This is such a comical strawman of what I wrote but I'll try to clarify nonetheless.

Say you have a startup that wants to disrupt inventory management, your secret sauce "idea" is a feature that allows you to cut stale inventory by 14% by implementing various just-in-time workflows that optimize shipping and reception of new goods.

If you are in the US (or most English-speaking countries but considering how combative you are let's say US), your MVP should be the most barebone system that you can ship that will give your customers that 14% reduction and make them want to sign a contract with you. Having a screen-reader-compatible interface might be a must ~1 year down the line when you reach a scale where conforming to ADA is a necessity or realistically when your customers ask for it because of the ADA.

It's not about caring or not, it's about assigning your meager startup resources towards the features that the prospective customer will say "yes we want your product". That's what an MVP is.

In addition (and perhaps even more importantly), MVPs are useful in determining your product-market fit. You might go over 5 or even 10 MVPs as you pivot and refine your inventory management idea, all without having a single customer keeping the lights on. You could implement accessibility in all of them, or you could implement it only in the version that actually gets adopted by a customer.