And as we saw, once a model is trained you need very little compute to run it and there is very little advantage in begin the 1st model and the 10th model.
Monopoly in this field is impossible, your product won't ever be so good that the competition does not make sense
I’m not so sure. Look for more gov regulations that make it hard for startups. Look for stricter enforcement of copyright (or even updates to laws) once the big players have secured licensing deals, to cut off the supply of cheap training data.
Don't forget the perfectly legal use of legislation and bureaucratic precedent that gives them "soft/lossy monopoly" power or all but forces people do to business with them.
True, yet all ills are blamed on the so-called "free market". There is nothing free about our economy, it's manipulation and lobbying all the way down.
Monopoly in this field is impossible, your product won't ever be so good that the competition does not make sense
Add to this that AGI is impossible with LLMs...