Lloyd E. Sponenburgh writes:

Assuming a perfect Stirling machine is about 54% efficient, how much
work do you think we'll have to add for losses to get the delta-T
desired?

My calculations for a conventional compressor assumed the efficiency of a
10 SEER air conditioner, which also has the losses of the electrical motor
and the air handler fan to overcome to get to that 10 SEER, which the beer
can chiller doesn't. I don't know if scaling things down makes them more
or less efficient, or how much so.