But, if your target market is US (or Europe) you've got shipping costs to get your finished shirts to the US (or European) market. If the machine is in the US (or Europe) those shipping costs are substantially lower (and with much less latency).
The lower latency (meaning you can sell shirts in a more "on-demand" manner) and much lower shipping costs might well outweigh the labor savings of the Bangladesh worker overseeing the machine.
I'm thinking the biggest advantage to outsourcing t-shirts is in a lot of places you can dump used dyes into the local river instead of dealing with rules on toxic waste.
China / Mexico are generally considered safe bets compared to much of the world and also have cheap labor.
i guess that time window is how long the automatic t-shirt making business remains viable in the US.
And one robot replaces 17 workers, it's probably around 1 robot tech to support 10-20 machines.
According to these stats [1], in Bangladesh there's 4800 factories with around 725 workers on average. So that would be 40 machines with one worker each to supervise and 3-4 robot techs. 5 people to replace 725!