Your talk about costs is basically speculation without evidence - there is nothing to suggest the costs are so high they'd risk bankrupting AMD, etc.
It is perfectly possible the explanation is that nVidia's implementation is less mature but just because AMD can support these monitors without issue (which isn't entirely true as some of them do exhibit problems on the AMD side) doesn't mean they aren't in some way implementing workarounds whether that is relaxing tolerances or special casing. It isn't a particularly bold claim when there are well known inconsistencies when it comes to range and other features are well known as well as that AMD when talking about FreeSync 2 talked about concerns over how monitor manufacturers were implementing it and the inconsistencies for customers - then there is the nature of how some of the monitors can be made to work on the nVidia side where the nature of the fixes and the odd ways it can be done in some cases by limiting features/range, etc. doesn't tend to suggest the problem is purely on the nVidia side as such.
Sure if the problem is varied levels of adherence to the spec nVidia can get it working via similar workarounds if that is the case but that isn't actually a good thing.
As an aside I don't tend to make claims like this just for the fun of it and have been proved right a lot of times in the past over similar stuff i.e. microstutter which for some reason people seem to quickly forget.
