Are OpenAI and Anthropic losing money on inference?

There's the usual issue of a CEO "talking their book" but there's also the fact that Sam has a rich, documented history of lying. That was the central issue of his firing. "Empire of AI" has a detailed account of this. He would outright tell board member A that "board member B said X", based on his knowledge of the social dynamics of the board he assumed that A and B would never talk. But they eventually figured it out, it unraveled, and they confronted him in a group. Specifically, when they confronted him about telling Ilya Sutskever that Tasha McCauley said Helen Toner should step off the board, McCauley said "I never said that" and Altman was at a loss for words for a minute before finally mumbling "Well, I thought you could have said that. I don't know."