Well, perhaps people see such success stories for what they are, well curated commodity flowers in the walled gardens of the major players, who will not hesitate to pluck them the instant they threaten to have any kind of uncontrolled growth. It's "ISVs" all over again, commoditization of complements etcetera, the tech molemen that serve the big machines.
AI looks to many as a wall buster, at least for the time being, so even if breakout success is unlikely you can't blame people for at least trying to escape the underground caverns where the "widely successful" ceiling is capped at perhaps reaching a FAANG manager level of compensation.
I think what that is demonstrating is that models are commodity objects. The model factory may have a value. I think it would need a specialized context. It would need a market large enough to support it and small enough to keep the context out of the mainstream.
My guess is this will always be a moving target. The consumer will choose models based on their value proposition.
AI looks to many as a wall buster, at least for the time being, so even if breakout success is unlikely you can't blame people for at least trying to escape the underground caverns where the "widely successful" ceiling is capped at perhaps reaching a FAANG manager level of compensation.