- "Linux" is not a unified desktop environment, there are many different configurations and supporting such variety is difficult. The Linux desktop landscape also changes more frequently than most (eg. Pipewire & Pulseaudio, Xorg & Wayland, Snap & Flatpak & AppImage & native distro package managers) which requires more development resources to keep up with.
- But suppose you try to cut costs by supporting only one blessed Linux configuration and constrain your Linux development budget. You still have another cost that you can't avoid: customer support, which is very expensive. It's especially expensive when you get a lot of Linux users who don't know or care that you technically only support one blessed Linux configuration, they'll have some wacko configuration and they'll take the time to complain to your customer support agents about it. Your constrained Linux development budget will only exacerbate your customer support costs as more users run into Linux bugs more often.
- Which isn't worth it because you know that Linux has a small user base. The actual sales bump you get from Linux support isn't worth the cost of maintaining it.
Frankly, I don't think Linux will ever solve the problem of a small user base. No one working on Linux cares enough about the normal-person-UX of its desktop to make it good enough for a majority of people to use, and many current Linux users even oppose measures that would trade off the power & flexibility that they enjoy now for normal-person-UX. This isn't going to change because Linux is largely a volunteer-led project.