A/B Test where half the audience gets email and half gets nothing

If you’re using the OOTB A/B test available in the email program, then you need to set the test sample size in the A/B test settings to determine the sample size for the qualified people to receive the test emails and the winner email. If you don’t want a portion of people to receive emails from this blast then why not filter them out in the smart list itself? 

Do you have any specific use-case/need where in you’re not able to suppress the people you don’t want to send any (test/winner) emails qualify for the send? Let’s back-track a bit to understand and analyze the use-case so as to potentially avoid an XY problem here. 

I'm curious what the use case for this is, so if you can expand on that I'd appreciate it.

However, you can do this via a smart campaign instead of an email program. If you create a smart campaign with your full list in the smart list, then add a flow step to send the email with a choice to select a random sample of 50% to get the email and the rest "do nothing"

you'll need to pull an email performance report to make any decisions.