DLSS is a technique of making a lower resolution frame into a higher one by using a trained dataset and a neural net. Unfortunately, as VR needs little latency (11ms), this is unlikely imo to drive the kind of resolutions that we want.
What you should be excited for is Variable Rate Shading (VRS)
In layman’s terms, this is an architecture-level change that will allow for rendering different parts of the screen at different resolutions effectively. Sound familiar? That’s what foveated rendering does. That is what we’re going to need, since we’re unlikely to progress seriously in GPUs for a long time.
If the latency induce by computing the scene and TAA is reduce by more with DLSS we still gaining something the real short term problem is that technique will not be used by small VR dev cie which create most of the games available now on steam.
Although preliminary from what I see in the FFXV demo a prefer the smoothing done by DLSS than TAA,