VR is like 3D films.
Every decade or two it is trotted out as THIS TIME THE GIMMICK WILL WORK, but the gimmick never worked to begin with and so it inevitably dies.
I wouldn't write off VR, right now the big barrier to entry is cost, but as time goes on, the cost will go down.
Another thing it could happen to Gabe and HL3 is that when video game tech advances, the old guard have difficultly adjusting to it or can't figure out how to make the game next gen. Maybe Gabe contributions and Steam hid the fact that its more difficult to make a HL3 like he wanted for the current market?
I think one thing that went wrong was that the PC AAA exclusive market dried up, Valve could no longer do what they did before and release an all new Half-Life first on PC and then later on console, instead they would have to release on both console and PC at the same, thus they couldn't "raise the bar" like they did before (at least until VR)
Now sure, that's what they did with L4D1 and 2 and Portal 1 and 2 plus Episode 2, but those were all still Source engine games, a Half-Life 3 would have required a new Source engine and I think having to do that on both console and PC at the same time simply didn't interest them.
In addition to that, Valve bet on the wrong horse with the episodic model for Half-Life, which episodic games fizzled out, but they wrote themselves into a corner, because Episode 2 did not end in a way that you could have really just picked up with a full new game without it being awkward.
It was a perfect storm of things to muck up that series and it would have been a bitch to figure out, which is why Valve just chose not to bother with the headache.
It's no excuse though, they could have done something, hell, just render the episodes non-canon and just pick up from where Half-Life 2 left off if it was that big of a problem.