The reality is all the money they make off Linux is with workstations cards or HPC. Neither of those market niches are likely demanding Wayland support since stability is a hard requirement and Wayland really isn't sufficiently mature yet (nor does it really have sufficient market adoption).
They're a business, so when their customers need it, they'll support it. Until then, only hobbyists or an insufficient part of their customer base is interested, so understandably, they're not investing in it.
Your forgetting the embedded device market that the Tegra K1 and newer are used in. Is Nvidia going to orphan this whole market segment they've built? I know they replaced a ton of WinXP boxes in labs across the world, and it seems as though they have captured most of the automakers on that same platform, from Toyota to BMW all running Xorg in their cars.
Will Toyota go for running legacy, unmaintained Xorg in their cars? How about BMW? I don't see Nvidia ready to dump this market they've spent a decade building, so short of them being complete idiots, they will probably walk back on this.
They're talking about not supporting XWayland, not Wayland proper.
Given this -- if Toyota doesn't want to run "legacy, unmaintained" X11 in their cars, then.... they don't have to, they can just move to Wayland, which will work, and NVidia supports Wayland just fine -- and they don't have to worry about X11 at all anymore?
So, everything would be fine? I'm not sure what point you're making here...
Those markets are also not demanding support because many of them have no display at all. This is definitely true for HPC GPUs and will be true for deep learning servers.
Wayland is probably going to be the default in all the next stable releases (RHEL, debian (is it already the case?), Ubuntu). Not supporting it sounds risky.
Risky, but for whom? If my GPU stops working with the Linux distro I use, I'm definitely more likely to change my distro. GPUs are expensive and there isn't much choice between vendors. Linux distributions are free and there's plenty of choice between vendors AND between versions.
Selling a GPU takes time and loses you money - and that's in PC case, because on a laptop you don't even have that option.
Moreover, if you have an NVIDIA GPU, you most likely need a decent GPU for something - high-end video games, 3D modelling work, GPU-accelerated computations, etc. Linux distros are, frankly, not much different for one another - they're pretty interchangeable for almost all tasks I can think of as a user/developer. GPUs, on the other hand, are a bit less interchangeable between vendors. You really have only two players to choose from (Intel GPUs don't count for anything serious) - so if both suddenly become incompatibile with a given Linux distro, then it's obviously bye bye to that Linux distro.
They're a business, so when their customers need it, they'll support it. Until then, only hobbyists or an insufficient part of their customer base is interested, so understandably, they're not investing in it.