing for AI is tokens. It’s not budgetable, it’s arbitrary, and eventually people will figure out ways to either game it or evade it.
It’s much more likely that they get crammed full of ads and become useless than it is that they become profitable charging via token IMO.
Free Version: Full of Ads
Premium Version: First adertised as Ad Free, then within a few years Full of Ads
Extreme Version: Ad Free for now, but features segregated into additional subscriptions.
Ad Free = we will not charge you for the ads
Eh, it’s close enough?
If they really wanted to, they could charge for “GPU time” directly. Usage gets a bit complicated with batching and context caching, but its definitely a metric that could be precisely derived.
Don’t get me wrong though. That whole business model is doomed.
The plan is for them to make agentic AI and charge a yearly salary per agent. They are obviously not close to that goal, or you would hear more about it.
That also will not work.
There is applications and ways to make money. But competition isn’t hard to come by and profit will be under expectations.
So my expectations is “skill-based programming” will become too expensive as a long-term solution and people who went that route will quickly become uncompetitive.
I’m unsure what you’re referring to. Tell me more if you will.
deleted by creator
Pretty much all cloud compute is charged by usage (CPU time, storage, ingress, egress, class X api calls, etc), and is notoriously unpredictable. Businesses deal with it because, despite the unpredictability, it is cheaper than any other model could viably be offered at.
Consumer facing products will likely adopt different models; but “AI” is not a product, and most AI products will be bought by businesses.
Sure but none of that is the same as tokens.