According to OpenAI, yes. They're "very profitable" on inference. It is the training and capex (driven by competition) that is driving loss.
According to OpenAI, yes. They're "very profitable" on inference. It is the training and capex (driven by competition) that is driving loss.
https://www.wheresyoured.at/oai_docs/ offers a different point of view on OAI's profitability on inference.
https://news.ycombinator.com/item?id=45050415
But isn't that their "moat"? If the moat is the source of losses, then yes, they're operating at a huge loss.