Also AMD seemly releases some "Pro" features for workstation on Linux first.
The problem of Linux gaming is not the graphics drivers, in fact this part has been working "ok" for a while now.
The problem with Linux gaming is:
1. Audio is still a mess, the fastest audio library still is OSS4 (not 3), that won't ever get in the mainstream distributions (sometimes it is even explicitly banned and will get you banned from forums and irc servers if you ask about it), JACK still is finnicky to work with, PA is still buggy and slow, and ALSA although much better now, is also too slow and has terrible API (while coding for OSS, any version, is very easy).
2. Desktop is still a pain to setup, even the most "GUI-happy" distros still have problems here and there, when I asked some random people about it, I got outright hostile replies, people telling me they deliberately don't want to make Linux easier, because they don't want "dumb people using something they are not worth using", mind you, this wasn't once, or in one community, more than one person shared this view with me.
3. Some people in the community are outright hostile to games, and even sabotage efforts to make games work better, I saw a flamewar where one guy was claiming games were purely dumb entertainment, and that people that played games weren't not only inferior humans, but people that deserved Windows, as punishment. This guy was of the opinion that no GUI server should ever support exclusive fullscreen properly, that instead they should have all features focused on "Work" features always.
4. Lots of other random hardware is a pain to setup on Linux, even things that usually don't cross people minds, for example I couldn't figure how to make Linux properly support my motherboard Super I/O chip, that is even a common one, I even convinced the manufacturer to give me the chip datasheet, but I couldn't even figure to what project I would need to contribute code. (as for what the Super I/O chip does: control the motherboard sensors, fans and power distribution, including voltage values and regulation commands on my motherboard)
This likely won't change (again, for me) because not only would more games need to be released on Linux into the future, but a significant portion of my existing gaming library would need to be ported. I have spent years and considerable money building my gaming library and with minor hiccups almost all of them can be run on modern Windows due to its focus on backwards compatibility along with the efforts of the community.
I suppose there is a breaking point where there are enough newer games being released, older ones ported, and continuing annoyances from Microsoft that I would use Linux as my main gaming OS. I wonder how long that will take? It's frustrating to say the least as gaming is far and away my favorite hobby.
The ONLY usable Linux "distros" at the moment are hobbyist projects. This includes Ubuntu, Debian, Mint, etc. etc.
No serious company or individual that relies on a predictable and reliable user experience is going to use any of the modern flavors of Linux. Someone has to spend the money to do this right, and do it right once and for all.
Obviously, Stallman was right and every closed source OS is going the way of 1984 by making the users the product and gathering every single keystroke and mouse click in order to monetize you. As a convenient side effect, they are creating a perfect turn-key totalitarian system that could be used to devastating effect with the necessary political will.
Greed is good? Apparently it isn't good enough, because Gabe Newel would rather hoard his billions than spend on critical infrastructure for his company's future.
A Linux OS needs to be built from the kernel by professional and well paid developers who don't fork every single repo if they don't like tab spacing.
Democracy doesn't work in engineering, it doesn't work in business, and it probably doesn't work in politics. Linux works because Linus is a "benevolent" dictator. That's why the kernel is such a marvel of engineering. Because he oversees every commit. The same needs to happen downstream.
Here's to dreams of the Linux desktop in 2050!