I don't disagree completely but underlying the growing excitement was for the past 18 months (since Abrash, etc. started talking about VR and presence as something incredible, game-changing and deliverable in around 2 years) was in large part based on the pace things were advancing and what was in store for a consumer release.
Plus many of his specific points are just so temporary in nature, like resolution complaints, heaviness/comfort, tethered, form factor, etc. These will be solved in just the next few years.
I think it is very unlikely any of that will be solved in a few years.
For example resolution, arguably the easiest one to attack, when we are talking 4k res at 120fps, that is barely possible on the highest end PC rigs today with 2 or even 3 video cards in SLI.
Isn't this only true if what you want is the equivalent of AAA FP games blasted directly at your retinas?
If you could solve all the problems of lag, resolution and artifacting, but only give me Nintendo 64-levels of polygons and shading, I bet I could still have some pretty compelling experiences.
On the Gear VR (dev kit, 2014) there was a 2.5k screen at a smooth 60fps. The consumer Gear VR coming out this fall will likely have a 2.5k screen but we'll likely see 4k screens on mobile headsets in 2016. 120fps will be a nice improvement but not required, with the Rift CV1 at 90fps and sufficient for presence. So we're talking a 2-4x difference.
The graphics card issue is not that relevant, Gear VR with a mobile GPU is showing that it's not really about some arbitrary graphical fidelity bar. Yes, some of the highest fidelity cool stuff will be on PC but multiple cards won't be required. Both the Vive and CV1 will run great on one high end card, and by the time they upgrade to ~4k/120Hz... they'll still require just 1 card.
4k resolution is roughly 4x as demanding as 1080p. A lot of games run at 30 fps (especially on consoles) but most target 60 fps. If we want to do 4k @ 120 Hz, we are talking about 8x as much work. Rendering the scene twice (once for each eye) also adds quite a bit of overhead (lighting, etc.) and so that's a least a 2x factor. We are now looking at roughly 16x as much work. That is a very rough estimate but it gives a good feel for how much more processing is required to render the same scene in VR.
Honestly, the consumer GPU market has been in a state of stagnation for the last 3 or 4 years because there's not a ton of demand for anything higher than 60fps/1080p. Hopefully VR will drive consumer demand for higher resolutions to bring prices down.
Plus many of his specific points are just so temporary in nature, like resolution complaints, heaviness/comfort, tethered, form factor, etc. These will be solved in just the next few years.