> All model training is in R&D Yes. But is all R&D model training? That was the actual question. Everybody that is presuming "inference is profitable" here is

> All model training is in R&D Yes. But is all R&D model training? That was the actual question. Everybody that is presuming "inference is profitable" here is just naively deducting "cost of revenue" from "revenue" and running with that.