Certainly part of the sales pitch. But so far it turns out humans are more efficient (cost less). I think the appeal to companies is the control (and the cost while it’s so heavily subsidized by the industry pushing it). The appeal to the major AI investors and execs is to… privatize the profits and socialize the losses. They will golden parachute themselves and leave the people with their mess.

The vast majority of the costs are HW and infra

I think they're hoping that reaches more of a steady state

https://www.wheresyoured.at/the-subprime-ai-crisis-is-here/

It could be, but it doesn’t look promising - and the fact that it’s pretty much impossible to know what the actual costs are is, in itself, very telling.

When you use these services, the company in question then pays for access to the AI models in question, either at a per-million-token rate to an AI lab, or (in the case of Anthropic and OpenAI) whatever cloud provider is renting them the GPUs to run the models. A token is basically ¾ of a word.

As a user, you do not experience token burn, just the process of inputs and outputs. AI labs obfuscate the cost of services by using “tokens” or “messages” or 5-hour-rate limits with percentage gauges, and you, as the user, do not really know how much any of it costs. On the back end, AI startups are annihilating cash, with up until recently Anthropic allowing you to burn upwards of $8 in compute for every dollar of your subscription. OpenAI allows you to do the same, though it’s hard to gauge by how much.

That’s ’embrace, extend, extinguish’ for you. Question is if there is a profitable model to come. The usual economies of scale don’t seem capable of adding up in this case. Even the maniacs on Wall Street are balking.