This makes no sense, their value in the marketplace is in usage and inflated promise, not actual revenues

> They do inference at a loss

They don't, inference is cheap, especially for agents because of cache hits. The API prices are just inflated.

Ive got a 'Claw interfacing with OpenAI and generating garbage questions and responses. I have an 8k context on mine.

Deletion with OpenAI isnt really deletion. So I'll waste their resources AND train on low quality slop on my side.

My work degrades theirs.