The majority of your argument seems to be Starbucks provides a sort of performance art, which can't be replaced by machines and, whilst partially true, the general idea of automation is "fire and forget".

I pay €29K (gotta love the exchange rate, right now) for a machine which will replace a worker that I probably pay for year the same service. Sure it's not "artisanal" but, given enough time to tweak the machine's behaviours, it could be.

For the lower-end markets, definitely, it would be a god-send (in the ROI sense); however, I think your arguments of dissuasion that Starbucks would never do this are akin to the arguments of high-end car manufacturers never moving to automation to build vehicles.

Put succinctly: If the end-goal of business is to increase the profit-marigin (which it is), then someone, somewhere will find a way to make that happen and this would be, in general, no exception to that rule. Eventually, they'll hit that sweet spot to make it work; even if it's slightly sub-par from the quality of today's product.