@CaedenV: That is also my view on things.
I personally would not mind ARM and Intel battling it on several fronts, not only the smartphone arena, but also desktop and even server markets. Competition is always a good thing, and we all know Intel is at its best when the competition is really biting at its heels.
However, while you can just pick up any x86-compatible CPU (generally speaking, any x86 computer) and just know it will work with just about any x86-ready OS you can throw at it in a predictable matter (sure, drivers might be a pain sometimes, but in general it works), with ARM you not only have several completely incompatible instruction sets but you can't even count on a single standard way of accessing hardware.
Which, in short, means OEMs need to spend time not only looking/coding drivers for the SoCs, but going around and making sure the whole OS plays nice with one particular SoC/SoC design, hence why for instance Android will only run on the Nexus smartphone it was designed for, and everyone else needs colossal amounts of time to not only to get drivers optimized and creating the custom OEM layers (which is expected and expectable), but also porting the whole OS to the other SoCs.
You can say what you want about x86: it's inefficient, bloated, too old, not designed for low-power operation, etc., but the thing is, the core (and most of the extras that have been added) has remained the same for decades now, and standardized in a way it's extremely easy to work with. Sure, you might get extra instructions from certain CPUs that speed up certain tasks, but most, if not all, of those instructions can be executed in another way, even if with a performance hit.
What I'm trying to say here is that ARM would only benefit if it followed a single line, like x86. You don't need to force a single way of doing things (AMD, Intel and VIA have been doing it differently for years without problems), you just need to make sure the ecosystem is coherent enough you don't have markets within the ARM market. It you think the best way of booting the system is having the GPU powering up first, handle the POST and then let the CPU take control, so be it, tell everyone that's the way it's going to work from now on and say exactly how it should work. If not, then define a way and stick with it. Same for other stuff going on within the SoC. Suddenly, ARM OS development time would drop like a stone, TtM of the OS too, meaning faster development cycles for finished products and better market penetration.
Now, that being said, when can I expect an ARM-based mini-ITX board for DIY builds? Like, "just add memory, an HDD/SSD, a case, mouse, keyboard and monitor, and you'll have a full-fledged PC". Heck, with the amount of board real estate available, you could even have a mini-ITX router platform (2x mini-PCIe, 2x Gigabit Ethernet) or a NAS platform (though this one would need a serious overhaul on the ARM storage subsystem capabilities...).
Miguel