Or it would create an incentive for reducing the cost of the hardware needed to run beefier AI models and passing that reduction down to consumers.
But if you wait a GPU generation or two, the performance has improved enough that you can do that with a much cheaper GPU.